I.

Impact of packaging material and storage conditions on

polyphenol stability, colour and sensory characteristics
of freeze-dried sour cherry (prunus cerasus var.
Marasca)
J Food Sci Technol. 2016 Feb; 53(2): 1247–1258.
Zoran Zorić, Sandra Pedisić, Danijela Bursać Kovačević, Damir Ježek, and Verica
Dragović-Uzelac
Author information ► Article notes ► Copyright and License information ►

Abstract

The aim of this study was to evaluate the influence of packaging materials and
storage conditions on polyphenols stability, colour and sensory characteristics of
freeze-dried sour cherry (Prunus cerasus var. Marasca). Freeze-dried sour cherries
were packed in high barrier metalized polypropylene and aluminium packaging
(PET/PPmet/PE and PET/Al/PE) for up to 12 months at 4, 20 and 37 °C.
Characterisation of polyphenol compounds was done by HPLC UV/Vis PDA and in
all samples individual anthocyanins (ANTs), flavonol-glycosides (FGs) and
hydroxycinnamic acids (HCAs) were determined. Polyphenol content was not
markedly affected by freeze-drying and decreases were amounted 1.5–5 %.
Furthermore, obtained results indicated that minimal loss of polyphenol content in
freeze dried sour cherries were achieved at 4 °C and 3 months of storage.
Regardless of the type of packaging materials, samples stored at lower temperature
during 12 months, retained the higher content of FGs (quercetin-3-glucoside,
kaempferol-3-glucoside, kaempferol-3-rutinoside) and HCAs (neochlorogenic,
chlorogenic, p-coumaric, caffeic and ferulic acid) than ANTs (cyanidin-3-
glucosylrutinoside, cyanidin-3-rutinoside, cyanidin-3-glucoside, cyanidin-3-
sophoroside). The same trend was confirmed with kinetic parameters, also. Sour
cherry products packed in both type of laminate and stored at lower temperature
retained characteristic dark red colour and sensory properties. This study showed
that freeze-dried cherry products have pleasant sensory and very good nutritional
properties, and storage in both type of laminates at 4 and 20 °C up to 6 months
ensured good product quality.

Keywords: Freeze dried sour cherries, Storage, Packaging, Polyphenols, Sensory

characteristics

Introduction

Sour cherry Marasca (Prunus cerasus var. Marasca) is very important seasonal crop
in Croatia, and from 2013 this autochthonous cultivar received the designation of
origin (Ministry of Agriculture, Republic of Croatia, 2013) and annual yield is
approximately 2000 t. The fruit has dark red colour and high dry matter,
approximately 27.3 % (Pedisić et al. 2007), specific sweet and sour aroma derived
from volatile compounds such as alcohols, carbonyls, esters and terpenes (Levaj et
al. 2010) and therefore is valuable raw material in food industry.

Regarding their phytochemical composition, sour cherries are rich source of

polyphenol compounds which strongly influence the quality and nutritional value of
the fruits and contributing to their sensorial attributes (Ferretti et al. 2010).
Anthocyanins (ANTs) [cyanidin-3-glucosylrutinoside (Cy-3-GR), cyanidin-3-
rutinoside (Cy-3-R), cyanidin-3-glucoside (Cy-3-G), cyanidin-3-sophoroside (Cy-3-
S)], flavonol glycosides (FGs) [quercetin-3-glucoside (Q-3-G), kaempferol-3-
glucoside (K-3-G), kaempferol-3-rutinoside (K-3-R)], hydroxycinnamic acids (HCAs)
[neochlorogenic (NChA), chlorogenic (ChA), p-coumaric (p-CA), caffeic (CA) and
ferulic acid (FA)] are the most abundant phenols in sour cherries (Ferretti et al. 2010),
including Marasca cultivar (Pedisić et al. 2010; Elez Garofulić et al. 2013; Zorić et al.
2014). These bioactive compounds have attracted the interest of many researchers
and the general public due to the potential beneficial actions on human health and
well-being (Ferretti et al. 2010). To avoid problems related to seasonality and short
postharvest life of sour cherries, various preservation methods are used to prolong
their shelf life and retain the amount of biologically active compounds (BACs). Fruit
drying is one of the most important methods to extend shelf life and to minimize
distribution of raw materials with high moisture content. However, the drying process
may cause physical, structural, chemical and biological changes that can affect
quality attributes like colour, texture, flavour and nutritional value, due to the loss of
nutrients and phytochemicals, which are relatively unstable to heat (Kwok et al.
2004). Therefore, it is important to select and to optimize drying method, as well as
storage conditions, in order to minimize all factors adversely affecting the quality of
the product.

Freeze-drying (FD), is one of the most effective methods of drying which ensures
maximum preservation of nutritional, bioactive and sensory properties of the product
and therefore is widely used for fruits with thermo-labile antioxidants such as
polyphenols. According to Sablani et al. (2011) FD improved retention of polyphenols
in berries during processing and in some cases showed enhancement of
polyphenols concentration compared to air drying. Despite these advantages,
freeze-dried samples exhibit strong hygroscopic properties, and they may be
adversely affected by moisture during storage period what may be reflected on the
quality of dried fruit. Food quality is also affected by storage temperature and may
affect polyphenol stability during storage. Major types of deteriorations of dried fruits
are browning, flavour and nutrient losses. Colour change of the dehydrated fruit
results from both enzymatic and non-enzymatic oxidation of polyphenol compounds
as well as from Maillard reaction (McEvily et al. 1992). Aside from appropriate
temperature and storage period, adequate packaging material also showed great
impact on quality and sensory properties of dried fruit (Gvozdenović et al. 2007;
Piasecka et al. 2013; Henríquez et al. 2013). Complex films such as laminated
materials and metalized films made from polyethylene and aluminium foils offer good
protection to dried foods because of improved barrier properties and therefore are
mainly used to package snacks and high value foods. According to Henríquez et al.
(2013) total phenols (TP) of apple peel powder samples were the best preserved in
metalized films of high barrier during storage. Several studies researched the impact
of drying processes and storage conditions on fruit antioxidants for periods till six
months (Del Caro et al. 2004; Henríquez et al. 2013). According to Ranđelović et al.
(2014) PET/Al/PE laminate provided the best protection of TP content in comparison
with other packaging laminates in dried apricot stored at room temperature during
12 months. To date, there is a lack of studies which concurrently determine the
nutritional and sensorial changes of freeze dried sour cherry Marasca (FDSCM)
fruits packaged in various packaging materials during storage. Due to high BAC’s
content and pleasant sensory profile, FDSCM have a great potential to be a valuable
ingredient for functional fruit products (Pedisić et al. 2010). Economic considerations
usually limit the widespread use of FD commercially, but above-mentioned Marasca
cherry properties could justify FD process.

Considering all the aforementioned, the objective of this study was to evaluate the
stability of polyphenol compounds, antioxidant capacity, colour and sensory
attributes of FDSCM fruit, packed in two different packaging materials (laminate PET
12 μm/PP 18 μm met/PE 100 μm and PET 12 μm/Al 9 μm/PE 85 μm), stored at the
different temperatures (4 °C, 20 °C and 37 °C) during 12 months (samples are
excluded at 0, 3, 6, 9, 12 months).

Additionally, kinetic parameters (reaction rate constant, half-life and activation

energy) for predicting the quality of stored FDSCM were applied.

Materials and methods

Chemicals and standards

Methanol, formic acid and acetonitrile were HPLC grade, purchased from BDH
Prolabo, VWR (Lutterworth, England). Water was Milli-Q quality (Millipore Corp.,
Bedford, USA). Standards Cyanidin-3-O-glucoside chloride, Q-3-G, p-CA, FA, CA
and ChA were purchased from Sigma (Steinheim, Germany), Cyanidin-3-O-
sophoroside chloride, Cyanidin-3-O-rutinoside chloride and K-3-R were obtained
from Extrasynthese (Lyon, France), 2,2-diphenyl-1-picrylhydrazyl (DPPH) was
purchased from Sigma and (±)-6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic
acid (Trolox) from Fluka (Neu-Ulm, Germany).

Plant material

The sour cherry fruits (Prunus cerasus var. Marasca) were harvested at the
plantation of Maraska factory (Zadar, Croatia) at technological stage of maturity.
After harvesting, the samples were immediately transported to a laboratory in a
portable refrigerator at 4 °C, subsequently deppited and frozen using liquid nitrogen,
and kept at −60 °C (ScanCool SCL210P, LaboGene™, Lynge, Denmark) until FD.

Freeze-drying of sour cherries

For FD of deppited sour cherries (previously frozen at −60 °C), 500 g of the sample
were placed on each tray. Five trays were placed in a laboratory freeze-dryer
(CoolSafe PRO, Labogene, Denmark) and the FD process was performed for 24 h
under high vacuum (0.13–0.55 hPa), with isothermal (heating) plate temperatures of
20 °C. The final water content of dried fruits was 9.7 %.

Packaging and storage conditions of freeze-dried sour cherries

Approximately 10 g of FDSCM were packed in bags (12 × 12 cm) of the two different
commercially available laminates: Pet 12 µm/PP 18 µm met/PE 100 µm
(Pet/PPmet/PE) and Pet 12 µm/Al 9 µm/PE 85 µm (Pet/Al/PE) (Folijaplast Ltd.,
Zadar, Croatia) and stored at three temperatures: 4 °C, 20 °C and 37 °C. Relative
humidity and temperatures were controlled (data logger EBI 20TH1, Ebro, Germany)
every 6 h for the samples stored at room temperatures. At the end of the storage
period average temperature and relative humidity was 19.73 °C and 62.35 %,
respectively. Temperature value is round up to 20 °C.

Extraction of phenols

The polyphenol compounds from all samples were extracted using procedure
previously described by Elez Garofulić et al. (2013). All extracts were prepared in
triplicate and analysed by HPLC.

HPLC analysis

Polyphenol compounds were determined by HPLC Agilent 1260 quaternary LC

Infinity system (Agilent Technologies, Santa Clara, CA, USA) equipped with diode
array detector (DAD), an automatic injector and ChemStation software on a
Nucleosil 100-5C18, 5 μm (250 × 4.6 mm i.d.) column (Macherey-Nagel). The
solvent composition and the used gradient conditions were described previously by
Zorić et al. (2014). Identification of phenols was carried out by comparing retention
times and spectral data with those of the authentic standards (anthocyanins were
identified at 520 nm, flavonol glycoside at 360 nm and phenolic acids at 280 nm).

Identification was confirmed using characteristic UV/Vis spectra, polarity, previous

literature reports (Elez Garofulić et al. 2013), and also by another study on sour
cherry Marasca products done with LC-MS/MS (data not yet published).

Quantitative determination was carried out using the calibration curves of the
standards. Cy-3-GR was determined according to Cy-3-G, kaempferol-3-glucoside
(K-3-G) according to K-3-R and neochlorogenic acid (NChA) according to ChA.

The concentration of all polyphenol compounds were expressed as mg per 100 g of

dry matter, as mean value ± standard deviation (N = 3 replicates).

Antioxidant capacity
The scavenging capacity was determined according to DPPH method as previously
reported in literature (Brand-Williams et al. 1995) with some modifications as follows:
an aliquot (2 mL) of phenolic extracts or methanol solution of Trolox (25–300 mM)
was mixed with 2 mL of methanol and 1 mL of 0.5 mM DPPH methanolic solution.
The mixture was kept in the dark for 20 min, after which the absorbance was
measured at 517 nm against a blank of methanol without DPPH. Measurements
were performed by the Uviline 9400 spectrophotometer (Secomam, France). The
results were calculated according to calibration curve for Trolox (25–300 mM). DPPH
values were expressed as mmol Trolox equivalents (TE) per 100 g of dry matter, as
mean value ± standard deviation (N = 3 replicates).

Colour analysis

Colour analysis was performed using a colorimeter (CM-3500 d, Konica Minolta,

Japan) at CIE Standard Illuminant D65 by 30 mm thick plate. The colour values were
expressed as L* (whiteness or brightness/darkness), a* (redness/greenness) and b*
(yellowness/blueness), respectively. Three measurements were made at different
points of the samples, and this procedure was repeated three times to get the
average values.

Hue angle (H), chroma (C) and total colour difference (∆E) were calculated from:

H∗=tan -1(b* / a*) 1

C * = √𝑎∗2 + 𝑏 ∗2 2
∗
∆𝐸𝑎𝑏 = √∆𝐿∗2 + ∆𝑎∗2 + ∆𝑏 ∗2 3

Where all ∆L*2, ∆a*2 and ∆b*2 were calculated in reference to the zero time of storage.

Sensory evaluation

For sensory evaluation of FDSCM samples quantitative descriptive analysis (QDA)

was chosen, since this method provides a detailed description of the tested samples
(Meillgaard et al. 1999). The procedure was performed according to methods
described in ISO 6564, ISO 8587 and ISO 11036 in a sensory laboratory equipped
according to ISO 8589. After the storage, sensory analysis was carried out by a
trained panel consisting of 21 members per session (11 female and 10 male).
Panellists were staff members of the Faculty of Food Technology and Biotechnology
(Zagreb, Croatia). After a series of discussion sessions, the panellists were required
to list the terms seemed appropriate for the description of colour, odour, flavour,
taste and texture. Finally, total of 17 descriptive terms as major sensory attributes
were evaluated: colour intensity, odour intensity, sour cherry odour, off-odour, sour
cherry flavour, flowery flavour, fruity flavour, sour taste, sweet taste, harmony taste,
sour cherry taste, astringent taste, aftertaste, off-taste, firmness, stickiness, and
chewiness. The panellists scored the samples for every sensory characteristic, using
a line intensity scale, with scores awarded on a scale of 1–7 to show the relative
intensity of each attribute, in which 1 indicated total absence (‘none’) of the sensory
attribute and 7 a very definite attribute (‘intense’).

Statistical analysis

Statistical analysis was carried out using the software Statistica ver. 10.0 (Statsoft
Inc., Tulsa, OK, USA). In order to explore the influence of storage time, storage
temperature and packaging material on FDSCM polyphenols, antioxidant capacity
and colour parameters multivariate analysis of variance (MANOVA) was performed
and marginal means were compared with Tukey HSD test. Differences at p ≤ 0.05
were considered statistically significant. For the sensory data principal component
analysis (PCA) was performed.

Calculation of kinetic parameters

Previous studies have shown that thermal degradation of anthocyanins followed

first-order reaction kinetics (Verbeyst et al. 2010; Zorić et al. 2014). This kinetic type
can generally be expressed as Eq. 4 with c(t) as the concentration at time t, c0 as
the initial concentration (mg/100 g dry matter of sample), t as the storage time
(months) and k as the first order degradation rate constant (month−1):

c(t) = c0 · exp.(−k · t) 4

The half-life (t1/2) of the reaction is obtained assuming the first-order kinetics as:

t1/2 = −ln0.5/k 5

The linear relationship between lnk and 1/T can be expressed by the Arrhenius Eq.
(3).

k(T)=k0⋅exp[−Ea /R⋅T] 6

where Ea is the activation energy (kJ/mol), T–absolute temperature (K) and R is the
universal gas constant (8.314·10−3 kJ/K·mol). The Arrhenius activation energy (Ea)
was calculated by plotting ln (k) against 1/T.

Results and discussion

The polyphenol content in fresh and freeze-dried Marasca sour cherries, and the
effect of packaging materials, temperature and storage time on polyphenol content
in FDSCM is summarized in Tables Tables 1 and and 2.

Table 1. Anthocyanins and flavonol glycosides in freeze-dried Marasca sour

cherry packed in two laminate and storage in three different temperature after
0, 3, 6, 9 and 12 months
 Storage Anthocyanins (mg/100 g dry matter) Flavonol glycosides (mg/100 g dry matter)
Packaging
condition Months
material
s Cy-3-SA Cy-3-GR Cy-3-G Cy-3-R TA Q-3-G K-3-G K-3-R TFG

Fresh
27.71 ± 0.17 231.60 ± 0.77 14.38 ± 0.35 99.15 ± 0.37 372.84 ± 1.67 16.79 ± 0.03 13.92 ± 0.01 10.98 ± 0.03 41.69 ± 0.07
fruit

Freeze-
dried
25.37 ± 0.33 219.00 ± 2.95 13.52 ± 0.10 99.01 ± 0.35 356.90 ± 3.72 16.62 ± 0.03 13.68 ± 0.03 10.78 ± 0.05 41.07 ± 0.11
(0/control
)

22.40 ± 0.11 215.35 ± 0.41 15.32 ± 0.08 103.98 ± 0.12 357.05 ± 0.73 14.08 ± 0.04 12.99 ± 0.03 11.92 ± 0.03 38.99 ± 0.09
3 a a a a a a a a a

20.79 ± 0.38 198.28 ± 0.05 14.01 ± 0.09 325.55 ± 0.71 13.68 ± 0.03 12.41 ± 0.03 10.54 ± 0.01 36.62 ± 0.07
6 b b b 92.47 ± 0.18b b b b b b

4 °C
20.16 ± 0.25 166.04 ± 0.25 12.30 ± 0.27 269.26 ± 0.95 12.79 ± 0.05 11.67 ± 0.04 10.16 ± 0.04 34.62 ± 0.13
9 b c c 70.77 ± 0.17c c c c c c

17.73 ± 0.12 162.10 ± 0.12 10.36 ± 0.14 258.61 ± 0.47 22.10 ± 0.08
12 c d d 68.42 ± 0.10d d 6.61 ± 0.02d 6.04 ± 0.03d 9.46 ± 0.03d d

10.14 ± 0.06 107.56 ± 0.11 174.52 ± 0.32 26.87 ± 0.07

3 a a 5.91 ± 0.05a 50.91 ± 0.10a a 9.60 ± 0.01a 9.61 ± 0.02a 7.66 ± 0.04a a

136.54 ± 0.46 22.23 ± 0.11

6 5.72 ± 0.17b 77.68 ± 0.08b 4.62 ± 0.13b 48.53 ± 0.08b b 8.33 ± 0.05b 7.92 ± 0.03b 5.98 ± 0.04b b
PET/PPmet/P
E 20 °C
19.73 ± 0.06
9 4.44 ± 0.17c 58.06 ± 0.11c 2.65 ± 0.10c 32.62 ± 0.08c 97.77 ± 0.46c 7.50 ± 0.01c 6.81 ± 0.03c 5.43 ± 0.03c c

2.21 ± 0.04c, 16.51 ± 0.09

12 3.10 ± 0.13d 39.53 ± 0.15d d 25.27 ± 0.01d 70.10 ± 0.33d 6.17 ± 0.04d 5.96 ± 0.04d 4.38 ± 0.01d d

11.49 ± 0.03 13.34 ± 0.03 11.33 ± 0.03 36.16 ± 0.08

3 0.25 ± 0.04a 4.02 ± 0.10a 0.65 ± 0.05a 5.71 ± 0.03a 10.63 ± 0.21a a a a a

20.55 ± 0.05
6 0.10 ± 0.01b 2.00 ± 0.02b 0.21 ± 0.01b 0.73 ± 0.03b 3.03 ± 0.07b 6.82 ± 0.01b 7.48 ± 0.02b 6.25 ± 0.02b b

37 °C

16.13 ± 0.05
9 – 0.02 ± 0.02c 0.03 ± 0.02c 0.03 ± 0.01c 0.08 ± 0.06c 3.71 ± 0.02c 6.65 ± 0.02c 5.78 ± 0.01c c

12 – – – – – 0.15 ± 0.02d 0.42 ± 0.03d 0.40 ± 0.02d 0.97 ± 0.08d

25.03 ± 0.09 214.53 ± 0.11 16.70 ± 0.07 351.59 ± 0.34 16.36 ± 0.03 12.79 ± 0.02 38.89 ± 0.08
3 a a a 95.33 ± 0.07a a a a 9.74 ± 0.03a a

23.60 ± 0.10 203.60 ± 0.11 13.61 ± 0.05 333.70 ± 0.31 15.28 ± 0.02 10.36 ± 0.02 35.19 ± 0.05
6 b b b 92.90 ± 0.05b b b a 9.55 ± 0.01b b

4 °C
21.20 ± 0.03 188.52 ± 0.05 13.16 ± 0.10 300.58 ± 0.25 13.35 ± 0.01 11.49 ± 0.22 33.80 ± 0.25
9 c c b 77.70 ± 0.07c c c b 8.96 ± 0.03c c

16.27 ± 0.12 162.96 ± 0.11 9.80 ± 0.13c, 262.18 ± 0.42 11.10 ± 0.04 24.11 ± 0.13
12 d d d 73.15 ± 0.06d d d 6.01 ± 0.08c 7.01 ± 0.01d d

125.12 ± 0.92 11.93 ± 0.03 10.67 ± 0.04 30.63 ± 0.08

PET/Al/PE 3 7.45 ± 0.06a 75.41 ± 0.77a 4.41 ± 0.03a 37.85 ± 0.05a a a a 8.02 ± 0.02a a

103.29 ± 0.27 19.79 ± 0.07

6 4.84 ± 0.01b 59.65 ± 0.20b 3.53 ± 0.04b 35.28 ± 0.02b b 8.09 ± 0.03b 6.63 ± 0.02b 5.08 ± 0.02b b

20 °C
14.19 ± 0.08
9 3.32 ± 0.05c 50.40 ± 0.09c 2.61 ± 0.02c 30.51 ± 0.04c 86.84 ± 0.19c 5.58 ± 0.02c 4.75 ± 0.04c 3.86 ± 0.02c c

2.57 ± 0.03c, 12.88 ± 0.12

12 2.22 ± 0.04d 40.01 ± 0.10d d 28.02 ± 0.05d 72.82 ± 0.22d 5.30 ± 0.03d 4.22 ± 0.04d 3.36 ± 0.05d d

10.57 ± 0.02 11.82 ± 0.03 31.63 ± 0.07

37 °C 3 – 7.25 ± 0.04a 1.00 ± 0.06a 8.30 ± 0.02a 16.55 ± 0.12a a a 9.25 ± 0.02a a
 Storage Anthocyanins (mg/100 g dry matter) Flavonol glycosides (mg/100 g dry matter)
Packaging
condition Months
material
s Cy-3-SA Cy-3-GR Cy-3-G Cy-3-R TA Q-3-G K-3-G K-3-R TFG

23.08 ± 0.10
6 – 0.10 ± 0.02b 0.07 ± 0.01b 0.11 ± 0.02b 0.28 ± 0.06b 8.99 ± 0.03b 6.90 ± 0.04b 7.19 ± 0.03b b

0.03 ± 0.00b, 16.75 ± 0.11

9 – 0.01 ± 0.01c c 0.03 ± 0.02b,c 0.07 ± 0.03c 5.04 ± 0.03c 4.86 ± 0.04c 6.85 ± 0.04c c

12 – – – – – 0.86 ± 0.03d 0.37 ± 0.03d 1.37 ± 0.03d 2.60 ± 0.10d

View it in a separate window

The results in the individual lines marked by the same letters do not differ statistically
at p ≤ 0.05

Acyanidin-3-sophoroside (Cy-3-S), cyanidin-3-glucosilrutinoside (Cy-3-GR),

cyanidin-3-glucoside (Cy-3-G), cyanidin-3-rutinoside (Cy-3-R); Total anthocyanins
(TA), quercetin-3-glucoside (Q-3-G); kaempferol-3-glucoside (K-3-G), kaempferol-3-
rutinoside (K-3-R), Total flavonol glycosides (TFG)

Table 2. Phenolic acids in freeze-dried Marasca sour cherry packed in two

laminate and storage in three different temperature after 0, 3, 6, 9 and
12 months

Phenolic acids (mg/100 g dry matter)

Packaging Storage
Months
material conditions
NChAA p-CA ChA CA FA TPA

Fresh fruit 37.24 ± 0.06 39.70 ± 0.67 3.90 ± 0.02 36.58 ± 0.02 7.81 ± 0.04 125.23 ± 0.81

Freeze-
dried 36.51 ± 0.05 39.43 ± 0.03 3.79 ± 0.03 35.94 ± 0.03 3.04 ± 0.01 118.94 ± 0.15
(0/control)

3 36.43 ± 0.03a 39.11 ± 0.04a 2.12 ± 0.02a 35.59 ± 0.03a 2.49 ± 0.02a 112.15 ± 0.19a

6 35.67 ± 0.03b 38.74 ± 0.05b 1.30 ± 0.05b 34.26 ± 0.04b 2.19 ± 0.03b 92.26 ± 0.30b
4 °C
9 29.93 ± 0.06c 31.69 ± 0.04c 0.86 ± 0.06c 28.41 ± 0.12c 1.37 ± 0.01c 87.01 ± 0.38c

12 29.52 ± 0.05d 29.21 ± 0.04d 0.81 ± 0.04c,d 26.43 ± 0.03d 1.03 ± 0.02d 70.25 ± 0.26d

3 22.08 ± 0.02a 22.82 ± 0.02a 0.78 ± 0.03a 23.13 ± 0.03a 1.44 ± 0.02a 93.85 ± 0.16a

PET/PPmet/PE 6 20.34 ± 0.03b 21.94 ± 0.05b 0.64 ± 0.01b 21.16 ± 0.02b 1.10 ± 0.03b 65.18 ± 0.21b
20 °C
9 16.47 ± 0.04c 16.32 ± 0.04c 0.58 ± 0.01b,c 18.19 ± 0.03c 0.77 ± 0.02c 52.32 ± 0.17c

12 13.42 ± 0.04d 13.16 ± 0.04d 0.54 ± 0.02c,d 13.64 ± 0.04d 0.55 ± 0.02d 41.31 ± 0.22d

3 10.71 ± 0.02a 11.99 ± 0.04a 3.56 ± 0.02a 17.96 ± 0.05a 1.23 ± 0.03a 45.45 ± 0.21a

37 °C 6 10.36 ± 0.01b 11.67 ± 0.04b 3.08 ± 0.03b 14.25 ± 0.00b 1.08 ± 0.03b 40.44 ± 0.17b

9 5.56 ± 0.03c 7.67 ± 0.02c 2.64 ± 0.02c 9.78 ± 0.02c 0.01 ± 0.01c 25.66 ± 0.16c
 Phenolic acids (mg/100 g dry matter)
Packaging Storage
Months
material conditions
NChAA p-CA ChA CA FA TPA

12 0.24 ± 0.04d 0.30 ± 0.04d 0.15 ± 0.01d 0.41 ± 0.03d - 1.56 ± 0.12d

3 36.69 ± 0.04a 30.05 ± 0.05a 1.79 ± 0.03a 34.80 ± 0.03a 2.28 ± 0.03a 105.61 ± 0.18a

6 36.15 ± 0.05b 28.30 ± 0.03b 1.21 ± 0.03b 34.74 ± 0.02a 2.18 ± 0.01a 102.58 ± 0.14b
4 °C
9 27.00 ± 0.07c 27.49 ± 0.01c 0.86 ± 0.01c 23.67 ± 0.01b 1.23 ± 0.03b 80.25 ± 0.12c

12 25.06 ± 0.04d 27.31 ± 0.03d 0.74 ± 0.06c,d 21.75 ± 0.05c,d 0.85 ± 0.02c,d 75.72 ± 0.20d

3 22.29 ± 0.03a 24.11 ± 0.04a 0.80 ± 0.02a 23.38 ± 0.02a 1.73 ± 0.01a 72.31 ± 0.12a

6 16.80 ± 0.01b 16.24 ± 0.03b 0.71 ± 0.02b 17.28 ± 0.03b 1.02 ± 0.03b 52.05 ± 0.12b
PET/Al/PE 20 °C
9 13.64 ± 0.03c 14.49 ± 0.04c 0.43 ± 0.02c 12.68 ± 0.03c 0.96 ± 0.01b 42.19 ± 0.13c

12 11.58 ± 0.05d 11.92 ± 0.02d 0.31 ± 0.02c,d 11.72 ± 0.03d 0.77 ± 0.03c,d 36.30 ± 0.15d

3 12.49 ± 0.02a 13.33 ± 0.05a 4.47 ± 0.04a 16.03 ± 0.03a 1.53 ± 0.05a 48.11 ± 0.18a

6 10.75 ± 0.04b 12.22 ± 0.03b 3.14 ± 0.04b 14.89 ± 0.04b 1.01 ± 0.03b 42.00 ± 0.17b
37 °C
9 9.01 ± 0.05c 12.07 ± 0.03c 2.11 ± 0.03c 14.42 ± 0.05c 0.17 ± 0.02c 37.77 ± 0.17c

12 2.70 ± 0.04d 3.91 ± 0.03d 1.86 ± 0.01d 5.08 ± 0.05d - 13.35 ± 0.13d

The results in the individual lines marked by the same letters do not differ statistically
at p ≤ 0.05
ANeochlorogenic acid (NChA), p-coumaric acid (p-CA), chlorogenic acid (ChA),
caffeic acid (CA), ferulic acid (FA),Total phenolic acids (TPA)

The influence of freeze-drying on polyphenols stability in Marasca sour

cherries

Marasca cherry phenols were characterised using HPLC with UV/Vis PDA detection
and include ANTs, FGs and HCAs. Fresh Marasca sour cherries had the substantial
levels of ANTs (69.08 % of total polyphenol content), while content of HCAs
(23.20 %) and FGs (7.72 %) were determined in lower levels. All researched
samples contained four ANTs: Cy-3-S, Cy-3-GR, Cy-3-G and Cy-3-R. The major
ANT in fresh sour cherry Marasca was Cy-3-GR which represented 62.12 % of total
ANTs, followed by Cy-3-R (26.59 %). The results obtained were similar with previous
data for Marasca sour cherries (Pedisić et al. 2010; Elez-Garofulić et al. 2013; Zorić
et al., 2014). Anthocyanin content is affected by numerous factors (agrotechnical
procedures, geographical origin, cultivation, stage of maturity, harvesting, climatic
and storage conditions etc. (Ferretti et al. 2010). Wojdyło et al. (2014b) researched
33 Polish cherry cultivars and results indicated cultivar influence on individual ANTs
content. The FGs in Marasca cherries included K-3-G, K-3-R and the most abundant
Q-3-G which represented 40.27 % of total FGs. Five phenolic acids, NChA, ChA,
CA, p-CA and FA were determined in range from 7.81 to 39.70 mg/100 g d.m. The
most abundant HCAs were p-CA, NChA and CA which individually represented
approximately 30 % of the total HCAs what is in accordance with previously reported
results for 33 Polish sour cherry cultivars, (NChA ∼47 %, ChA ∼ 30 %) (Wojdyło et
al. 2014a, b).

FD did not considerable affect the polyphenol content in cherry samples, therefore
decrease in total HCAs, ANTs and FG was only 5.02 %, 4.28 % and 1.49 %,
respectively. In FDSCM among HCAs, the highest decrease was for FA almost
40 %, while the most stable was ChA which decrease of 2.82 %.

ANTs in FDSCM were retained in high proportion between 94.56 % (Cy-3-GR) to

99.86 % (Cy-3-R). The results of this study are consistent with those reported in the
literature which showed that FD provided greater retention of polyphenol compounds
with respect to other drying techniques (Michalczyk et al. 2009; Wojdyło et al.
2014b). Previous study on FDSCM paste showed that the heating temperature and
duration affect the anthocyanins considerably more than the other phenols in terms
of degradation what is linked with their cation structure (Zorić et al. 2014).

The influence of packaging material and storage conditions on polyphenols

stability and AOC in freeze-dried Marasca sour cherries

The optimal combination choice of the packaging material and storage conditions is
very important for retaining the quality of FDSCM during storage. Multi-layer polymer
materials such as PET/Al/PE and PET/PPmet/PE laminates used in this study have
good barrier properties against light, gas and vapour (Gvozdenović et al. 2007).

After 6 months of storage total ANT content decreased in the cherry products packed
in PET/PPmet/PE laminate less than 10 % at 4 °C, while at 20 °C and at 37 °C
decrease was for 61.75 % and 99.15 %, respectively. At same storage conditions,
samples packed in PET/Al/PE laminate showed similar trend, also. Compared to
control samples (after FD) the content of total ANTs was slightly higher in cherry
samples stored in PET/Al/PE laminate at 4 °C after 12 month of storage
(262.18 mg/100 g d.m.) than in samples stored in PET/PPmet/PE laminate
(258.61 mg/100 g d.m.). Same trend was observed for samples stored at room
temperature (20 °C). Individual ANTs were stable in both types of packaging at 4 °C,
while greater stability was observed in PET/Al/PE laminate at 20 °C. In order to
determine the level of degradation in extreme storage conditions, the samples were
also stored at 37 °C and regardless of the packaging type more than 90 % of total
ANTs were lost after 3 months of storage. ANT degradation can be result of different
oxidation processes especially at elevated temperature and cleavage of a covalent
bond. At elevated storage temperature, ANT glycosides loss sugar moieties and in
next step aglycons degrade to aldehydes and benzoic acid derivatives (Zhang et al.
2012). According to Fracassetti et al. (2013) storage at different temperatures
reduced individual and total ACN content of freeze-dried blueberry powder; it was
slower at 25 °C (−3 % after 2 weeks), whereas it was faster at 60 °C (−60 %) and at
80 °C (−85 %) after 3 days. Total ANTs of freeze dried blackcurrant packed in PE/PA
bags at 18 °C for 5 months decreased about 21 % (Kampuse et al. 2009). Piasecka
et al. (2013) reported losses of ANTs about 6 % in convectively dried sour cherry
samples packed hermetically in seal laminated foil bags at 18 °C after 12 months of
storage. FGs of FDSCM were more stable than ANTs during storage and after
12 months of storage, the lowest degradation was observed at 4 °C, while maximal
degradation was at 37 °C, as expected. Comparing packaging materials at the same
storage conditions, samples had similar total FG content. Degradation of FGs was
more than 60 % in samples packed in both laminates and stored at 20 °C during
12 months of storage. The most stable individual FG at all storage temperatures
after 12 months was K-3-R. According to Pérez-Gregorio et al. (2011) all flavonols
were stable in freeze-dried onion powder stored at room temperature in air- and
water-tight glass bottles for up to 6 months.

Stability of FGs was similarly as the stability of HCAs at same storage conditions
(Table (Table2).2). In samples stored at 20 °C decreases in total HCAs were more
intensively in first 3 months of storage especially in samples packed in PET/Al/PE
(57 %) and after 12 months of storage was about 65 % in both packaging material.
Piasecka et al. (2013) reported significant decrease of total HCAs of dried sour
cherry samples packed in seal laminated foil bags at 18 °C during the first 3 months
of storage.

It could be explain that degradation is more rapid in samples with higher dry matter
due to reacting molecules become closer to polyphenol compounds and accelerate
the rate of chemical reactions (Wang and Xu 2007).

The most stable HCAs at 4 °C and 20 °C after 12 months of storage were NChA, p-
CA and CA which decreased in PET/PPmet/PE in range from 19.15 % - 62.05 %
and in PET/Al/PE in range from 30.74 % - 69.77 %. At the same storage period ChA
had the lowest decrease in samples stored at 37 °C in PET/Al/PE (50.92 %), while
in samples stored in PET/PPmet/PP laminate decreased for 96.04 %.

At least 50 % of initial polyphenol content (expressed as sum of ANTs, FGs and

HCAs) in cherry samples stored at 20 °C packed in PET/PPmet/PE was retained
during 6 months of storage, whereas samples packed in PET/Al/PE retained the
same percentage during 3 months. Ranđelović et al. (2014) researched quality
changes of dried apricot packed in different packaging materials during 12-months
storage period at room temperature and results showed the highest polyphenol
stability in PET/Al/PE laminate. According to Henriquez et al. (2013) in the apple
peel powder stored at 38 °C for 120 days the loss of polyphenols was higher when
packed in high density polyethylene than in metalized films of high barrier. In dried
Moringa oleifera leaves packed in PET/Al/PE total phenols decreased for 19 % at
15 °C during up to 6 months of storage (Potisate et al. 2015). Del Caro et al. (2004)
showed that different polyphenol groups may differ significantly in stability during
storage.

In this study, ANOVA showed that storage time and temperature significantly
affected polyphenol stability during storage (p ≤ 0.05).
The antioxidant capacity (AOC) of the FDSCM decreased significantly during
storage (Fig. (Fig.1).1). The lowest AOC in FDSCM packed in both type of packaging
materials was determined in products stored at 37 °C after 12 months, while the
highest AOC values was determined in products stored at 4 °C after 3 months.
Significant reduction of the AOC in FDSCM packed in PET/PPmet/PE, stored at 4 °C
after 9 months was observed, whereas in the samples packaged in PET/Al/PE
significant reduction of AOC is determined after 6 months of storage. It seems that
the decrease in AOC do not follows the same pattern as polyphenol (ANTs)
reduction in the stored samples what is in accordance with other reports (Zafrilla et
al. 2003). FDSCM are very complex matrix, and correlation between individual
groups of polyphenols and AOC revealed that FGs (r = 0.93–0.95) make greater
contribution to the AOC of FDSCM than ANTs (r = 0.79–0.96) and HCAs (r = 0.77–
0.83).

Fig. 1. Antioxidant capacity of freeze-dried cherry Marasca before and after

packaging in two types of laminates (PET/PPmet/PE and PET/Al/PE) during
12 months of storage at three different temperature

According to MANOVA, storage conditions (temperature and time), affected AOC in

FDSCM, regardless of the type of packaging material, (p ≤ 0.05). This variation in
the AOC is related with structural differences of the antioxidant compounds involved
 (Rice-Evans et al. 1996). Some authors reported that delphinidin and Cy-3-R are
less active in the DPPH scavenging activity than the corresponding monoglucosides
(Kähkönen and Heinonen 2003). According to Razzaghi-Asl et al. (2013) the
presence of an unsaturated bond on the side chain of HCAs is vital to their activity.
During storage polyphenol compounds are susceptible to oxidation what probably
have major impact on AOC in FDSCM products.

The influence of packaging material and storage conditions on colour

characteristics in freeze-dried Marasca sour cherries

Colour is one of the important quality parameters that affect the acceptability of the
final product and depend on processing and storage conditions. The colour of
FDSCM during storage was reported as CIE L*, a*, b* and H values and the results
are shown in Table Table3.3. Marasca sour cherry products have intense dark red
almost black colour. The results showed similar changes in the colour of FDSCM
packed in both packaging materials and they had high values of colour parameter a
(redness of samples), which is attributed to ANTs content. The highest decrease in
a was in samples stored at 37 °C in both packaging materials after 9 and 12 months
of storage. Parameter L*, which represents lightness of sample, increased among
cherry samples during storage regardless of storage conditions. Decrease in a* and
increase in L* values is an indication of discoloration, which could be influenced by
the degradation of ANTs (Estupiñan et al. 2011).

Table 3. Colour parameters of freeze-dried Marasca sour cherry packed in two

laminate and storage in three different temperature after 0, 3, 6, 9 and
12 months

Packaging material PET/PPmet/PE PET/Al/PE

Storage Storage
L*(D65) a*(D65) b*(D65) H(D65) C*(D65) ΔE L*(D65) a*(D65) b*(D65) H(D65) C*(D65) ΔE
conditions (months)

Freeze-dried sour cherries 15.13 ± 0.03 10.37 ± 0.03 0.65 ± 0,03 3.54 ± 0.03 10.43 ± 0.03 15.13 ± 0.03 10.37 ± 0.03 0.65 ± 0.03 3.54 ± 0.03 10.43 ± 0.03

3 15.2 ± 0.02d 14.6 ± 0.02b 3.64 ± 0.01b 13.98 ± 0.02d 15.03 ± 0.01b 5.19 ± 0.03d 11.41 ± 0.01d 16.92 ± 0.02a 4.63 ± 0.01a 15.23 ± 0.03b 17.57 ± 0.02a 8.52 ± 0.02d

6 21.7 ± 0.01c 9.89 ± 0.02c 2.74 ± 0.02d 15.58 ± 0.01c 10.24 ± 0.01c 7.00 ± 0.03c 23.39 ± 0.02b 16.60 ± 0.01b 4.49 ± 0.02a 15.05 ± 0.03c 17.17 ± 0.01b 11.03 ± 0.02b
4 °C
9 26.63 ± 0.02a 16.03 ± 0.02a 4.77 ± 0.01a 16.52 ± 0.01b 16.76 ± 0.03a 13.46 ± 0.00a 22.06 ± 0.01c 3.99 ± 0.04d 0.97 ± 0.03c,d 13.47 ± 0.01d 14.06 ± 0.02d 9.43 ± 0.02c

12 25.25 ± 0.03b 9.61 ± 0.01d 3.15 ± 0.01c 18.08 ± 0.02a 10.09 ± 0.02d 10.45 ± 0.05b 25.92 ± 0.02a 15.23 ± 0.01c 4.23 ± 0.03a 15.61 ± 0.02a 15.82 ± 0.01c 12.36 ± 0.03a

3 22.07 ± 0.02d 14.47 ± 0.01a 3.72 ± 0.02a 14.45 ± 0.01c,d 14.94 ± 0.02a 8.62 ± 0.03d 12.80 ± 0.01d 13.13 ± 0.02a 4.15 ± 0.03a 17.59 ± 0.01a 13.79 ± 0.01a 5.03 ± 0.05d

6 22.99 ± 0.01c 5.76 ± 0.05d 1.52 ± 0.01d 15.02 ± 0.02b 5.91 ± 0.02d 9.16 ± 0.00c 17.31 ± 0.02c 4.09 ± 0.01d 0.97 ± 0.01d 13.37 ± 0.01d 4.22 ± 0.02d 6.66 ± 0.01c
20 °C
9 24.66 ± 0.01a 9.79 ± 0.02c 2.65 ± 0.02c 15.00 ± 0.02b 10.16 ± 0.02c 9.75 ± 0.02a 24.10 ± 0.02a 9.65 ± 0.02c 2.85 ± 0.01c 16.46 ± 0.01b 10.05 ± 0.01c 9.26 ± 0.00a

12 24.13 ± 0.02b 11.85 ± 0.01b 3.27 ± 0.01b 15.34 ± 0.01a 12.31 ± 0.01b 9.48 ± 0.05b 22.39 ± 0.01b 11.35 ± 0.01b 3.12 ± 0.02b 15.30 ± 0.01c 11.77 ± 0.01b 7.73 ± 0.02b

3 18.01 ± 0.01d 2.35 ± 0.01b 0.69 ± 0.02c 15.90 ± 0.01c 2.43 ± 0.03b 8.52 ± 0.00d 21.90 ± 0.01c 9.05 ± 0.01b 2.35 ± 0.02b 14.62 ± 0.03d 9.34 ± 0.01b 7.10 ± 0.00d

37 °C 6 21.22 ± 0.02b 3.06 ± 0.03a 1.05 ± 0.01a 19.23 ± 0.01b 3.21 ± 0.01a 9.52 ± 0.00c 23.44 ± 0.00a 12.47 ± 0.01a 5.66 ± 0.01a 24.42 ± 0.01c 13.70 ± 0.01a 9.93 ± 0.01b

9 22.01 ± 0.01a 1.45 ± 0.02d 0.29 ± 0.01d 11.89 ± 0.01d 1.48 ± 0.01d 11.27 ± 0.02a 22.61 ± 0.03b 2.35 ± 0.01c 1.72 ± 0.02c 36.35 ± 0.02a 2.95 ± 0.03c 11.02 ± 0.03a
Packaging material PET/PPmet/PE PET/Al/PE

Storage Storage
L*(D65) a*(D65) b*(D65) H(D65) C*(D65) ΔE L*(D65) a*(D65) b*(D65) H(D65) C*(D65) ΔE
conditions (months)

Freeze-dried sour cherries 15.13 ± 0.03 10.37 ± 0.03 0.65 ± 0,03 3.54 ± 0.03 10.43 ± 0.03 15.13 ± 0.03 10.37 ± 0.03 0.65 ± 0.03 3.54 ± 0.03 10.43 ± 0.03

12 20.09 ± 0.02c 1.77 ± 0.03c 0.95 ± 0.02b 27.41 ± 0.01a 2.04 ± 0.02c 9.93 ± 0.02b 17.41 ± 0.02d 1.03 ± 0.03d 0.77 ± 0.01d 35.77 ± 0.01b 1.28 ± 0.02d 9.61 ± 0.03c

View it in a separate window

L* whiteness or brightness/darkness, a* redness/greenness, b*

yellowness/blueness, H hue angle, C chroma, ∆E total colour difference
a-dThe results in the individual lines marked by the same letters do not differ
statistically at p ≤ 0.05

The chroma value (C) represents colour intensity, while hue angle (H°) depicts how
an average person will perceive that colour (Siddiq et al. 2011). The highest
decrease of C values was in samples stored at 37 °C regardless of the packaging
material what suggests instability of red colour in stored samples. The greatest
increase of H° was in samples stored at 37 °C after 9 and 12 months of storage, but
stored samples maintained H° < 40° what indicates that samples conserved dark red
colour (McGuire R.G. 1992). Presented results of the ΔE values indicate that all
stored samples are darker than the initial samples of FDSCM what is consistent with
previous reports (Wojdyło et al. 2014a). Formation of dark brown pigment during
storage is possible as a result of non-enzymatic browning reactions or to a lesser
extent as result of Maillard reaction and caramelisation (Sloan et al. 1969).

The TA in packed FDSCM products decreased linearly during storage with losses of
>60 % after 6 month of storage and change of Lab colour values indicate the
formation of dark coloured pigments as result of condensation (polymerization)
reactions of anthocyanins with other phenolic compounds. Consistent with the
showed results storage of FDSCM products at 4 °C is a suitable way to stabilize the
colour (redness).

The influence of packaging material and storage conditions on sensory

characteristics in freeze-dried Marasca sour cherries

Currently, consumer demand has increased for processed products that keep more
of their original characteristics with preserved nutritional value. Results of sensory
evaluation were summarised and displayed by Principal Component Analysis (PCA)
to explore the inter-relationship between storage conditions and sensory
parameters; in other words, which sensory attributes were characteristic in
describing the sensory profile of the samples investigated. The variables were 17
sensory attributes and the cases were freeze dried samples. The first two factors
(PC1 and PC2) represent 76.38 % of the initial data variability (Fig. (Fig.2).2). Figure
Figure22 gives a visual representation of the differences between FDSCM packed
in two different laminates, storage at 4 °C, 20 °C and 37 °C during 3, 6, 9 and
12 months. First principal component (PC1, 60.99 %) separated samples
considering storage time, and second principal component (PC2, 74.91 %)
separated samples considering storage temperature. Consequently, samples stored
9 and 12 months were mostly positioned on the left, and samples stored at 3 and
6 months as well as control sample, on the right side of the PC1. Samples stored at
4 °C were the most characterized by colour intensity. That is reasonable considering
that FD has been shown to positively affect the stability of the colour, thus freeze-
dried materials are usually more colourful, than the conventional air-dried samples
(Ratti 2001). Sensory attributes of odour intensity, sour cherry odour, sour cherry
flavour, fruity flavour, sour and harmony taste showed strong correlation with the
PC1 thus are attributes that best characterize samples stored at 4 °C. Sensory
attributes such as flowery flavour and astringent taste were more observed in
samples stored at 37 °C.

Fig. 2. PC1 vs. PC2 for sensory properties of freeze-dried sour cherry Marasca (a)
and for storage conditions (b). A – samples packed in PET/Al/PE; P – samples
packed in PET/PPmet/PE

The influence of packaging material and storage conditions on kinetic

parameters in freeze-dried Marasca sour cherries

The results of determined kinetic parameters, the rate constant (k), half-life values
(t1/2) and activation energy (Ea) confirmed also significant influence of storage
temperature on the stability of BAC’s in FDSCM samples (Table (Table 4).

Table 4. Influence of storage temperature and packaging materials on the k,

t1/2 and E a values of phenolic compounds degradation in in freeze-dried sour
cherry Marasca paste at 4 and 37 °C
Packaging material PET/PPmet/PE PET/Al/PE

k b·
t 1/2 c E a d k · 10−2 t 1/2 E a
Compoun Temperatur 10−2 2 − 2
R (week (kJ/mol (week) R (week (kJ/mol
d e (°C) (week)− 1
1 ) ) ) )

0.9 0.7 111.4 140.78

4 0.7114 97.43 86.254 0.6217
3 9 9 4
Cy-3-Sa
38.338 1.0 416.646 1.0
37 1.81 0.17
7 0 7 0

0.8 124.6 0.8 158.8

4 0.5563 88.378 0.4363 90.276
5 0 5 7
Cy-3-GR
33.072 0.9 1.0
37 2.10 28.3165 2.45
1 9 0

0.3 247.7 0.2 265.2

4 0.2798 97.295 0.2613 95.607
8 3 3 7
Cy-3-G
25.119 0.9 1.0
37 2.76 21.6978 3.19
8 9 0

0.7 106.5 0.8 126.4

4 0.6503 77.802 0.548 78.625
2 9 7 9
Cy-3-R
23.711 1.0 1.0
37 2.92 20.7568 3.34
7 0 0

0.7 0.8 110.4

4 1.1951 57.99 26.289 0.6278 36.784
7 6 1
Q-3-G
0.9 0.9
37 4.0287 17.21 3.4376 20.16
6 3

0.6 0.9
4 0.9049 76.99 22.210 1.8507 66.49 23.090
5 6
K-3-G
0.7 0.8
37 2.5262 27.49 2.176 22.87
8 1

0.3 416.0 0.8 101.9

K-3-R 4 0.1666 56.679 0.6796 23.528
7 5 8 9
Packaging material PET/PPmet/PE PET/Al/PE

k b·
t 1/2 c E a d k · 10−2 t 1/2 E a
Compoun Temperatur 10−2 2 − 2
R (week (kJ/mol (week) R (week (kJ/mol
d e (°C) (week)− 1
1 ) ) ) )

0.7 0.7
37 2.2882 30.29 2.0164 34.38
3 8

0.7 178.8 0.7 114.0

4 0.3876 63.263 0.6079 49.039
5 3 0 2
NChA
0.9 0.8
37 7.2172 9.60 5.8652 11.82
0 6

0.7 142.4 0.2

4 0.4866 56.793 0.9805 70.69 36.744
5 5 3
p-CA
0.8 0.7
37 6.7187 10.32 5.3589 12.93
8 7

0.9 0.9
4 4.1237 16.81 37.46 4.6179 15.01 32.74
7 5
ChA
0.6 0.5
37 1.791 38.70 1.083 64.00
6 8

0.8 131.8 0.7

4 0.5259 46.474 0.8354 82.97 32.647
2 0 4
CA
0.9 0.7
37 4.5067 15.38 3.7781 18.35
4 7

0.9 0.9
4 1.9054 36.38 26.305 2.1766 31.85 20.929
5 2
FA
0.9 0.9
37 6.4277 10.78 5.7271 12.10
2 7

acyanidin-3-sophoroside (Cy-3-S), cyanidin-3-glucosilrutinoside (Cy-3-GR),

cyanidin-3-glucoside (Cy-3-G), cyanidin-3-rutinoside (Cy-3-R); quercetin-3-
glucoside (Q-3-G); kaempferol-3-glucoside (K-3-G), kaempferol-3-rutinoside (K-3-
R), neochlorogenic acid (NChA), p-coumaric acid (p-CA), chlorogenic acid (ChA),
caffeic acid (CA), ferulic acid (FA)
bRate constant
cHalf life
dActivation energy

Comparing k-values in samples stored at 37 °C and at 4 °C, for Cy-3-GR as

predominant ANT, the k-values increased about 60 folds and amounted
33.0721 weeks−1 for samples packed in PET/PPmet/PE packaging and
28.316 weeks−1 in samples packed in PET/Al/PE, respectively. The t1/2 of the
individual ANTs in samples stored at 37 °C was from 0.17 (Cy-3-S in samples
packed in PET/Al/PE packaging) to 3.34 weeks (Cy-3-R in samples packed in
PET/Al/PE packaging). The t1/2 were higher in samples stored at 4 °C and amounted
97.43 weeks (Cy-3-S in samples in PET/PPmet/PE) to 126.49 weeks (Cy-3-R in
samples in PET/Al/PE), respectively (Table (Table44).

FGs have similar values for kinetic parameters as ANT in samples stored at 4 °C.
However, in samples stored at 37 °C all FGs showed more stability compared to the
ANTs in both types of packaging. Half-life (t1/2) was from 17.21 weeks (Q-3-G in
samples packed in PET/PPmet/PE) to 34.38 weeks (K-3-R) in samples packaged in
PET/Al/PE), respectively. Van der Sluis et al. (2005) reported that the FGs and their
derivatives are submissive to degradation during storage.

According to kinetic parameters, HCAs showed the highest stability compared to

other groups of polyphenol compounds. In samples stored at 37 °C, the most stable
HCA was ChA (in samples packed in PET/Al/PE), while NChA was the most unstable
in samples packed in PET/PPmet/PE, respectively (Table (Table44).

The Ea in stored cherry samples was from 20.929 (FA) to 140.784 kJ/mol (Cy-3-S)
what is approximately 1–3 folds higher than in FDSCM paste samples where
degradation of polyphenol compounds during heat treatments was monitored (Zorić
et al. 2014). These differences indicate positive effect of used packaging material on
polyphenol stability in stored FDSCM.

Kinetic parameters indicate that phenolic stability depending on structure properties

and composition of compounds present in FDSCM products as well as storage
conditions.

Conclusion

The results of this study showed that PET/PPmet/PE and PET/Al/PE can be
effectively used for packaging of FDSCM during storage at 4 and 20 °C, with
preservation BACs, colour and sensory properties. However, PET/PPmet/PE foil
offer better retention of almost all group of polyphenols in FDSCM during 12 months
of storage. Storage at lower temperatures prolongs shelf-life of cherry products and
sensory and colour properties are better maintained, what is important for
consumer’s perception.

Acknowledgments

This study was part of the project co-financed by the European Union from the
Science and Innovation Investment Fund, Component IIIC, Operational Programme
of Regional Competitiveness: Marasca Sour Cherry (Prunus cerasus var. Marasca)
as an Ingredient for Functional Food (IPA 2007/HR/16IPO/001-040302).

References

1. Brand-Williams W, Cuvelier ME, Berset C. Use of a free radical method to

evaluate antioxidant activity. Lebensm Wiss Technol. 1995;28:25–30. doi:
10.1016/S0023-6438(95)80008-5. [Cross Ref]
2. Del Caro A, Piega A, Pinna I, Fenu PM, Agabbio M. Effect of drying conditions
and storage period on polyphenolic content, antioxidant capacity, and
ascorbic acid of prunes. J Agric Food Chem. 2004;52(15):4780–4784. doi:
10.1021/jf049889j. [PubMed] [Cross Ref]
3. Elez Garofulić I, Dragović-Uzelac V, Režek Jambrak A, Jukić M. The effect of
microwave assisted extraction on the isolation of anthocyanins and phenolic
acids from sour cherry marasca (prunus cerasus var. Marasca) J Food Eng.
2013;117:437–442. doi: 10.1016/j.jfoodeng.2012.12.043. [Cross Ref]
4. Estupiñan DC, Schwartz SJ, Garzón GA. Antioxidant activity, total phenolics
content, anthocyanin, and color stability of isotonic model beverages colored
with Andes berry (rubus glaucus benth) anthocyanin powder. J Food Sci.
2011;76:26–34. doi: 10.1111/j.1750-3841.2010.01935.x. [PMC free article]
[PubMed] [Cross Ref]
5. Ferretti G, Bacchetti T, Belleggia A, Neri D. Cherry antioxidants: from farm to
table. Molecules. 2010;15:6993–7005. doi: 10.3390/molecules15106993.
[PubMed] [Cross Ref]
6. Fracassetti D, DelBo C, Simonetti P, Gardana C, Klimis-Zacas D, Ciappellano
S. Effect of time and storage temperature on anthocyanin decay and
antioxidant activity in wild blueberry (vaccinium angustifolium) powder. J Agric
Food Chem. 2013;61:2999–3005. doi: 10.1021/jf3048884. [PubMed] [Cross
Ref]
7. Gvozdenović JJ, Aljilji AR, Lazić VL, Tepić AN, Svrzić GV. Influence of
protective characteristics of packaging material on packed dried fruits. Acta
Periodica technologica. 2007;38:21–28. doi: 10.2298/APT0738021G. [Cross
Ref]
8. Henríquez C, Córdova A, Lutz M, Saavedra J. Storage stability of apple peel
powder in two different packaging materials: high-density polyethylene and
metalized films of high barrier. Ind Crop Prod. 2013;45:121–127. doi:
10.1016/j.indcrop.2012.11.032. [Cross Ref]
9. Kähkönen MP, Heinonen M. Antioxidant activity of anthocyanins and their
aglycons. J Agric Food Chem. 2003;51:628–633. doi: 10.1021/jf025551i.
[PubMed] [Cross Ref]
10. Kampuse S, Volkova I, Seglina D, Krasnova I. Effects of packaging and
preparation method on the quality of freeze-dried blackcurrant products.
Chem Technol. 2009;3(52):37–42.
11. Kwok BHL, Hu C, Durance T, Kitts D.D. Dehydration techniques affect
phytochemical contents and free radical scavenging activities of Saskatoon
berries (amelanchier alnifolia Nutt.) J Food Sci. 2004;69(3):122–126.
12. Levaj B, Dragović-Uzelac V, Delonga K, Kovačević Ganić K, Banović M,
Bursać Kovačević D. Polyphenols and volatiles in sour cherries, berries and
jams. Food Technol Biotechnol. 2010;48(4):538–547.
13. McEvily AJ, Iyengar R, Otwell WS. Inhibition of enzymatic browning in foods
and beverages. Crit Rev Food Sci Nutr. 1992;32(3):253–273. doi:
10.1080/10408399209527599. [PubMed] [Cross Ref]
14. McGuire RG. Reporting of objective color measurements. Hortic Sci.
1992;27(12):1254–1255.
15. Meillgaard M, Civille GV, Carr BT. Sensory evaluation techniques. CRC:
Press LLC, Boca Raton, FL, USA; 1999.
16. Michalczyk M, Macura R, Matuszak I. The effect of air-drying, freeze-drying
and storage on the quality and antioxidant activity of some selected berries.
J Food Process Preserv. 2009;33:11–21. doi: 10.1111/j.1745-
4549.2008.00232.x. [Cross Ref]
17. Pedisić S, Dragović-Uzelac V, Levaj B, Škevin D. Effect of maturity and
geographical region on anthocyanin content of sour cherries (prunus cerasus
var. marasca) Food Technol Biotechnol. 2010;48:86–93.
18. Pedisić S, Levaj B, Dragović-Uzelac V, Kos K. Physicochemical composition,
phenolic content and antioxidant activity of sour cherry cv. Marasca during
ripening. Agric Conspec Sci. 2007;72:295–300.
19. Pérez-Gregorio MR, Regueiro J, González-Barreiro C, Rial-Otero R, Simal-
Gándara J (2011) Changes in antioxidant flavonoids during freeze-drying of
red onions and subsequent storage. Food Control 22:1108–1113
20. Piasecka E, Uczciwek M, Klewicki R, Konopacka D, Mieszczakowska-Frąc
M, Szulc M, Bonazzi C (2013) Effect of long-time storage on the content of
polyphenols and ascorbic acid in osmo-convetively dried and osmo-freeze-
dried fruits. J Food Process Preserv 198-209. doi:10.2478/cpe-2013-0005
21. Potisate Y, Kerr WL, Phoungchandang S. Changes during storage of dried
moringa oleifera leaves prepared by heat pump-assisted dehumidified air
drying. Int J Food Sci Technol. 2015;50(5):1224–1233. doi:
10.1111/ijfs.12744. [Cross Ref]
22. Ranđelović D, Lazić V, Tepić A, Mošić I. The influence of packaging materials
protective properties and applying modified atmosphere on packed dried
apricot quality changes. Hem Ind. 2014;68(3):289–295. doi:
10.2298/HEMIND130226053R. [Cross Ref]
23. Ratti C. Hot air and freeze-drying of high-value foods: a review. J Food Eng.
2001;49:311–319. doi: 10.1016/S0260-8774(00)00228-4. [Cross Ref]
24. Razzaghi-Asl N, Garrido J, Khazraei H, Borges F, Firuzi O. Antioxidant
properties of hydroxycinnamic acids: a review of structure- activity
relationships. Curr Med Chem. 2013;20(36):4436–4450. doi:
10.2174/09298673113209990141. [PubMed] [Cross Ref]
25. Rice-Evans CA, Miller NJ, Paganga G. Structure-antioxidant activity
relationships of flavonoids and phenolic acids. Free Radic Biol Med.
1996;20:933–956. doi: 10.1016/0891-5849(95)02227-9. [PubMed] [Cross
Ref]
26. Sablani SS, Andrews PK, Davies NM, Walters T, Saez H, Bastarrachea L.
Effects of air and freeze drying on phytochemical content of conventional and
organic berries. Dry Technol. 2011;29(2):205–216. doi:
10.1080/07373937.2010.483047. [Cross Ref]
27. Siddiq M, Iezzoni A, Khan A, Breen P, Sebolt AM, Dolan KD, Ravi R.
Characterisation of new tart cherry (prunus cerasus L.) selections based on
fruit quality, total anthocyanins, and antioxidant capacity. Int J Food Prop.
2011;14:471–480. doi: 10.1080/10942910903277697. [Cross Ref]
28. Sloan JL, Bills DD, Libbey LM. Heat-induced compounds in strawberries. J
Agric Food Chem. 1969;17(6):1370–1372. doi: 10.1021/jf60166a020. [Cross
Ref]
29. van der Sluis AA, Dekker M, van Boekel MA. Activity and concentration of
polyphenolic antioxidants in apple juice. 3. Stability during storage. J Agric
Food Chem. 2005;53:1073–1080. doi: 10.1021/jf040270r. [PubMed] [Cross
Ref]
30. Verbeyst L, Oey I, Van der Plancken I, Hendrickx M, Van Loey A. Kinetic
study on the thermal and pressure degradation of anthocyanins in
strawberries. Food Chem. 2010;123:269–274. doi:
10.1016/j.foodchem.2010.04.027. [Cross Ref]
31. Wang WD, Xu SY. Degradation kinetics of anthocyanins in blackberry juice
and concentrate. J Food Eng. 2007;82:271–275. doi:
10.1016/j.jfoodeng.2007.01.018. [Cross Ref]
32. Wojdyło A, Figiel A, Lech K, Nowicka P, Oszmiański J. Effect of convective
and vacuum–microwave drying on the bioactive compounds, color, and
antioxidant capacity of sour cherries. Food Bio Tech. 2014;7(3):829–841. doi:
10.1007/s11947-013-1130-8. [Cross Ref]
33. Wojdyło A, Nowicka P, Laskowski P, Oszmiański J. Evaluation of sour cherry
(prunus cerasus L.) fruits for their polyphenol content, antioxidant properties,
and nutritional components. J Agric Food Chem. 2014;62(51):12332–12345.
doi: 10.1021/jf504023z. [PubMed] [Cross Ref]
34. Zafrilla P, Morillas J, Mulero J, Cayuela J, Martinez-Cacha A, Pardo F, Lopez
Nicolas JM. Changes during storage in conventional and ecological wine:
phenolic content and antioxidant activity. J Agric Food Chem.
2003;51(16):4694–4700. doi: 10.1021/jf021251p. [PubMed] [Cross Ref]
35. Zhang L, Zhou J, Liu H, Khan MA, Huang K, Gu Z. Compositions of
anthocyanins in blackberry juice and their thermal degradation in relation to
antioxidant activity. Eur Food Res Technol. 2012;235:637–645. doi:
10.1007/s00217-012-1796-6. [Cross Ref]
36. Zorić Z, Dragović-Uzelac V, Pedisić S, Kurtanjek Ž, Elez Garofulić I. Kinetics
of the degradation of anthocyanins, phenolic acidsand flavonols during heat
treatments of freeze-dried sour cherry marasca paste. Food Technol
Biotechnol. 2014;52(1):101–108.
II. Production and evaluation of mineral and nutrient
contents, chemical composition, and sensory
properties of ice creams fortified with laboratory-
prepared peach fibre
Food Nutr Res. 2016; 60: 10.3402/fnr.v60.31882.
Filiz Yangılar*

Abstract

Background

In the coming years, a nutraceutical food may provide both physical and mental
benefits that are commonly attributed to the active components of the food.

Objective

In this study, we determined the nutrient and mineral contents, sensory properties,
and physical and chemical characteristics of ice creams manufactured using peach
fibre at different concentrations (1 and 2%).

Method

A total of five experimental groups were formed: two types (from peach peel and
pulp) of flour, two fibre concentrations (1 and 2%), and a control group without fibres.

Results

Flour obtained from peach pulp and peel was found to have a significant (p<0.05)
effect on the chemical composition and elemental composition of ice cream samples,
especially the rates of Ca, K, Mg, and P, which increased in the samples depending
on the content of peach fibre. Sensory ratings and acceptability of ice creams
decreased significantly with increasing peach peel fibre, whereas ice creams made
with C (control) and B1 (ice creams made from 1% peach pulp fibre) was the highest
scored by the panellists.

Conclusions
Peach fibre concentrates might be used as a good source of nutraceutical
ingredients.

Keywords: ice cream, peach fibre, chemical composition

Peach (Prunus persica L.) fruits have high economic and nutritional value.
Carbohydrates, organic acids, minerals, and dietary fibre (DF) are among the major
components of peach fruit, which contribute to the nutritional quality of both fruits
and juices (1, 2). Peach fruits have aperient properties, are appropriate to prevent
costiveness, and are used for the treatment of duodenum ulcers. Phenolic acids,
flavonoids, and anthocyanin compounds serve as major sources of potential
antioxidants in peach fruit, which might be responsible for these therapeutic
functions (2). Peaches are a tropical fruit, which means that the ripening process is
triggered and driven by a plant hormone called ethylene. This is responsible for its
short shelf-life and represents a serious constraint for its efficient handling and
transportation (3, 4).

DF can provide a multitude of functional properties when they are incorporated into
food systems (5). But, there are little data dealing with the study of the functionality
of DF in ice creams (6). To the best of our knowledge, no data have been published
regarding the effect of drying technique on physicochemical characteristics of fibre
obtained from peach or about possible differences on physical properties and
functionality of peach fibre because of the type of tissue used (peel or pulp) (7). Fibre
addition contributes to the modification and improvement of the texture, sensory
characteristics, and shelf-life of food because of its water-binding capacity, gel-
forming ability, and fat mimetic, antisticking, anticlumping, texturising, and thickening
effects (6).

Peach fibre is a valuable nutrition component and its non-use for human diet
constitutes a real nutritional loss because peach fibres contain extractable bioactive
compounds which can be used as value-added materials. The objective of the
present study is to evaluate the functional properties of fibres obtained from peach
peel and pulp for possible enrichment of the quality and nutritional contents of ice
creams.

Materials and methods

Materials

Cow's milk and cream were purchased from the Research and Application Farm of
Atatürk University, Erzurum, Turkey. The peach fruits were purchased from a
supermarket in Erzurum, Turkey, and classified visually for colour and physical
damage. Sugar, salep, and emulsifier (mono- and diglycerides) were obtained from
the local market. Skimmed milk powder was purchased from a commercial company
(Pınar Dairy Products Co., Istanbul, Turkey).
Preparation of peach peel and pulp fibre

The total weight of the peaches used was 4 kg, and the average diameter of each
peach was 7 cm. Pulp and peel were stripped off after each peach fruit was dipped
in water and washed. Pulp and peel were submitted to a juice extraction using a
domestic appliance with the purpose of eliminating a great part of the water content
and its solubility. The remaining solids were contacted with ethanol (96%, v/v) and
stirred (600 rpm) for 15 min. Finally, ethanol was discarded and a portion of each
remnant (pulp remnant or peel remnant) was dried under 30°C forced air convection
for 7 h. As a result of the process, two types of flour were produced: type A flour,
obtained from peach peel, and B flour, obtained from peach pulp. The two types of
flour were milled in a Mill Laboratory at 2,890 rpm and then at 5,000 rpm until they
could pass through a 0.2-mm sieve to recover the peach fibre concentrates and
stored at −18°C for subsequent analyses and incorporation studies (7).

Flour yield

Yield was calculated by dividing the amount of flour produced by the amount of
peach used, and the results were converted to g/kg (g of flour per kg of peach).

Chemical and physical composition of DFs

Moisture content was measured according to the Association of Official Analytical

Chemists (AOAC) method (8). Ash was analysed by combusting the sample in a
muffle furnace at 550°C for 4 h. The residue was dissolved in HNO3 and the mineral
constituents (Ca, K, Na, Mg, P, and Fe) were determined using an inductively
coupled plasma optical emission spectrophotometer (Optima 2100 DV, ICP/OES,
Perkin-Elmer, Shelton, CT). The Kjeldahl method was used to evaluate the protein
content (9). The flour's pH was determined using a pH meter (Mettler-Toledo AG
8603, Schwerzenbach, Switzerland) (10). The colour of the flour was monitored
using a Minolta colorimeter (Chroma Meter CR-200, Osaka, Japan). DF, that also
accounts for structural variability and associated health benefits, has recently been
defined by CODEX Alimentarius to be ‘carbohydrate polymers’, which have been
obtained from food raw material by physical, enzymatic, or chemical means and
which have been shown to have a physiological effect of benefit to health as
demonstrated by generally accepted scientific evidence to competent authorities
(11, 12). Insoluble dietary fibres (IDFs) and soluble dietary fibres (SDFs) were
assessed using the method of Prosky et al. (13). In this method, samples were
enzymatically digested under the same conditions as used in the AOAC official
method, and the total dietary fibre (TDF) was calculated as the sum of IDFs and
SDFs. Total phenolic content was determined for peach peel and pulp flour
according to Bunzel et al. (14). The total phenolic contents were evaluated using the
Folin–Ciocalteu method (15).

Water- and oil-holding capacity

A volume of 25 mL of distilled water or commercial olive oil was mixed with 1 g of
dry sample. The mixture tubes were centrifuged at 3,000g for 20 min and then the
supernatant was poured out. After that, the tubes were drained for 10 min by putting
them at an angle of 45°. The residue was weighed and water-holding capacity (WHC,
g of water per 100 g of sample) and oil-holding capacity (OHC, g of oil per 100 g of
sample) were calculated according to Gould et al. (16) and Caprez et al. (17),
respectively.

Manufacture of ice creams

The ice cream samples were prepared at the Seref patisserie, Erzurum, Turkey.
First, the fat content of cow's milk was adjusted to 6% by adding cream, and the
prepared milk was separated into five equal amounts of 2 kg each. Skimmed milk
powder (125 g), sugar (405 g), a stabiliser (salep; 16.2 g), and emulsifiers (mono-
and diglycerides; 6.75 g) were added to these milk samples. Flour samples obtained
both from peach peel and pulp were also added to the milk samples (65°C) at two
different mass fractions (1 and 2%). The mixtures were subjected to pasteurisation
at 85°C for 25 min and stored at 4°C for 24 h, after which they were placed in an ice
cream machine to freeze and then hardened at −22°C for 24 h. They remained at
−18°C during all physical, chemical, mineral, and sensory analyses.

Ice cream analysis

Dry matter, fat and ash contents, acidity (°SH), and pH of ice cream samples were
determined using the method of Demirci and Gündüz (18). Mineral contents (Ca, K,
Na, P, S, Mg, Fe, Mn, Zn, and Ni) of ice cream samples were described by Guler
(19). Overrun was detected by the method proposed by Jimenez et al. (20) and
calculated using the following equation:

Overrum=[V(ice cream)−V(mix)/V(mix)]×100(V:volume)

Time period (sec) from the initial dripping to complete melting was determined using
the method of Güven and Karaca (21). Samples of 25 g were heated to melt at room
temperature (20°C) by putting them in a beaker capped with a 0.2-cm wire mesh
screen. Initial dripping and complete melting times of the samples were calculated
in seconds. Viscosity of the ice cream samples was measured at 4°C using a digital
Brookfield viscometer (22).

The colour parameters of ice cream samples were obtained by measuring L*

(brightness, 0: black, 100: white), a* (+: red, −: green), and b* (+: yellow, −: blue)
values using the method of described in (23).

Sensory assessment

Eight professional panellists participated in the study and evaluated the ice cream
samples using a score test for flavour, body and texture, colour and appearance,
resistance to melting, and general acceptability. Hardened ice cream samples were
tested at a serving temperature of −10°C and given scores for their sensory
characteristics using a scale ranging from 1 (poor) to 9 (excellent). All panellists were
preferred to be non-smokers (24).

Statistical analysis

In total, five experimental groups were formed: two types (from peach peel and pulp)
of flour, two fibre concentrations (1 and 2%), and a control group without fibres.
Statistical analysis was carried out using SPSS 15.0 (SPSS, Inc., Chicago, IL) (25)
software. Data were subjected to a multiple analysis of variance, and the average
values were compared using Duncan's multiple range test.

Results and discussion

The mean yield of peach peel and peach pulp flour was calculated to be 57.1 and
100.5 g/kg, respectively. Pulp was found to give a larger amount of flour than peel.

Chemical and physical characteristics of DFs

Chemical compositions of fibre samples produced from peach peel and pulp are
given in Table 1. Peach pulp flour was found to have higher moisture content
(11.36%) than peach peel flour (9.21%). Abdul Aziz et al. (26) stated that mango
pulp flours have consistently higher moisture content than mango peel flours, and
moisture content increased in the pulp as the ripening process progressed and are
in agreement with our report.

Table 1. The chemical and physical properties of peach peel and pulp flour

Peach peel (type A) Peach pulp (type B)

Chemical analysis

Moisture (g/100 g) 9.21±0.50 11.36±0.20

Ash (g/100 g) 4.3±0.28 4.11±0.37

pH 4.21±0.43 3.58±0.64

Protein (g/100 g) 7.51±0.71 6.90±0.20

Total phenolic contentc 1002±1.91 1566±0.91

Total dietary fibre (g/100 g) 57.48±0.25 51.05±0.24

Insoluble dietary fibre (g/100 g) 53.15±0.41 46.21±0.63

 Peach peel (type A) Peach pulp (type B)

Soluble dietary fibre (g/100 g) 12.00±1.10 8.53±0.72

Minerals (mg/kg)

Ca 88.36±0.18 53.60±0.37

K 1269±0.83 1744±0.03

Mg 97.72±0.15 55.30±0.43

Zn 0.54±0.27 0.27±0.52

Fe 7.14±0.29 2.43±0.50

Physical analysis

L* 58±1.00 79.7±0.06

a* 11.42±0.17 6.99±0.10

b* 37.49±0.71 28.71±0.46

WHCa 13.10±0.92 21.87±0.70

OHCb 2.88±0.64 1.76±0.32

L*, lightness; a*, redness (+); b*, blueness (−).

aWater-holding capacity (g water per g sample).
bOil-holding capacity (g oil per g sample).
cmg/100 g of fibre concentrates.

As observed in Table 1, the ash content of peach peel and pulp flour was found to
be 4.3 and 4.11%, respectively. Rodríquez-Ambriz et al. (27) found that banana flour
had an ash content of 4.4%, whereas Juarez-Garcia et al. (28) found this rate to be
4.7%. These findings were in agreement with the findings of the present study.

The WHC value obtained for the peach peel flour was lower than that obtained for
the peach pulp flour, and that found (21.87 g water per g dry sample) in the present
study was lower than that reported by Grigelmo-Miguel and Martín-Belloso (29) in
peach DF (between 9.2 and 12.1 g of water per g of dry sample). de Escalada Pla
et al. (7) found that the WHC value of peach pulp fibre was 24 g water per g dry
sample, whereas the WHC value of peach peel fibre was 25 g water per g dry
sample. OHC is among the important functional properties of peach flour and was
found in the peach peel (2.88 g oil per g dry sample) and pulp flour samples (1.76 g
oil per g dry sample) in our study. When compared with the results of previous
studies, the peach pulp flour values were lower than the value of 2.2 g of oil per g of
dry sample found in the study of Rodríquez-Ambriz et al. (27). It was stated by de
Escalada Pla et al. (7) that the OHC value in peach pulp fibre was 1.81 g oil per g
dry sample and in peach peel fibre was 2.03 g oil per g dry sample. Calvache et al.
(30) reported high values of OHC in DF from peach bagasse P. persica L., which is
in agreement with the results obtained in this study.

The mean L*, a*, and b* values were found to be 58, 11.42, and 37.49 in peach peel
flour, and 79.7, 6.99, and 28.71 in peach pulp flour, respectively. It was found by
monitoring the samples in the present study that the flour obtained from peach peel
had a darker colour than that produced from pulp. Because of the possible existence
of some browning contributor enzymes, such as polyphenol oxidase, and the
occurrence of the Maillard reaction (31), a significant colour change was observed
during the peel drying process in the present study, from which dark brown powder
was produced from the peel. The L*, a*, and b* values were found to be 76, 5.5, and
29.39 in peach pulp flour, and 56, 12.2, and 34 in peach peel flour, respectively (7),
and these results are in agreement with the results of the present study.

Antioxidants play an important role in the prevention of oxidative stress-related

diseases (32). Quantitatively, the main dietary antioxidants are polyphenols,
followed by vitamins, and carotenoids (33). Goñi et al. (34) stated that polyphenols
associated with polysaccharides and proteins in cell walls are significant constituents
of DF. Table 1 shows the polyphenol contents of the peach peel and pulp flour types.
The pulp flour (1,566 mg/100 g) was found to have the highest content of
polyphenols. The total phenolics of the pulp and peel of three different peach fruit
varieties (Golden, Shireen, and Shahpasand) was determined by Manzoor et al. (2).
Among the different peach varieties tested, the peel and pulp of cv. Golden exhibited
the highest phenolic contents (1,354.5 and 881.3 mg gallic acid equivalents (GAE)
per 100 of DW, respectively). The findings given above are in accordance with those
found in the present study.

As observed in Table 1, peach peel fibres are rich in protein (7.51 g/100 g). Grigelmo-
Miguel and Martín-Belloso (29) reported that the protein content (5.44–6.29%) of
peach fibre concentrate was also low and that it was the only component, among
those studied, that decreased throughout the peach harvest time. Ahmed et al. (35)
reported that the sodium content was significantly lower than other minerals (55–86
mg/kg); however, the fibres were rich in potassium.

TDF content in the peach peel flour was found to be 57.48% (Table 1), the majority
of which was represented by IDF (53.15%) and the rest by SDF (12.00%). Compared
with this finding, in a fibre-rich fraction of chia (Salvia hispanica), Alfredo et al. (36)
reported TDF, IDF, and SDF contents to be 56.46, 53.45, and 3.01 g per 100 g
respectively, whereas in orange peel were found, determined by Chau and Huang
(37), to be 57, 47.6, and 9.4 g per 100 g, respectively. These rates were 64.1, 55.2,
and 8.9 g per 100 g, respectively, found by Ruales and Zumba (38) in guava and
were 66.8, 58.6, and 8.2 g per 100 g, respectively, found by Yangılar (39) in green
banana flour. The findings given above are in agreement with those found in the
present study.

Physical and chemical characteristics of ice cream samples

The results of some physical and chemical analyses and mineral contents of ice
cream samples are given in Tables 2 and and3.3. The dry matter content of the
control sample was lower than that of other samples at significant levels (p<0.05).
The dry matter rate of ice creams increased with the addition of peach fibre. As
observed in Table 2, the addition of DF significantly affected the fat and acidity
values (p<0.05). The pH values of ice cream samples ranged from 5.89 to 6.51 in
samples with fibre.

Table 2. Some chemical and physical properties of ice cream samples with
peach fibre

Analysis C A1 A2 B1 B2

Dry matter (%) 40.01±0.01a 40.45±0.04a 41.21±0.16b 40.33±0.30a 41.05±0.05b

Ash (%) 0.98±0.00a 1.01±0.02ab 1.14±0.03d 1.07±0.02bc 1.11±0.02cd

Fat (%) 6.55±0.07b 6.11±0.01a 5.98±0.02a 6.00±0.03a 6.06±0.08a

Acidity (%) 0.30±0.00ab 0.31±0.02c 0.34±0.03d 0.28±0.01a 0.30±0.00ab

pH 6.51±0.02c 6.10±0.14b 6.15±0.04b 6.04±0.04ab 5.89±0.04a

L* 87.17±0.24e 78.18±0.28b 74.85±0.37a 83.87±0.98d 80.30±0.43c

a* 3.26±0.09a 4.02±0.03bc 4.13±0.01c 3.81±0.10b 4.00±0.00bc

b* 9.69±0.05a 10.84±0.09c 11.07±0.21d 10.12±0.17ab 10.55±0.57bc

*Mean values followed by different letters in the same row are significantly different
(p<0.05).

C, control without fibre; A1, ice cream made with 1% (w/w) peach peel fibre; A2, ice
cream made with 2% (w/w) peach peel fibre; B1, ice cream made with 1% (w/w)
peach pulp fibre; B2, ice cream made with 2% (w/w) peach pulp fibre.

Table 3. Elemental composition (mg/kg) of the ash in ice creams with peach
fibre
Concentrati
ons of C A1 A2 B1 B2
minerals

1981.11±7.
Ca 2157±26.82c 2256±8.57d 2101±2.18b 2104±5.65b
96a

2050±14.14 2174.05±25. 2435.33±49. 2372.82±32.

K a 2263±36.76c
18b 96e 97d

488.30±3.8 438.78±0.82 413.96±5.60

Na d 463.24±4.58c b a 459.49±2.24c
1

1038.45±9. 2224.21±7.1 2342.78±37. 2518.21±36. 2618.67±29.

P
12a 9b 85c 18d 25e

541.32±1.8 628.60±2.12 590.18±1.61 494.32±4.70 513.56±4.73

S
7c e d a b

124.65±0.4 161.45±1.47 188.01±2.80 192.85±2.04

Mg a b d 179.35±0.91c d
9

Fe 10.47±0.16a 12.18±0.26b 12.58±0.18c 11.01±0.04a 14.43±1.72d

Mn 0.46±0.01a 0.54±0.03b 0.69±0.01c 0.55±0.05ab 0.66±0.04bc

112.21±3.12 100.52±0.29
Zn 73.54±0.65a 99.85±0.21b d b 106.77±3.50c

Ni 1.59±0.01a 1.60±0.02a 1.69±0.02c 1.61±0.04b 1.70±0.01c

*Mean values followed by different letters in the same row are significantly different
(p<0.05).

C, control without fibre; A1, ice cream made with 1% (w/w) peach peel fibre; A2, ice
cream made with 2% (w/w) peach peel fibre; B1, ice cream made with 1% (w/w)
peach pulp fibre; B2, ice cream made with 2% (w/w) peach pulp fibre.

Viscosity, accepted to be among the significant characteristics of an ice cream

mixture, may result in good body and texture properties in the production process of
ice cream. From this point of view, it is important to measure viscosity to determine
how peach flour may affect the characteristics of an ice cream mixture. As shown in
Fig. 1, a lowest viscosity value was obtained in the control sample (3,100 cP) and
the highest in sample A (with 2% of peach peel flour added; 4,600 cP). These
findings are in agreement with the results of Hwang et al. (40) for ice cream samples
with grape wine lees, Dervisoglu and Yazici (5) for ice cream samples with citrus
fibre, Cakmakci et al. (41) for ice creams with oleaster (Elaeagnus angustifolia L.)
flour, and Yangılar (42) for ice cream samples containing date fibre.

Fig. 1. Viscosity values (20 rpm) of ice cream mixes produced using peach fibre.
Different letters above the bars indicate significant differences by Duncan multiple
comparison test (p<0.05).

The lightness (L*) values of ice cream samples were close to each other, but they
were significantly higher for the A1 and A2 samples than for others (Table 2). All of
the samples taken into consideration were found to have negative greenness rates,
whereas B1 and B2 samples seemed to be similar to and sometimes higher than the
others. The colour rates of the samples were affected favourably by an increase in
the concentration of peach fibre (p<0.05). Samples had negative a* (greenness)
values, and the A2 (4.13) sample was significantly higher than the other samples.
The b* values were increased by the addition of peach peel fibre. Samples C (9.69)
gave the lowest b* rate, whereas the highest rate was received from the A2 (11.07)
samples. Dervisoglu and Yazici (5) reported that the addition of citrus fibre increased
the colour properties; these results are in agreement with the results of the present
study.

Overrun and melting are associated with the volume of air involved in the
manufacturing process. This property can shape the structure of the final product
because the air present in the ice cream can provide it with a light texture and affect
some physical properties, such as melting and hardness (43). All of the ice cream
samples in the present study showed much lower overrun values (31.8–42.35%)
compared with the values reported in the literature (80–120%). Although the addition
of peach fibre lowered the overrun rate of the ice cream samples (p>0.05), the
control samples showed higher overrun rates than the peach flour samples (Fig. 2).
Because the viscosity of the ice cream increased with fibre samples, it was possible
that less air was incorporated into the ice cream mixture with fibre during batch
freezing, which resulted in a lower overrun compared with the control (without fibre).
The results of the study carried out by El-Samahy et al. (44) showed that overrun
decreased in ice cream as cactus pulp was added, which could be dependent on the
increase in the viscosity of the mixture. Hwang et al. (40) reported that the overrun
values of ice cream samples decreased significantly with the addition of grape wine
lees. It was found by Sun-Waterhouse et al. (45) that the overrun rate of ice cream
containing green kiwifruit was 90.5%, which is higher than that in the present study.
Results similar to those found in the present study were found by Cakmakci et al.
(41) who studied producing ice cream with oleaster (E. angustifolia L.) flour.

Fig. 2 Overrun values of ice cream mixes produced using peach fibre. Different
letters above the bars indicate significant differences by Duncan multiple comparison
test (p<0.05).

As observed in Fig. 3, the length of the time period required for the melting process
to be completed was found to be significantly longer in B2 samples with the addition
of fibre content. Such a situation may result from the addition of some components
with the ability to absorb water in B. The ice cream with B2 revealed the longest
complete melting time (0.43 g/min), whereas the shortest complete melting time
(0.37 g/min) was shown by the A1 sample. It was stated by Akin et al. (22) that the
reason for a decrease in the melting rate of ice cream with added inulin might be
because of the ability of inulin to prevent water molecules from moving freely. B (1
and 2%) concentrations affected the initial dripping time positively (Table 2).
Fig. 3. First dripping and complete melting times of ice cream samples with added
peach fibre. Different letters above the bars indicate significant differences by
Duncan multiple comparison test (p<0.05).

The results of the present study indicate that the length of the time period until the
initial dripping became prolonged as the fibre content of the ice cream samples
increased (p<0.05). Dervisoglu and Yazici (5) reported that ice cream samples with
citrus fibre had longer dripping times and a similar result was also reported by
Yangılar (39) in ice cream with green banana peel flour.

Table 3 shows the changes in the mineral contents of the ice cream samples. The
contents of K, Mg, and P in ice cream samples significantly increased, and
statistically significant differences were found in terms of the major element contents,
such as Ca, K, P, and Mg, between the samples, except for the S concentration in
all ice cream samples. The increase in these mineral contents may be because of
the high K, Mg, and P concentrations in peach fruit. As observed in Table 3, the K
content of peach peel was between 2174.05 and 2435.33 mg/kg, and the A2 sample
in the present study revealed the highest K rate. Manzoor et al. (2) reported that
mineral analysis results were in agreement with those of Basar (46) who reported
that K was the most abundant nutrient in fruits of different peach varieties, followed
by Mg, Ca, Fe, Zn, and Mn. The A1 sample gave the highest S rate (628.60 mg/kg),
whereas the B1 sample had the lowest (494.32 mg/kg). In our study, the Na values
of ice cream were found to be decreased in peach fibre-added samples. If we
evaluate the results only in the study, we see that the decrease is seen in the fibre
additive rather than in the control; however, the highest increase is seen in the Na
rate coming from the peel flour. A similar result was reported by Dağdemir (47) for
vegetable marrow (Cucurbita pepo L.) added to ice cream. Decreasing Na in the
human diet may provide protection from hypertension in people who are sensitive to
high levels of Na. A2 revealed the highest Zn content (112.21 mg/kg). It was stated
by Wu et al. (48) that Zn could play some vital roles by serving as a non-enzymatic
antioxidant and protecting cells from oxidative damage. Even if peach fibre contains
small doses of Fe, Zn, Ni, and Mn, which carry the capability of contributing to the
antioxidant activity of fruit (49), in the present study, when added, fibre seems to rise
significantly the Fe, Zn, and Mn contents of the ice cream samples (p<0.05). Similar
results were reported for Cape gooseberry (Physalis peruviana L.) added to ice
cream by Erkaya et al. (50) and for the ice cream samples with date fibre by Yangılar
(42).

Sensory evaluation of ice creams

The results of the sensory evaluation of the ice cream samples on a scale from 1
(poor) to 9 (excellent) are shown in a radar plot in Fig. 4. Fortifying ice cream with
DF had a significant effect on all sensory properties overall. All of the fibre-enriched
samples received higher scores for total evaluation in terms of sensory
characteristics (p<0.05). Colour scores were significantly different, between 6.95 and
8.20. The lowest score was determined for sample A2. Peach fibre gave a slight
yellowish colour to ice cream samples. The A1 and A2 samples containing peach
peel fibre showed relatively high scores in terms of organoleptic characteristics such
as body and texture, resistance to melting, and mouth feeling compared with the
control group. The B2 sample had a similar mouth feeling, showed resistance to
melting, and gave the same generally acceptable ratings as the control sample. The
highest values for general acceptability belonged to the C, B1, B2, A1, and A2
samples. Processing of peach into fibre presents an excellent opportunity for use as
a functional ingredient to ensure its extended consumption and reduce wastage.

Fig. 4. Some sensorial properties of ice cream samples.

Conclusions

The enrichment of ice cream with peach fibre is an effective way to enhance
nutritional and physiological aspects by influencing the rheological and thermal
properties of the final product. Peach fibre alone or with ice cream stabilisers was
successfully used in ice cream production. The addition of peach peel fibre affected
moisture, fat, acidity, ash, and viscosity positively; on the contrary, meltdown, colour,
and overrun were affected negatively. Given peach fibre's nutritive value and
pleasant flavour, it may be used as a suitable source of natural additive in ice cream
production to enhance nutritional values. In our study, among the two parts (peel
and pulp) analysed, the present results also revealed that peach pulp exhibited
higher moisture, phenolic content, and WHC value compared with that of the peel,
indicating that removal of peel from such fruits may induce significant nutrient losses.
Therefore, the intake of functional foods along with their fibres can be more beneficial
to increased nutritional properties of foods such as ice cream. Therefore, our study
may provide a base of knowledge for future research.

Conflict of interest and funding

There is no conflict of interest to declare.

References

1. Versari A, Castellari M, Parpinello GP, Riponi C, Galassi S. Characterization of

peach juices obtained from cultivars Redhaven, Suncrest and Maria Marta grown in
Italy. Food Chem. 2002;76:181–5.
2. Manzoor M, Anwar F, Mahmood Z, Rashid U, Ashraf M. Variation in minerals,
phenolics and antioxidant activity of peel and pulp of different varieties of peach
(Prunus persica L.) fruit from Pakistan. Molecules. 2012;17(6):6491–506. [PubMed]
3. Hussain PR, Meena RS, Dar MA, Wani AM. Studies on enhancing the keeping
quality of peach (Prunus persica Bausch) Cv. Elbert a by gamma-irradiation. Radiat
Phys Chem. 2008;77(4):473–81.
4. Kim KH, Kim MS, Kim HG, Yook HS. Inactivation of contaminated fungi and
antioxidant effects of peach (Prunus persica L. Batsch cv Dangeumdo) by 0.5–2 kGy
gamma irradiation. Radiat Phys Chem. 2010;79(4):495–501.
5. Dervişoğlu M, Yazıcı F. Note. The effect of citrus fibre on the physical, chemical
and sensory properties of ice cream. Food Sci Technol Int. 2006;12(2):159–64.
6. Soukoulis C, Lebesi D, Tzia C. Enrichment of ice cream with dietary fibre: effects
on rheological properties, ice crystallisation and glass transition phenomena. Food
Chem. 2009;115(2):665–71.
7. de Escalada Pla MF, González P, Sette P, Portillo F, Rojas AM, Gerschenson LN.
Effect of processing on physico-chemical characteristics of dietary fibre concentrates
obtained from peach (Prunus persica L.) peel and pulp. Food Res Int.
2012;49(1):184–92.
8. AOAC. Official methods of analysis. Washington, DC: Association of Official
Analytical Chemists; 1997.
9. AOAC. Official methods of analysis. 15th ed. Washington, DC: Association of
Official Analytical Chemists; 1990.
10. Suntharalingam S, Ravindran G. Physical and biochemical properties of green
banana flour. Plant Foods Hum Nutr. 1993;43(1):19–27. [PubMed]
11. McCleary BV, DeVries JW, Rader JI, Cohen Gerald, Prosky leon C, Mugford D,
et al. Collaborative study report: determination of insoluble, soluble, and total dietary
fibre (Codex definition) by an enzymatic-gravimetric method and liquid
chromatography. Cereal Food World, 2011;56(6):238.
12. Jones JM. Dietary fiber future directions: integrating new definitions and findings
to inform nutrition research and communication. Adv Nutr. 2013;4(1):8–15. [PMC
free article] [PubMed]
13. Prosky L, Asp NG, Schweizer TF, DeVries JW, Furda I. Determination of
insoluble, soluble, and total dietary fibre in foods and food products: inter laboratory
study. J Assoc Off Anal Chem. 1988;71(5):1017–23. [PubMed]
14. Bunzel M, Ralph J, Marita J, Steinhart H. Identification of 4-O-5′-coupled diferulic
acid from insoluble cereal fibre. J Agric Food Chem. 2000;48(8):3166–9. [PubMed]
15. Singleton VL, Rossi JA. Colorimetry of total phenolics with phosphomolybdic-
phosphotungstic acid reagents. Am J Enol Viticul. 1965;16(3):144–58.
16. Gould JM, Jasberg BK, Dexter LB, Hsu JT, Lewis SM, Fahey GC. High-fibre,
noncaloric flour substitute for baked foods. Properties of alkaline peroxide-treated
lignocellulose. Cereal Chem. 1989;66(3):201–5.
17. Caprez A, Arrigoni E, Amado R, Zeukom H. Influence of different types of thermal
treatment on the chemical composition and physical properties of wheat bran. J
Cereal Sci. 1986;4(3):233–9.
18. Demirci M, Gündüz H. Süt teknolojisi el kitabı. Hasad Yayıncılık. Istanbul,
Turkey: 1997. p. 66.
19. Güler Z. Levels of 24 minerals in local goat milk, its strained yoghurt and salted
yoghurt (tuzlu yoğurt) Small Ruminant Res. 2007;71(1–3):130–7.
20. Jimenez-Florez R, Klipfel NJ, Tobias J. Ice cream and frozen desserts. In: Hui
YH, editor. Dairy science and technology handbook: 2. Product manufacturing. New
York: VCH Publishers; 1993. p. 57.
21. Güven M, Karaca OB. The effects of varying sugar content and fruit
concentration on the physical properties of vanilla and fruit ice-cream-type frozen
yogurts. Int J Dairy Technol. 2002;55(1):27–31.
22. Akın MB, Akın MS, Kırmacı Z. Effects of inulin and sugar levels on the viability
of yogurt and probiotic bacteria and the physical and sensory characteristics in
probiotic ice-cream. Food Chem. 2007;104(1):93–9.
23. Anonymous. Farbmetrische Bestimmung von Farbabstanden bci Korperfarben
nach der CIELAB Formol. Berlin, Germany: Beuth-Vertrieb GMbH; 1979. p. 30.
24. Roland AM, Phillips LG, Boor KJ. Effects of fat content on the sensory properties,
melting, colour, and hardness of ice cream. J Dairy Sci. 1999;82(1):32–8.
25. SAS. SAS/STAT guide for personal computers. version 6.12 ed. Cary, NC: SAS
Institute; 1998.
26. Abdul Aziz NA, Wong LM, Bhat R, Cheng LH. Evaluation of processed green
and ripe mango peel and pulp flours (Mangifera indica var. Chokanan) in terms of
chemical composition, antioxidant compounds and functional properties. J Sci Food
Agric. 2012;92(3):557–63. [PubMed]
27. Rodríguez-Ambriz SL, Islas-Hernández JJ, Agama-Acevedo E, Tovar J, Bello-
Pérez LA. Characterization of a fibre-rich powder prepared by liquefaction of unripe
banana flour. Food Chem. 2008;107(4):1515–21.
28. Juarez-Garcia E, Agama-Acevedo E, Sáyago-Ayerdi SG, Rodrìguez-Ambriz SL,
Bello-Pėrez LA. Composition, digestibility and application in bread making of banana
flour. Plant Foods Hum Nutr. 2006;61(3):131–7. [PubMed]
29. Grigelmo-Miguel N, Martín-Belloso O. Comparison of dietary fibre from by-
products of processing fruits and greens and from cereals. LWT-Food Sci Technol.
1999;32:503–8.
30. Calvache JEN, Fissore EN, Latorre ME, Soria M, Pla MFDE, Gerschenson LN.
Obtention of dietary fibre enriched fractions from peach bagasse using ethanol pre-
treatment and microwave drying. LWT-Food Sci Technol. 2015;62(2):1169–76.
31. Thipayarat A. Quality and physiochemical properties of banana paste under
vacuum dehydration. Int J Food Eng. 2007;3(4):1556–3758.
32. Uttara B, Singh AV, Zamboni P, Mahajan RT. Oxidative stress and
neurodegenerative diseases: a review of upstream and downstream antioxidant
therapeutic options. Curr Neuropharmacol. 2009;7:65. [PMC free article] [PubMed]
33. Pérez-Jiménez J, Arranz S, Tabernero M, Díaz-Rubio ME, Serrano J, Goñi I, et
al. Updated methodology to determine antioxidant capacity in plant foods, oils and
beverages: extraction, measurement and expression of results. Food Res Int.
2008;41(3):274–85.
34. Goñi I, Díaz-Rubio ME, Pérez-Jiménez J, Saura-Calixto F. Towards an updated
methodology for measurement of dietary fibre, including associated polyphenols, in
food and beverages. Food Res Int. 2009;42(7):840–6.
35. Ahmed J, Almusallam A, Al-Hooti SN. Isolation and characterization of insoluble
date (Phoenix dactylifera L.) fibres. LWT-Food Sci Technol. 2013;50(2):414–19.
36. Alfredo VO, Gabriel RR, Luis CG, David BA. Physicochemical properties of a
fibrous fraction from chia (Salvia hispanica L.) LWT-Food Sci Technol.
2009;42(1):168–73.
37. Chau C, Huang Y. Comparison of the chemical composition and
physicochemical properties of different fibres prepared from the peel of Citrus
sinensis L. Cv. Liucheng. J Agric Food Chem. 2003;51(9):2615–18. [PubMed]
38. Ruales J, Zumba J. Cuantificación y caracterización de fibra dietética en frutas
y hortalizas ecuatorianas. Temas en Tecnología de Alimentos. 1998;2:55–9. (in
Spanish)
39. Yangılar F. Effects of green banana flour on the physical, chemical and sensory
properties of ice cream. Food Technol Biotechnol. 2015;53(3):315–23.
40. Hwang JY, Shyu YS, Hsu CK. Grape wine lees improves the rheological and
adds antioxidant properties to ice cream. LWT-Food Sci Technol. 2009;42(1):312–
18.
41. Çakmakçı S, Topdaş EF, Kalın P, Han H, Şekerci P, Köse LP, et al. Antioxidant
capacity and functionality of oleaster (Elaeagnus angustifolia L.) flour and crust in a
new kind of fruity ice cream. Int J Food Sci Technol. 2015;50(2):472–81.
42. Yangılar F. Mineral contents and physical, chemical, sensory properties of ice
cream enriched with date fibre. Italian J Food Sci. 2015;27(3):397–406.
43. Sofjan RP, Hartel RW. Effects of overrun on structural and physical
characteristics of ice cream. Int Dairy J. 2004;14(3):255–62.
44. El-Samahy SK, Youssef KM, Moussa-Ayoub TE. Producing ice cream with
concentrated cactus pear pulp: a preliminary study. J Prof Assoc Cactus. 2009;11:1–
12.
45. Sun-Waterhouse D, Edmonds L, Wadhwa SS, Wibisono R. Producing ice cream
using a substantial amount of juice from kiwifruit with green, gold or red flesh. Food
Res Int. 2013;50:647–56.
46. Başar H. Elemental composition of various peach cultivars. Sci Hort.
2006;107(3):259–63.
47. Dağdemir E. Effect of vegetable marrow (Cucurbita pepo L.) on ice cream quality
and nutritive value. Asian J Chem. 2011;23(10):4684–8.
48. Wu SJ, Ng LT, Huang YM, Lin DL, Wang SS, Huang SN, et al. Antioxidant
activities of Physalis peruviana. Biol Pharm Bull. 2005;28(6):963–6. [PubMed]
49. Puente LA, Pinto-Muñoz CA, Castro ES, Cortés M. Physalis peruviana Linnaeus,
the multiple properties of a highly functional fruit: a review. Food Res Int.
2011;44(7):1733–40.
50. Erkaya T, Dağdemir E, Sengül M. Influence of cape gooseberry (Physalis
peruviana L.) addition on the chemical and sensory characteristics and mineral
concentrations of ice cream. Food Res Int. 2012;45(1):331–5.

III. Salt reduction in vegetable soup does not affect

saltiness intensity and liking in the elderly and
children
Food Nutr Res. 2014; 58: 10.3402/fnr.v58.24825.
Carla Gonçalves,1,* Sérgio Monteiro,1 Patrícia Padrão,1,2 Ada Rocha,1,3 Sandra
Abreu,4 Olívia Pinho,1,3 and Pedro Moreira1,4

Abstract

Study background

Reduction of added salt levels in soups is recommended. We evaluated the impact

of a 30% reduction of usual added salt in vegetable soups on elderly and children's
saltiness and liking evaluation.

Methods

Subjects were elderly and recruited from two public nursing homes (29 older adults,
79.7±8.9 years), and preschool children recruited from a public preschool (49
children, 4.5±1.3 years). This study took place in institutional lunchrooms. Through
randomization and crossover, the subjects participated in two sensory evaluation
sessions, on consecutive days, to assess perceived saltiness intensity (elderly
sample) and liking (elderly and children samples) of a vegetable soup with baseline
salt content and with a 30% salt reduction. Elderly rated perceived liking through a
10 cm visual analogue scale [‘like extremely’ (1) to ‘dislike extremely’ (10)] and
children through a five-point facial scale [‘dislike very much’ (1) to ‘like very much’
(5)].

Results

After 30% added salt reduction in vegetable soup, there were no significant
differences in saltiness noted by the elderly (p=0.150), and in perceived liking by
children (p=0.160) and elderly (p=0.860).

Conclusions

A 30% salt reduction in vegetable soup may be achieved without compromising

perceived saltiness and liking in children and the elderly.

Keywords: elderly, preschool children, salt, hedonic evaluation, soup

High sodium intake has been associated with the etiology and pathogenesis of
hypertension (1) and cardiovascular diseases (CVD), which is the leading cause of
mortality and morbidity worldwide (2).

The World Health Organization (WHO) recommends the consumption of less than
2,000 mg sodium per day for adults and the adjustment of 2,000 mg sodium
downward based on the energy requirements of children (2). In this global scenario,
the reduction of population sodium intake is one of the most urgent strategies to put
into practice (3). However, in Portugal, the intake levels are well above these
recommendations, reaching an average of 4,800 mg per day in adults (4), and
children and the elderly are two important population subgroups to target for lower
sodium intake (5).

Since most sodium is consumed in the form of sodium chloride, which is table salt
(approximately 40% sodium), the European Salt Framework established a
benchmark of a minimum of 16% salt reduction over 4 years for all food products,
also encompassing salt consumed in restaurants and catering, including vegetable
soup (6). In Portugal, the amount of salt in a 100 g portion of vegetable soup may
reach 1,073 mg in nursery homes, and 1,098 mg in kindergarten, and one of the key
approaches to decrease sodium intake is to target added salt in soup preparation
(7). In order to achieve this goal, a reduction benchmark of 16% against the
individual baseline food levels was established by the European Commission (8),
but strong controversy exists regarding the choice of cutback levels in each food
group. In France (3), United Kingdom (3), United States (9), and Czech Republic (6),
the targets for sodium reduction in soups are 7, 15, 8, and 50%, respectively.
Therefore, the establishment of a sodium reduction benchmark for vegetable soup
should be worked on a country level with highest priority in order to accomplish a
significant reduction of sodium intake without compromising sensorial
characteristics, such as liking and saltiness perception, which may compromise food
acceptance (10). It has been reported that the reduction of added sodium to foods
ranging between 10 and 48% of baseline levels may not be detected by taste
receptors (11–15) ; however, the large majority of these studies involves trained
panelists, and commercial soups or soups with salt substitutes. On the other hand,
the new directions to help consumers to eat healthier focus on training in food
preparation practice, including in the catering sector, to reduce salt usage in their
kitchens. In this scenario, different benchmarks of salt reduction should be tested
across the life cycle, in order to significantly reduce sodium intake without affecting
consumer acceptability (10, 11, 13, 15, 16).

Thus, the objective of this study was to assess the perceived saltiness intensity and
liking of a vegetable soup after 30% reduction of the usual sodium content, in a
sample of institutionalized elderly and community preschool children.

Materials and methods

Sampling

Elderly

Elderly subjects (n=35) who attended from two public nursing homes (NH1 and NH2)
were invited to participate in the study, and interviewed through a structured
questionnaire about general health conditions. Subjects with renal diseases or
hyponatremia were excluded (n=6), resulting in 29 elderly subjects remaining.
Cognitive status was evaluated using the mini-mental state examination (MMSE)
(17), and all had MMSE ranked >18. The elderly provided written informed consent.

Children

All children from a public preschool were also invited to participate (n=75). Letters
were distributed to all parents outlining the aims of the study along with a consent
form. Forty-nine parents signed and returned the filled-out form, and all of these
children gave verbal consent, therefore the sample consisted of 49 children.

Study design

Each set of participants – elderly and children – was randomly divided into one of
two groups to perform the saltiness and liking evaluation of vegetable soups in two
sessions on separated days.

Each group tasted and evaluated the soups following a crossover design, in which
the subjects were randomly assigned to two different arms of the study, one
consuming a vegetable soup with baseline sodium content and the other having the
soup with 30% salt reduction, switching the soups with the two different levels of
sodium content afterwards. Other ingredients in the soups besides salt were the
same in both groups. The study was also single blinded such that the subjects were
unaware of the salt content of the soup they were assessing.

The sensory evaluation took place in the usual lunchrooms at the institutions, in
order to minimize the effects due to changes in the physical environment, and at the
typical lunch period – between 12:00 and 13:00 am – which is also the time when
participants are most alert to perform sensory testing. Furthermore, the researchers
visited the institutions several times before the test so that participants became
familiar with them.

Salt composition of vegetable soups

According to our previous experiences (7), no specific amount of added salt is used
for different kinds of vegetable soups, and the added quantities widely vary even for
a soup with the same recipe. Hence, for the estimation of the baseline sodium
content of soup, we computed the mean sodium content of the consecutive days
prior to the trial (7 consecutive days in nursing home and 5 days in kindergarten),
and estimated the added salt reduction of 30% in order to obtain a soup in the trial.

To estimate baseline sodium content, four samples of soup were collected, two
samples before adding salt in confection by a food handler and two samples after
added salt, in each day. Baseline added salt was calculated subtracting average
sodium before adding salt (sodium from vegetables) to average sodium after added
salt, and the average sodium content of the 7 days/5 days was considered as the
baseline.

Sodium content was determined using flame photometry method, the intern
reference method to analyze sodium in food matrices (18). Samples preparation
procedure was adapted to soups from one validated method proposed to quantify
sodium content in bread (19). All soup samples were stored in plastic containers at
4°C until analysis. After homogenization of each soup (Robot 300 IX, Taurus, Oliana,
Spain), 2 g was sampled and 2 ml of nitric acid was added. The mixture was shaken
during 90 min to allow the food matrix's to complete hydrolysis. Then, 20 ml of water
was added, and the mixture was again homogenized using an electric homogenizer
(Ultra Turrax blender T25, Sotel, Staufen, Germany). Volume was completed up to
40 ml and shaken for 30 min, followed by centrifugation (4,000 rpm, 15 min;
Labofuge 6000® Haerus model, Burladingen, Germany). Finally, 1 ml of aqueous
supernatant was diluted up to 40 ml of deionized water before reading in the flame
photometer (Model PFP7, JenWay, Staffordshire, England).

For sensory evaluation, the vegetable soups were presented in a white ceramic plate
at 68–72°C. Soups were served with nectar consistency and stored in the first day
for the second refrigerated at 4°C, without added salt. Before the test soups were
heated, soups were divided in two groups: in one group, the amount of added salt
corresponded to the usual amount in baseline soups, and in the other group of
soups, the amount of added salt was reduced by 30% compared to baseline. For the
vegetable soups prepared for the elderly, a traditional recipe was selected,
consisting of 1.5 kg of potatoes, 2.0 kg of white cabbage, 1.5 kg of Portuguese
cabbage, 0.5 kg of onions, and 1.0 kg of carrots. For preschool children, the usual
recipe for the soup was prepared with 7.0 kg of potatoes, 1.3 kg of onions, 2.3 kg of
carrots, 2.5 kg of leek, and 30 g of garlic.

Sensory evaluation

A sensory description of the two soups with the two different sodium contents was
performed; elderly evaluated saltiness and liking using a visual analogue scale
(VAS) with 10 cm line scale [from ‘extremely’ (1) to ‘not at all’ (10) for salt perception
and ‘like extremely’ (1) to ‘dislike extremely’ (10) for hedonic perception], and
children evaluated liking through perceived liking ranking using a five-point facial
scale (FS), by means of smiles icons that express feelings from ‘dislike very much’
(1) to ‘like very much’ (5). Given the low educational attainment in the elderly, a prior
explanation of how to fulfill the VAS was given by a trained researcher.

Sociodemographic and anthropometric data

A questionnaire was applied to elderly and to the children's parents in order to collect
data on age, and the highest education qualification completed. Weight and height
measurements were performed in the elderly using standardized procedures (20); in
the case of children, these anthropometric measures were reported by parents. Body
mass index (BMI) was calculated as weight (kg) divided by the square of height (m).
Elderly were classified as overweight if their BMI exceeded 25 kg/m2 and obese if
their BMI exceeded 30 kg/m2, and children were classified as at risk of overweight if
BMI at or above the 85th percentile and lower than the 95th percentile and obese if
BMI at or above the 95th percentile for children of the same age and sex according
to United States Centers for Disease Control and Prevention criteria (21).

Statistical analysis

Mean and standard deviations (SD) were used to describe continuous variables with
normal distribution; otherwise, medians, minimum and maximum were presented.

Data analysis was performed with SPSS (version 17, Chicago, IL). The Shapiro–
Wilk test was used to assess the assumption of normality, as VAS and FS variables
did not reach normality, its median scores were compared using a non-parametric
test for paired data, the Wilcoxon signed rank test. Spearman correlation coefficient
was used to measure the degree of the association between pairs of variables. A p
value of <0.05 was regarded as significant.

The impact of reduced added salt level on sensory evaluation is described

separately for children and elderly.
Results

Elderly subjects (n=29, 20 females) were 79.7±8.9 years old with BMI of 25.7±3.9
kg/m2 (41.4% overweight and 13.8% obese); 62.1% had basic education (1–4 years)
and 34.5% could not read or write. Children (n=49, 26 girls) were 4.5±1.3 years old
with a BMI of 16.0±1.5 kg/m2 (16.3% overweight and 12.2% obese). Almost half
(49%) of the children's parents reported to have attained a high education level;
however, 18.3% had nine or less schooling years.

The baseline sodium concentrations in soups from each institution are shown in
Table 1. The values of added sodium to soups (mg of sodium/100 g of soup) were
300.7±6.6 mg in NH1, 206.7±7.3 mg in NH2, and 147.0±8.5 mg in preschool.

Table 1. Sodium content in vegetable soups (mg sodium/100 g soup)a

Salt added during the

Before adding salt After adding salt
cooking process

Mean± Me Mi Mean±S Mean±S

Max Med Min Max Med Min Max
SD d n D D

Sou
p 212.4±4. 207.9±4.
4.5±0.1
day 6 5
1

Sou
p 315.2±2 309.7±2
5.5±1.6
day 3.7 2.1
2

NH Sou
1 p 290.8±1 289.5±1
1.3±1.2
day 2.2 1
3

Sou
p 308.3±1 302.8±9.
5.5±1.5
day 1.4 9
4

Sou 17.2±2. 334.4±1. 317.2±0.

p 6 9 7
 Salt added during the
Before adding salt After adding salt
cooking process

Mean± Me Mi Mean±S Mean±S

Max Med Min Max Med Min Max
SD d n D D

day
5

Sou
p 267.8±1 260.4±8.
7.4±4.2
day 2.3 1
6

Sou
p 420.0±1 417.3±1
2.7±0.8
day 2.3 1.5
7

Mea
n of
the 7 307.0±1 308. 212. 420. 300.7±6. 302. 207. 417.
6.3±4.6 4.6 1.3 17.2
days 1.2 3 4 0 6b 8 9 3
soup
s

Sou
p 358.1±1 355.7±9.
2.4±0.6
day 0.2 6
1

Sou
NH p 13.5±2. 185.4±4. 171.9±1.
2 day 9 1 2
2

Sou
p 33.9±2. 166.4±1 132.5±8.
day 0 0.2 2
3
 Salt added during the
Before adding salt After adding salt
cooking process

Mean± Me Mi Mean±S Mean±S

Max Med Min Max Med Min Max
SD d n D D

Sou
p 18.9±0. 175.3±1 156.4±1
day 2 4.2 4.0
4

Sou
p 21.4±1. 204.9±5. 183.5±3.
day 3 0 7
5

Sou
p 130.1±1 300.9±6. 170.9±5.
day .0 5 5
6

Sou
p 226.9±9. 226.3±8.
0.6±0.4
day 1 7
7

Mea
n of
31.5±6. 15. 130. 231.1±1 212. 166. 300. 206.7±7. 196. 132. 355.
the 7 0.6
2 7 1 3.5 5 4 9 3b 8 5 7
days
soup

Sou
p 164.6±3. 161.6±2.
3.0±1.1
day 5 4
1
PS
Sou
p 107.6±2. 104.6±2.
3.0±0.5
day 9 5
2
 Salt added during the
Before adding salt After adding salt
cooking process

Mean± Me Mi Mean±S Mean±S

Max Med Min Max Med Min Max
SD d n D D

Sou
p 110.2±8. 107.2±8.
3.0±0.5
day 4 0
3

Sou
p 18.5±0.
67.2±5.9 48.7±5.0
day 9
4

Sou
p 18.5±0. 318.0±1 299.5±1
day 9 9.5 8.6
5

Mea
n of
153.5±8. 131. 318. 144.3±7. 125. 299.
the 5 9.2±0.7 6.1 3.0 18.5 67.2 b 48.7
0 9 0 3 8 5
days
soup
aSodium obtained using flame photometry methodology; vegetable soups included
four vegetables and potatoes.
bValues used to perform a 30% salt reduction.

SD=standard deviation; Med=median; Min=minimum; Max=maximum; NH1=

Nursing Home 1; NH2= Nursing Home 2; and PS= preschool.

The sensory evaluation of the two soups (one with baseline salt content and another
with 30% added salt reduction) is presented in Table 2. There were no significant
differences in saltiness and liking between the two soups. Furthermore, no significant
correlation was found between saltiness and liking in the elderly (ρ=−0.714,
p=0.475).

Table 2. Evaluation tests according to the salt content in the soup (hedonic
and perceived salt ratings)
 Vegetable soups with
Vegetable soups with
30% reduction of added
usual added salt
salt

Hedonic Median Median

Mean±SD Mean±SD p
perception (Min; Max) (Min; Max)

Elderly 5.0 (3.4;

Saltiness 5.7±3.2 6.9±2.5 7.8 (4.9; 9.3) 0.150
(n=29) 9.1)

0.8 (0.2;
Liking 2.1±2.8 2.2±2.9 0.7 (0.3; 4.0) 0.860
2.7)

Preschool
4.0 (4.0;
children Liking 4.4±0.5 4.41±0.54 4.0 (4.0; 5.0) 0.160
5.0)
(n=49)

SD=standard deviation; Min=minimum; Max=maximum.

Discussion

The overweight and obesity prevalence values found in the elderly and children
samples were similar to values reported in the same age groups in Portugal (22, 23).

A 30% added salt reduction did not change the saltiness and liking perception of a
vegetable soup in both preschool children and elderly subjects. These results
encourage for achieving higher sodium reductions than those typically advised,
namely with decreases between 4 and 10% per year (9), without affecting hedonic
response, particularly among children and the elderly. Nevertheless, these results
should be considered with caution, and this reduction ought to be tested in soups
with other vegetables, since optimum added sodium levels to increase saltiness or
sweetness, or decrease bitterness, (24) may vary according to vegetable
ingredients. Previous studies (10, 14, 25, 26) addressed sodium reductions and the
combinations of extra aroma and salt replacers to substitute sodium (11) in adults
although more research is needed to understand and control the human liking for
salt (27), particularly among children and the elderly. Malherbe et al. found similar
results in adults and conclude that it is possible to reduce the sodium content in soup
by about 30% without significantly changing acceptability and pointed to the masking
of other ingredients, partial adaptation, as well as the decreased absolute sensitivity
and increased differential sensitivity as possible reasons (13).

Strong concerns about the consumption of salt in vegetable soup exist in Portugal.
The soups analyzed to establish the baseline (7 days in institutionalized elderly and
5 days for preschool children) take part in the usual menu cycle in both institutions,
and regardless of the recipe, they are usually composed of four vegetables plus
potatoes. Thus, the soup used in the trial symbolizes a general traditional vegetable
composition of any of the soups analyzed in the baseline, with no expected
significant variations in the nutritional composition, other than added sodium. The
latter may vary according to the food handlers/cookers since no standardized
amount of added sodium is established for each recipe.

Accordingly, in our study, salt added in soups during the cooking process (Table 1)
varies widely between different collection days and between institutions, probably
because the sodium added to soups varied according to intrapersonal and
interpersonal food preparation practices without standardization (7). Moreover, the
levels of salt found in the set of days used as ‘baseline’ are similar, or even lower
than others reported in the literature (7, 28). For that reason, it is not expected that
soup used in the trial was ‘over-salted’.

According to a study that evaluated the sodium content of Portuguese vegetable

soups in NH, elementary schools and kindergartens, subjects eating two portions of
soup per day may reach 67% of the upper limit of sodium intake, when considering
the mean values of sodium in vegetable soups (7) and the values found in this study
were similar to ours. Given the low content of sodium in non-processed vegetables,
nutrition education approaches to target sodium reduction should focus on training
in cooking practice to significantly decrease added salt (according to the present
study, about one third reduction in relation to usual levels may be acceptable) in
vegetable soups produced both at home and in catering industries. A major
challenge is to ensure that salt reductions do not exceed consumer expectations
from a hedonic perspective which is a difficult task considering that bliss point
depends on the individual and on his hedonic relation with a particular food (29), with
some products largely punished on their hedonic evaluation even with small sodium
reductions (9). However, sensitivity, perception of the intensity, preference and
hedonic response to salt are independent measures; thus even individuals who are
able to detect salt reduction may still maintain high ranking hedonic responses (30).
Our results suggested that in vegetable soup, it is possible to moderately decrease
the added salt without affecting both the consumer salty perception and the hedonic
value attributed.

Salt is used in the food industry and during the cooking process due to its capacity
to improve sensory properties of food (31), and each food preparation process or
technology of processing has its own specific challenges as salt has multiple
functions. In the case of soup preparation, adding salt may be a very cheap way to
positively influence the taste of soup. For individuals who are used to tasting high
levels of salt, its sudden reduction may cause the rejection of food (9, 32). For this
reason, a small reduction of added salt, which is not detected by salty taste
receptors, might be a valuable strategy to reduce sodium intake, and this study adds
30% as a possible benchmark for reduction.

Few studies have assessed the impact of a reduction of the sodium content of
vegetable soup on hedonic perception of consumers in different age groups.
Drewnowski et al. (33) showed that older adults prefer less salty soups than young
adults; these results are also supported by Kremer et al. (34). However, it is largely
assumed that people aged over 70 may have suffered changes in the function of ion
channels and receptors of taste buds, with a consequent decrease of the threshold
for detection and identification of flavor (35). Accordingly, Murphy and Withee have
shown that the elderly have difficulty in detecting sweet and salty tastes (36), leading
to an increased preference for salt (37), which could result in a higher sodium intake
with the associated adverse health consequences. However, our findings showed
that the elderly and children did not differently evaluate the soups with different
sodium contents, which may be a good indicator for the process of reduction of
added salt in vegetable soups in these age groups, leaving soup as a perfect vehicle
of micronutrients. Vegetable soup is part of the food culture in southern Europe (38),
being a culinary preparation with a suitable consistency for anyone with weak
dentition (common in the elderly and children populations), with easy digestibility,
high content of vitamins, minerals and fiber, that may also contribute to lower risk of
overweight and obesity (39, 40). The nutritional and metabolic advantages of
vegetable soup intake may further include hydration and anti-inflammatory
properties (41), being strongly advised to consume it twice a day, in order to
counteract the low intake of vegetables in Portuguese adults (42) and children (43)
and prevent non-communicable diseases (44).

The results of the present study should be interpreted while taking into account its
strengths and limitations. The study was carried out in the habitual familiar context
for participants and during the period that they usually take their meals, leaving these
variables constant over the intervention. Another strength of the study was the use
of a single-blinded crossover model. On the other hand, participants were
institutionalized; that is, the cooking process was the responsibility of the institution.
Thus, the results should not be speculated to other population groups, neither the
food used to taste stimuli; the perception of sodium reduction in vegetable soup
should not be generalized to other foods because the effect of salt appears to be
food specific. Another limitation of this study is that salt reduction was carried out a
particular kind of vegetable soup recipe; however, we can achieve other values of
reduction of salt content in other types of vegetable soup with other ingredients
without affecting saltiness and hedonic perception.

For that reason, in the future, it will be important to consider other vegetable
ingredients and recipes, and other values for the reduction of added salt in order to
formulate targeted measures to control sodium intake, and set realistic and culturally
relevant goals in food policy.

In conclusion, a 30% salt reduction in soups can be achieved without affecting

perceived saltiness intensity and liking among the elderly and children, considering
the reported mean baseline content of sodium.

Conflict of interest and funding

The authors declare no conflict of interest.

References

1. Meneton P, Jeunemaitre X, De Wardener HE, Macgregor GA. Links between

dietary salt intake, renal salt handling, blood pressure, and cardiovascular diseases.
Physiol Rev. 2005;85:679–15. [PubMed]
2. World Health Organization. Geneva: World Health Organization; 2012. Guideline:
sodium intake for adults and children.
3. World Health Organization. Paris: World Health Organization; 2006. Reducing salt
intake in populations: report of a WHO forum and technical meeting 5–7 October
2006.
4. Polónia J, Maldonado J, Ramos R, Bertoquini S, Duro G, Almeida C, et al.
Determinação do consumo de sal numa amostra da população portuguesa adulta
pela excreção urinária de sódio. Sua relação com a rigidez arterial. Rev Port Cardiol.
2006;25:801–17. [PubMed]
5. Institute of Medicine. Washington, DC: The National Academies Press; 2013.
Sodium intake in populations: assessment of evidence.
6. European Commission. Brussels, Belgium: European Commission; 2012.
Implementation of the EU salt reduction framework – results of member states
survey.
7. Gonçalves C, Silva G, Pinho O, Camelo S, Amaro L, Teixeira V, et al. Sodium
content in vegetable soups prepared outside the home: identifying the problem. In:
Arezes P, Baptista J, Barroso M, Carneiro P, Cordeiro P, Costa N, et al., editors.
Occupational safety and hygiene – sho 2012 – book of abstracts. Vila Nova de Gaia:
Portuguese Society of Occupational Safety and Hygiene (SPOSHO); 2012. pp. 213–
5.
8. European Commission. Brussels, Belgium: European Commission; 2009.
National salt initiatives implementing the EU framework for salt reduction initiatives.
9. Institute of Medicine (US) Washington, DC: The National Academies Press; 2010.
Strategies to reduce sodium intake in the United States.
10. Mitchell M, Brunton NP, Wilkinson MG. The influence of salt taste threshold on
acceptability and purchase intent of reformulated reduced sodium vegetable soups.
Food Qual Prefer. 2013;28:356–60.
11. Batenburg M, Velden R. Saltiness enhancement by savory aroma compounds.
J Food Sci. 2011;76:S280–88. [PubMed]
12. Feng J, MacGregor G. Salt in food. Lancet. 2005;365:844–5. [PubMed]
13. Malherbe M, Walsh CM, Van der Merwe CA. Consumer acceptability and salt
perception of food with a reduced sodium content. J Fam Ecol Consum Sci.
2003;31:12–20.
14. Mitchell M, Brunton NP, Wilkinson MG. Impact of salt reduction on the
instrumental and sensory flavor profile of vegetable soup. Food Res Int.
2011;44:1036–43.
15. Adams SO, Maller O, Cardello A. Consumer acceptance of foods lower in
sodium. J Am Diet Assoc. 1995;95:447–53. [PubMed]
16. Beauchamp G, Bertino M, Engelman K. Failure to compensate decreased
dietary-sodium with increased table salt usage. JAMA. 1987;258:3275–8. [PubMed]
17. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state.” A practical method
for grading the cognitive state of patients for the clinician. J Psychiatr Res.
1975;12:189–98. [PubMed]
18. Ferreira I, Lima J, Rangel A. Flow injection sequential determination of chloride
by potentiometry and sodium by flame emission spectrometry in instant soups. Anal
Sci. 1994;10:801–5.
19. Vieira E, Soares M, Ferreira I, Pinho O. Validation of a fast sample preparation
procedure for quantification of sodium in bread by flame photometry. Food Anal
Methods. 2012;5:430–4.
20. World Health Organization. Report of a WHO expert committee. Geneva: World
Health Organization; 1995. Physical status: the use and interpretation of
anthropometry. 1995/01/01 ed. [PubMed]
21. Kuczmarski RJ, Ogden CL, Guo SS, Grummer-Strawn LM, Flegal KM, Mei Z, et
al. 2000 CDC growth charts for the United States: methods and development. Vital
Health Stat 11. 2002;246:1–190. [PubMed]
22. Antunes A, Moreira P. [Prevalence of overweight and obesity in Portuguese
children and adolescents] Acta Med Port. 2011;24:279–84. [PubMed]
23. Wanderley FA, Moreira A, Sokhatska O, Palmares C, Moreira P, Sandercock G,
et al. Differential responses of adiposity, inflammation and autonomic function to
aerobic versus resistance training in older adults. Exp Gerontol. 2013;48:326–33.
[PubMed]
24. Keast RSJ, Breslin PAS. An overview of binary taste–taste interactions. Food
Qual Prefer. 2003;14:111–24.
25. Liem DG, Miremadi F, Zandstra EH, Keast RS. Health labelling can influence
taste perception and use of table salt for reduced-sodium products. Public Health
Nutr. 2012;15:2340–47. [PubMed]
26. Methven L, Langreney E, Prescott J. Changes in liking for a no added salt soup
as a function of exposure. Food Qual Prefer. 2012;26:135–40.
27. Leshem M. Biobehavior of the human love of salt. Neurosci Biobehav Rev.
2009;33:1–17. [PubMed]
28. Viegas C, Graça P, Martins M, Torgal J. Lisboa: Universidade de Lisboa; 2014.
Salt in the youngsters diet: evaluation and perception of salt in meals.
29. McBride RL. The bliss point as a measure of pleasure. Pleasure, the politics and
the reality. In: Warburton DM, editor. New York: John Wiley & Sons; 1994. pp. 67–
76.
30. Pangborn RM, Pecore SD. Taste perception of sodium chloride in relation to
dietary intake of salt. Am J Clin Nutr. 1982;35:510–20. [PubMed]
31. Gillette M. Flavor effects of sodium-chloride. Food Technol. 1985;39:47–56.
32. Mattes RD. The taste for salt in humans. Am J Clin Nutr. 1997;65(Suppl
69):692S–7S. [PubMed]
33. Drewnowski A, Henderson S, Driscoll A, Rolls B. Salt taste perceptions and
preferences are unrelated to sodium consumption in healthy older adults. J Am Diet
Assoc. 1996;96:471–4. [PubMed]
34. Kremer S, Bult JHF, Monet J, Kroeze JHA. Food perception with age and its
relationship to pleasantness. Chem Senses. 2007;35:591–602. [PubMed]
35. Mistretta CM. Aging effects on anatomy and neurophysiology of taste and smell.
Gerodontology. 1984;3:131–6. [PubMed]
36. Murphy C, Withee J. Age-related differences in the pleasantness of
chemosensory stimuli. Psychol Aging. 1986;1:312–18. [PubMed]
37. Campos M, Monteiro J, Ornelas A. Factores que afectam o consumo alimentar
e a nutrição do idoso. Rev Nutr. 2000;13:157–65.
38. Peres E. Sopa: Comida de pobre ou de filósofo? Há tanta ideia perdida.
2002;2:3.
39. Moreira P, Padrão P. Educational, economic and dietary determinants of obesity
in Portuguese adults: a cross-sectional study. Eat Behav. 2006;7:220–8. [PubMed]
40. Sanchez-Moreno C, Pilar Cano M, De Ancos B, Plaza L, Olmedilla B, Granado
F, et al. Intake of Mediterranean vegetable soup treated by pulsed electric fields
affects plasma vitamin c and antioxidant biomarkers in humans. Int J Food Sci Nutr.
2005;56:115–24. [PubMed]
41. Sanchez-Moreno C, Cano MP, de Ancos B, Plaza L, Olmedilla B, Granado F, et
al. Consumption of high-pressurized vegetable soup increases plasma vitamin c and
decreases oxidative stress and inflammatory biomarkers in healthy humans. J Nutr.
2004;134:3021–25. [PubMed]
42. Instituto Nacional de Estatística. Lisboa: Instituto Nacional de Estatística (INE);
2010. Balança alimentar portuguesa 2003–2008.
43. Yngve A, Wolf A, Poortvliet E, Elmadfa I, Brug J, Ehrenblad B, et al. Fruit and
vegetable intake in a sample of 11-year-old children in 9 European countries: the pro
children cross-sectional survey. Ann Nutr Metab. 2005;49:236–45. [PubMed]
44. World Health Organization. Geneva: World Health Organization; 2002. Diet,
nutrition and the prevention of chronic diseases.
IV. Aroma Profile of Montepulciano d'Abruzzo Wine
Fermented by Single and Co-culture Starters of
Autochthonous Saccharomyces and Non-
saccharomyces Yeasts
Front Microbiol. 2016; 7: 610.
Rosanna Tofalo,1,* Francesca Patrignani,2 Rosalba Lanciotti,2 Giorgia Perpetuini,1
Maria Schirone,1 Paola Di Gianvito,1 Daniel Pizzoni,1 Giuseppe Arfelli,1 and
Giovanna Suzzi1

Abstract

Montepulciano d'Abruzzo is a native grape variety of Vitis vinifera L., grown in central
Italy and used for production of high quality red wines. Limited studies have been
carried out to improve its enological characteristics through the use of indigenous
strains of Saccharomyces cerevisiae. The main objective of the present work was to
test two indigenous strains of S. cerevisiae (SRS1, RT73), a strain of Starmerella
bacillaris (STS12), one of Hanseniaspora uvarum (STS45) and a co-culture of S.
cerevisiae (SRS1) and S. bacillaris (STS12), in an experimental cellar to evaluate
their role in the sensory characteristic of Montepulciano d'Abruzzo wine. A S.
cerevisiae commercial strain was used. Fermentations were conducted under
routine Montepulciano d'Abruzzo wine production, in which the main variables were
the yeast strains used for fermentation. Basic winemaking parameters, some key
chemical analysis and aroma compounds were considered. S. cerevisiae strain
dynamics during fermentation were determined by molecular methods. The musts
inoculated with the co-culture were characterized by a faster fermentation start and
a higher content of glycerol after 3 days of fermentation, as well as the musts added
with strains S. bacillaris (STS12) and H. uvarum (STS45). At the end of fermentation
the parameters studied were quite similar in all the wines. Total biogenic amines
(BA) content of all the wines was low. Ethanolamine was the predominant BA, with
a concentration ranging from 21 to 24 mg/l. Wines were characterized by esters and
alcohols. In particular, 2-phenylethanol, 3-methylbut-1-yl methanoate, and ethyl
ethanoate were the major aroma volatile compounds in all wines. Statistical analysis
highlighted the different role played by aroma compounds in the differentiation of
wines, even if it was impossible to select a single class of compounds as the most
important for a specific yeast. The present study represents a further step toward the
use of tailored autochthonous strains to impart the specific characteristics of a given
wine which are an expression of a specific terroir.

Keywords: aroma compounds, autochthonous yeast strains, Saccharomyces

cerevisiae, non-Saccharomyces, Montepulciano d'Abruzzo wine

Introduction

Wine fermentations constitute complex microbial ecosystems consisting of highly

dynamic yeast communities which play a key role in shaping wine quality (Fleet,
2003). This complex array of relations influences the nutritional, hygienic, and
aromatic features of the product through the consecutive growth and death of
different species and strains within each species, during the fermentation process
(Fleet, 2003; Liu et al., 2015). Many studies have been focused on the nature of
these relations improving the knowledge about ecology, physiology, biochemistry,
and molecular biology of the microrganisms involved in wine fermentation process
underlying the ecological complexity and variability of these fermentations that
extend beyond the species level (for a review see Liu et al., 2015).

Yeasts mainly impact on the wine flavor producing a large array of volatile
substances (Howell et al., 2006). In this context the existing commercial yeast strains
present some limits, especially because they reduce the uniqueness of wine bouquet
(Alves et al., 2015). In fact, different yeast species and even different genotypes of
Saccharomyces cerevisiae produce different wine aroma profiles (Alves et al., 2015;
Barbosa et al., 2015; Vernocchi et al., 2015). This awareness opened new issues to
meet wine-maker demand for “special yeasts for special traits” (Schuller and Casal,
2005; Sadoudi et al., 2012). Recently, the role of indigenous yeast strains has gained
importance, as a tool to impart regional characters to wines. Indeed, the use of a
“microarea-specific” starter culture highlighted the association between the volatile
profile of wine and the geographical origin of the yeast used for the fermentation
process (Tufariello et al., 2014).

The role of non-Saccharomyces (NS) yeasts in winemaking has been re-evaluated,

leading to a more complex “flavor phenotype” producing more than 1300 volatile
compounds e.g., esters, higher alcohols, acids, and monoterpenes (Swiegers et al.,
2005; for a review see Jolly et al., 2014). Moreira et al. (2005) and Medina et al.
(2013) demonstrated that Hanseniaspora uvarum increased the quantity of some
desirable compounds, such as higher alcohols and esters, while Rantsiou et al.
(2012) showed that inoculation with selected couples of S. cerevisiae and
Starmerella bacillaris resulted in a decrease of about 0.3 g/l of acetic acid,
maintaining high ethanol and glycerol levels.
Montepulciano d'Abruzzo is a red wine grape variety of Vitis vinifera L., grown widely
in central Italy, most notably in Abruzzo, Marche, and Molise regions. However, it is
mainly identified with Abruzzo, the region in which it is also the most common and
cultivated red variety for over two centuries. The first report of the Montepulciano
grape in Abruzzo is found in “Saggio Itinerario Nazionale nel Paese dei Peligni,”
written by Torcia (1972). It currently accounts for around 50% of the regional
vineyard, that is, about 18.500 hectares (Regione Abruzzo,
http://www.regione.abruzzo.it/). Montepulciano d'Abruzzo is used for production of
high quality red wines characterized by fruity notes (apple, pear, cherry, etc.). The
most famous example is Montepulciano d'Abruzzo “Colline Teramane” DOCG wine
(recognition in 2003) produced in the Teramo province.

Despite the economic importance of Montepulciano d'Abruzzo “Colline Teramane”

few studies have been performed to identify its enological characteristics. In a
previous study Tofalo et al. (2011) highlighted that the major NS yeasts present
during must fermentation of Montepulciano cultivar were H. uvarum, Metschnikowia
fructicola, and S. bacillaris, representing 43, 31, and 11%, respectively, of the total
NS population isolated. Selected strains of H. uvarum (STS45), S. bacillaris
(STS12), and S. cerevisiae (SRS1 and RT73) were then studied to evaluate their
fermentation performance and interactions in microvinifications (Suzzi et al.,
2012a,b).

The aim of this study was to establish the role and the inter-strains variability of two
indigenous strains of S. cerevisiae (SRS1, RT73), a strain of S. bacillaris (STS12),
one of H. uvarum (STS45) and a co-culture of S. cerevisiae (SRS1), and S. bacillaris
(STS12) in shaping Montepulciano d'Abruzzo wine aroma profile in an experimental
cellar. A S. cerevisiae commercial strain was used. Vinifications were conducted
under routine Montepulciano d'Abruzzo wine production. Basic winemaking
parameters (residual sugar, glycerol, organic acids, etc.), biogenic amines (BA) and
volatile metabolites were determined. S. cerevisiae strain dynamics were also
determined by microsatellite analysis.

Materials and methods

Yeast strains and media

Non-Saccharomyces (H. uvarum, STS45 and S. bacillaris, STS12) and S. cerevisiae

autochthonous strains (RT73 and SRS1) have been previously characterized for
their oenological performances in Montepulciano d'Abruzzo microvinification trials
(Suzzi et al., 2012a,b). A commercial strain (CS) of S. cerevisiae (Flower Fresh,
Tecnofood, Pavia, Italy) was also used. All strains belong to the Culture Collection
of the Faculty of BioScience and Technology for Food, Agriculture, and Environment
(University of Teramo, Italy). Non-Saccharomyces and S. cerevisiae strains were
routinely grown in YPD medium (1% w/v yeast extract, 2% w/v peptone, and 2% w/v
glucose) for 48 h under aerobic conditions. All strains were stored at −80°C in YPD
broth supplemented with glycerol (20% v/v final concentration; Sigma-Aldrich, Milan,
Italy).
Cellar vinifications

Vinifications were carried out in a cellar of Consorzio per la Ricerca Viticola ed

Enologica in Abruzzo (CRIVEA), during the vintage 2011. Montepulciano d'Abruzzo
must (235 g/l fermentable sugars, 8.17 titratable acidity (TTA) and pH 3.44) was
separated in tanks of 50 l, after destemming and crushing and added with 100 mg/l
potassium metabisulfite.The fermentations were performed in maceration with the
skins. The tanks were inoculated with 106 cells/ml from 24 h pre-cultures grown in
the same pasteurized must. Two S. cerevisiae strains (SRS1, RT73), a strain of S.
bacillaris (STS12), one of H. uvarum (STS45), and a co-culture of SRS1+STS12
were used to conduct fermentations. All fermentations were carried out in triplicate
at room temperature (maximum temperature variation from 9 to 19°C). When the
fermentation ended, the yeast lees were allowed to settle for 7 days and then wines
were racked in 40 l tanks and stored at controlled temperature in the cellar for 3
months. Then the wines were placed into glass bottles (750 ml), crown-sealed, and
stored at 15–20°C for up to 6 months until sensorial analyses were performed.

Enumeration and yeast isolation

Total viable yeast counts were performed after 3, 5, 7, 10, and 15 days, using
Wallerstein Laboratory Nutrient Agar (WLN, Oxoid, Milan, Italy), according to
Pallmann et al. (2001).

Analytical determinations

The main wine analytical components (ethanol, reducing sugar, pH, volatile acidity,
TTA, citric, lactic, malic, and tartaric acids, glycerol) were determined using a FOSS
WineScan (FT-120) rapid scanning Fourier Transform Infrared Spectroscopy with
FOSS WineScan software version 2.2.1. Samples were firstly centrifuged at 8000 g
for 10 min and then analyzed following the manufacturer's instructions.

Microsatellite PCR fingerprinting

Total DNA was extracted directly from musts and wines using the PowerSoil DNA
Isolation Kit (MoBio Laboratories). Ten milliliter of each sample were centrifuged to
collect cells. The DNA was then extracted according to manufacturer's protocol.
Quantification of total DNA was achieved using a VersaFluor fluorimeter and a
Fluorescent DNA Quantitation Kit (Bio-Rad, Milan Italy). DNA was used as a
template for microsatellite PCR fingerprinting, as described by Vaudano and Garcia-
Moruno (2008). PCR amplifications were performed in a thermocycler (MyCycler,
Bio-Rad Laboatories, Milan, Italy) with the following PCR programme: 4 min of initial
denaturation at 94°C, 28 cycles of 30 s at 94°C, 45 s at 56°C, 30 s at 72°C and,
finally, 10 min at 72°C. The products were run on a 2.5% (w/v) agarose gel 1 × TAE
buffer at 100 V for 80 min. Gels were stained with ethidium bromide. 1-kb plus DNA
ladder (Life Technologies, Milan, Italy) was used as marker for the gel normalization.

Volatile profiles
Volatile compounds were determined by solid phase microextraction coupled with
gas chromatography (GC/MS-SPME) according to Suzzi et al. (2012a). Molecule
identification was based on comparison of their retention times with those of pure
compounds (Sigma-Aldrich, Milan, Italy) analyzed in the same conditions. The
identification was further confirmed by comparing mass spectra of compounds with
those contained in the available database (NIST version 2005). The data were
expressed as the relative peak area (%) calculated from head space SPME
(HS/SPME) gas chromatograms of the identified peaks. All determinations were
performed in triplicate.

Biogenic amines determination

Biogenic amines (BA) were determined according to Manetta et al. (2016). BA were
analyzed using an HPLC system consisting of an Alliance (Waters, Milford, MA,
USA), equipped with a Waters 2695 separation module connected to a Waters 2996
photodiode array detector (PDA), set at 254 nm. A Supelcosil LC-18 column (5 μm
particle size, 250 × 4.6 mm i.d.) from Sigma was used. The system was governed
by Waters Empower personal computer software. All analyses were performed in
triplicate.

Sensory analysis

Sensory tests were performed at room temperature (20°C). Wine samples were
coded with 3-digit numbers, were evaluated in triplicate and presented according to
a completely randomized block design. Skilled judges (n = 13) were trained as stated
in the ISO 8586-1: 1993 rules (ISO, 1993).

Descriptive analysis was carried out in only one session. Sensory profile was
determined using nine descriptors (fruity, persistence, body, astringency, grassy,
reduced, floral, tropical fruits, drupaceous fruits) as previously reported (Suzzi et al.,
2012a). Samples were scored for selected descriptors on a 4 cm scale anchored
with “low” and “high” intensity.

Statistical analysis

All data were processed using Excel 2016 (Microsoft, USA) and MatLab 2009b
(Mathworks, Natick, MA, USA) softwares. In particular, a Principal Component
Analysis (PCA) was performed on SPME–GC data after auto-scaling. The volatile
molecule data were used to build up a single matrix, which was submitted to a two-
way hierarchical clustering analysis. A heat map, visualizing metabolite levels was
then obtained in which values are represented by cell colored according to the Z-
scores, where Z is the mean value of different vinifications with the same yeast strain
(Ferrara et al., 2008; Serrazanetti et al., 2011). The significant differences of the
main enological characteristics were determined by F-test.

Results
Viable counts and strain dynamics

In order to improve the quality of Montepulciano d'Abruzzo wine through the use of
autochthonous wine yeasts, Saccharomyces and non-Saccharomyces strains
isolated from the terroir “Colline Teramane” and characterized for their enological
aptitudes (Suzzi et al., 2012a,b) were chosen for experimental cellar vinifications as
reported in Materials and Methods. Six vinifications were carried out, two inoculated
with single S. cerevisiae strains (SRS1 and RT73), two with single NS strains (H.
uvarum STS45 and S. bacillaris STS12) and one with the simultaneous presence of
SRS1 and STS12.

Fermentation trials inoculated with SRS1, RT73, and the co-culture (SRS1+STS12)
started the fermentation quickly (Figure (Figure1),1), reaching higher values of viable
cells after 5 days. At the end of fermentation lower values were observed in must
inoculated with S. bacillaris STS12 and H. uvarum STS45, even if a faster growth
was observed during the first fermentation days.

Figure 1. Growth kinetic profiles of pure and mixed fermentation trials.

To verify the dominance of inoculated strains on natural yeast population in the must,
microsatellite analysis on total DNAs was performed. The S. cerevisiae SRS1, RT73,
and CS were present during the whole fermentation process, confirming a clear
dominance of these S. cerevisiae strains (Figure (Figure2).2). As expected more
complex profiles were detected in Montepulciano d'Abruzzo must inoculated with S.
bacillaris STS12 and H. uvarum STS45 probably due to the presence of different
indigenous S. cerevisiae strains.
Figure 2. Yeast strains electrophoretic patterns of microsatellite multiplex PCR
(SC8132X, YOR267C and SCPTSY7) at the end of fermentation (15 days).
Similar profiles were obtained after 3 days in inoculated fermentations. M: 1-kb plus
DNA ladder (Life Technologies).

Wine characteristics

During the first days of fermentation a higher production of ethanol by NS and mixed
cultures was observed (Table (Table1),1), whereas no differences were registered
at the end of fermentation. In fact in all the six different conditions, must
fermentations were completed according to reducing sugar concentration. The NS
strains formed higher levels of glycerol up to 3.39 g/l after 3 days of fermentation,
whereas the S. cerevisiae strains ranged from 1.38 to 2.20 g/l. The co-culture
produced a wine with an intermediate glycerol content of 2.93 g/l. At the end of
fermentation, the dominance of S. cerevisiae strains (Figures (Figures1,1, ,2)2)
made uniform all the wines with a glycerol content of about 10 g/l and an ethanol
concentration of about of 14% (v/v). Similar behaviors were observed for other
parameters such as volatile acidity, pH, TTA, and organic acids concentration. In all
the samples the consumption of malic acid started before alcoholic fermentation was
completed. This fact could be related to an high number of malolatic bacteria on
grapes, as reported by Renouf et al. (2006), who found Oenococcus oeni and other
lactic acid bacteria at the beginning of alcoholic fermentation. On the other hand,
Nehme et al. (2010) reported simultaneous fermentations by inoculated yeasts and
malolactic acid bacteria. Obtained wines were also analyzed for the presence of BA.
In all wines cadaverine, tryptamine, β- phenylethylamine, tyramine, and histamine
were below the limit of detection for the method used (Manetta et al., 2016). The
levels of ethanolamine, ethylamine, isoamilamine, and putrescine had no significant
changes. Their content was quite similar in all samples with ethanolamine which was
the most abundant amine found ranging from 21 to 24 mg/l (data not shown).

Table 1. Montepulciano d'Abruzzo wine characteristics fermented with

autochthonous S. cerevisiae strains (SRS1 and RT73), commercial strain (CS), co-
culture (SRS1+STS12) and non-Saccharomyces strains (STS12 and STS45) after
3 and 15 days (in parentheses) of fermentation.

Paramete Strains
rs

CS SRS1 RT73 SRS1+STS STS12 STS45

12

Ethanol* 0.36±0.18a 0.41±0.08a 0.91±0.68a 2.44±1.69b 2.77±0.15b 2.05±0.09c

(% v/v) c

(14.40±0.0 (14.26±0.1 (14.24±0.0 (14.21±0.0 (14.24±0.1 (14.10±0.1

4)d 9)d 8)d 8)d 3)d 1)d

Reducing 215±10a 220±14a 197±17a 166±34ab 167±12b 174±12b

sugar (g/l)

(2.07±0.13) (2.18±0.28 (2.06±0.22 (1.81±0.10) (1.96±0.14 (1.95±0.12

c )c )c d )d )d

pH 3.29±0.09a 2.94±0.61a 3.33±0.05a 3.36±0.01a 3.33±0.10a 3.35±0.09a

(3.30±0.02) (3.34±0.04 (3.31±0.04 (3.33±0.06) (3.33±0.07 (3.31±0.10

a )a )a a )a )a
Paramete Strains
rs

CS SRS1 RT73 SRS1+STS STS12 STS45

12

Volatile 0.07±0.04a 0.05±0.02a 0.09±0.02a 0.14±0.06bc 0.14±0.02c 0.16±0.05c

b
acidity**
(g/l)

(0.47±0.06) (0.58±0.15 (0.55±0.12 (0.51±0.08) (0.57±0.09 (0.49±0.09

d )d )d d )d )d

Titratable 6.03±0.30a 5.97±0.33a 5.89±0.14a 6.10±0.19a 6.20±0.21a 6.27±0.22a

acidity***
(g/l)

(7.02±0.19) (6.55±0.26 (6.80±0.18 (6.78±0.38) (7.02±0.27 (7.48±0.13

b )b )b b )b )c

Citric acid 0.15±0.01a 0.06±0.01b 0.14±0.02a 0.20±0.03c 0.19±0.04b 0.27±0.04c

(g/l) c

(0.59±0.09) (0.55±0.02 (0.57±0.07 (0.59±0.03) (0.63±0.04 (0.62±0.05

de )d )d d )e )de

Lactic 0.15±0.08a 0.30±0.05b 0.15±0.02a 0.03±0.01c − 0.16±0.02a

acid

(1.10±0.22) (1.04±0.11 (1.15±0.19 (1.14±0.13) (1.01±0.17 (0.77±0.15

d )d )d d )d )e

Malic acid 1.00±0.10a 1.08±0.21a 1.17±0.06b 1.18±0.09b 1.20±0.08b 1.17±0.12a

(g/l) b b

(0.34±0.21) (0.10±0.04 (0.21±0.09 (0.21±0.08) (0.39±0.11 (0.55±0.12

cde )c )cd cd )d )de

Tartaric 6.88±0.21a 7.22±0.38a 6.78±0.31a 6.59±0.38b 6.79±0.21a 6.30±0.19b

acid (g/l) b b b

(3.0±0.06)c (2.98±0.18 (3.11±0.27 (2.87±0.27) (2.96±0.17 (2.99±0.15

)c )c c )c )c

Glycerol 2.2±1.5ab 1.38±0.19a 1.67±0.65a 2.9±1.2b 3.39±0.27b 3.02±0.24b

(g/l)
Paramete Strains
rs

CS SRS1 RT73 SRS1+STS STS12 STS45

12

(10.96±0.2 (10.12±0.0 (10.0±0.17 (10.16±0.4 (10.22±0.1 (10.71±0.5

8)c 8)d )d 1)d 9)d 4)d
*ml of alcohol/100 ml of wine,
**expressed as acetic acid,
***expressed as tartaric acid.

Data are expressed as average ± SD. Same letters indicate samples in the same
line with non-significant differences (p < 0.05).

Volatile compounds

The volatile metabolites of the Montepulciano d'Abruzzo wines obtained with

autochthonous strains of S. cerevisiae and NS and a mixed culture have been
identified for a total of 101. Aroma compounds belonged to eight different families
such as alcohols, aldehydes, ketones, esters, acids, terpenes, phenols, and
aromatic compounds. The number of metabolites ranged from 63 for the wine
produced by strain SRS1, 53 by STS45, 51 by CS, 49 by RT73, and mixed culture
SRS1+STS12 and 47 by STS12 (Table (Table2).2). Table Table22 shows the main
volatile molecules identified in relation to starter culture used. In the table only the
main components of each aroma profile are reported. Nevertheless their presence
represented at least the 95% of the total area in all the wine. Esters represented the
major group for all the wines, followed by alcohols. The wine obtained with the co-
culture showed the lowest relative percentage of alcohols in the heat space (about
14%), while those produced with S. cerevisiae CS and SRS1 were characterized by
the highest ones, about 24.78 and 28.08%, respectively. In particular 2-
phenylethanol (line 30, rose odor) had the highest relative percentage, ranging from
8.07% (SRS1+STS12) to 21.2% (CS). Differences were observed also for 2-methyl-
1-propanol (line 15) and 1-hexanol (line 28, fruity and erbal odor) prevailing in wines
fermented by NS and co-culture. Regarding esters the relative percentage in the
heat space ranged from 57.42% (CS) to 77.38% (mixed culture), with more
differences on the relative quantities of compounds among strains, as it can be easily
evaluated from Figure Figure33 built in order to better visualize the wine
characterizing volatile molecules in relation to the starter used. The main esters
present in the wines were 3-methylbut-1-yl methanoate (line 45, fruit aroma), ranging
from 19.97% (CS) to 41.7% (RT73) followed by ethyl ethanoate (line 60), due to the
large quantities of ethanol present. Isoamyl acetate was produced in relevant
quantities only by the S. cerevisiae strains (line 44, banana aroma), whereas ethyl
octanoate (line 65) and ethyl decanoate (line 59), generally associated to fruity
aroma, were produced only by CS and SRS1. Ethyl hexanoate (line 63), related to
red apple, fruity apple or estery flavor was completely absent in the co-culture wine.
Acids ranged from 2.8% (STS45) to 4.73% (RT73). The other compounds were
present only in low amount or absent (Table (Table22).

Table 2. Main volatile compounds identified (expressed as percentage of the peak

area of each compound compared to the total area) ad in the wines produced by S.
cerevisiae and non-Saccharomyces strains.

Line Compounds CS SRS1 RT73 SRS1+STS12 STS12 STS45

ALCOHOL

1* (2E)-3,7-dimethylocta- – 0.1 ± – – – –
2,6-dien-1-ol 0.09

2 3,7,11-Trimethyl-1,6,10- – 0.07 ± 0.2 ± – – –

dodecatrien-3-ol 0.01 0.09

3 1-butanol 0.3 ± 0.3 ± 0.7 ± 0.53 ± 0.05 0.6 ± 0.6 ±

0.0 0.1 0.11 0.1 0.02

4 1-dodecanol – 0.07 ± – – – –
0.05

5 1-nonanol – – 0.03 ± – – –
0.02

6 1-octanol 0.37 ± 0.4 ± – – – –

0.15 0.1

7 Oct-1-en-3-ol – 0.03 ± 0.07 ± 0.1 ± 0.01 0.1 ± 0.1 ±

0.01 0.03 0.02 0.01

8 1-pentanol – – 0.07 ± 0.17 ± 0.02 0.2 ± 0.1 ±

0.01 0.01 0.01

9 1-undecanol 0.03 ± – – – – –
0.01

10 2,2 ethoxyethoxy – – – – 0.1 ± –

ethanol 0.01

11 2,3-butandiol 0.35 ± 0.27 ± 0.7 ± 0.3 ± 0.1 0.5 ± 0.2 ±

0.10 0.11 0.11 0.02 0.02
Line Compounds CS SRS1 RT73 SRS1+STS12 STS12 STS45

12 2,3-dimethyl-2-hexanol – 0.1 ± – – – –
0.03

13 2-decen-1-ol – – – – 0.2 ± 0.1 ±

0.02 0.01

14 2-ethyl-1-hexanol 0.2 ± 0.17 ± – – – –

0.05 0.03

15 2-methyl 1-propanol 0.97 ± 1.73 ± 1.93 ± 2.17 ± 0.15 2.5 ± 2.8 ±

0.12 0.60 0.77 0.7 0.8

16 2-octanol – – – 0.13 ± 0.05 0.1 ± 0.1| ±

0.01 0.01

17 2-pentanol – – 0.03 ± 0.13 ± 0.01 0.2 ± –

0.01 0.01

18 3,4-dimethyl-2-hexanol 0.23 ± 0.1 ± 0.03 ± 0.07 ± 0.02 – –

0.09 0.04 0.01

19 3-hexen-1-ol – 0.07 ± 0.13 ± 0.2 ± 0.08 0.2 ± 0.1 ±

0.01 0.04 0.01 0.01

20 3-methyl-1-pentanol – – 0.03 ± 0.07 ± 0.01 0.1 ± –

0.01 0.01

21 3-(methylthio)-1- 0.07 ± 0.03 ± 0.03 ± – – –

propanol 0.01 0.01 0.01

22 5-methyl-2-hexanol – – 0.03 ± 0.4 ± 0.12 0.7 ± 0.1 ±

0.01 0.03 0.01

23 5-methoxy-1-pentanol 0.17 ± 0.07 ± – – – –

0.07 0.02

24 6,10,13-trimethyl-1- 0.03 ± 0.03 ± – – – 0.2 ±

tetradecanol 0.01 0.01 0.01

25 3,7-dimethyloct-6-en-1- – 0.1 ± – – – –
ol 0.02

26 Phenylmethanol 0.23 ± 0.23 ± 0.13 ± 0.13 ± 0.05 0.1 ± 0.1 ±

0.06 0.04 0.04 0.01 0.01
Line Compounds CS SRS1 RT73 SRS1+STS12 STS12 STS45

27 2-ethoxyethanol – – 0.03 ± – – –
0.01

28 1-hexanol 0.63 ± 0.83 ± 1.5 ± 2.27 ± 0.15 1.7 ± 1.3 ±

0.05 0.32 0.7 0.7 0.6

29 1-heptanol – 0.07 ± – – – –
0.01

30 2-phenylethanol 21.2 ± 16.03 12.77 8.07 ± 1.59 11.7 ± 10.4 ±

7.59 ± 5.72 ± 5.98 2.5 2.68

Total 24.78 20.80 18.41 14.74 19.0 16.2

ALDEHYDES

31 3-furaldehyde – – – 0.1 ± 0.01 0.3 ± 0.4 ±

0.08 0.2

32 Benzaldehyde 0.83 ± 0.87 ± 0.17 ± 0.63 ± 0.15 1.0 ± 1.0 ±

0.05 0.15 0.05 0.3 0.33

33 2-Phenylacetaldehyde – – – – – 0.1 ±
0.03

34 Carbaldeide – – – – 0.1 ± 0.1 ±

0.01 0.02

35 Decanal 0.17 ± 0.17 ± – – – –

0.02 0.08

36 Furan-2-carbaldehyde 0.17 ± 0.23 ± 0.07 ± 0.07 ± 0.01 – –

0.05 0.05 0.01

37 Heptanal 0.07 ± – – – – –
0.01

38 Nonanal 0.27 ± 0.3 ± – – – –

0.01 0.16

Total 1.51 1.57 0.24 0.80 1.4 1.6

KETONS
Line Compounds CS SRS1 RT73 SRS1+STS12 STS12 STS45

39 2,3-butanedione – – 0.17 ± 0.13 ± 0.01 – –

0.0.01

40 (E)-1-(2,6,6-Trimethyl-1- 0.1 ± 0.07 ± – – – 0.1 ±

cyclohexa-1,3- 0.01 0.01 0.01
dienyl)but-2-en-1-one

41 3-hexanone – – 0.03 ± 0.07 ± 0.01 0.1 ± 0.1 ±

0.01 0.01 0.03

42 3-hydroxy-2-butanone – 0.07 ± 0.03 ± 0.07 ± 0.01 0.1 ± 0.1 ±

0.01 0.01 0.01 0.02

Total 0.1 0.14 0.23 0.27 0.2 0.3

ESTERS

43 2-methylbut-1-yl – – 2.53 ± 5.4 ± 0.52 3.8 ± 4.1 ±

ethanoate 0.38 0.7 1.59

44 3-methylbut-1-yl 3.37 ± 3.47 ± 2.13 ± 0.63 ± 0.23 0.6 ± 0.6 ±

ethanoate 0.8 1.02 0.73 0.2 0.23

45 3-methylbut-1-yl 19.97 31.1 ± 41.7 ± 37.33 ± 3.13 36.4 ± 39.8 ±

methanoate ± 1.05 11.19 2.68 3.78 3.89

46 Ethyl furan-2- 0.07 ± 0.13 ± – – 0.1 ± –

carboxylate 0.01 0.01 0.01

47 Pentan-2-yl methanoate – 0.03 ± – – – –

0.01

48 3,7-dimethyloct-6-enyl 0.07 ± – – – – –
methanoate 0.01

49 2-methylpropyl acetate – – 0.23 ± 0.43 ± 0.05 0.2 ± 0.4 ±

0.1 0.04 0.2

50 2-phenylethyl ethanoate 1.0 ± 0.73 ± 0.13 ± 0.1 ± 0.01 0.1 ± 0.2 ±

0.2 0.16 0.05 0.01 0.1

51 Hexyl ethanoate 0.07 ± 0.67 ± – – – –

0.01 0.26
Line Compounds CS SRS1 RT73 SRS1+STS12 STS12 STS45

52 Ethyl phenylacetate 0.1 ± 0.1 ± – – – –

0.01 0.01

53 Ethyl 3- – 0.07 ± – – – –
phenylpropanoate 0.01

54 Benzyl 2- 0.6 ± 0.67 ± 0.27 ± 0.17 ± 0.05 0.2 ± 0.2 ±

hydroxybenzoate 0.1 0.17 0.15 0.03 0.08

55 Butanedioic acid, diethyl 2.0 ± 1.8 ± 1.3 ± 1.5 ± 0.65 1.3 ± 0.9 ±
ester 0.26 0.40 0.78 0.5 0.03

56 Ethyl butanoate 0.63 ± 1.23 ± 1.37 ± 1.3 ± 0.12 1.2 ± 1.2 ±

0.05 0.41 0.11 0.37 0.2

57 E
Fig. 3. Heatmap representing volatile profile of autochthonous S. cerevisiae
strains (SRS1 and RT73), commercial strain (CS), co-culture (SRS1+STS12),
and non-Saccharomyces strains (STS12 and STS45).

Compounds were organized by chemical families, and with the indication of the
number of compounds per family. Each line corresponds to one metabolite, and each
column corresponds to each strain. For the correspondence between number and
volatile compound see Table Table2.2. *The quantitative analysis of wine aroma
compounds was carried out on the basis of the relative peak area (Qi) calculated
from head space SPME (HS/SPME) gas chromatograms after addition of known
amounts of analyte standards.

In order to understand the variability among the strains, 101 aroma compounds data
were submitted to PCA analysis (Figures 4A–C) to generate a visual representation
of the wine discrimination on the basis of the specific aroma profiles generated by
the strains used. The first three principal components were able to explain >50% of
the total variances. Wines showed similar aroma profiles with differences for some
compounds as reported above. The first 3 PCs score plot (Figure (Figure4A)4A)
highlighted an overlapping of wines produced with SRS1+STS12, STS12, and RT73,
while wines obtained with CS, SRS1, and STS45 were well differentiated. For a
clearer comprehension of the loadings plot (Figure (Figure4C),4C), only the first 2
PCs of it were reported along with the first 2 PCs scores plot (Figure (Figure4B).4B).
When collapsing the scores plot in two dimensions, separations between the
different strains remained the same, except for an overlapping of CS and SRS1.
Looking at the loading plot it was impossible to select a single class of compounds
as the most important for a specific yeast (even observing the 3rd component, data
not shown). However, H. uvarum STS45 was characterized by sulfur compounds
(thiolane and 2-thiophene-acetic acid), some aromatic compounds (2-
phenylacetaldeide, 5H-dibenzo[b,f]azepine, toluene and 1,3 dimethyl, 2-ethyl
benzene) and hydocarbons such as 3-heptene and 2-heptamethyl nonene.
Moreover, most of aldehydes such as heptanal, nonanal, and decanal can be found
in the first quadrant of the loading plot correlated with CS and SRS1. Most of alcohols
with even number of C atoms such as 1-butanol, 1-hexanol, and 3-mehyl 1 pentanol
were present in the 3rd quadrant related to RT73 and to the co-culture SRS1+STS12.
Fig. 4. Score plot of the first 3 PCs (A), score (B) and loading plot (C) of the
first and second PCs after PC analysis on volatile compounds GC/MS-SPME
data for autochthonous S. cerevisiae strains (SRS1, RT73), commercial strain
(CS), non-Saccharomyces strains (STS12 and STS45), and co-culture
(SRS1+STS12).

Sensory analysis

Sensory analysis revealed the influence of yeast strains on some of the considered
descriptors. The wines fermented with SRS1 and the co-culture were characterized
by a good floral and a highest persistence (Figure (Figure5).5). Moreover negative
attributes such as reduced and grassy were not very pronounced (significantly lower
compared to STS12 and STS45 respectively). In particular, the co-culture had the
lowest reduced aroma of all theses. Wines obtained with STS12 and STS45 were
mainly characterized by grassy and reduced aroma. RT73 produced balanced wines
with negative and positive attributes arranged in good proportions. Wines fermented
with CS presented significantly low persistence, unwanted characteristic for
Montepulciano d'Abruzzo wine. However these wines showed good aroma
descriptors. In general, sensory analysis highlighted that the most interesting wines
were those produced with SRS1 and the co-culture since they were characterized
by a good floral, a highest persistence and, above all, have the reduced and grassy
not too marked, as often it happens also in high quality Montepulciano d'Abruzzo
wines.

Figure 5. Descriptive analysis of obtained wines. *p < 0.05, **p < 0.01.

Discussion

Montepulciano d'Abruzzo is a native grape variety of V. vinifera L., grown in central

Italy and used for production of high quality red wines. Limited studies have been
carried out to improve its enological characteristics through the use of indigenous
wine yeasts. The interest for autochthonous strains as single or mixed cultures in
combination with S. cerevisiae is gaining more and more importance since they are
potentially associated to a particular terroir and therefore adapted to a specific grape
must reflecting the biodiversity of a particular area (Bokulich et al., 2014; Capozzi et
al., 2015). For this reason, the application of indigenous mixed non-
Saccharomyces/Saccharomyces starter, able to mimic wine biodiversity, could be a
valid alternative to spontaneous fermentations, since the multi-starter ability to
increase the organoleptic properties of wine and to minimize the microbial spoilage
(Comitini et al., 2011; Ciani and Comitini, 2015).

In this study the organoleptic properties of Montepulciano d'Abruzzo wine and the
fermentation of two indigenous strains of S. cerevisiae (SRS1, RT73), a strain of S.
bacillaris (STS12), one of H. uvarum (STS45), and a co-culture of S. cerevisiae
(SRS1) and S. bacillaris (STS12) were evaluated. The data highlighted that at 3 days
faster fermentations were obtained in the musts inoculated with NS yeasts, in
agreement with other authors (Mendoza et al., 2007; Fleet, 2008; Ciani et al., 2010;
Suzzi et al., 2012b). Also the co-culture SRS1+STS12 showed a good fermentation
kinetic in comparison with SRS1. The positive interaction between S. cerevisiae and
S. bacillaris has been highlighted by other authors (Rantsiou et al., 2012; Suzzi et
al., 2012b). The sugar consumption was faster in SRS1+STS12 co-culture than in
S. cerevisiae pure cultures probably because of the osmotolerant and fructophilic
character of this non-Saccharomyces yeast. In fact, it consumes sugars at the early
stage of the fermentation, alleviating the S. cerevisiae from the osmotic stress,
thereby improving also the fermentation kinetics (Rantsiou et al., 2012; Englezos et
al., 2015). H. uvarum STS45 showed a good fermentation kinetic at the beginning
however at the end of fermentation it showed the lowest viable count values. The
disappearance of Hanseniaspora yeasts can be associated to their low ethanol
tolerance or to the production of other toxic compounds besides ethanol (Egli et al.,
1998; Fleet, 2003). S. bacillaris STS12 showed better fermentation kinetic than
STS45 and a higher number of viable cells at the end of fermentation. Some authors
reported that S. bacillaris was able to complete Macabeo must fermentation even if
with a slight delay compared to the S. cerevisiae fermentation (Andorrà et al., 2010).

The enological parameters during the first days of fermentation highlighted the
metabolic cooperation between inoculated and indigenous strains, although at the
end of fermentation all wines showed similar characteristics due to the dominance
of S. cerevisiae strains. In fact, also wines inoculated with NS wine yeasts showed
low values of residual sugar and an ethanol concentration of about 14%, probably
due to the contribution of indigenous Saccharomyces population present in the must
at the start of fermentation. The wine organoleptic properties are related to the
presence of several compounds deriving from the yeast metabolism (Capozzi et al.,
2015) and the dominance or competitiveness of a starter strain could have an
influence on the sensorial quality of wine by imposing its aromatic profile or deleting
the collaborative role of natural S. cerevisiae populations. In this study the
microsatellites analysis performed directly on the must allowed to establish the
dominance of all S. cerevisiae strains (SRS1, RT73, and CS) during all the
fermentation process shaping wine aroma and the presence of other non-starter
yeasts during fermentation with NS strains. In S. cerevisiae, microsatellites have
been described as abundant and highly polymorphic in length (Richards et al., 2009),
and for this reason, they are used as a reproducible and portable typing method
(Hennequin et al., 2001; Schuller et al., 2004; Bradbury et al., 2005; Legras et al.,
2005; Tofalo et al., 2013).

In all wines, the volatile acidity was below the legal limit of 1.2 g/l of acetic acid (Office
Internationale de la Vigne et du Vin, 2009), since higher values can confer to wine a
detrimental acidic flavor (Bely et al., 2003). In this context it is interesting to underline
that despite acetic acid production is considered as a common pattern in apiculate
yeasts (Romano et al., 2003), we found that wines inoculated with H. uvarum STS45
did not show an increased volatile acidity, in agreement with other authors (Andorrà
et al., 2010; Suzzi et al., 2012b). In addition all wines showed low quantity of BA
indicating the low decarboxylase activity of wine yeasts and indigenous malolactic
bacteria (Marcobal et al., 2006; Smit et al., 2008; Suzzi et al., 2012b).

Esters were the most representative compounds in all wines according to Ferreira
et al. (1995) and according to Suzzi et al. (2012a) the fruity character attributed to
the aroma of Montepulciano wines is mainly related to apple, pear, and cherry notes.
In fact, esters are a group of volatile compounds, arise from yeast metabolic activity,
that impart a mostly pleasant smell (Capozzi et al., 2015). The wines produced with
SRS1 and CS were well differentiated by other wines as shown by PCA and sensory
analyses acquiring the aromatic fingerprinting of the strain.

Specific features were also shown by wines produced with STS45. These wines
were characterized by the presence of sulfur compounds. Sulfur compounds have
different sensory properties and, although most of them could negatively affect the
wine aroma, they can also give a positive contribute by adding fruity notes (Swiegers
and Pretorius, 2005).

The wines produced with RT73 and SRS1+STS12 clustered together in the PCA
analysis, however sensory analysis revealed that wines obtained with the co-culture
showed interesting olfactory and tasting properties such as fruity, good body, and
persistence which are important characteristics for red wines. In addition the
simultaneously malolactic and alcoholic fermentation suggested a possible impact
of lactic acid bacteria on the final wines. In fact it is well known as the role of
malolactic fermentation is more than a deacidification, affecting the quality of wine
positively, such as volatile acids and negatively such BA production (Liu, 2002;
Renouf et al., 2006). In all the wines the content of BA was lower than the detection
limits, confirming that lactic acid bacteria vary on the production of these compounds
(Lonvaud-Funel, 2001).

The data obtained in this study highlighted that the use of NS autochthonous yeasts
positively influence wine aroma profile. In particular STS45 produced wines with a
specific aroma fingerprinting. In conclusion the natural cultures applied in cellar
vinification in this study can be considered as a useful tool that take the advantages
of the spontaneous fermentation, enhancing the chemical and organoleptic
characteristics of the wine and avoiding the risk of stuck fermentations and microbial
contamination.

Author contributions

Conceived and designed the experiments: GS, RT. Performed the experiments: GP,
MS, PG, FP. Analyzed the data: MS, DP, GA, RL. Wrote the paper: GS, RT, MS.

Conflict of interest statement

The authors declare that the research was conducted in the absence of any
commercial or financial relationships that could be construed as a potential conflict
of interest.
Acknowledgments

This work is part of the project “Oenological microbiota: selection to identify the wine
character and to improve the competitiveness of Montepulciano d'Abruzzo wineries”
supported by grant from Cassa di Risparmio di Teramo.

References

1. Alves Z., Melo A., Figueiredo A. R., Coimbra M. A., Gomes A. C., Rocha S.
M. (2015). Exploring the Saccharomyces cerevisiae volatile metabolome:
indigenous versus commercial strains. PLoS ONE 11:e0143641.
10.1371/journal.pone.0143641 [PMC free article] [PubMed] [Cross Ref]
2. Andorrà I., Berradre M., Rozès N., Mas A., Guillamón J. M., Esteve-Zarzoso
B. (2010). Effect of pure and mixed cultures of the main wine yeast species
on grape must fermentations. Eur. Food Res. Technol. 231, 215–224.
10.1007/s00217-010-1272-0 [Cross Ref]
3. Barbosa C., García-Martínez J., Pérez-Ortín J. E., Mendes-Ferreira A.
(2015). Comparative transcriptomic analysis reveals similarities and
dissimilarities in Saccharomyces cerevisiae wine strains response to nitrogen
availability. PLoS ONE 10:e0122709. 10.1371/journal.pone.0122709 [PMC
free article] [PubMed] [Cross Ref]
4. Bely M., Rinaldi A., Dubourdieu D. (2003). Influence of assimilable nitrogen
on volatile acidity production by Saccharomyces cerevisiae during high sugar
fermentation. J. Biosci. Bioeng. 96, 507–512. 10.1016/S1389-
1723(04)70141-3 [PubMed] [Cross Ref]
5. Bokulich N. A., Thorngate J. H., Richardson P. M., Mills D. A. (2014).
Microbial biogeography of wine grapes is conditioned by cultivar, vintage, and
climate. PNAS 111, E139–E148. 10.1073/pnas.1317377110 [PMC free
article] [PubMed] [Cross Ref]
6. Bradbury J. E., Richards K. D., Niederer H. A., Lee S. A., Dunbar P. R.,
Gardner R. C. (2005). A homozygous diploid subset of commercial wine yeast
strains. Antonie Leeuwenhoek 89, 27–37. 10.1007/s10482-005-9006-1
[PubMed] [Cross Ref]
7. Capozzi V., Garofalo C., Chiriatti M. A., Grieco F., Spano G. (2015). Microbial
terroir and food innovation: the case of yeast biodiversity in wine. Microbiol.
Res. 181, 75–83. 10.1016/j.micres.2015.10.005 [PubMed] [Cross Ref]
8. Ciani M., Comitini F. (2015). Yeast interactions in multi-starter wine
fermentation. Curr. Opin. Food Sci. 1, 1–6. 10.1016/j.cofs.2014.07.001 [Cross
Ref]
9. Ciani M., Comitini F., Mannazzu I., Domizio P. (2010). Controlled mixed
culture fermentation: a new perspective on the use of non- Saccharomyces
yeasts in winemaking. FEMS Yeast Res. 10, 123–133. 10.1111/j.1567-
1364.2009.00579.x [PubMed] [Cross Ref]
10. Comitini F., Gobbi M., Domizio P., Romani C., Lencioni L., Mannazzu I., et al.
. (2011). Selected non-Saccharomyces wine yeasts in controlled multistarter
fermentations with Saccharomyces cerevisiae. Food Microbiol. 28, 873–882.
10.1016/j.fm.2010.12.001 [PubMed] [Cross Ref]
11. Egli C. M., Edinger W. D., Mitrakul C. M., Henick-Kling T. (1998). Dynamics
of indigenous and inoculated yeast populations and their effect on the sensory
character of Riesling and Chardonnay wines. J. Appl. Microbiol. 85, 779–789.
10.1046/j.1365-2672.1998.00521.x [PubMed] [Cross Ref]
12. Englezos V., Rantsiou K., Torchio F., Rolle L., Gerbi V., Cocolin L. (2015).
Exploitation of the non-Saccharomyces yeast Starmerella bacillaris (synonym
Candida zemplinina) in wine fermentation: physiological and molecular
characterizations. Int. J. Food Microbiol. 199, 33–40.
10.1016/j.ijfoodmicro.2015.01.009 [PubMed] [Cross Ref]
13. Ferrara C. T., Wang P., Neto E. C., Stevens R. D., Bain J. R., Wenner B. R.,
et al. . (2008). Genetic networks of liver metabolism revealed by integration
of metabolic and transcriptional profiling. PLoS Genet. 4:e1000034.
10.1371/journal.pgen.1000034. [PMC free article] [PubMed] [Cross Ref]
14. Ferreira B., Hary C., Bard M. H., Táisant C., Olsson A., Lefur Y. (1995). Effect
of skin-contact and setting on the level of the C18:2, C18:3 fatty acids and C6
compounds in Burgundy Chardonnay musts and wines. Food Qual. Prefer. 6,
35–41. 10.1016/0950-3293(94)P4210-W [Cross Ref]
15. Fleet G. H. (2003). Yeast interactions and wine flavour. Int. J. Food Microbiol.
86, 11–22. 10.1016/S0168-1605(03)00245-9 [PubMed] [Cross Ref]
16. Fleet G. H. (2008).Wine yeast for the future. FEMS Yeast Res. 8, 979–995.
10.1111/j.1567-1364.2008.00427.x [PubMed] [Cross Ref]
17. Hennequin C., Thierry A., Richard G. F., Lecointre G., Nguyen H. V.,
Gaillardin C., et al. . (2001). Microsatellite typing as a new tool for
identification of Saccharomyces cerevisiae strains. J. Clin. Microbiol. 39, 551–
559. 10.1128/JCM.39.2.551-559.2001 [PMC free article] [PubMed] [Cross
Ref]
18. Howell K. S., Cozzolino D., Bartowsky E. J., Fleet G. H., Henschke P. A.
(2006). Metabolic profiling as a tool for revealing Saccharomyces interactions
during wine fermentation. FEMS Yeast Res. 6, 91–101. 10.1111/j.1567-
1364.2005.00010.x [PubMed] [Cross Ref]
19. ISO (1993). Sensory Analysis—General Guidance for the Selection, Training,
and Monitoring of Assessors. Part 1: Selected Assessors. International
Organization for Standardization, ISO; 8586–1:1993.
20. Jolly N. P., Varela C., Pretorius I. S. (2014). Not your ordinary yeast: non-
Saccharomyces yeasts in wine production uncovered. FEMS Yeast Res. 14,
215–237. 10.1111/1567-1364.12111 [PubMed] [Cross Ref]
21. Legras J. L., Ruh O., Merdinoglu D., Karst F. (2005). Selection of
hypervariable microsatellite loci for the characterization of Saccharomyces
cerevisiae. Int. J. Food Microbiol. 102, 73–83.
10.1016/j.ijfoodmicro.2004.12.007 [PubMed] [Cross Ref]
22. Liu S. Q. (2002). A review Malolactic fermentation in wine-beyond
deacidification. J. Appl. Microbiol. 92, 589–601. 10.1046/j.1365-
2672.2002.01589.x [PubMed] [Cross Ref]
23. Liu Y., Rousseaux S., Tourdot-Maréchal R., Sadoudi M., Gougeon R.,
Schmitt-Kopplin P., et al. . (2015). Wine microbiome, a dynamic world of
microbial interactions. Crit. Rev. Food Sci. Nutr.
10.1080/10408398.2014.983591 [Epub ahead of print]. [PubMed] [Cross Ref]
24. Lonvaud-Funel A. (2001). Biogenic amines in wines: role of lactic acid
bacteria. FEMS Microbiol. Lett. 199, 9–13. 10.1111/j.1574-
6968.2001.tb10643.x [PubMed] [Cross Ref]
25. Manetta A. C., Di Giuseppe L., Tofalo R., Martuscelli M., Schirone M.,
Giammarco M., et al. (2016). Evaluation of biogenic amines in wine:
determination by an improved HPLC-PDA method. Food Control 62, 351–
356. 10.1016/j.foodcont.2015.11.009 [Cross Ref]
26. Marcobal Á., Martín-Álvarez P. J., Polo M.,C.,, Muñoz R., Moreno-Arribas M.
V. (2006). Formation of biogenic amines throughout the industrial
manufacture of red wine. J. Food Prot. 69, 397–404. [PubMed]
27. Medina K., Boido E., Fariña L., Gioia O., Gomez M. E., Barquet M., et al. .
(2013). Increased flavour diversity of Chardonnay wines by spontaneous
fermentation and co-fermentation with Hanseniaspora vineae. Food Chem.
141, 2513–2521. 10.1016/j.foodchem.2013.04.056 [PubMed] [Cross Ref]
28. Mendoza L. M., Mancade Nadra M. C., Farías M. E. (2007). Kinetics and
metabolic behavior of a composite culture of Kloeckera apiculata and
Saccharomyces cerevisiae wine related strains. Biotechnol. Lett. 29, 1057–
1063. 10.1007/s10529-007-9355-0 [PubMed] [Cross Ref]
29. Moreira N., Mendes F., Hogg T., Vasconcelos I. (2005). Alcohols, esters and
heavy sulphur compounds production by pure and mixed cultures of apiculate
wine yeasts. Int. J. Food Microbiol. 103, 285–294.
10.1016/j.ijfoodmicro.2004.12.029 [PubMed] [Cross Ref]
30. Nehme N., Mathieu F., Taillandier P. (2010). Impact of the co-colture of
Saccharomyces cerevisiae- Oenococcus oeni on malolactic fermentetion and
partial characterization of a yeast –derived inihibitory peptidic fraction. Food
Microbiol. 27, 150–157. 10.1016/j.fm.2009.09.008 [PubMed] [Cross Ref]
31. Office Internationale de la Vigne et du Vin (2009). Compendium of
International Methods of Wine and Must Analysis. Paris: International
Organisation of Vine and Wine (OIV).
32. Pallmann C. L., Brown J. A., Olineka T. L., Cocolin L. S., Mills D. A., Bisson
L. F. (2001). Use of WL medium to profile native flora fermentations. Am. J.
Enol. Vitic. 52, 198–203.
33. Rantsiou K., Dolci P., Giacosa S., Torchio F., Tofalo R., Torriani S., et al. .
(2012). Candida zemplinina can reduce acetic acid production by
Saccharomyces cerevisiae in sweet wine fermentations. Appl. Environ.
Microbiol. 78, 1987–1994. 10.1128/AEM.06768-11 [PMC free article]
[PubMed] [Cross Ref]
34. Renouf V., Claisse O., Miot-Sertier C., Lonvaud-Funel A. (2006). Lactic acid
bacteria evolution during winemaking: use of rpoB gene as a target for PCR-
DGGE analysis. Food Microbiol. 23, 136–145. 10.1016/j.fm.2005.01.019
[PubMed] [Cross Ref]
35. Richards K. D., Goddard M. R., Gardner R. C. (2009). A database of
microsatellite genotypes for Saccharomyces cerevisiae. Antonie
Leeuwenhoek 96, 355–359. 10.1007/s10482-009-9346-3 [PubMed] [Cross
Ref]
36. Romano P., Fiore C., Paraggio M., Caruso M., Capece A. (2003). Function of
yeast species and strains in wine flavour. Int. J. Food Microbiol. 86, 169–180.
10.1016/S0168-1605(03)00290-3 [PubMed] [Cross Ref]
37. Sadoudi M., Tourdot-Maréchal R., Rousseaux S., Steyer D., GallardoChacón
J. J., Ballester J., et al. . (2012). Yeast-yeast interactions revealed by aromatic
profile analysis of Sauvignon Blanc wine fermented by single or co-culture of
non-Saccharomyces and Saccharomyces yeasts. Food Microbiol. 32, 243–
253. 10.1016/j.fm.2012.06.006 [PubMed] [Cross Ref]
38. Schuller D., Casal M. (2005). The use of genetically modified Saccharomyces
cerevisiae strains in the wine industry. Appl. Microbiol. Biotechnol. 68, 292–
304. 10.1007/s00253-005-1994-2 [PubMed] [Cross Ref]
39. Schuller D., Valero E., Dequin S., Casal M. (2004). Survey of molecular
methods for the typing of wine yeast strains. FEMS Microbiol. Lett. 231, 19–
26. 10.1016/S0378-1097(03)00928-5 [PubMed] [Cross Ref]
40. Serrazanetti D. I., Ndagijimana M., Sado S. L., Corsetti A., Vogel R. F.,
Ehrmann M., et al. . (2011). Acid stress-mediated metabolic shift in
Lactobacillus sanfranciscensis LSCE1. Appl. Environ. Microbiol. 77, 2656–
2666. 10.1128/AEM.01826-10 [PMC free article] [PubMed] [Cross Ref]
41. Smit A. Y., du Toit W. J., du Toit M. (2008). Biogenic amines in wine:
understanding the headache. South Afr. J. Enol. Vitic. 29, 109–127.
42. Suzzi G., Arfelli G., Schirone M., Corsetti A., Perpetuini G., Tofalo R. (2012a).
Effect of grape indigenous Saccharomyces cerevisiae strains on
Montepulciano d'Abruzzo red wine quality. Food Res. Int. 46, 22–29.
10.1016/j.foodres.2011.10.046 [Cross Ref]
43. Suzzi G., Schirone M., Sergi M., Marianella R. M., Fasoli G., Aguzzi I., et al. .
(2012b). Multistarter from organic viticulture for red wine Montepulciano
d'Abruzzo production. Front. Microbiol. 3, 135. 10.3389/fmicb.2012.00135
[PMC free article] [PubMed] [Cross Ref]
44. Swiegers J. H., Bartowsky E. J., Henschke P. A., Pretorius I. S. (2005). Yeast
and bacterial modulation of wine aroma and flavour. Aust. J. Grape Wine Res.
11, 139–173. 10.1111/j.1755-0238.2005.tb00285.x [Cross Ref]
45. Swiegers J. H., Pretorius I. S. (2005). Yeast modulation of wine flavor. Adv.
Appl. Microbiol. 57, 131–175. 10.1016/S0065-2164(05)57005-9 [PubMed]
[Cross Ref]
46. Tofalo R., Perpetuini G., Schirone M., Fasoli G., Aguzzi I., Corsetti A., et al. .
(2013). Biogeographical characterization of Saccharomyces cerevisiae wine
yeast by molecular methods. Front. Microbiol. 4, 166.
10.3389/fmicb.2013.00166 [PMC free article] [PubMed] [Cross Ref]
47. Tofalo R., Schirone M., Telera G. C., Manetta A. C., Corsetti A., Suzzi G.
(2011). Influence of organic viticulture on non-Saccharomyces wine yeast
populations. Ann. Microbiol. 61, 57–66. 10.1007/s13213-010-0102-8 [Cross
Ref]
48. Torcia M. (1972). Saggio Itinerario Nazionale pel Paese de' Peligni. Abruzzo:
Edizioni digitali del CISVA.
49. Tufariello M., Chiriatti M., Grieco F., Perrotta C., Capone S., Rampino P., et
al. (2014). Influence of autochthonous Saccharomyces cerevisiae strains on
 volatile profile of Negroamaro wines. LWT Food Sci. Technol. 58, 35–48.
10.1016/j.lwt.2014.03.016 [Cross Ref]
50. Vaudano E., Garcia-Moruno E. (2008). Discrimination of Saccharomyces
cerevisiae wine strains using microsatellite multiplex PCR and band pattern
analysis. Food Microbiol. 25, 56–64. 10.1016/j.fm.2007.08.001 [PubMed]
[Cross Ref]
51. Vernocchi P., Patrignani F., Ndagijimana M., Chaves Lopez C., Suzzi G.,
Gardini F., et al. (2015). Trebbiano wine produced by using Saccharomyces
cerevisiae strains endowed with β-glucosidase activity. Ann. Microbiol. 65,
1565–1571. 10.1007/s13213-014-0995-8 [Cross Ref]
V. Change in Color and Volatile Composition of Skim Milk
Processed with Pulsed Electric Field and
Microfiltration Treatments or Heat Pasteurization †

Foods. 2014 Jun; 3(2): 250–268.

Anupam Chugh,1 Dipendra Khanal,1 Markus Walkling-Ribeiro,1,2,* Milena Corredig,1
Lisa Duizer,1 and Mansel W. Griffiths1

Abstract

Non-thermal processing methods, such as pulsed electric field (PEF) and tangential-
flow microfiltration (TFMF), are emerging processing technologies that can minimize
the deleterious effects of high temperature short time (HTST) pasteurization on
quality attributes of skim milk. The present study investigates the impact of PEF and
TFMF, alone or in combination, on color and volatile compounds in skim milk. PEF
was applied at 28 or 40 kV/cm for 1122 to 2805 µs, while microfiltration (MF) was
conducted using membranes with three pore sizes (lab-scale 0.65 and 1.2 µm TFMF,
and pilot-scale 1.4 µm MF). HTST control treatments were applied at 75 or 95 °C for
20 and 45 s, respectively. Noticeable color changes were observed with the 0.65 µm
TFMF treatment. No significant color changes were observed in PEF-treated, 1.2
µm TFMF-treated, HTST-treated, and 1.4 µm MF-treated skim milk (p ≥ 0.05) but
the total color difference indicated better color retention with non-thermal
preservation. The latter did not affect raw skim milk volatiles significantly after single
or combined processing (p ≥ 0.05), but HTST caused considerable changes in their
composition, including ketones, free fatty acids, hydrocarbons, and sulfur
compounds (p < 0.05). The findings indicate that for the particular thermal and non-
thermal treatments selected for this study, better retention of skim milk color and
flavor components were obtained for the non-thermal treatments.

Keywords: pulsed electric field (PEF), microfiltration (MF), thermal pasteurization,

skim milk, volatile compounds, color degradation, non-thermal processing, hurdle
technology

1. Introduction

Milk is a widely consumed beverage due to its nutritional importance, a pleasant

aroma and mouth-feel, and a slightly sweet taste. Color and flavor are important
sensorial attributes of milk that are influenced by several factors, such as milk
composition, cow’s feed and metabolism, environmental factors and processing
conditions [1,2]. Few studies are available on the impact of conventional thermal
treatments and emerging processing technologies like pulsed electric field (PEF),
high hydrostatic pressure (HHP) and ultra high pressure (UHP) on volatile
compounds in milk or milk based beverages [3,4,5,6]. Pulsed electric field (PEF) and
microfiltration (MF) are emerging innovative technologies that could meet the
increasing consumer demand for “fresh-like” minimally processed foods. The mode
of action that underlies PEF is based on short electric pulses of high voltage applied
to a product that is placed between a pair of electrodes, thereby, bringing about
electroporation of the bacterial cell wall and its subsequent breakdown [7]. In addition
to the type of microorganism and its innate resistance [8], the efficacy of PEF for
food preservation depends on processing factors such as electric field strength,
number of pulses applied and the treatment time [9], as well as product parameters,
including electrical conductivity, viscosity, and pH [10].

MF is a membrane-driven separation process typically employing membranes with

a pore size of 1.2 to 1.4 µm that allows removal of bacteria in vegetative and spore
forms from milk [11,12]. Commercially available thermal processing methods
promise a high degree of microbial safety but can adversely affect other food
properties such as color and volatile compounds. Several studies have
demonstrated a comparable impact of PEF and MF processes on microbial
inactivation in foods [13,14,15,16,17]. Milk has its natural color due to the reflectance
of light by dispersed milk fat globules, proteins, and natural milk pigments like
riboflavin and carotenoids [18,19]. However, milk color is altered at high processing
temperatures as a result of Maillard browning or a temporary increase in lightness
due to denaturation of soluble whey proteins [20,21]. The changes in milk
compounds that are responsible for change in milk color may, at the same time,
affect the perception of flavor in milk due to intensification of some volatile
components. It is generally accepted that prolonged or severe heat treatments cause
the degradation of the volatile profile of milk, with the increase in concentration of
volatile compounds being positively correlated to intensity of the heating [22,23]. The
development of such volatile compounds may alter consumer acceptance as
suggested by Gandy and others [24]. In contrast, non-thermal processing may have
a minimal effect on the concentration of volatiles in foods due to shorter processing
times at temperatures below those used for pasteurization. Zhang et al. [5] observed
increased levels of aldehydes and ketones in pasteurized milk as compared to raw
and PEF processed milk. PEF-treated and HHP-treated orange juice-milk beverage
showed a reduced loss of volatiles as compared to thermally processed beverage
[3]. From a quality and safety perspective, it may be advantageous to use hurdle
technology rather than using a single technology to achieve sufficient product safety
and maximized quality while maintaining minimal processing. Microfiltration is one
of the non-thermal technologies that have been commercially allowed for extending
shelf life of milk, although its use as a stand-alone treatment is restricted due to
regulatory requirements [25]. Based on microbiological data, previous studies have
indicated that the safety of skim milk processed with a combination of PEF and MF
is comparable to that achieved by HTST pasteurization [13,17], potentially
representing a non-thermal alternative to the well-established heat pasteurization.
However, the effect of this novel process on color and volatile compounds has not
been determined. Thus, the objective of the present study was to study the effect of
different PEF, MF, and hurdle (PEF/MF) processing conditions on the color and
flavor profiles of skim milk. In addition, the retention of color and flavor following
treatments with these non-thermal technologies was compared to that obtained by
heat pasteurization.
2. Experimental Materials and Methodology

2.1. Supply, Preparation, and Storage Conditions of Skim Milk

Raw milk obtained from a local dairy processing plant was separated at the
Canadian Research Institute for Food Safety at the Department of Food Science,
University of Guelph, using a cream separator (STsM-100-18, Motor Sich JSC,
Zaporozhye, Ukraine). Skim milk was selected for processing in this study due to its
lower fat content, allowing efficient microfiltration without membrane blockage, and
because it is common research model for whole milk and has grown in consumer
popularity.

2.2. Pulsed Electric Field (PEF) Treatment of Skim Milk

Raw skim milk processing was carried out in a PEF treatment chamber designed at
the University of Guelph using an exponential decay pulse generator (PPS 30,
University of Waterloo, Waterloo, ON, Canada) to form monopolar pulses with an
average width of 1.5 µs. PEF parameters were measured and monitored as
described previously [17]. The single PEF chamber used consists of two co-axial
stainless steel electrodes encased in an insulating plastic casing. Liquid passes
between electrodes through a gap of 0.21 cm. The chamber has a hold up volume
of 44 cm3 of which 25 cm3 is the volume between the electrodes where milk is treated
by means of electric pulses. An external jacket with a volume of 790 cm 3 surrounds
the outer electrode for additional temperature control, allowing the PEF chamber to
be heated or cooled during operation. Skim milk was pumped through Masterflex
silicone tubing (size L/S 16) using a peristaltic pump (Masterflex pump drive 7524-
40 and pump head 77201-60, Cole Parmer Instrument Co., Vernon Hills, IL, USA).
The product was pumped at a flow rate of 20 to 35 mL/min and the temperature at
the entry and exit of the PEF chamber was recorded using T-type thermocouples
(TMTSS-040G-6, Omega, Stamford, CT, USA) connected to a wireless temperature
data logger (OM-SQ2020-2F8, Omega, Stamford, CT, USA). After treatment the
outgoing product was cooled by flow through tubing submerged in a cooling water
bath (NESLAB RTE 7, Thermo Fisher Scientific Inc., Newlington, NH, USA), set at
2.5 °C. For hurdle treatment, involving subsequent microfiltration, samples were kept
in ice in a cold room at 4 °C before further processing to maintain the cold chain and
thus, avoid growth of mesophilic microorganisms and potential quality degradation
in line with common practice used in the dairy industry. Experiments were conducted
at three PEF intensities that reflect different processing demands for skim milk with
regard to food safety and energy consumption, which can vary considerably, based
on production facilities and locations, and have not been investigated for comparable
effects on milk quality parameters, such as color and volatile composition, based on
these treatment intensities. The skim milk was processed using electric field strength
and treatment time combinations of 28 kV/cm and 2805 µs for low-intensity PEF
(PEF-L), 40 kV/cm and 1122 µs for moderate-intensity PEF (PEF-M), and 40 kV/cm
and 1571 µs for high-intensity PEF (PEF-H). These processing conditions
corresponded to energy densities of 83, 157 and 198 kJ/L, respectively. A pulse
frequency of 25 Hz was applied for PEF-L, whereas 17.5 Hz were used for PEF-M
and PEF-H treatments. Milk temperature at the inlet of the PEF chamber was 17 °C
on average and depending on the treatment intensity applied maximum
temperatures of 37 (PEF-L), 56 (PEF-M), and 65 (PEF-H) °C were obtained at the
outlet. Following PEF and the cooling step the product temperature was 10 °C and
samples were collected at this temperature.

2.3. Microfiltration (MF) of Skim Milk

Bearing in mind that smaller sized membrane pores allow for enhanced microbial
reduction in skim milk, the latter was microfiltered at different pore size diameters to
compare their effect on the product quality. Moreover, lab and pilot scale systems
featuring different membrane materials and designs were used for MF to determine
whether or not the processing scale and/or membrane materials and/or membrane
designs affect color and volatile compound composition in skim milk. Tangential-flow
microfiltration (TFMF) was performed using a laboratory scale Supor tangential-flow
filtration system featuring either 0.65 µm (medium screen, 1.4 cm thick) or 1.2 µm
(suspended screen, 1.8 cm thick) pore-sized polyethersulfone membranes (Pall
Corporation, Port Washington, NY, USA). For laboratory-scale TFMF, a flow rate of
300 mL/min was generated using a Masterflex EW 77200-60 pump (Cole Parmer
Instrument Co., Vernon Hills, IL, USA) and the skim milk was delivered by silicone
tubing (Masterflex 96400-25, Cole Parmer Instrument Co., Vernon Hills, IL, USA)
through either of the TFMF membrane cassettes, with trans-membrane pressures
ranging from 0.3 to 0.6 kPa. The permeate flow rate was between 7 and 13 mL/min
for milk passing through the TFMF 0.65 µm membrane (MF-0.65), whereas it was
between 37 and 53 mL/min for the TFMF 1.2 µm (MF-1.2) membrane. In addition,
skim milk processing was also carried out on a pilot scale 1.4 μm pore size, ceramic
membrane, cross-flow microfiltration (CFMF) unit (MFS-1, Tetra Pak Filtration
Systems, Aarhus, Denmark). The CFMF unit was used at a permeate flow rate (100
L/h) ten times that of retentate (10 L/h), using filter module Type-SCT Membralox
(Tetra Pak Filtration Systems). The trans-membrane pressure of the ceramic
membrane module was 106 kPa. All microfiltration treatments were carried out at a
constant temperature of 35 °C after pre-heating skim milk in a water bath (Isotemp
210, Thermo Fisher Scientific Inc., Newlington, NH, USA) set to this temperature
and equipped with a thermometer. In agreement with membrane and equipment
manufacturer specifications milk volumes were selected for TFMF and CFMF
processing that allowed proper milk circulation through the membranes and process
stability before sampling. To avoid membrane clogging over time, laboratory scale
membranes were cleaned by recirculating 0.5 N sodium hydroxide followed by 0.1
N nitric acid and 2% bleach at 50 °C, and a final rinse with water immediately after
each run. Following the same approach but using different cleaning agents the pilot
plant CFMF ceramic membrane was cleaned by recirculation of commercially
available mixed acid detergent descaler (Diversey Divos2 (VM13), Diversey,
Oakville, ON, Canada) and caustic detergent (Diversey Liquid Bril Tak (VC85))
followed by a final rinse with water.

2.4. PEF/MF Processing of Skim Milk

For hurdle processing, milk was processed using PEF as described above and
stored at 4 °C prior to processing with MF. All three PEF treatments: PEF-L, PEF-M
and PEF-H were used for hurdle processing followed by MF-0.65 µm and MF-1.2
µm membranes (PEF/MF). However, due to the difficulty in achieving the high
sample volumes required for pilot scale microfiltration (MF-1.4) using the laboratory
scale PEF unit, only skim milk processed using the PEF-M treatment was coupled
with the MF-1.4 µm treatment.

2.5. High-Temperature Short-Time Pasteurization of Skim Milk

High-temperature short-time (HTST) processing of raw skim milk was carried out
with a pilot scale, dual stage heat exchanger unit (UHT/HTST Lab-25 EDH,
Microthermics Inc., Raleigh, NC, USA) at the Guelph Food Technology Centre
(GFTC), Guelph, ON, Canada. Skim milk was kept at 4 °C before processing in the
tubular heat exchanger. Milk was processed at two different temperatures of 75
(HTST-75) and 95 °C (HTST-95) for respective holding times of 20 and 45 s,
corresponding to flow rates of 0.9 and 0.4 L/min, respectively. The higher intensity
HTST pasteurization (95 °C, 45 s) for milk takes into consideration an additional
safety margin and extended shelf life in part applied by the industry, but at the same
time increased thermal load also renders changes in color and volatile compounds
more likely. The heat exchanger pre-heated milk to 55 °C prior to reaching the
respective temperatures and cooled the product to 10 °C after the treatments and
prior to sampling.

2.6. Color Analysis

Color attributes were measured in Hunter Lab color space using a CM 3500-d
spectrophotometer (Konica Minolta Sensing Inc., Mahwah, NJ, USA) equipped with
SpectraMagic NX CM-S 100 software (Konica Minolta Sensing Inc.). Reflectance
measurements were collected over a wavelength range from 400 to 700 nm. The
color values were expressed in Hunter Lab color space as lightness (L), redness (a),
and yellowness (b). In order to compare the total color difference (ΔE), between the
color properties of untreated samples to those of values obtained after subjecting
milk to different treatments, the following equation [26,27] was utilized:

(1)

where, ΔL = Lstandard − Lsample, Δa = astandard − asample, Δb = bstandard − bsample.

In addition, the whiteness was calculated for the skim milk samples by converting
Hunter Lab to CIE 1931 XYZ color space values using following formulas:
(2)

(3)

(4)

and then using American Society for Testing and Materials (ASTM) E313 standard
practice for calculating Whiteness index (WI):

WI = (3.388 × Z) - (3 × Y)
(5)

For each treatment, samples were collected in triplicate. Color measurements were
done in duplicate for each sample.

2.7. Analysis of Volatile Compounds

Headspace volatiles in skim milk were analyzed using SIFT-MS (Selected In Flow
Tube-Mass Spectrometry), which is a direct mass spectrometer that applies
chemical ionization for the analysis of volatile components, yielding product ions that
are analyzed with SIFT-MS technology (Voice200®, Syft Technologies Ltd.,
Christchurch, New Zealand). The latter identifies and quantifies volatile compounds
with a built-in scientific library, that is experimentally determined with product ions
and reaction rate coefficients, in combination with the flow tube geometry, the ionic
reaction time, the measured flow rates and the pressure applied during sample
analysis and requires no chemical standards for calibration. Skim milk samples (35
mL) were placed in 300 mL glass jars with lids closed and then warmed to room
temperature before analysis. An empty jar was used as a blank. Twenty-one flavor
compounds associated with milk were quantified using selected ion mode with 100
ms time limit, count limit of 10,000, 30 s background scan time and a product scan
time of 60 s. Helium was applied as the carrier gas (250 kPa) and the scan was
performed following charge transfer from three positively-charged reagent ions
(H3O+, NO+, and O2+) to the analyte in the flow tube. The temperature of the inlet
arm extension was 120 °C and flow tube pressure was 113.6 ± 1.7 mTorr.
Samples were collected in triplicate for each treatment with a duplicate analysis for
each sample.

2.8. Statistical Data Analysis

Color and flavor data were statistically analyzed with Sigmaplot version 11.0 (Systat
Software Inc., London, UK). One-way analysis of variance (ANOVA) was carried out
for three different batches of skim milk (n = 3), analyzing all treatments at a
confidence interval of 95%. When statistical differences between treatment means
were observed, the Holm-Sidak method for multiple pairwise comparisons was used.
For adequate reproducibility each treatment was repeated three times and analysis
was performed in duplicate.

3. Results

3.1. Color Attributes of Non-Thermally and Thermally Treated Skim Milk

No significant difference in color was found in untreated skim milk and when
processed using PEF-L, PEF-M, PEF-H, MF-1.2 and MF-1.4 and different
combinations of PEF with MF treatments. The values for the color attributes L, a,
and b following various treatments are provided in Table 1.

Table 1. Comparison of Hunter color space attributes L (lightness), a (redness) and

b (yellowness) measured in raw skim milk and after processing with high-
temperature short-time pasteurization at 75 (HTST-75) and 95 (HTST-95) °C, pulsed
electric field at low (PEF-L), moderate (PEF-M) and high (PEF-H) intensities, 0.65
and 1.2 μm pore size tangential flow microfiltration (0.65 TFMF and 1.2 TFMF,
respectively), 1.4 μm-pore size cross-flow microfiltration (1.4 CFMF), and different
pulsed electric field-based combinations with the latter three membrane filtration
treatments.

Treatments L a b ΔE

Raw Skim Milk 64.50 (±0.87) a −3.92 (±0.08) a −0.15 (±0.15) a -

HTST-75 65.00 (±0.61) a −4.01 (±0.10) a −0.08 (±0.02) a 0.51

HTST-95 65.21 (±0.10) a −4.03 (±0.01) a 0.19 (±0.04) a 0.79

0.65 TFMF 53.90 (±1.25) b −3.92 (±0.03) a −3.21 (±0.28) b 11.03

1.2 TFMF 64.22 (±0.05) a −3.94 (±0.05) a −0.28 (±0.17) a 0.31

1.4 CFMF 64.71 (±0.08) a −3.89 (±0.03) a −0.33 (±0.04) a 0.10

PEF-L 64.26 (±0.68) a −3.88 (±0.02) a −0.20 (±0.17) a 0.25

Treatments L a b ΔE

PEF-M 64.58 (±0.51) a −3.88 (±0.01) a −0.16 (±0.26) a 0.10

PEF-H 64.85 (±0.60) a −3.96 (±0.05) a −0.07 (±0.16) a 0.37

PEF-L/0.65 TFMF 53.07 (±0.65) b −3.99 (±0.04) a −3.12 (±0.11) b 11.81

PEF-M/0.65 TFMF 53.37 (±1.23) b −3.99 (±0.04) a −3.52 (±0.03) b 11.63

PEF-H/0.65 TFMF 53.44 (±1.38) b −4.02 (±0.04) a −3.28 (±0.24) b 11.49

PEF-L/1.2 TFMF 64.51 (±0.50) a −3.92 (±0.02) a −0.15 (±0.22) a 0.02

PEF-M/1.2 TFMF 64.36 (±0.42) a −3.92 (±0.01) a −0.23 (±0.03) a 0.15

PEF-H/1.2 TFMF 64.65 (±0.32) a −4.00 (±0.02) a −0.20 (±0.06) a 0.18

PEF-M/1.4 CFMF 64.57 (±0.18) a −3.91 (±0.02) a −0.34 (±0.07) a 0.20

P1 * NS * -

SEM 2 0.539 0.010 0.148 0.711

a, b
Different superscripted letters in the same column indicate statistical significance
between the two means. Values in parentheses following the mean values of the
color attributes indicate the standard deviations. 1 P stands for the statistical
probability. * Refers to a statistical significance of p < 0.05. NS indicates no statistical
significance. 2 SEM abbreviates the standard error of the mean.

Total color difference, ΔE, representing an overall color difference of treated samples
from the untreated skim milk was calculated using equation 1 and presented in Table
1. Color difference classification was adopted from Cserhalmi and others [27]. Based
on this classification system ΔE can be categorized as: 0 to 0.5 = “not noticeable”,
0.5 to 1.5 = “slightly noticeable” and >1.5 = “noticeable”. Slightly noticeable color
differences of 0.51 and 0.79 were observed in skim milk treated with HTST treatment
at 75 °C or 95 °C, respectively. However, a noticeable ΔE of 11.03 was obtained in
skim milk processed with MF-0.65 (p < 0.05). Similar values (p ≥ 0.05) were obtained
when combination treatments were applied consisting of PEF (PEF-L, PEF-M, and
PEF-H) and subsequent MF-0.65 for hurdle treatment of skim milk, yielding ΔEs of
11.83, 11.63 and 11.49, respectively. Overall, noticeably better color retention was
achieved with PEF-L, PEF-M and PEF-H treatments; 1.2 µm TFMF and 1.4 µm
CFMF-based processing, achieving lower ΔE values (in the range between 0.10 and
0.37 (p < 0.05)) than by the HTST and 0.65 µm TFMF treatments.
While skim milk pasteurized at 95 °C showed a positive yellowness value, no
significant difference in any of the color attributes was observed (p ≥ 0.05) between
the heat-treated skim milk compared to that processed by PEF, MF-1.2 and MF-1.4,
and their combinations. Lightness was 16%–18% lower for all treatments that
involved skim milk processing through 0.65 µm MF and the corresponding samples
also exhibited a higher blueness as indicated by a lower b value (p < 0.05). As a
result significantly lower whiteness index values were obtained for MF-0.65 and
hurdle treatments combining PEF of different processing intensities with MF-0.65 as
shown in Figure 1 (p < 0.05).

Fig. 1. Whiteness obtained in untreated raw skim milk and by processing the latter
with low-intensity (PEF-L), medium-intensity (PEF-M) and high-intensity (PEF-H)
pulsed electric field, 0.65 (MF-0.65), 1.2 (MF-1.2), and 1.4 (MF-1.4) μm
microfiltration, high-temperature short-time (HTST) pasteurization treatments and
different pulsed electric field-based combinations with microfiltration treatments
(after conversion from Hunter Lab color space data to CIE 1931 XYZ color space
values). Different letters (i.e., a, b) above bars indicate significant differences
between treatment means (p < 0.05).

For the remaining non-thermal or thermal treatments no significant differences in

whiteness were determined when compared among one another or with raw skim
milk (p ≥ 0.05). Overall, non-thermal treatments using MF-1.2 MF-1.4 and PEF-L,
PEF-M, PEF-H, and heat pasteurization did not have a significant impact on the
single color properties of skim milk but, with the exception of MF-0.65-based
treatments, thermally pasteurized skim milk was in a higher ΔE category than non-
thermally treated skim milk.
3.2. Flavor Attributes of Non-Thermally and Thermally Treated Skim Milk

The effect of heat, pulsed electric field and microfiltration treatments on the volatile
profile of skim milk was measured using SIFT-MS. Volatile compounds generated
as a result of heat treatment and non-enzymatic browning reactions in skim milk
were of main interest and, therefore, 21 compounds were selected for this study.
Quantities of acetic acid and acetaldehyde were not detected by the mass
spectrometer, so that data for 19 compounds (Table 2, Table 3 and Table 4) was
obtained. Concentrations of ketones were found to be significantly higher in skim
milk treated with HTST at 95 °C for a holding time of 45 s when compared to other
treatments (p < 0.05). The major ketones contributing to this were 2-hexanone, 2-
heptanone, 2-octanone, acetone, and butanone. However, there was no marked
increase in concentration of these ketones in skim milk pasteurized at 75 °C for 20
s, nor for PEF-treated and MF-treated skim milk (p ≥ 0.05).

Table 2. Retention of in 3-methyl butanal, heptanal, decanal, butanoic acid, hexanoic

acid, dimethyl sulfide, and hydrogen sulfide obtained in skim milk after high-
temperature short-time pasteurization (HTST75 = 75 °C for 20 s; HTST95 = 95 °C for
45 s) and lab-scale tangential-flow microfiltration (TFMF0.65 = 0.65 µm; TFMF1.2 =
1.2 µm), pilot scale cross-flow microfiltration (CFMF1.4 = 1.4 µm), pulsed electric field
at low (PEF-L: 28 kV/cm; 2805 µs), moderate (PEF-M: 40 kV/cm, 1122 µs), and high
(PEF-H: 40 kV/cm, 1571 µs) processing intensities and selected non-thermal
combination treatments using PEF and subsequent microfiltration.

3-Methyl Dimethyl Hydrogen

Heptanal Decanal Butanoic Hexanoic
Treatments butanal sulfide sulfide
[ppb] [ppb] acid [ppb] acid [ppb]
[ppb] [ppb] [ppb]

Raw Skim 2.7 (±0.5)

3.6 (±1.1) a 15 (±6.2) a a 14 (±6.0) a 2.3 (±0.8) a 12 (±5.9) a 1.4 (±0.7) a,b
Milk

3.3 (±0.0)
HTST75 3.3 (±0.1) a 15 (±1.9) a a
13 (±1.4) a 9.7 (±0.3) d 17 (±2.2) a,b 4.2 (±0.1) c

4.8 (±1.6)
HTST95 6.3 (±0.7) b 24 (±5.2) a a 44 (±6.0) b 4.8 (±0.2) c 25 (±6.1) b 5.5 (±0.2) d

14 (±10.3) 2.2 (±0.3)

TFMF0.65 3.9 (±0.0) a a a
15 (±5.8) a,b 1.1 (±0.3) a 11 (±0.7) a,b n.d.

3.4 (±1.4)
TFMF1.2 4.5 (±1.7) a 17 (±7.3) a a 13 (±1.8) a 2.4 (±0.2) a,b 12 (±0.9) a,b 1.1 (±0.5) a,b

2.3 (±0.8)
CFMF1.4 3.5 (±1.5) a 11 (±1.8) a a
11 (±3.1) a 2.5 (±0.3) a,b 11 (±0.5) a,b 1.1 (±0.4) a,b

2.5 (±0.7)
PEF-L 2.4 (±0.4) a 10 (±0.7) a a 12 (±3.3) a 1.6 (±0.3) a,b 9.1 (±3.4) a 0.8 (±0.1) a
 3-Methyl Dimethyl Hydrogen
Heptanal Decanal Butanoic Hexanoic
Treatments butanal sulfide sulfide
[ppb] [ppb] acid [ppb] acid [ppb]
[ppb] [ppb] [ppb]

3.4 (±0.8)
PEF-M 3.0 (±1.7) a 17 (±2.9) a a 16 (±7.2) a,b 1.8 (±0.5) a,b 15 (±7.0) a,b 1.3 (±0.2) a, b

2.6 (±1.1)
PEF-H 3.9 (±2.8) a 14 (±2.2) a a 15 (±5.1) a,b 2.7 (±0.1) b 18 (±2.1) a,b 2.1 (±0.1) b

PEF- 3.9 (±2.3)

2.5 (±0.4) a 20 (±0.7) a a 10 (±3.4) a 1.7 (±0.2) a,b 5.0 (±0.8) a 1.1 (±0.3) a,b
L/TFMF0.65

PEF- 3.3 (±1.6)

3.6 (±2.8) a 21 (±1.0) a a
13 (±5.2) a 1.7 (±0.5) a,b 7.5 (±1.3) a 0.9 (±0.3) a,b
M/TFMF0.65

PEF- 3.2 (±0.5)

4.8 (±2.5) a 16 (±2.1) a a 9.1 (±2.3) a 1.5 (±0.5) a,b 7.9 (±0.7) a 1.1 (±0.2) a,b
H/TFMF0.65

PEF- 2.5 (±0.1)

2.1 (±0.5) a 14 (±2.7) a a
8.6 (±1.9) a 1.7 (±0.1) a,b 9.2 (±1.9) a 0.9 (±0.2) a,b
L/TFMF1.2

PEF- 1.5 (±1.2)

5.6 (±0.2) a 20 (±0.4) a a 16 (±0.3) a,b 2.5 (±0.1) a,b 17 (±2.7) a,b 1.6 (±0.2) a,b
M/TFMF1.2

PEF- 1.8 (±0.2)

2.7 (±1.3) a 13 (±3.0) a a
8.9 (±3.2) a 2.1 (±0.5) a,b 12 (±2.4) a,b 1.2 (±0.3) a,b
H/TFMF1.2

PEF- 3.5 (±0.1)

2.0 (±0.2) a 17 (±1.2) a a 9.1 (±0.7) a 1.4 (±0.4) a,b 14 (±2.3) a,b 1.0 (±0.0) a,b
M/CFMF1.4

P1 * NS NS * * * *

SEM 2 0.19 0.64 0.12 1.53 0.20 0.87 0.11

a, b, c
Different superscripted letters in the same column indicate statistical
significance between the two means. Values in parentheses following the mean
values of compound indicate the standard deviations. 1 P stands for the statistical
probability. * Refers to a statistical significance of p < 0.05. NS indicates no statistical
significance. 2 SEM abbreviates the standard error of the mean.

Table 3. Retention of in ethanol, 1-propanol, 1-octene, toluene, p-xylene, and methyl

acetate obtained in skim milk after high-temperature short-time pasteurization
(HTST75 = 75 °C for 20 s; HTST95 = 95 °C for 45 s) and lab-scale tangential-flow
microfiltration (TFMF0.65 = 0.65 µm; TFMF1.2 = 1.2 µm), pilot scale cross-flow
microfiltration (CFMF1.4 = 1.4 µm), pulsed electric field at low (PEF-L: 28 kV/cm;
2805 µs), moderate (PEF-M: 40 kV/cm, 1122 µs), and high (PEF-H: 40 kV/cm, 1571
µs) processing intensities and selected non-thermal combination treatments using
PEF and subsequent microfiltration.
 Methyl
Ethanol 1-Propanol 1-Octene Toluene p-Xylene
Treatments acetate
[ppb] [ppb] [ppb] [ppb] [ppb]
[ppb]

Raw Skim 4.2 (±0.8) 1.6 (±0.9)

65 (±18) a 10 (±2.0) a 34 (±8.3) a a a 6.7 (±1.3) a
Milk

57 (±9.9) 4.6 (±0.8) 1.5 (±0.3)

HTST75 a,b 11 (±1.8) a,b 31 (±2.1) a a a 10 (±3.0) a,b

131 (±1.8) 6.9 (±1.6) 2.0 (±0.8)

HTST95 c 16 (±2.1) b 38 (±1.5) a b a 19 (±4.1) b

51 (±6.0) 4.0 (±0.7) 1.6 (±1.1)

TFMF0.65 a,b 11 (±0.8) a,b 15 (±1.7) a a a 6.7 (±0.8) a,b

60 (±0.9) 4.6 (±0.6) 1.3 (±0.3)

TFMF1.2 a,b 10 (±1.3) a,b 33 (±8.3) a a a 6.3 (±0.5) a,b

61 (±27) 4.5 (±1.8) 2.7 (±0.5)

CFMF1.4 a,b 9.2 (±3.3) a,b 22 (±3.7) a a a 5.6 (±2.3) a,b

2.8 (±0.2) 0.9 (±0.5)

PEF-L 39 (±2.0) a 7.1 (±1.8) a,b 28 (±7.8) a a a 3.5 (±0.1) a

64 (±16) 5.5 (±0.2) 1.4 (±0.4)

PEF-M a,b 9.3 (±2.8) a,b 39 (±8.0) a a a 7.8 (±3.1) a,b

61 (±7.1) 2.4 (±0.1) 2.5 (±0.2)

PEF-H a,b 9.6 (±4.3) a,b 29 (±4.7) a a a 7.5 (±0.7) a,b

PEF- 2.6 (±0.0) 1.9 (±0.0)

18 (±13) a 4.8 (±0.2) a 31 (±11) a a a 5.4 (±0.4) a
L/TFMF0.65

PEF- 3.9 (±0.3) 2.1 (±0.6)

39 (±36) a 6.1 (±1.7) a,b 26 (±5.7) a a a 3.8 (±0.4) a
M/TFMF0.65

PEF- 2.2 (±0.2) 1.2 (±0.5)

44 (±2.9) a 5.2 (±2.2) a 24 (±3.4) a a a 5.2 (±0.5) a
H/TFMF0.65

PEF- 58 (±2.6) 2.0 (±0.4) 0.9 (±0.3)

a,b 5.2 (±0.1) a 28 (±7.7) a a a 4.1 (±0.6) a
L/TFMF1.2

PEF- 84 (±5.3) 5.6 (±0.7) 3.0 (±1.1)

b,c 12 (±0.8) a,b 37 (±1.8) a a a 6.4 (±0.5) a,b
M/TFMF1.2
 Methyl
Ethanol 1-Propanol 1-Octene Toluene p-Xylene
Treatments acetate
[ppb] [ppb] [ppb] [ppb] [ppb]
[ppb]

PEF- 52 (±3.4) 3.0 (±0.9) 1.1 (±0.6)

a,b 7.4 (±3.2) a 29 (±2.2) a a a 3.8 (±2.2) a
H/TFMF1.2

PEF- 42 (±12) 2.2 (±0.4) 1.5 (±0.1)

a,b 6.6 (±1.5) a,b 25 (±5.8) a a a 4.1 (±0.1) a
M/CFMF1.4

P1 * * N.S. * N.S. *

SEM 2 3.06 0.34 0.95 0.17 0.13 0.37

a, b, c
Different superscripted letters in the same column indicate statistical
significance between the two means. Values in parentheses following the mean
values of compound indicate the standard deviations. 1 P stands for the statistical
probability. * Refers to a statistical significance of p < 0.05. NS indicates no statistical
significance. 2 SEM abbreviates the standard error of the mean.

Table 4. Retention of in acetone, butanone, 2-hexanone, 2-heptanone, 2-octanone

and 2-nonanone obtained in skim milk after high temperature short time
pasteurization (HTST75 = 75 °C for 20 s; HTST95 = 95 °C for 45 s) and lab-scale
tangential-flow microfiltration (TFMF0.65 = 0.65 µm; TFMF1.2 = 1.2 µm), pilot scale
cross-flow microfiltration (CFMF1.4 = 1.4 µm), pulsed electric field at low (PEF-L: 28
kV/cm; 2805 µs), moderate (PEF-M: 40 kV/cm, 1122 µs), and high (PEF-H: 40
kV/cm, 1571 µs) processing intensities and selected non-thermal combination
treatments using PEF and subsequent the microfiltration.

2- 2- 2- 2-
Acetone Butanone
Treatments Hexanone Heptanone Octanone Nonanone
[ppb] [ppm]
[ppb] [ppb] [ppb] [ppb]

Raw Skim 2.7 (±

a 174 (±51) a 6.8 (±3.1) a 4.9 (±2.5) a 9.7 (±6.5) a 4.2 (±1.1) a
Milk 0.8)

4.0 225 (±23) 9.0 (±1.1) a, 4.8 (±1.7) a, 11 (±1.8) a,

HTST75 a, b a, b b b b 5.5 (±0.8) a
(±0.4)

6.2 344 (±0.7)

HTST95 9.3 (±0.3) b 12 (±2.6) b 16 (±3.6) b 5.9 (±0.3) a
(±1.0) b b

2.4 4.6 (±1.1) a, 8.9 (±1.4)

TFMF0.65 169 (±23) a 2.8 (±1.6) a 3.3 (±0.1) a
(±0.4) a b a, b
 2- 2- 2- 2-
Acetone Butanone
Treatments Hexanone Heptanone Octanone Nonanone
[ppb] [ppm]
[ppb] [ppb] [ppb] [ppb]

2.6 a, 4.9 (±4.0) a,

TFMF1.2 188 (±56) a 3.4 (±1.8) 9.5 (±3.5)
3.7 (±2.7) a
(±0.2) a, b b b a, b

2.3 7.0 (±1.6) a, 3.7 (±0.3) a,

CFMF1.4 140 (±51) a b 6.5 (±0.6) a 5.0 (±2.4) a
(±0.9) a b

1.8 5.7 (±1.5) a, 3.9 (±0.2) a, 7.2 (±1.4)

PEF-L 120 (±32) a b 2.6 (±0.4) a
(±1.0) a b a, b

2.6 a, 5.1 (±2.8) a,

PEF-M 135 (±74) a 7.0 (±0.7) 8.8 (±3.5)
3.5 (±1.9) a
(±1.9) a, b b b a, b

3.1 a, 4.8 (±2.2) a,

PEF-H 147 (±58) a 7.8 (±0.3) 7.9 (±4.2)
5.4 (±0.9) a
(±2.2) a, b b b a, b

PEF- 1.5 5.0 (±0.1) a, 6.0 (±1.2) a, 8.7 (±1.8)

54 (±8.0) a 5.2 (±3.8) a
L/TFMF0.65 (±0.9) a b b a, b

PEF- 1.9 5.7 (±0.2) a, 6.5 (±1.4) a, 8.8 (±1.8)

75 (±40) a 5.4 (±3.6) a
M/TFMF0.65 (±1.1) a b b a, b

PEF- 1.9 6.9 (±0.6) a, 4.7 (±3.1) a, 7.4 (±2.8)

80 (±41) a 4.5 (±0.9) a
H/TFMF0.65 (±1.1) a b b a, b

PEF- 1.5 5.0 (±0.1) a, 7.5 (±0.3)

56 (±18) a 3.5 (±0.3) a 1.9 (±0.5) a
L/TFMF1.2 (±0.4) a b a, b

PEF- 3.0 a, 6.6 (±1.0) a, a,

74 (±26) a 7.6 (±0.4) 12 (±0.0)
5.7 (±0.4) a
M/TFMF1.2 (±0.2) a, b b b b

PEF- 2.2 5.4 (±1.3) a, 4.8 (±1.2) a, 8.3 (±0.9)

84 (±60) a 2.9 (±0.1) a
H/TFMF1.2 (±0.6) a b b a, b

PEF- 2.1
81 (±11) a 8.9 (±0.2) b 3.4 (±0.3) a 8.8 (±2.4) a 4.0 (±0.7) a
M/CFMF1.4 (±0.4) a

P1 * * * * * NS

SEM 2 0.12 8.01 0.38 0.36 0.54 0.23

a, b, c
Different superscripted letters in the same column indicate statistical
significance between the two means. Values in parentheses following the mean
values of compound indicate the standard deviations. 1 P stands for the statistical
probability. * Refers to a statistical significance of p < 0.05. NS indicates no statistical
significance. 2 SEM abbreviates the standard error of the mean.

The aldehydes in skim milk were altered in a different way as compared to ketones
in skim milk. Thermal treatments had a significant impact on 3-methylbutanal levels.
The concentration of 3-methylbutanal was increased significantly upon heat
treatment at 95 °C (p < 0.05). By contrast, microfiltration did not have an impact on
aldehyde concentrations in skim milk. HTST treatments at 75 and 95 °C, PEF and
microfiltration treatments did not have a significant impact on the concentration of
heptanal or decanal (p ≥ 0.05).

Short chain fatty acids were the other group of volatiles analyzed in skim milk. Of
these, hexanoic acid and butanoic acid were detected in raw skim milk, heat-treated,
PEF-treated and microfiltered skim milk. Increased intensity of thermal treatments
resulted in an increased concentration of hexanoic acid and the values were
significantly higher in skim milk subjected to both HTST-75 and HTST-95 (p < 0.05)
than in raw skim milk. Pasteurization at the higher temperature (HTST-95) led to a
significant increase (p < 0.05) in the concentration of butanoic acid, whereas no
impact (p ≥ 0.05) on its concentration was found for skim milk pasteurized at 75 °C.
Moreover, no significant difference (p ≥ 0.05) in hexanoic and butanoic acid
concentrations was observed between raw, PEF, MF, and hurdle treated skim milk.

The concentrations of both ethanol and propanol increased in skim milk following
heat pasteurization at 95 °C (p < 0.05). Contrasting results (p ≥ 0.05) were obtained
for PEF, MF, hurdle processing and HTST-75 treatments, indicating that their
concentrations did not change in skim milk subjected to these processing conditions.
Heat treatment at 95 °C resulted in increased levels of toluene (p < 0.05), while there
was no significant impact by HTST-75, PEF, and MF treatments on this volatile
compound (p ≥ 0.05).

The methyl acetate concentration was significantly higher (p < 0.05) in skim milk
processed at 95 °C. None of the other processes affected the concentration of this
volatile component (p ≥ 0.05). Furthermore, two sulfur compounds, hydrogen sulfide
and dimethyl sulfide, were detected by the SIFT-MS equipment. Hydrogen sulfide
concentrations were found to be at significantly higher levels in HTST-75 and HTST-
95 treated skim milk, reaching the highest levels with the latter treatment (p < 0.05).
The highest concentration of dimethyl sulfide occurred in skim milk that underwent
the HTST-95 treatment. PEF processing and microfiltration did not increase the
content of sulfur compounds.

4. Discussion

For Hunter L, a, b color space values, no significant differences were observed in

lightness, redness, and yellowness for skim milk treated with PEF, HTST and, MF-
1.2 and MF-1.4 (p ≥ 0.05). A significant reduction in L, b and whiteness index values
for skim milk that was microfiltered through a 0.65 µm pore size membrane indicates
that higher cut-off membranes partially retain suspended milk components [28,29].
Pafylias et al. [30] suggested the use of larger pore size membranes of about 1.4
µm to maintain a balance between reduction of bacterial cells and retention of milk
components. However, we did not see any significant differences in color attributes
for skim milk processed through 1.2 and 1.4 µm pore size MF. Silva et al. [31]
compared instrumental color differences between microfiltered and pasteurized skim
milk and found smaller changes in color coordinates of microfiltered skim milk due
to a lack of reactions caused by heating skim milk to pasteurization temperatures.
HTST treatments caused a small increase in lightness values for all the samples,
which may be due to an increase in number of dispersed components as a result of
denaturation of milk proteins at high temperatures [20]. However, the increase in
lightness was not significant (p ≥ 0.05). The Hunter-b value was positive for HTST-
95 treatment indicating yellowness in skim milk. This might be attributed to heat
induced browning reactions in milk between lactose and amino acids [32,33,34].
However, PEF and MF-treated skim milk retained milk color better than heat
treatments as indicated by “not noticeable” total color difference values compared to
raw skim milk. As indicated by ∆E values, there was a lower impact on color of skim
milk produced with the non-thermal technologies, with the exception of the smallest
pore size MF that produced “unacceptable” color values. However, our results are
not in agreement with results of Bermúdez-Aguirre et al. [35] who observed
significant changes in L, a, b values of PEF-treated skim milk. The authors, however,
did not observe any trends in color change and attributed this change partly to the
wearing down of the electrode due to arcing.

Among the 21 volatile compounds analyzed, ketones represented a major class

based on their expected contribution to flavor. Ketones in milk can originate from
different sources. Some of them may be present in raw milk as a result of feed, e.g.,
acetone and 2-butanone, while others may be generated or accumulated as a result
of β-oxidation of saturated fatty acids followed by decarboxylation or by β-ketoacid
decarboxylation [36,37]. A significant increase in concentration of all methyl ketones,
except 2-nonanone, was observed for samples subjected to heat treatment (p <
0.05). Increased levels of methyl ketones in milk due to thermal treatments have also
been reported by other researchers [23,38]. The methyl ketone content of skim milk
treated with either PEF or MF alone was similar to that of raw skim milk (p ≥ 0.05),
indicating that they have a minimum effect on the concentration of lipids in skim milk.
Comparable results (p ≥ 0.05) were obtained by hurdle processing of skim milk using
both PEF and MF, suggesting a non-significant impact of combined processes on
ketones in skim milk.

Another important class of compounds identified and relevant to this study were the
sulfur compounds: dimethyl and hydrogen sulfide. These compounds are formed in
milk as a result of thermal denaturation of milk whey proteins during processing and
are responsible for cooked flavors in milk. Hydrogen sulfide is produced mainly from
Strecker degradation of sulfhydryl groups of sulfur-containing amino acids (i.e.,
cysteine and methionine) in whey proteins. Its production is also attributed to
denatured proteins associated with the milk fat globule membrane and the
rearrangement of a thiazole group during thermal degradation of thiamine [36,39,40].
In our research, the concentration of hydrogen sulfide in raw skim milk was found to
be 1.39 ± 0.68 µg/L. Its concentration was higher when skim milk was pasteurized
at 95 °C for 45 s (5.53 µg/L) as compared to skim milk subjected to 75 °C for 20 s
(4.21 µg/L). Thus, the hydrogen sulfide concentration was found to increase with the
severity of heat treatment. In contrast, PEF treatment applied at the highest intensity,
which is known to generate increased temperature in the treatment chamber to about
65 °C by means of Joule heating, however, did not cause any significant increase in
hydrogen sulfide concentration (2.05 µg/L) (p ≥ 0.05). The PEF-L treatment that
resulted in the longest exposure (2805 µs) of skim milk to electric pulses also did not
cause any significant changes in hydrogen sulfide concentration (0.82 µg/L) when
compared to raw skim milk (p ≥ 0.05). The increase in concentration of sulfur
compounds has been associated with growing denaturation of sulfur containing
polypeptides affected to a large degree by increasing exposure to heat [39,41].
Dimethyl sulfide has been reported to occur naturally in cow’s milk as a result of diet
and at low levels it contributes to milk flavor but at high concentrations it may impart
an off-flavor to the milk [42,43]. It can also be formed as a result of thermal
denaturation of sulfur containing amino acids in milk. Increased dimethyl sulfide
content in milk subjected to heat treatment at pasteurization temperatures and ultra-
high-temperatures has been reported by different research groups [22,44,45]. In this
study, dimethyl sulfide concentration was observed to be significantly higher in skim
milk processed at 95 °C for 45 s (23.6 µg/L) as compared to raw skim milk (6.17 to
17.95 µg/L). However, no significant differences in concentrations of dimethyl sulfide
from those in raw skim milk were observed for milk that underwent HTST-75, PEF,
and MF treatments (p ≥ 0.05). Hurdle processing of PEF with MF also did not cause
any significant changes in concentrations of sulfur compounds in comparison to raw
skim milk (p ≥ 0.05).

Among aldehydes, acetaldehyde concentrations were below the detection limit of

SIFT-MS for raw as well as treated skim milk. Aliphatic aldehydes are produced as
a result of auto-oxidation of unsaturated fatty acids as well as by breakdown of amino
acids in milk [37,40]. No significant differences were found in heptanal and decanal
concentrations among raw skim milk, HTST-treated, PEF-treated, and MF-treated
skim milk (p ≥ 0.05). In contrast, Zhang and others [5] observed significantly
increased heptanal and decanal levels in pasteurized milk in comparison to raw milk
while no significant differences were found in PEF-treated skim milk. However, the
increase in aldehyde concentrations in skim milk upon exposure to heat treatment
at 95 °C, though not significant (p ≥ 0.05) could contribute to significant changes in
aroma of skim milk due to lower threshold values for some aldehydes [44]. 3-Methyl
butanal is a product of Strecker degradation of leucine during the non-enzymatic
browning reaction, which occurs during heat treatment of milk [36,37]. No significant
increase in its concentration was found after non-thermal PEF and MF treatments (p
≥ 0.05). Its presence in skim milk treated at 95 °C was found to be significantly higher
(p < 0.05) than in raw skim milk and in skim milk processed by all other treatments
tested. Contarini and Povolo [23] also observed a positive correlation between
increased levels of 3-methyl butanal and severity of heat treatment, while no
significant differences were found in heptanal concentrations.
Short-chain and medium-chain fatty acids (C-4 to C-14 and some C-16) comprise
45% of total fatty acids in milk fat and are produced in the mammary gland from
acetate and β-hydroxybutyrate, the products of bacterial fermentation in the rumen.
Activation of acetyl to acetyl CoA and its carboxylation to malonyl CoA results in a
stepwise addition of two CH2 groups and hence, increase in fatty acid chain length
from short- to medium-chain fatty acids [46]. Out of these straight-chained, even
numbered fatty acids, butanoic and hexanoic acids were quantified in this study. An
increase in concentration of these two short-chain fatty acids as a result of heat
exposure has been reported by Gandy and others [24]. Significantly higher amounts
of both butanoic acid and hexanoic acid were observed in skim milk treated at 95
°C. However, HTST pasteurization at 75 °C showed a greater increase in hexanoic
acid and, interestingly, the increase was twice the increase observed for the 95 °C
treatment. Increased levels of butanoic and hexanoic acid in heat treated milk may
impart cheesy, sour, and cream notes to milk as classified by Zhang and others [5]
under the category of high olfactometric intensity odorants.

Formation of alcohols by reduction of corresponding carbonyl compounds [29] was

observed for milk stored at refrigeration temperatures, where alcohol concentration
increased as result of reduction of carbonyl compounds [30,47]. HTST-75, PEF, and
MF treatments did not have any significant impact on the concentration of ethanol
and 2-propanol in skim milk. Similarly, Zhang et al. [5] reported no marked change
in concentrations of alcohols subsequent to PEF and HTST processing. However,
at a processing temperature of 95 °C, a marked increase in concentration of both
these alcohols was observed, which indicates some reducing reactions take place
at this temperature. Toluene and p-xylene are considered as products of thermal
degradation of β-carotene [48] and these compounds have been associated with
heat treated milk [24,38,49]. Esters are formed by reaction of carboxylic acids with
alcohols. Methyl acetate was the only ester quantified in this study and its
concentration increased three-fold in HTST-95 treated skim milk. As observed in
similar studies [3,50,51] non-thermal processing had a lesser impact on the
concentration of volatile compounds in skim milk than thermal processing and
indicates that skim milk of better quality results from non-thermal processing. Last
but not least, it is worthy of mention that different scales (lab, pilot plant) of MF
processing equipments (1.2 µm TFMF and 1.4 µm CFMF, respectively) used in this
study indicated comparable effects on color and composition of volatiles in skim milk,
thereby, also substantiating that different designs and materials of the MF membrane
modules had no significant effect on these organoleptical quality parameters.

5. Conclusions

MF (1.2 and 1.4 µm) and PEF alone and in combination (hurdle technology) caused
no significant changes to skim milk color and its volatile compound composition,
thus, the hurdle approach can be regarded as promising for preservation of skim
milk of very good sensory quality. No noticeable color differences from raw skim milk
were observed by application of PEF, and MF with a pore size of 1.2 µm or above.
Smaller pore size microfiltration (0.65 µm) was found to be unsatisfactory when
applied alone or in combination with PEF, resulting in significant changes to skim
milk color attributes. Different scale, membrane materials, and membrane designs
of the MF systems showed no effect on color and volatile compounds. A significant
increase in ketones, fatty acids, hydrocarbons and sulfur compounds was obtained
following heat treatment of skim milk at the higher intensity treatment (95 °C, 45 s)
applied in this study, while lower intensity heat treatment (75 °C, 20 s) also resulted
in increased concentration of selected volatile compounds. However, increased PEF
intensity did not significantly alter concentrations of volatile compounds. These
findings indicate PEF/MF is an emerging technology with potential to produce skim
milk of high sensory quality. Further research on this promising hurdle technology is
envisaged such as the study of enzyme activity and vitamin retention in PEF/MF-
treated skim milk as well as its application for processing other liquid dairy products.

Acknowledgments

The authors are thankful for the financial support by the Dairy Farmers of Ontario
and the Natural Sciences and Engineering Research Council of Canada that both
made the present study possible. Moreover, we are grateful for the supply of raw
milk by the Gay Lea Foods Co-operative Limited and for the use of the pilot scale
heat exchanger and microfiltration units provided by the Guelph Food Technology
Centre.

Author Contributions

The present manuscript was written by Anupam Chugh, who carried out the
experimental work together with Dipendra Khanal under the guidance and
supervision of Markus Walkling-Ribeiro, Milena Corredig, and Mansel W. Griffiths.
The experimental design of the study was devised by Walkling-Ribeiro, also
responsible for the article correspondence, and Griffiths, the principal investigator of
the research project. Due to her expertise in sensory analyses Lisa Duizer, also a
co-investigator of the research project, mentored Chugh and Khanal during the
course of their volatile compound analyses.

Conflicts of Interest

The authors declare no conflict of interest.

References

1. Doan F.J. The color of cow’s milk and its value. J. Dairy Sci. 1924;7:147–153. doi:
10.3168/jds.S0022-0302(24)94004-5. [Cross Ref]
2. Thomas E.L. Trends in milk flavors. J. Dairy Sci. 1981;64:1023–1027. doi:
10.3168/jds.S0022-0302(81)82680-X. [Cross Ref]
3. Sampedro F., Geveke D.J., Fan X., Zhang H.Q. Effect of PEF, HHP and thermal
treatment on PME inactivation and volatile compounds concentration of an orange
juice-milk based beverage. Innov. Food Sci. Emerg. Technol. 2009;10:463–469. doi:
10.1016/j.ifset.2009.05.006. [Cross Ref]
4. Pereda J., Jaramillo D.P., Quevedo J.M., Ferragut V., Guamis B., Trujillo A.J.
Characterization of volatile compounds in ultra-high-pressure homogenized milk. Int.
Dairy J. 2008;18:826–834. doi: 10.1016/j.idairyj.2007.12.002. [Cross Ref]
5. Zhang S., Yang R., Zhao W., Hua X., Zhang W., Zhang Z. Influence of pulsed
electric field treatments on the volatile compounds of milk in comparison with
pasteurized processing. J. Food Sci. 2011;76:C127–C132. doi: 10.1111/j.1750-
3841.2010.01916.x. [PubMed] [Cross Ref]
6. Chugh A., Khanal D., Duizer L., Walkling-Ribeiro M., Griffiths M.W. Effect of non-
thermal pulsed electric fields and tangential-flow microfiltration processing on flavor
and color properties of skim milk; Proceedings of the IFT Annual Meeting and Food
Expo; Chicago, IL, USA. 14 July 2013; Jul 14, 2013. session 074-04.
7. Zimmermann U. The effect of high intensity electric field pulses on eukaryotic cell
membranes: Fundamentals and applications. In: Zimmermann U., Neil G.A., editors.
Electromanipulation of Cells. CRC Press; Boca Raton, FL, USA: 1996. pp. 1–106.
8. Waite-Cusic J.G., Diono B.H.S., Yousef A.E. Screening for Listeria
monocytogenes surrogate strains applicable to food processing by ultra high
pressure and pulsed electric field. J. Food Prot. 2011;74:1655–1661. doi:
10.4315/0362-028X.JFP-11-099. [PubMed] [Cross Ref]
9. Grahl T., Markl H. Killing of microorganisms by pulsed electric fields. Appl.
Microbiol. Biotechnol. 1996;45:148–157. doi: 10.1007/s002530050663. [PubMed]
[Cross Ref]
10. Griffiths M.W., Walkling-Ribeiro M. Implementation of microbially safe foods with
pulsed electric fields. In: Boziaris I.S., editor. Food Microbiology: Novel Food
Preservation and Microbial Assessment Techniques. Science Publishers—Taylor &
Francis CRC Press; Boca Raton, FL, USA: 2014.
11. GEA Filtration. [(accessed on 1 September 2013)]. Available online:
http://www.geafiltration.com/applications/microfiltration_whole_skim_milk.asp.
12. Gésan-Guiziou G. Removal of bacteria, spores and somatic cells from milk by
centrifugation and microfiltration techniques. In: Griffiths M.W., editor. Improving the
Safety and Quality of Milk, Volume 1—Milk Production and Processing. CRC Press;
Boca Raton, FL, USA: 2010. pp. 356–369.
13. Rodríguez-González O., Walkling-Ribeiro M., Jayaram S., Griffiths M.W. Factors
affecting the inactivation of the natural microbiota of milk processed by pulsed
electric fields and cross-flow microfiltration. J. Dairy Res. 2011;78:270–278. doi:
10.1017/S0022029911000367. [PubMed] [Cross Ref]
14. Sobrino-López A., Martín-Belloso O. Enhancing the lethal effect of high-intensity
pulsed electric field in milk by antimicrobial compounds as combined hurdles. J.
Dairy Sci. 2008;91:1759–1768. doi: 10.3168/jds.2007-0979. [PubMed] [Cross Ref]
15. Swanson B.G., Barbosa-Cánovas G.V. Inactivation of Listeria innocua in skim
milk by pulsed electric fields and nisin. Int. J. Food Microbiol. 1999;51:19–30. doi:
10.1016/S0168-1605(99)00069-0. [PubMed] [Cross Ref]
16. Odriozola-Serrano I., Bendicho-Porta S., Martín-Belloso O. Comparative study
on shelf life of whole milk processed by high-intensity pulsed electric field or heat
treatment. J. Dairy Sci. 2006;89:905–911. doi: 10.3168/jds.S0022-0302(06)72155-
5. [PubMed] [Cross Ref]
17. Walkling-Ribeiro M., Rodríguez-González O., Jayaram S., Griffiths M.W.
Microbial inactivation and shelf life comparison of “cold” hurdle processing with
pulsed electric fields and microfiltration, and conventional thermal pasteurisation in
skim milk. Int. J. Food Microbiol. 2011;144:379–386. doi:
10.1016/j.ijfoodmicro.2010.10.023. [PubMed] [Cross Ref]
18. Solah V.A., Staines V., Honda S., Limley H.A. Measurement of milk colour and
composition: Effect of dietary intervention on Western Australian Holstein-Friesian
cow’s milk quality. J. Food Sci. 2007;72:S560–S566. doi: 10.1111/j.1750-
3841.2007.00491.x. [PubMed] [Cross Ref]
19. Nozière P., Graulet B., Lucas A., Martin B., Grolier P., Doreau M. Carotenoids
for ruminants: From forages to dairy products. Anim. Feed Sci. Technol.
2006;131:418–450. doi: 10.1016/j.anifeedsci.2006.06.018. [Cross Ref]
20. Dunkerley J.A., Ganguli N.C., Zadow J.G. Reversible changes in the colour of
skim milk on heating suggest rapid reversible heat-induced changes in casein
micelle size. Aust. J. Dairy Technol. 1993;48:66–70.
21. Browning E., Lewis M., Macdougall D. Predicting safety and quality parameters
for UHT-processed milks. Int. J. Dairy Technol. 2001;54:111–120. doi:
10.1046/j.1364-727x.2001.00021.x. [Cross Ref]
22. Datta N., Elliott A.J., Perkins M.L., Deeth H.C. Ultra-high-temperature (UHT)
treatment of milk: Comparison of direct and indirect modes of heating. Aust. J. Dairy
Technol. 2002;57:221–227.
23. Contarini G., Povolo M. Volatile fraction of milk: Comparison between purge and
trap and solid phase microextraction techniques. J. Agric. Food Chem.
2002;50:7350–7355. doi: 10.1021/jf025713a. [PubMed] [Cross Ref]
24. Gandy A.L., Schilling M.W., Coggins P.C., White C.H., Yoon Y., Kamadia V.V.
The effect of pasteurisation temperature on consumer acceptability, sensory
characteristics, volatile compound composition, and shelf-life of fluid milk. J. Dairy
Sci. 2008;91:1769–1777. doi: 10.3168/jds.2007-0833. [PubMed] [Cross Ref]
25. Lewis M. Improving pasteurized and extended shelf-life milk. In: Griffiths M.W.,
editor. Improving the Safety and Quality of Milk: Volume One, Milk Production and
Processing. CRC Press; Boca Raton, FL, USA: 2010. pp. 277–301.
26. Walkling-Ribeiro M., Noci F., Riener J., Cronin D.A., Lyng J.G., Morgan D.J. The
impact of thermosonication and pulsed electric fields on Staphylococcus aureus
inactivation and selected quality parameters in orange juice. Food Bioprocess
Technol. 2009;2:422–430. doi: 10.1007/s11947-007-0045-7. [Cross Ref]
27. Cserhalmi Z., Sass-Kiss Á., Tóth-Markus M., Lechner N. Study of pulsed electric
field treated citrus juices. Innov. Food Sci. Emerg. Technol. 2006;7:49–54. doi:
10.1016/j.ifset.2005.07.001. [Cross Ref]
28. Sachdeva S., Buchheim W. Separation of native casein and whey proteins using
micrfiltration. Aust. J. Dairy Technol. 1997;52:92–97.
29. Vivekanand V., Kentish S.E., Connor O., O’Connor A.J., Barber A.R., Stevens
G.W. Microfiltration offers environmentally friendly fractionation of milk proteins.
Aust. J. Dairy Technol. 2004;59:186–188.
30. Pafylias I., Cheryan M., Mehaiab M.A., Saglam N. Microfiltration of milk with
ceramic membranes. Food Res. Int. 1996;29:141–146. doi: 10.1016/0963-
9969(96)00007-5. [Cross Ref]
31. Silva R.C.S.N., Vasconcelos C.M., Suda J.Y., Minim V.P.R., Pires A.C.S.,
Carvalho A.F. Acceptance of microfiltered milk by consumers aged from 7 to 70
years. Rev. Inst. Adolfo Lutz. 2012;71:481–487.
32. Pagliarini E., Vernile M., Peri C. Kinetic study on colour changes in milk due to
heat. J. Food Sci. 1990;55:1766–1767. doi: 10.1111/j.1365-2621.1990.tb03625.x.
[Cross Ref]
33. Patton S. Browning and associated changes in milk and its products: A review.
J. Dairy Sci. 1955;38:457–478. doi: 10.3168/jds.S0022-0302(55)95000-1. [Cross
Ref]
34. Boekel M.A. Effect of heating on Maillard reactions in milk. Food Chem.
1998;62:403–414. doi: 10.1016/S0308-8146(98)00075-2. [Cross Ref]
35. Bermúdez-Aguirre D., Fernández S, Esquivel H., Dunne P.C., Barbosa-Cánovas
G.V. Milk processed by pulsed electric fields: Evaluation of microbial quality,
physicochemical characteristics, and selected nutrients at different storage
conditions. J. Food Sci. 2011;76:S289–S299. doi: 10.1111/j.1750-
3841.2011.02171.x. [PubMed] [Cross Ref]
36. Fennema O.R. In: Food Chemistry. 3rd ed. Fennema O.R., editor. CRC Press;
New York, NY, USA: 1996. pp. 254–273.
37. Calvo M.M., de la Hoz L. Flavour of heated milks: A review. Int. Dairy J.
1992;2:69–81. doi: 10.1016/0958-6946(92)90001-3. [Cross Ref]
38. Contarini G., Povolo M., Leardi R., Toppino P.M., Lattiero-Caseario I.S.,
Lombardo V., Salerno V.B. Influence of heat treatment on the volatile vompounds of
milk. J. Agric. Food Chem. 1997;45:3171–3177. doi: 10.1021/jf960849s. [Cross Ref]
39. Hutton J.T., Patton S. The origin of sulfhydrylgroups in milk proteins and their
contributions to “cooked” flavour. J. Dairy Sci. 1952;35:699–705. doi:
10.3168/jds.S0022-0302(52)91658-5. [Cross Ref]
40. Zabbia A., Buys E.M., de Kock H.L. Undesirable sulphur and carbonyl flavour
compounds in UHT milk: A review. Crit. Rev. Food Sci. Nutr. 2012;52:21–30. doi:
10.1080/10408398.2010.487166. [PubMed] [Cross Ref]
41. Al-Attabi Z., D’Arcy B.R., Deeth H.C. Volatile sulphur compounds in UHT milk.
Crit. Rev. Food Sci. Nutr. 2009;49:28–47. doi: 10.1080/10408390701764187.
[PubMed] [Cross Ref]
42. Toso B., Procida G., Stefanon B. Determination of volatile compounds in cows’
milk using headspace GC-MS. J. Dairy Res. 2002;69:569–577. [PubMed]
43. Patton S., Forss D.A., Day E.A. Methyl sulfide and the flavour of milk. J. Dairy
Sci. 1956;39:1469–1470. doi: 10.3168/jds.S0022-0302(56)94872-X. [Cross Ref]
44. Vazquez-Landaverde P.A., Velazquez G., Torres J.A, Qian M.C. Quantitative
determination of thermally derived off-flavour compounds in milk using solid-phase
microextraction and gas chromatography. J. Dairy Sci. 2005;88:3764–3772. doi:
10.3168/jds.S0022-0302(05)73062-9. [PubMed] [Cross Ref]
45. Vazquez-Landaverde P.A., Torres J.A., Qian M.C. Quantification of trace volatile
sulfur compounds in milk by solid-phase microextraction and gas chromatography-
pulsed flame photometric detection. J. Dairy Sci. 2006;89:2919–2927. doi:
10.3168/jds.S0022-0302(06)72564-4. [PubMed] [Cross Ref]
46. Fox P.F., McSweeney P.L., editors. Advanced Dairy Chemistry. 3rd ed. Kluwer
Academic/Plenum; New York, NY, USA: 2003–2006. p. 4.
47. Nursten H.E. The flavour of milk and dairy products: I. Milk of different kinds, milk
powder, butter and cream. Int. J. Dairy Technol. 1997;50:48–56. doi: 10.1111/j.1471-
0307.1997.tb01735.x. [Cross Ref]
48. Mader I. Beta-Carotene: Thermal degradation. Science. 1964;144:533–534.
[PubMed]
49. Valero E., Villamiel M., Miralles B., Sanz J., Martínez-Castro I. Changes in
flavour and volatile components during storage of whole and skimmed UHT milk.
Food Chem. 2001;72:51–58. doi: 10.1016/S0308-8146(00)00203-X. [Cross Ref]
50. Vallverdú-Queralt A., Bendini A., Barbieri S., di Lecce G., Martin-Belloso O.,
Toschi T.G. Volatile profile and sensory evaluation of tomato juices treated with
pulsed electric fields. J. Agric. Food Chem. 2013;61:1977–1984. doi:
10.1021/jf3051126. [PubMed] [Cross Ref]
51. Sampedro F., Geveke D.J., Fan X., Rodrigo D., Zhang Q.H. Shelf-life study of
an orange juice-milk based beverage after PEF and thermal processing. J. Food Sci.
2009;74:S107–S112. doi: 10.1111/j.1750-3841.2009.01068.x. [PubMed] [Cross
Ref]

VI. Quality attributes and consumer acceptance of new

ready-to-eat frozen restructured chicken
J Food Sci Technol. 2015 May; 52(5): 2869–2877.
Marcio Aurelio de Almeida, Nilda Doris Montes Villanueva, José Ricardo Gonçalves,
and Carmen J. Contreras-Castillo

Abstract

The aim of the present study was to develop a new restructured product, cooked
and frozen ready-to-eat product that was prepared with boneless chicken meat
(breast and drumstick) and mechanically separated chicken meat (MSCM). Non-
meat ingredients, such as transglutaminase (TG) and egg albumin powder, were
tested to obtain a better strength of adhesion between the meat particles. Five
formulations for restructured chicken were developed as follows: T1 (1 %
transglutaminase), T2 (1 % transglutaminase and 15 % MSCM), T3 (1 % egg
albumin powder), T4 (1 % egg albumin powder and 15 % MSCM) and T5 (1 %
transglutaminase, 1 % egg albumin powder and 15 % MSCM). The results of the
experiment showed a greater luminosity (L*) in the treatments with TG (T1) and
albumin (T3). The treatments without MSCM (T1 and T3) presented significantly
lower mean values for redness (a*) when compared to treatments with MSCM (T2,
T4 and T5) (p ≤ 0.05). No significant differences were noted between the treatments
(p  ≥ 0.05) when analyzing the percentage of total saturated fatty acids (SFA),
polyunsaturated fatty acids (PUFA) and cholesterol content. Consumer testing
showed a high acceptance of the restructured products in all evaluated attributes.
Similarly, with regard to the purchase intention, consumers mostly expressed that
they would probably or certainly buy the products, for treatments T1, T2, T3 and T5.
Moreover, the meat cuts with no commercial value, can transform into new ready-
to-eat products that have a high probability of success in the market.

Keywords: Processed food, Sensory analysis, Transglutaminase, Albumin,

Mechanically separated chicken meat

Introduction

The poultry production chain is of utmost importance to the Brazilian economy, with
a total production of thirty million tons of chicken meat in 2012. Annual Brazilian
chicken exports exceed 3.9 million tons (1.57 million tons of whole chickens and 2.2
million tons of meat cuts), and annual Brazilian per capita chicken consumption is
40.8 kg. These productivity and consumption levels make Brazil the largest exporter
of poultry meat and the world’s second largest poultry consumer market.

Special and industrialised cuts (represented by sausages and ready-to-eat meals)

are the products that drive the expansion of chicken consumption, and they are
designed for the consumer who seeks new ready-to-eat food options to reduce the
time spent on food preparation (Mielnik et al. 2002; Perlo et al. 2006). This need has
demanded adjustments from the poultry industry for the development of ready-to-
eat products that are easy to prepare and that meet consumer requirements while
offering attractive prices compared to other types of meat (Contreras-Castillo et al.
2008; Keeton 2001). Therefore, one aim of the meat processing industry is to
develop quality products with a higher added value to accompany the changes in
the patterns of consumption, which require products with lower processing costs with
positive attributes related to food safety, practicality and appearance. An illustration
of this market is the wide range of options for processed meat, as represented by
sausages and bologna sausages (Arora and Kempkes 2008; Pearson and Gillett
1996).

Formed products stand out among those offered by the meat industry. These
products are moulded in several ways from whole muscles or parts with the purpose
of using available muscle pieces to add a higher commercial value (Barbut 2002).
However, the formed product must have similar characteristics to those of the intact
muscle, which is a feature desired by consumers and the main obstacle to
transforming secondary parts of the carcass into formed products.

Salt and tumbling are traditionally required to produce formed meat, thus allowing
for the extraction of myofibrillar proteins from the meat. This process (salt + tumbling)
reduces firmness and increases juiciness, but the addition of salt leads to an
increased amount of sodium in the food. To meet the market’s interest for reducing
salt content, non-meat ingredients are added to these products to improve the
functional properties of the natural proteins as well as to increase its water-holding
capacity and gel formation, which results in products with enhanced juiciness,
structure and texture that are similar to those of intact muscles (Barbut 2002;
Dimitrakopoulou et al. 2005; Lu and Chen 1999).

The gelling ability and physical properties of proteins are important for processed
foods (Lanier, 1998). Egg albumin (EA), as an example of isolated proteins, has
been used as functional ingredient in ground, formated and emulsified products due
to meat binding properties (Tabilo-Munizaga and Barbosa-Canovas 2004) and
positive effect on texture. However, other studies on EA effect have been
contradictory regarding its binding and textural properties (Carballo et al., 1995).
Other authors (Lu and Chen 1999) showed EA powder is an excellent non-meat
protein because its application increased the binding strength of chicken meat and
probably contributed to the formation of a harder protein matrix (Carballo et al.,
1995). Its addition causes minor changes in the color of restructured meat (Lu and
Chen 1999).

Transglutaminase (TG) showed capacity to cross-link between different food

proteins, peptides and some other primary amines (Payne, 2000). The microbial
transglutaminase (MTG) is an enzyme that catalyzes covalent cross-link formation
of food proteins (Carballo et al., 2006). When the ε-amino group of lysine residues
in proteins can mediate as an acyl-acceptor, both intra-and intermolecular ε-(γ-
glutamyl) lysine cross-links are formed between food proteins (Kuraishi et al., 1997).
These TG are active within the temperature range of about 0–60 °C with an optimal
activity around 50–55 °C (Motoki and Seguro, 1998). This enzyme plays an
important role in food processing due to its aggregation and gelation properties and
can improve of rheological properties of food proteins (Carballo et al., 2006; Lennon
et al., 2010). Comercial microbial TG can be mass produced by aerobic fermentation
procedures and show stability between pH 5–9 (Boles, 2011 and Kuraishi et al.,
1997). This MTG could be a cold-binder agent in restructured beef, chicken, fish,
and other meats and low sodium restructured meats and generally used as a liquid
or as a powder application. After application in both ways, the product is refrigerated
for 4–24 h to allow the chemical reaction to occur. TG binding results from cross-
linking myosin and actin (Ramirez-Suarez and Xiong, 2003; Tammatinna et al. 2007)
and it has also been used to improve texture or mechanical properties of products
that contain soluble proteins for reaction substrates (Boles, 2011). The addition of
this enzyme does not modify appearance, color and flavor of chicken restructured
products (Tseng et al., 2000).

Consumption of poultry meat and poultry meat products is currently growing and
increased production of cut-up and processed meat has provided considerable
quantities of parts with lower commercial value. The mechanically separated meat
(MSM) is obtained by mechanical grinding and separation of bones, carcasses or
parts of carcasses of poultry, beef and swine intended for the preparation of specific
meat products. Poultry slaughterhouses produce a considerable amount of residues
and the alternative for this (e.g., carcasses, back and neck) is mechanically deboned
chicken meat (Perlo et al., 2006). MSM of poultry can be used to process several
products, such as sausages, mortadella, nuggets, cooked hamburgers, meatballs
and other cooked meat products (Field, 1998). The MSM, particularly, is used in the
manufacture of emulsion products where the emulsifying capacity is the most
important property for meat products. Another property of MSM, the water holding
capacity (WHC), is important because MSM retains water within the matrix due to
the high pH, presence of mineral from bones, skin and collagen, in addition to the
cooking and freezing processes used (Al-Najdawi and Abdullah 2002).

Other reports (Serrano et al., 2004; Canto et al., 2014) showed that the products
were mechanically suitable (meat particle binding) for handling in the raw state, but
when used in fresh and cooked products, water and binding properties were
inadequate.

The current challenge of the meat industry is to achieve a cost reduction by using
meat cuts of no commercial value, such as the dorsum back of chicken, to obtain
products of a higher commercial value. Therefore, the present study aimed to
develop a new, restructured, cooked and frozen ready-to-eat product that has a
similar sensorial quality to that of the intact chicken fillet using boneless breasts,
thighs, breast shavings, thigh shavings, mechanically separated chicken meat
(MSCM) and non-meat ingredients such as transglutaminase (TG) (which has the
ability to improve the physical properties of the food and promote the adherence of
meat pieces) and egg albumin.

Material and methods

Raw material

Furcula bones obtained from the boning of the breast and bones with meat trimmings
from the back and neck, without skin and following the maximum removal of visible
fat, were used to produce MSCM from recently slaughtered poultry at the Fricock
slaughterhouse (Rio Claro, São Paulo). These bones (back, neck and furcula bones)
were frozen and sent to the pilot plant of the University of São Paulo in Pirassununga,
SP on the same day for the removal of the MSCM, which was then placed in a
deboning machine (model HT 250; HIGH TECH SOLUÇÕES, Chapecó, Santa
Catarina) with an axis that has a setting at the tip to determine the size of the bone
fragments in the mass. In this experiment, the setting at the output of the machine
was standardised for all the extractions.

After the extraction, the MSCM was manually homogenised, packed in polyethylene
bags, placed in an ultra-fast freezer at−40 °C and transported frozen to the pilot plant
of the Laboratory of Meat Quality and Processing of University São Paulo/ESALQ in
Piracicaba, SP to process the restructured product. Boneless and skinless poultry
meat (breast and drumstick), which originated from the same slaughterhouse from
animals slaughtered the previous day (approximately 42 days of age), was
transferred under refrigeration to the laboratory’s pilot plant for the processing of
restructured products.
Ingredients for increasing the strength of adhesion (transglutaminase and egg
albumin)

Activia BP® transglutaminase (Ajinomoto do Brazil Ind. e Com. de Alimentos Ltda,

São Paulo, Brazil) was used, and pasteurised dehydrated egg whites with 80 %
albumin (Salto’s Alimentos Ltda, Salto, São Paulo, Brazil) were used as the source
of albumin.

Formulation

The following non-meat ingredients were used for the restructured product: 0.15 %
onion powder, 0.10 % garlic powder, 0.03 % white pepper, 0.20 % sugar, 0.80 %
sodium lactate, 0.25 % sodium erythorbate, 0.5 % smoke flavour, 1.5 % salt and
10 % water. The non-meat ingredients were kept constant for all treatments. The
amounts of MSCM, TG and albumin varied according to each treatment as follows:
T1 and T3 had 42.74 % chicken breast, 42.74 % deboned drumsticks and 13.53 %
non-meat ingredients with 1 % transglutaminase added to T1 and 1 % egg albumin
powder added to T3; T2 and T4 had 35.74 % chicken breast, 35.74 % deboned
drumsticks, 13.53 % non-meat ingredients and 15 % MSCM with 1 %
transglutaminase added to T2 and 1 % egg albumin powder added to T4; and T5
had 34.74 % chicken breast, 34.74 % deboned drumsticks, 13.53 % non-meat
ingredients, 15 % MSCM, 1 % transglutaminase and 1 % egg albumin powder.

Restructured preparation

The final cleaning for the removal of pieces of cartilage and excess visible fat from
the chicken meat (drumstick and breast) was performed at the Laboratory of Meat
Quality and Processing. The meats were ground separately in a grinder (HOBART
4B22–2) using a #12 disk to grind the thighs and drumsticks and a #3 disk to grind
the still-frozen, cubed MSCM. To reduce the impact on colour change resulting from
the addition of MSCM, chicken breast was added at a ratio of 1:1 during grinding.
The restructured product was prepared in a mixer for a mixing period of 5 min in the
following order: ground chicken meat, MSCM, seasoning, treatment ingredients, salt
(to lower the extraction of myofibrillar proteins) and sodium erythorbate. The
products were then packed in nylon polyethylene bags using a manual packaging
machine (17 cm × 12 cm), producing portions of 100 g each, which were then
vacuum sealed.

Thermal treatment

The pre-packaged restructured products were subjected to cooking in water (82 °C)
until an internal temperature of 82 °C was achieved. This process was designed in
order to ensure the microbial inactivation standard proposed by Thippareddi and
Sanchez (2005), as well as guarantee the appropriate sensorial and cooking
properties. After the heating and retention periods, the products were then cooled to
32 °C under running water and immediately placed in a freezer at−18 °C. The
internal temperature was recorded using PT-100 sensors, installed both in the
product cold spot and in the processing environment, which were coupled to a
datalogger (My PCLab Datalogger; NOVUS Produtos Eletrônicos, Porto Alegre, Rio
Grande do Sul). To obtain the thermal destruction curve, the following parameters
were set.

Experimental design

To collect the physicochemical data, a completely randomised design was used,

which considered each processing as a replication. The five experimental treatments
(formulations shown in Table 1) were as follows: T1 (1 % transglutaminase), T2 (1 %
transglutaminase and 15 % MSCM), T3 (1 % egg albumin powder), T4 (1 % egg
albumin powder and 15 % MSCM) and T5 (1 % transglutaminase, 1 % egg albumin
powder and 15 % MSCM). Four replications of each treatment were performed, and
the following analyses were performed to determine the instrumental colour, pH, fatty
acid profile, cholesterol content and microbiological evaluation.

Table 1 Colour CIELab* (parameters L* and a*) and pH for each treatment

Treatments L* a* pH

T1 68.72b ± 2.11 3.81c ± 0.79 6.34b ± 0.07

T2 62.69e ± 3.48 6.48ab ± 1.18 6.40ª ± 0.07

T3 72.24a ± 1.95 3.72c ± 0.61 6.30c ± 0.05

T4 65.41c ± 3.69 6.06b ± 1.02 6.38ª ± 0.05

T5 63.89d ± 4.26 6.73a ± 1.17 6.39a ± 0.09

Results are the mean ± standard deviation. The means followed by the same
lowercase letters in the columns comparing parameters L*, a* and pH among
treatments do not differ according to Tukey’s test (p > 0.05)

T1 Restructured with transglutaminase; T2 Restructured with MSCM +

transglutaminase; T3 Restructured with egg albumin powder; T4 Restructured with
MSCM + egg albumin powder; T5 Restructured with MSCM + transglutaminase +
egg albumin powder

To generate sensory data, a Latin Square experimental design was used to control
the effects of the sample presentation order. Consumers assessed the acceptance
of the colour, flavour, tenderness, juiciness and overall quality of the restructured
products as well as their intention to purchase the products.

Physicochemical and microbiological analysis

After cooking and freezing, the samples from the five restructured product
formulations were thawed in a commercial refrigerator for 12 h (5 ± 2 °C) so that the
following analyses could be performed at room temperature. The analysis was
performed on five samples from each treatment after 48 h of freezing with five
readings for each sample.

Instrumental colour

The instrumental colour was determined by a colorimeter (Konica Minolta, CR-400

chromameter, Mahwah, New Jersey, USA) using the following parameters: L*
(luminosity) and a* (redness) (CIELAB). The parameters were calibrated in a
standard white porcelain container with Y = 93.7, x = 0.3160 and y = 0.3323 as well
as a measurement area of 8 mm in diameter, an observation angle of 2° and a C
illuminant.

pH

The pH was determined directly on samples using a pH metre (Oakton pH 300,

series 35618, Vernon Hills, Illinois, USA) with automatic temperature compensation
and a glass penetration electrode (Digimed, São Paulo, Brazil). The analysis was
performed on five processed samples of each treatment after 48 h of freezing, using
five readings for each sample.

Fatty acid profile

Lipid extraction was performed according to the method described by Folch et al.
(1957). The fatty acids were converted into fatty acid methyl esters using the method
described by Hartman and Lago (1973). The fatty acid profile was determined by
high-resolution gas chromatography (GC) using a gas chromatograph (HP 5890)
equipped with a SUPELCO SP-2560 capillary column (100 mm × 0.25 mm) coupled
to a flame ionisation detector. The temperature program was set as follows: 130 °C
(1.0 min) to 170 °C (6.5°/min), 170 °C to 215 °C (2.75 °C/min), 215 °C (12 min),
215 °C to 230 °C (40°/min) and 230 °C (6 min). The injector and detector
temperatures were 270 °C and 280 °C, respectively. The samples (0.3 μl) were
injected by the direct injection technique. Saturated and unsaturated fatty acids
containing 6, 8, 10, 12, 14, 15, 16 (cis and trans), 17, 18 (cis and trans), 20, 22 and
24 carbon atoms were identified by comparison with the data obtained for the GC of
authentic methylated standards eluted under the same conditions.

Cholesterol content

The cholesterol content was measured by direct saponification with 2 % KOH in

absolute ethanol (Bragagnolo and Rodriguez-Amaya 2003; Jiang et al. 1991).
Following the extraction, the samples were dried under nitrogen gas and
resuspended in 2 ml of petroleum ether, and 5 or 10 μl was injected into a Shimadzu
liquid chromatography with an automated injector that was coupled to a photodiode
array detector at 210 nm and a C18 reversed-phase column (250 mm × 4.6 mm) with
a particle size of 5 μm. The mobile phase consisted of acetonitrile/isopropanol in
isocratic mode.

Texture profile analysis (TPA)

The TPA of the samples was performed at room temperature (22 °C) using a
TA.XTA2i texture analyser (Stable Micro Systems, Godalming, UK) equipped with a
cylindrical probe with a 50 mm diameter. This procedure involved cutting slices
approximately 2.5 cm thick, which were then compressed twice to 30 % of their
original height. Force-time curves were recorded at a crosshead speed of 2 mm/s.
Hardness (N) and chewiness (Ncm) were evaluated 48 h after processing.

Thiobarbituric acid reactive substances (TBARS)

The TBARS values were determined in duplicate using the extraction method
described by Vyncke (1970) and Jorgensen and Sorensen (1996), with
modifications. The absorbance was measured at 532 nm and 600 nm by a
spectrophotometer (Shimadzu, UV–Vis mini 1240, Chiyoda-ku, Tokyo, Japan). The
difference (A532 nm–A600 nm) was used as the absorbance value, corrected for
turbidity. The results were calculated from the standard curve of TEP and expressed
as mg of malonaldehyde per kg of meat (MDA/kg). The TBARS value determination
was performed 48 h after processing.

Microbiological evaluation

Microbiological evaluations were made after the samples were cooked to verify the
hygienic quality of the sample processing according to the limits specified by
Brazilian law for raw, cooked, cooled or frozen meat products as follows: coliforms
at 45 °C/g (5 × 103 CFU/g), sulphite-reducing clostridia at 46 °C/g (3 × 103 CFU/g)
and Salmonella sp. (absence in 25 g). Analyses for Staphylococcus aureus and
Listeria sp. as well as the total count of mesophilic bacteria were also performed.
Microbiological analyses were performed after cooking and 48 h of storage at−18 °C.

Sensory analysis

Consumer panel

Habitual chicken fillet and/or nugget consumers of different ages and from different
regions were recruited from the Institute of Food Technology (ITAL) in Campinas,
Brazil. Among the 52 consumers recruited, 69 % were female and 31 % were male,
with ages ranging from 21 to 60 years. The frequency of chicken fillet and/or nugget
consumption of the recruited consumers was as follows: 48.10 % consumed chicken
fillet and/or nugget every 15 days, 23.10 % consumed chicken fillet and/or nugget
once a week, 19.20 % consumed chicken fillet and/or nugget twice a week, and
9.60 % consumed chicken fillet and/or nugget at least 3 times a week. To determine
whether the addition of MSCM, albumin or transglutaminase affected the sensory
characteristics of the restructured steaks, an acceptance test was performed.

Consumer testing

The sensory tests were performed in the Sensory Analysis Laboratory (LAFISE),
ITAL. The consumers were placed in individual tasting booths, where they received
instructions on the use of the scale, the nature of the products and the type of
evaluation to be carried out. The five grilled restructured samples, which were served
in a monadic way according to the sample presentation order, were evaluated under
white light on disposable white plastic plates coded with random three-digit numbers.
The respondents were instructed to cleanse their pallets with mineral water and
unsalted crackers before each sample evaluation.

After the colour acceptance evaluation, the consumers were requested to taste the
product and evaluate how much they liked or disliked each sample with respect to
the odour, flavour, tenderness, juiciness and overall acceptance by using a nine-
point hedonic scale (1 = disliked extremely, 5 = neither liked/nor disliked and 9 = liked
extremely). Finally, the consumers evaluated their purchase intention of the tested
products by using a five-point structured scale (1 = certainly will not buy, 3 = may or
not may buy and 5 = certainly will buy).

Statistical analysis

The physiochemical measurements of the experimental treatments were analysed

using a one-way ANOVA (p ≤ 0.05). The acceptance responses of the evaluated
attributes were analysed using a three-way ANOVA (p  ≤ 0.05), which considered the
effects of the consumers, sample presentation order and samples. To evaluate the
differences in physicochemical characteristics and acceptance among the samples,
paired comparisons of the means were carried out using Tukey’s HSD test (p ≤ 0.05).
The statistical data analysis was performed using SAS and StatisticaTM software
(Statsoft Inc., Tulsa, Oklahoma, USA).

Results and discussion

pH and colour

In the pH analysis, treatments containing MSCM (T2, T4 and T5) presented

statistically significant differences (p  ≤ 0.05) compared to treatments without MSCM
(T1 and T3). However, the differences observed in the pH values did not represent
changes in this parameter (Table 1).

The ANOVA results for luminosity (L*) and redness (a*) showed significant
differences (p ≤ 0.05) among treatment means (Table 1). The restructured products
without MSCM (T1 and T3) presented the highest luminosity and the lowest redness
values in comparison to the restructured products with MSCM (T2, T4 and T5), which
presented the lowest luminosity and highest redness values. Thus, MSCM addition
affected the colour parameters. These results suggested that the addition of MSCM
should occur in small proportions to prevent the colour from darkening in restructured
products. The darkness caused by the addition of MSCM was due to the release of
bone marrow and large amounts of haemoglobin when the bones were broken in the
production of MSCM. The presence of haemoglobin in this meat (MSCM) confers a
brownish colour after cooking and increased redness (Contreras-Castillo et al. 2008;
Hecer and Sozen 2011; McMindes and Siedler 1988).

When comparing the treatments without MSCM (T1 and T3), the formulation
containing egg albumin (T3) presented higher values of luminosity than the
formulation containing transglutaminase (T1), which may be due to the water-
retaining and gel-forming properties of albumin, thereby conferring a lighter colour
to the product, as also reported by Lu and Chen (1999). Thus, due to this
characteristic, it was expected that the addition of egg albumin powder in
combination with MSCM would produce restructured products with redness and
lightness close to a product without MSCM. In this study, however, there was greater
whitening (p < 0.05) when only egg albumin powder (T3) was added and less
luminosity when egg albumin was added in combination with MSCM. Thus, further
studies will be required to optimise the best combination of egg albumin powder and
MSCM.

Fatty acid profile, cholesterol content and thiobarbituric acid reactive

substances (TBARS)

For the analysis of the total percentage of saturated fatty acids (SFAs),
polyunsaturated fatty acids (PUFAs) and cholesterol, no statistically significant
differences were noted among treatments (p ≥ 0.05), thereby verifying that the
addition of MSCM did not negatively affect the fatty acid profile of the restructured
products in T2, T4 and T5 when compared to the restructured products based on
the drumstick and breast fillet (T1 and T3). The mean values of SFAs, PUFAs and
cholesterol ranged from 28.31 to 30.57 %, 68.94 to 71.26 % and 65.04 to
76.84 mg/100 g, respectively (Table 2). These results were similar to values
indicated by the United States Department of Agriculture (USDA) for “Chicken,
roasting, light meat, meat only, cooked, roasted”, which prescribes values of 69.4 %
for PUFAs, 30.6 % for SFAs and 74 mg/100 g for cholesterol. Studying the quality of
chicken meat under different rearing systems, (Aguiar et al. 2008) found SFA and
PUFA values of 36.3 % and 62.9 %, respectively, in fillets from chickens reared
using the conventional method. Therefore, the results showed that the fatty acid
profile and cholesterol content of the restructured products were within acceptable
levels for products made with chicken meat. This result is favourable because the
food industry aims to develop new products with low levels of SFAs and higher
contents of PUFAs to increase the nutritional value of food.

Table 2. Values of fatty acids, cholesterol content and TBA in the experimental
treatments
 Fatty acids (%) Cholesterol
Treatments TBARS (mg MDA/kg)
Saturated Polyunsaturated (mg/100 g)

29.70ª 65.37ª
T1 69.78ª (±1.72) 0.34a (±0.19)
(±1.64) (±12.90)

28.85ª
T2 70.64ª (±1.68) 74.70ª (±7.68) 0.24a (±0.10)
(±2.08)

28.63ª
T3 70.31ª (±1.61) 65.04ª (±7.58) 0.23a (±0.06)
(±1.25)

28.31ª
T4 71.26ª (±1.26) 70.74ª (±8.67) 0.20a (±0.10)
(±1.48)

30.57ª
T5 68.94ª (±3.13) 76.84ª (±7.97) 0.22a (±0.10)
(±3.07)

Overall 29.21
70.19 (±1.94) 70.54 (±9.69) 0.25 (±0.19)
mean (±1.97)

Results are the mean ± standard deviation. The means followed by the same
lowercase letters in the columns comparing fatty acids, cholesterol and TBA among
treatments do not differ according to Tukey’s test (p > 0.05)

T1 1 % transglutaminase; T2 1 % transglutaminase and 15 % MSCM; T3 1 % egg

albumin powder; T4 1 % egg albumin powder and 15 % MSCM; T5 1 %
transglutaminase, 1 % egg albumin powder and 15 % MSCM

Regarding the lipid oxidation of the restructured products, non-significant differences

(p ≥ 0.05) were found among the treatments (Table 2). This analysis was conducted
to ascertain the impact of MSCM addition on the oxidation of the restructured
products after cooking. The results indicated that MSCM addition did not adversely
affect the sensory quality with respect to rancidity. These results showed low values
of TBARS, indicating that the product was preserved while maintaining its distinctive
flavour. Rhee et al. (2005) verified a correlation between an increase in the TBARS
value and the perception of off-flavours by consumers. Moreover, Campo et al.
(2006) defined the limiting threshold with respect to beef lipid oxidation as 2.0 mg
malonaldehyde/kg sample, which was a relatively high value compared to the value
found in the present study (0.22–0.34 mg MDA/kg).

Thus, the results showed that removing excess fat and skin in the production of
MSCM from drumsticks and breasts for the preparation of restructured products
reduced TBA values, which produced higher oxidative stability. This fact was verified
by the consumer sensory evaluation results (Table 2), in which the smell and flavour
attributes had high acceptance, and no negative comments were made about rancid
odour or taste with regard to the restructured products.

Texture profile analysis (TPA)

The use of TG alone or in combination with egg albumin (T1, T2 and T5) produced
restructured products with significantly higher values (p ≤ 0.05) of hardness and
chewiness when compared to restructured products containing egg albumin powder
(T3 and T4) (Table 3).

Table 3. Texture profile analysis of hardness and chewiness in the

experimental treatments

Texture Profile Analysis

Treatments
Hardness (N) Chewiness

T1 36.91a (±8.09) 2,544.45a (±711.41)

T2 32.64ab (±6.94) 2,258.90a (±503.22)

T3 25.93b (±5.66) 1,644.54b (±542.02)

T4 25.06b (±2.42) 1,682.84b (±375.78)

T5 32.96ab (±6.52) 2,337.74a (±548.22)

Results are the mean ± standard deviation. The means followed by the same
lowercase letters in the columns comparing the texture profile analysis among
treatments do not differ according to Tukey’s test (p > 0.05)

T1 1 % transglutaminase; T2 1 % transglutaminase and 15 % MSCM; T3 1 % egg

albumin powder; T4 1 % egg albumin powder and 15 % MSCM; T5 1 %
transglutaminase, 1 % egg albumin powder and 15 % MSCM

As the objective of this study was to obtain restructured products with texture
characteristics similar to those of a chicken fillet, the results showed that addition of
the TG enzyme refined the MSCM by cross-linking actomyosin (Akamittath and Ball
1992). This result corroborated the results described by Keeton (2001), who reported
that TG influences the texture of meat products, and these authors also stated that
TG is used to improve the rheological properties of food and can be applied in
combination with other ingredients, such as salt, phosphates and alkali curing.

Microbiological evaluation
The results from the analyses of the processed products with all treatments were in
agreement with Brazilian legislation. The results indicated the absence of
Salmonella and Listeria sp. Moreover, sulphite-reducing clostridia
(<1.0 × 10 CFU/g), coliforms at 45 °C (<3.0 MPN/g) and Staphylococcus aureus
(<1.0 × 102 CFU/g) were not detected. With regard to total aerobic mesophiles,
scores between 1.0 × 10 and 1.9 × 102 CFU/g were observed, indicating that the heat
treatment process at 85 ° C for 34 min (Fig. 1) had the desired effect. Further, the
calculated microbial reduction, based on the logarithmical reduction proposed by
Thippareddi and Sanchez (2005; inactivation of at least seven logarithm cycles of
Salmonella spp.) also confirms the final product safety.

Fig. 1 Heat treatment parameters of the restructured product. Red line represents
the temperature of the cooking water, and the green line represents the temperature
inside the restructured product

Sensory analysis

The analysis of variance (ANOVA) of acceptance data (Table 4) showed non-

significant differences (p ≥ 0.05) among the mean values of acceptance for colour
(p = 0.088345), odour (p = 0.1236), flavour (p = 0.0710), tenderness (p = 0.1170),
juiciness (p = 0.6165) and overall quality (p = 0.1174). As shown in Table 4, all
attributes had a mean acceptance of 7 (moderately liked), with the exception of
colour, which obtained a mean acceptance of 6.3 (slightly liked), thereby indicating
that the restructured poultry products had a high acceptance among the consumers
who evaluated the product. The reason for the lower scores for colour acceptance
was likely the combination of the meat ingredients (drumsticks, chicken breast and
MSCM) in the formulation of the restructured product, which gave it a slightly non-
uniform colour. The consumers in the present study were accustomed to eating
chicken products as nuggets, fillet and hamburger. In the case of chicken nuggets,
the coating masks the colour of the product, and in the case of chicken hamburger,
the greater comminution of the chicken meat with fat gives the product a more
uniform colour. It is interesting to note that among all the attributes assessed, there
were no significant differences between the acceptance of the restructured products
containing MSCM and those without MSCM, thereby suggesting that the MSCM did
not alter the sensory characteristics of the restructured products. This result
indicated that the meat-processing industry would be able to produce ready-to-eat
products for the consumer market that are safe, nutritious and highly accepted in
addition to having a high commercial value and considerably lower costs relative into
currently available products.

Table 4. Acceptance values for colour, odour, flavour, tenderness, juiciness,

overall quality and purchase intention for the five restructured chicken meat
product formulations

Overall
T1 T2 T3 T4 T5
Average

Overall
7.1ª ± 1.3 6.7a ± 1.6 6.8ª ± 1.5 7.1ª ± 1.2 6.8ª ± 1.5 6.9ª ± 1.4
Quality

Colour 6.4ª ± 1.6 5.9ª ± 1.8 6.6ª ± 1.6 6.4ª ± 1.8 6.4ª ± 1.9 6.3ª ± 1.8

Odour 6.9ª ± 1.3 6.6ª ± 1.6 6.9ª ± 0.9 7.0a ± 1.2 7.1ª ± 0.9 6.9ª ± 1.2

Flavour 7.1ª ± 1.2 6.6ª ± 1.6 6.9ª ± 1.5 7.1ª ± 1.2 7.0a ± 1.3 6.9ª ± 1.4

Tenderness 7.2ª ± 0.9 6.7ª ± 1.5 7.0a ± 1.2 7.2ª ± 1.1 7.1ª ± 1.2 7.0ª ± 1.2

Juiciness 6.9ª ± 1.1 6.7ª ± 1.7 6.8ª ± 1.3 7.0a ± 1.3 6.7ª ± 1.5 6.8ª ± 1.4

Purchase
3.9ª ± 1.1 3.3bc ± 1.3 3.8ab ± 1.2 3.7ab ± 1.2 3.7ab ± 1.0
intention

The means followed by the same letters in the row comparing the acceptance among
treatments do not differ according to Tukey’s test (p > 0.05)

T1 Restructured with transglutaminase; T2 Restructured with MSCM +

transglutaminase; T3 Restructured with egg albumin powder; T4 Restructured with
MSCM + egg albumin powder; T5 Restructured with MSCM + transglutaminase +
egg albumin powder

In restructured chicken meat the substitution of 15 % of chicken meat (breast and

drumstick) by MSCM results in a reduction of 14 % (approximately) in the final price
of the product, because the price of kg of breast and drumstick for the industry is
US$ 1,68 and US$ 1,88 respectively and the cost of MSCM elaborated with neck
and back (including furcula bone) is US$ 0,20. Many authors (Ozkececi et al 2003,
Guerra-Daros et al. 2005) describe the benefits of using MSCM in the formulation of
industrialized meat products because of the low prices when compared with other
kinds of meat.

Purchase intention

The results of the ANOVA showed significant differences (p  ≤ 0.05) among the mean
purchase intention values of the restructured products. The results showed that most
of the treatments obtained mean purchase intention values of approximately 4
(probably will buy), indicating that the probability of purchase for the restructured
products was high, except for treatment T2, which obtained an average of 3.3 (may
or not may buy). The lower purchase intention for treatment T2 may be associated
with the addition of MSCM because MSCM gave the product a darker colour, as
evidenced by the instrumental analysis (Table 1), which showed that treatment T2
had the lowest average luminosity (L*) among the all treatments. These results
indicated that colour is an important attribute for consumers and may influence their
purchase intention of these products.

Conclusions

The results of the present study showed that transglutaminase and egg albumin
powder are useful ingredients in the formulation of restructured chicken meat
products because they do not alter the physicochemical and microbiological quality
and their use results in high sensory acceptance among consumers. The use of
processed trimmings in MSCM together with albumin and/or TG was satisfactory;
the restructured product containing both MSCM and albumin showed a slight
advantage, as albumin positively influenced the colour of the product. TG addition
was advantageous because it produced restructured products with greater firmness
and binding capacity. For the sensory evaluation, there were no differences
regarding the acceptance of texture and other evaluated attributes. One advantage
of the addition of albumin is related to the low cost of this ingredient and the resulting
improvement in colour for restructured products. Moreover, the use of MSCM in
restructured chicken meat products produced positive results because it did not
interfere with the sensory quality of the restructured product and it contributed to a
considerable reduction in the product cost.

Therefore, the products developed in the present study are a viable alternative for
the meat processing industry. These products would allow the transformation of cuts
with no commercial value into new, ready-to-eat products with high added value and
high sales potential in the consumer market.

Acknowledgments

Funding for this project was provided by Fundação de Amparo à Pesquisa do Estado
de São Paulo (FAPESP, Project n 2010/08182–2) and Coordenação de
Aperfeiçoamento de Pessoal do Ensino Superior (CAPES) for a scholarship to
Marcio Almeida. We would also like to thank Fricok S.A., Ajinomoto do Brazil Ind. e
Com. de Alimentos Ltda and Salto Alimentos Ltda. for donating the meat and
materials used in this study.

References

1. Aguiar APS, Contreras-Castillo CJ, Baggio SR, Vicente E. Meat Quality of

Broilers from Different Rearing Systems. Ital J Food Sci. 2008;20(2):213–223.
2. Akamittath JG, Ball HR. Transglutaminase Mediated Polymerization of Crude
Actomyosin Refined from Mechanically Deboned Poultry Meat. J Muscle
Foods. 1992;3(1):1–14. doi: 10.1111/j.1745-4573.1992.tb00666.x. [Cross
Ref]
3. Al-Najdawi R, Abdullah B (2002) Proximate composition, select minerals,
cholesterol content and lipid oxidation mechanically and hand deboned
chickens from Jordanian market. Meat Sci. 61:243–247. doi:10.1016/S0309-
1740(01)00186-3 [PubMed]
4. Arora VK, Kempkes M. Industry Perspective and Roles. Food Sci Technol
Inter. 2008;14(5):455–457. doi: 10.1177/1082013208098832. [Cross Ref]
5. Barbut S. Poultry products: formulations and gelation. In: Barbut S, editor.
Poultry products processing: an industry guide. 1. USA: CRC Press. Boca
Raton; 2002. pp. 249–289.
6. Boles JA. Use od cold-set binders in meat systems. In: Kerry JP, Kerry JF,
editors. Processed meats Improving safety, nutrition and quality. Cambridge,
UK: Woodhead Publishing Limited; 2011. pp. 270–298.
7. Bragagnolo N, Rodriguez-Amaya DB. Comparison of the Cholesterol Content
of Brazilian Chicken and Quail eggs. J Food Comp Anal. 2003;16(2):147–
153. doi: 10.1016/S0889-1575(02)00129-1. [Cross Ref]
8. Campo MM, Nute GR, Hughes SI, Enser M, Wood JD, Richardson RI. Flavour
Perception of Oxidation in Beef. Meat Sci. 2006;72(2):303–311. doi:
10.1016/j.meatsci.2005.07.015. [PubMed] [Cross Ref]
9. Canto ACVCS, Costa Lima BRC, Suman SP, Lazaro CA, Monteiro MLG,
Conte-Junior CA, Freitas MQ, Cruz AG, Santos EB, Silva TJP. Physico-
chemical and Sensory Attributes of Low-sodium restructured Caiman Steaks
Containing Microbial Transglutaminase and Salt Replacers. Meat Sci.
2014;96:623–632. doi: 10.1016/j.meatsci.2013.08.003. [PubMed] [Cross Ref]
10. Carballo J, Barreto G, Jiménez Colmenero F. Starch and Egg White Influence
on Properties of Bologna Sausage as Related to Fat Content. J Food Sci.
1995;60(4):673–677. doi: 10.1111/j.1365-2621.1995.tb06204.x. [Cross Ref]
11. Carballo J, Ayo J, Jiménez-Colmenero F. Microbial transglutaminase and
caseinate as cold set binders: Influence of meat species and chilling storage.
Food Sci Technol-LEB. 2006;39(6):692–699. doi: 10.1016/j.lwt.2005.03.020.
[Cross Ref]
12. Contreras-Castillo CJ, Trindade MA, de Felicio PE. Physical and Chemical
Characterisation of Spent Hens Mechanically Separated Meat (MSHM) from
the Brazilian Production. Acta Aliment Hung. 2008;37(2):283–291. doi:
10.1556/AAlim.37.2008.2.13. [Cross Ref]
13. Daros FG, Masson ML, Amico SC. The influence of the addition of
mechanically deboned poultry meat on the rheological properties of sausage.
 J Food Eng. 2005;68(2):185–189. doi: 10.1016/j.jfoodeng.2004.05.030.
[Cross Ref]
14. Dimitrakopoulou MA, Ambrosiadis JA, Zetou FK, Bloukas JG. Effect of Salt
and Transglutaminase (JG) Level and Processing Conditions on Quality
Characteristics of Phosphate-free, Cooked, Restructured Pork Shoulder.
Meat Sci. 2005;70(4):743–749. doi: 10.1016/j.meatsci.2005.03.011.
[PubMed] [Cross Ref]
15. Field RA. Mechanically separated meat, poultry and fish. In: Pearson AM,
Dutson TR, editors. Edible Meat By-Products (Advances in Meat Research)
New York: Elsevier Applied Science; 1998. pp. 83–128.
16. Folch J, Lees M, Stanley GHS. A Simple Method for the Isolation and
Purification of Total Lipides from Animal Tissues. J Biol Chem.
1957;226(1):497–509. [PubMed]
17. Hartman L, Lago R. Rapid Preparation of Fatty Acid Methyl Esters From
Lipids. Laboratory practice. 1973;22(6):475. [PubMed]
18. Hecer C, Sozen BHU. Microbiological Properties of Mechanically Deboned
Poultry Meat that Applied Lactic Acid, Acetic Acid and Sodium Lactate.
African J Agricul Res. 2011;6(16):3847–3852.
19. Jiang ZR, Fenton M, Sim JS (1991) Comparison of four different methods for
egg cholesterol determination. Poul Sci.
https://www.biol.unlp.edu.ar/analisisdealimentos/paper11.pdf
20. Jorgensen SS, Sorensen G. A Combined Sampling and Delay Unit for Flow
Injection Analysis. The Automated Determination of 2-Thiobarbituric acid
Reactive Substances in Foods. Anal Chim Acta. 1996;322(1–2):69–76. doi:
10.1016/0003-2670(95)00604-4. [Cross Ref]
21. Keeton JT (2001) Formed and emulsion products. In: Sams AR (ed) Poultry
Meat Processing, 1th edn. CRC press LLC Boca Raton, pp 195–226
22. Kuraishi C, Sakamoto J, Yamazaki E, Susa Y, Kuhara C, Soeda T. Production
of restructured meat using micribial transglutaminase without salt or cooking.
J Food Sci. 1997;62(3):488–490. doi: 10.1111/j.1365-2621.1997.tb04412.x.
[Cross Ref]
23. Lanier TC. High Pressure Processing Effects on Fish Proteins. In: Shahidi F,
Ho CT, van Chuyen N, editors. Process-Induced Chemical Changes in Food.
1. New York: Springer Science + Business Media; 1998. pp. 45–56.
24. Lennon AM, McDonald K, Moon SS, Ward P, Kenny TA. Performance of
Cold-set Binding Agents in Re-formed Beef Steaks. Meat Sci. 2010;85:620–
624. doi: 10.1016/j.meatsci.2010.03.014. [PubMed] [Cross Ref]
25. Lu GH, Chen TC. Application of Egg White and Plasma Powders as Muscle
Food Binding Agents. J Food Eng. 1999;42(3):147–151. doi: 10.1016/S0260-
8774(99)00112-0. [Cross Ref]
26. McMindes MK, Siedler AJ. Nitrite Mode of Action - Inhibition of Yeast Pyruvate
Decarboxylase (EC 4.1.1.1) and Clostridial Pyruvate - Ferredoxin
Oxidoreductase (EC 1.2.7.1) by Nitric-oxide. J Food Sci. 1988;53(3):917–
919. doi: 10.1111/j.1365-2621.1988.tb08985.x. [Cross Ref]
27. Mielnik MB, Aaby K, Rolfsen K, Ellekjaer MR, Nilsson A. Quality of
Comminuted Sausages Formulated from Mechanically Deboned Poultry
 Meat. Meat Sci. 2002;61(1):73–84. doi: 10.1016/S0309-1740(01)00167-X.
[PubMed] [Cross Ref]
28. Motoki M, Seguro K. Transglutaminase and Ist Use for Food Processing.
Trends Food Sci Tech. 1998;9(5):204–210. doi: 10.1016/S0924-
2244(98)00038-7. [Cross Ref]
29. Ozkececi RB, Karakaya M, Yilmaz MT, Saricoban C, Ockerman HW. The
Effect of Carcass Part and Packaging Method on the Storage Stability of
Mechanically Deboned Chicken Meat. J Muscle Foods. 2003;19(3):288–301.
doi: 10.1111/j.1745-4573.2008.00118.x. [Cross Ref]
30. Payne CA (2000) Non-thermal gelation. Proc. 53rd Reciprocal Meat
Conference. 53:25–26
31. Pearson AM, Gillett TA. Restrucutred meat products. In: Pearson AM, Gillett
TA, editors. Processed Meats. 3. New York: Chapman & Hall; 1996. pp. 414–
425.
32. Perlo F, Bonato P, Teira G, Fabre R, Kueider S. Physicochemical and
Sensory Properties of Chicken Nuggets with Washed Mechanically Deboned
Chicken Meat: Research Note. Meat Sci. 2006;72(4):785–788. doi:
10.1016/j.meatsci.2005.09.007. [PubMed] [Cross Ref]
33. Ramirez-Suarez JC, Xiong YL. Effect of Transglutaminase-induced Cross-
linking on Gelation of Myofribrillar/Soy protein Mixtures. Meat Sci.
2003;65:899–907. doi: 10.1016/S0309-1740(02)00297-8. [PubMed] [Cross
Ref]
34. Rhee KS, Anderson LM, Sams AR. Comparison of Flavor Changes in
Cooked-refrigerated Beef, Pork and Chicken Meat Patties. Meat Sci.
2005;71(2):392–396. doi: 10.1016/j.meatsci.2005.03.016. [PubMed] [Cross
Ref]
35. Serrano A, Cofrades S, Jiménez Colmenero F. Transglutaminase as Binding
Agent in Fresh Restructured Beef Steak with Added Walnuts. Food Chem.
2004;85(3):423–429. doi: 10.1016/j.foodchem.2003.07.021. [Cross Ref]
36. Tabilo-Munizaga G, Barbosa- Canovas GV. Ultra High Pressure Technology
and its use in Surimi Manufacture: An Overview. Food Sci Technol Int.
2004;10(4):207–222. doi: 10.1177/1082013204045687. [Cross Ref]
37. Tammatinna A, Benjakul S, Visessanguan W, Tanaka M. Gelling Properties
of White Shrimp (Penaeus vannamei) Meat as Influenced by Setting
Condition and Microbial Transglutaminase. Food Sci Technol-LEB.
2007;40:1489–1497. doi: 10.1016/j.lwt.2006.11.017. [Cross Ref]
38. Thippareddi H, Sanchez M (2005) Thermal Processing of Meat Products. In:
Sun DW (ed)Thermal Food Processing New Technologies and Quality
Issues, CRC Press Taylor & Francis Group, Boca Raton, pp155–196
39. Tseng TF, Liu DC, Chen MT. Evaluation of Transglutaminase on the Quality
of Low-salt Chicken Meat Balls. Meat Sci. 2000;55:427–431. doi:
10.1016/S0309-1740(99)00172-2. [PubMed] [Cross Ref]
40. Vyncke W. Influence of Biological and Environmental Factors on Nitrogenous
Extractives of Spurdog Squalus-acanthias. Mar Biol. 1970;6(3):248. doi:
10.1007/BF00347233. [C
VII. Sensory, Microbiological and Chemical Changes in
Vacuum‐Packaged Blue Spotted Emperor (Lethrinus sp),
Saddletail Snapper (Lutjanus malabaricus), Crimson
Snapper (Lutjanus erythropterus), Barramundi (Lates
calcarifer) and Atlantic Salmon (Salmo salar) Fillets
Stored at 4°C
Food Sci Nutr. 2016 May; 4(3): 479–489.
Luisa Fernanda Fuentes‐Amaya, 1 , 2 Steve Munyard, 3 Judith Fernandez‐Piquer, 4
and Janet Howieson 1

Abstract

Quality assessment of finfish fillets during storage is important to be able to predict

the shelf life of the fresh product during distribution. Microbial, chemical (pH, TMA,
and TVB‐N), and sensory (Quality index assessment QIA, Torry scheme) changes
in vacuum‐packaged blue‐spotted emperor (Lethrinus sp), saddletail (Lutjanus
malabaricus), crimson snapper (Lutjanus erythropterus), barramundi (Lates
calcarifer), and Atlantic salmon (Salmo salar) fillets stored at 4°C were evaluated for
5 days. Microbiological study included evaluation of TVC (total viable counts), total
psychrotrophic organisms, and H2S‐producing bacteria. Numbers increased during
storage time and reached an average of 8.5, 8.5, and 9.2 log 10 cfu/g, respectively,
for the five different fish species. These levels were above accepted microbiological
limits for fish fillets. Although the sensory analyses showed a decrease in quality,
none of the finfish fillets were considered unacceptable at the end of the storage trial.
Chemically, there was a slight pH increase, but trimethylamine (TMA) levels
remained low. However, total volatile basic nitrogen (TVB‐N) levels increased over
time, reaching levels above 35 mg/100 g for blue spotted emperor, saddletail
snapper, and crimson snapper by the end of the storage period. Results show that
the deterioration of finfish fillet quality is a complex event of biochemical, sensory,
and microbial factors, and multiple analyses may be required to define acceptability.

Keywords: Finfish, quality assessment, specific spoilage organisms, spoilage, total

viable count

Introduction

Quality monitoring of seafood is an important consideration due to the short shelf life
of fresh product. The cost of product loss due to spoilage is a big concern for the
seafood industry: it is estimated that 30% of landed fish are lost every year due to
spoilage as a result of chemical and microbial activity (Baird‐Parker 2000). In
Australia, combined wild and finfish aquaculture harvest is an important part of the
economy with 113.310 tonnes produced during 2011–12, valued at approximately
$451 million (ABARES 2013). Although there are preservation techniques available,
there is still a lack of understanding about the factors that could be addressed to
optimize the shelf life of fish (Ghaly et al. 2010).

Microbial growth in combination with enzymatic autolysis and oxidation are the major
causes of fish spoilage. The bacterial species that contribute to spoilage are known
as SSO (Specific Spoilage Organisms). These organisms produce unacceptable
flavors, odors, texture, and/or color, thus reducing the quality of the seafood. Typical
SSO of fish include Shewanella, Pseudomonas, Photobacterium, Aeromonas,
and/or Enterobacteriaceae (Gram et al. 2002). The SSO found in fish depend on the
initial bacterial load, type of processing, preservation methods, and storage
conditions. The spoilage of whole cod and sea salmon in ice is explained by the
presence of Shewanella putrefaciens (Jørgensen and Huss 1989; Hozbor et al. 2006).
Photobacterium phosphoreum has been associated with spoilage of haddock fillets
stored at temperatures in the range of 0–15°C (Olafsdottir et al. 2006), and in spoilage
of vacuum‐ or MAP (modified atmosphere packed) cod fillets stored at 0°C (Dalgaard
et al. 1993). In the case of Mediterranean fish stored between 0°C and 15°C and MAP
Atlantic salmon, Pseudomonas spp. have been identified during spoilage
(Koutsoumanis and Nychas 1999; Milne and Powell 2014).

The traditional microbiological assessment used to evaluate finfish fillet quality is

TVC (total viable count). The media allows the growth of aerobic mesophilic
organisms present in the finfish samples and is commonly used as a measure of
quality by food safety authorities and retailers. A TVC level greater than 10 7 cfu/g in
fresh fish is considered unacceptable, (ICMSF 1986) while TVC level of 106 cfu/g
corresponds to the maximum level set by some Australian supermarkets. However,
it is well documented that not the entire bacterial load enumerated using TVC is
responsible for spoilage. Several studies have shown that fish with TVC levels above
107 cfu/g are still acceptable by sensory evaluation. Some examples include
haddock fillets stored at 0°C, 7°C, and 15°C (Olafsdottir et al. 2006) and the European
sea bass stored in ice (Kyrana and Lougovois 2002).

Specific Spoilage Organisms are present in low quantities initially but as storage
progresses, SSO grow faster than the rest of the bacteria present in the fish (Huis
In't Veld 1996). Consequently, specific microbial measurements to determine and
quantify SSO are more reliable to evaluate accurately the freshness of seafood.
Microbiological media such as Iron Agar (IA) and Long and Hammer Agar (LH) have
emerged as being more suitable to determine and quantify the presence of SSO
responsible for deterioration of fish and seafood (Gram 1992).

Chemical indexes widely used to evaluate fish quality deterioration include TVB‐N
(total volatile basic nitrogen) and TMA (trimethylamine). While TVB‐N measures the
overall volatile nitrogen present in the fish, TMA focuses on the reduction in TMAO
(trimethylamine oxide) by SSO activity (Howgate 2009).

Sensory evaluation is an important method to monitor the effect of storage conditions

on the changes in sensory attributes of the tested food. Sensory evaluation methods
such as Quality Index Methods (Erkan and Özden 2008; Vaz‐Pires et al. 2008) or Torry
Scheme (Kyrana and Lougovois 2002; Olafsdottir et al. 2006) are useful techniques to
measure the quality and freshness for different fish species. Given that consumer
acceptability is such an important driver for finfish quality, it has been discussed that
instrumental methods should be directly correlated with the sensory assessment
(Olafsdottir 1997). The aim of this study was to compare and contrast the various
techniques used to evaluate fish quality and freshness to understand better the
relationship between the different quality evaluation techniques.

Accordingly, the study evaluated the quality changes of five different vacuum‐packed
fillets of common finfish species after 1, 3, and 5 days stored at 4°C. Quality
evaluation included the use of different microbiological media, and traditional
biochemical and sensory quality assessment analyses.

Materials and Methods

Fish source, processing, storage conditions, and sampling

The target fish and the experimental period of storage were chosen based on those
species most commonly sold as fillets in local Australian supermarkets and
fishmongers, and the normal five‐day shelf life accorded to the raw fillet product by
the processors. Fish used for the study included blue‐spotted emperor (Lethrinus
sp.) harvested in Exmouth Gulf, Western Australia (WA) and saddletail snapper
(Lutjanus malabaricus) and crimson snapper (Lutjanus erythropterus) harvested in
the Timor Sea and WA. Barramundi (Lates calcarifer) from a WA sea cage operation
and Atlantic salmon (Salmo salar) aquacultured in Tasmania were also assessed.
The fish were refrigerated and transported in 25 kg plastic‐covered tubs to the
processing plant in Perth, with the exception of the Atlantic salmon, which was
transported in polystyrene eskies. Triplicate samples of each finfish were filleted
upon arrival, aseptically divided into 6 × 200 g portions, vacuum packed and stored
at 4°C ± 1°C. Two random samples of each vacuum‐packed finfish were collected
after 1, 3, and 5 days from filleting, and sent to the testing laboratories in a box
equipped with ice packs. For each finfish species, one pack was used for
microbiology analyses at the Food Microbiology laboratories (L Block, University of
Western Australia, Perth) and a second pack was used for sensory assessments at
the Food Sensory laboratories (Building 400, Curtin University, Perth). Two
independent experimental trials were performed in February and May 2012.
Samples from both trials were used for microbiological, chemical, and sensorial
assessments. Both trials were performed under the same conditions, with no
significant differences noted between them. Accordingly, data from same fish fillet
type was combined for analysis against the study objective.

Microbiological analysis

Duplicate fish fillet samples (10 g) were aseptically weighted and aseptically
transferred to sterile plastic bags. Samples were diluted 1:10 with MRD (Maximum
Recovery Diluents) (peptone saline water containing 0.1% peptone and 0.85%
saline), and homogenized for 1 min at the highest settings in a Seward Stomacher®
(model 80, Seward Ltd. West sussex, BN14 8HQ, United Kingdom). Appropriate 10‐
fold serial dilutions were prepared in MRD and spiral plated (0.05 mL) in duplicate
onto selected media using a Whitley Automated Spiral Plater. TVC agar plates were
incubated at 30°C for 48 h and results were expressed as total viable bacteria counts
(Buchbinder et al. 1953; Anon 1994). Iron agar (IA) plates were incubated at 25°C for
48 h and used to enumerate hydrogen sulphide‐producing bacteria (Gram et al. 1987;
Nordic Committee on Food Analysis, 2006). Long and Hammer agar (LH) (Van
Spreekens (1974) and Nordic Committee on Food Analysis (2006)), were prepared and
incubated at 15°C for 5 days for quantitative determination of psychotrophic bacteria.

Chemical analyses

For each analysis, the vacuum package for each fish species was opened and the
sample was minced using sterile scissors to an approximate size of 0.5 cm2. A
portion of 10 g was transferred to a 10 mL glass beaker, then homogenized with
25 mL of 7.5% aq TCA (trichloroacetic acid, Sigma‐Aldrich, Saint Louis, MO 63101,
United States). The homogenate mixture was then passed through a Whatman No.
1 filter paper. The filtrates were stored at −20°C until TMA and TVB‐N levels were
determined following the procedures described by Baixas–Nogueras and others.
(Baixas‐Nogueras et al. 2001). The TMA analysis included a traditional Dyer
colorimetric as described by the Association of Official Analytical Chemists (AOAC
1995). Absorbance at 410 nm wavelength was measured using a Konica Minolta

Spectrophotometer (CM‐500i/CM‐500C). The TVB‐N test was performed according

to the official EU method and involved the use of a Kjeltec System (model 1002
Distilling Unit, Tecator Inc., Boulder, CO). Results were expressed as mg/100 g
sample.

The pH of 1:5 diluted samples was measured at room temperature with a glass
electrode connected to a Hanna pH meter (Hanna HI 9321, Hanna Instruments Inc,
Rhode Island, 02895, United States).

Sensory analyses

Sensory analyses included in this study had ethics approval from Curtin University
(HREC Number RD‐47‐10).

A Quality Index Assessment Scheme (QIA) for the raw fish fillets from each of the
target species was developed using previously described procedures for quality
assessment of fish (Boulter et al. 2006; Martinsdóttir et al. 2009). Each scheme
consisted of six attributes (appearance, transparency, texture, bloodlines, odor, and
gaping) and a score system ranking from 0 (high freshness) to 3 (low freshness).
For sensory analyses, each fish fillet was placed on a separate plastic tray labeled
with the corresponding name of the fish. Every day during the development of the
scheme, two trained panelists assessed the fish on sterile, neutral colored desks
using the draft list of attributes. Separate species schemes were developed based
on this initial assessment. The developed QIA scheme for each species is shown in
Table 1.

Table 1. Quality index method schemes for sensory evaluation of vacuum‐

packaged barramundi, Atlantic salmon, blue‐spotted emperor, saddletail
snapper, and crimson snapper fillets stored at 4°C

Fish species pH TMA (mg/100 g) TVB‐N (mg/100 g)

1a 3 5 1 3 5 1 3 5

Barramundi 6.3 6.5 6.6 0.37 0.30 0.38 27.44 22.96 31.64

Atlantic salmon 6.2 6.3 6.5 0.19 0.28 0.28 20.44 26.88 28.56

Blue‐spotted emperor 6.5 6.6 6.6 0.23 0.39 0.18 25.90 27.16 41.44

Saddletail snapper 6.4 6.5 6.7 0.21 0.40 0.40 22.68 27.30 58.52

Crimson snapper 6.3 6.4 6.6 0.33 0.33 0.43 25.48 30.52 58.10
aStorage time (days).

The Torry Scheme (Martinsdóttir et al. 2009) was used for sensory (odor and flavor)
evaluation of the cooked fish fillets. The maximum score is 10 indicating the highest
freshness in flavor and odor. An average score of 5.5 is recognized as the minimum
acceptable for human consumption (Boulter et al. 2006; Martinsdóttir et al. 2009).

For the experiments, sensory assessments (both QIA on raw fillets and Torry
assessment on cooked fillets) were undertaken at the School of Public Health at
Curtin University in Western Australia. Seven volunteers formed the sensory panel
and as all seven had previously been trained in QIA and Torry scheme, there was
need for only one training session at the facilities of the industrial partner. Before
participating in the sensory analysis, all the panelists were given a consent form to
sign, which described the background to the study and their sensory evaluation
schedule.

At each session, each panelist was given a Torry scoresheet, “Freshness evaluation
of cooked lean fish” on which to record their assessment of the odor and flavor for
each of the five types of cooked fish fillets. All fillets were a standardized size of
2 cm × 2 cm and were cooked in a microwave oven (1800 Watts) for 20 sec in the
kitchen located behind the assessing area. Each panelist was seated in a separate
sensory booth and access to the kitchen was supplied by a small window, which
could be opened only from the kitchen side. A random order of presentation of
samples was used throughout the assessments, and all panelists assessed a piece
of the same fish sample, each sample was placed on a white plate, and labeled with
an individual random number. The panelists were provided with water and plain
crackers to allow them to assess each sample with a fresh palate.

Statistical analysis

The five fish fillets in the study from both trials were analyzed under the same
conditions and the data for both trials were combined. The data were analyzed using
the statistical software SPSS 10.0 for Windows (SPSS Inc., Chicago, IL), and the
results were statistically computed, graphed, and tabulated. As the data were not
normally distributed, a nonparametric test Kruskal–Wallis test (one‐way ANOVA)
with significance level of 0.01 was applied.

Results and Discussion

Microbiological analysis

Changes in microbiological numbers with time in filleted barramundi, Atlantic salmon,

blue‐spotted emperor, saddletail snapper, and crimson snapper during storage at
4°C are shown in Figure 1. As observed for cod fillets (Dalgaard et al. 1993), vacuum
packaging did not limit bacterial growth. TVC levels were an average of
5.8 ± 0.3 log10 cfu/g (Day 1), 7.2 ± 0.5 log10 cfu/g (Day 3), and 8.5 ± 0.8 log10 cfu/g
(Day 5) for the five different fish species (Fig. 1A). Lower initial mesophilic aerobic
bacteria, ranging between 2.5 to 4 log10 cfu/g, have been previously been reported
during microbiological assessment of whole rainbow trout, whole sardines, and sea
salmon (Chytiri et al. 2004; Hozbor et al. 2006; Erkan and Özden 2008). Differences in
microbiological levels among finfish species may be related to a diverse initial
composition depending on the harvest water temperature, handling, and storage
situation (Huis In't Veld 1996; Chytiri et al. 2004). In addition, spoilage mechanisms
have been observed to be faster in fillets compared to whole fish (Huss 1995; Chytiri
et al. 2004). The higher initial TVC levels found in this study may also be because first
sampling was performed after 1 day of chilled storage (day 1)., a 10‐fold higher TVC
levels was observed for haddock fillets tested after 1 day of cooled storage in
Styrofoam boxes compared with fillets processed 1 day postcatch (Olafsdottir et al.
2006). In fact, the TVC levels in all finfish fillets analyzed in this study were found to

be above the acceptability limits (6 log10 cfu/g) of some Australian supermarkets and
close to the microbiological limit for seafood of 7 log10 cfu/g (ICMSF 1986) after 3 days.
Fig. 1. Microbiological changes in vacuum‐packaged barramundi (black), Atlantic
salmon (horizontal lines), blue‐spotted emperor (gray), saddletail snapper (diagonal
lines), and crimson snapper (white) fillets stored at 4°C. Bars indicate standard
deviations.

Other studies have shown longer storage periods to reach these acceptability limits
(Kyrana and Lougovois 2002; Chytiri et al. 2004; Hozbor et al. 2006; Erkan and Özden
2008). This difference may be due to the use of lower storage temperatures as it has

been observed that the seafood deteriorates twice as fast at 4°C than it does at 0°C
(Boulter et al. 2006). Research has shown that TVC growth rates in haddock fillets are
found to be approximately three times faster for fish stored at 7°C compared with
fish stored at 0°C (Olafsdottir et al. 2006).

For all the species included in this study, TVC on fillets increased more than 2 logs
after 5 days at 4°C. Atlantic salmon showed the highest bacterial increase (3.7 logs)
and likewise, crimson snapper presented the lowest bacterial increase (2 logs) at the
end of the period of study. However, there were no significant difference (P > 0.05)
found among the five fish fillet species during the entire storage.

Sulphide producers such as Shewanella putrefaciens often constitute a major

portion of the microbial composition of spoiling fish (Jørgensen and Huss 1989).
Sulphide‐producing bacteria levels in the five fish species ranged from 4.8 log10 cfu/g
to 5.9 log10 cfu/g on Day 1 (Fig. 1B). These bacterial levels are higher than the 4
log10 cfu/g observed in sardines and sea salmon stored in ice, (Hozbor et al. 2006;
Erkan and Özden 2008) but, in the range of the 3–8 log10 cfu/g was found in seafood
at retail (Gorczyca et al. 1984). As reported in other studies (Huis In't Veld 1996; Kyrana
and Lougovois 2002), H2S‐producing bacteria levels are lower than TVC levels at the
beginning of storage, but levels become less different after a few days of storage. A
similar bacterial growth rate pattern was observed for each of the different finfish
species in this study and the average of hydrogen sulphide‐producing bacteria levels
were 7.3 ± 0.2 log10 cfu/g and 8.5 ± 0.1 log10 cfu/g after 3 and 5 days, respectively.

The use of sulphide‐producing bacteria as a spoilage indicator has been found as

suitable for freshwater Rohu fish stored at ambient temperature, (Madhusudana Rao
and Imam Khasim 2009) but not for sea bass stored in melting ice (Kyrana and
Lougovois 2002). Different SSO are able to grow in diverse fish in either synergism or
antagonism and produce different spoilage indicators or metabolites (Huss 1995).
Although, H2S‐producing bacteria and TVC levels were higher in Atlantic salmon,
there was no significant difference (P > 0.05) among the five fish fillet species and
storage time, however, a positive correlation (R 2 = 0.93) was observed. Thereby,
the five different finfish species showed similar microbial deterioration during storage
when evaluated with TVC or IA media.

Psychrotrophic bacteria levels were generally higher than TVC or H 2S‐producer

bacteria levels (Fig. 1C). The average levels of psychrotrophic bacteria were
6.1 ± 0.4 log10 at Day 1, 7.4 ± 0.2 log10 at Day 3, and 9.2 ± 0.3 log10 cfu/g after
5 days' storage. Compared with the TVC results, psychrotrophic bacteria counts
were 0.5–1.0 log higher. Similar results have been found in sea salmon studies
(Hozbor et al. 2006) and therefore, highlight the importance of these cold tolerant
organisms in the product's shelf life.

However, the use of TVC as an effective indicator of fish fillet spoilage is uncertain
(Gram 1992; Dalgaard 2000). Our study supports the use of media such as LH in
combination with low temperature incubation to enable better recovery of the
bacteria found in finfish fillets stored at low temperature.
 Chemical analyses

Total volatile basic nitrogen measures the content of ammonia, TMA, and DMA
(dimethylamine) (Howgate 2009) in fish. The TVB‐N levels in the five finfish species
were in the range of 20.44–27.44 mg/100 mg initially and increased to 28.56–
58.52 mg/100 mg after 5‐days storage at 4°C (Table 2). Previous studies classified
TVB‐N levels of 10 mg/100 or lower for fresh fish, 20–30 mg/100 g for beginning of
spoilage, and above 30 mg/100 g for spoiled fish (Kimura and Kiamukura 1934).
Moreover, it has been suggested that the quality classification of fish and fish
products based on TVB‐N as “high quality” up to 25 mg/100 g, “good quality” up to
30 mg/100 g, “limit of acceptability” up to 35 mg/100 g, and “spoilt” above
35 mg/100 g (Gulsun et al. 2009). EU document 2074/2005 established limits of
acceptability between 25 mg/100 g to 35 mg/100 g. However and possibly being
predominantly Australasian species, of the species studied here, only the Atlantic
salmon is included in the EU documents in which 35 mg/100 g is the limit of
acceptability. TVB‐N values above the highest EU limit of acceptability were found
for blue‐spotted emperor, saddletail snapper, and crimson snapper after 5‐days
storage at 4°C. Of the five varieties, studied saddletail snapper and crimson snapper
presented the highest concentrations (58.52 and 58.10 TVB‐N mg/100 mg,
respectively) by Day 5.

Table 2. Chemical changes in vacuum‐packaged barramundi, Atlantic salmon,

blue‐spotted emperor, saddletail snapper, and crimson snapper fillets stored
at 4°C

Atlantic salmon Barramundi Blue‐spotted Crimson snapper Saddletail

emperor snapper

Quality parameter Description Sco Description Sco Description Sco Description Sco Description Sco
re re re re re

Bright color Gray and Creamy pink,

0 0 Pink tinge 0 Bright sheen 0 0
sheen pink bright sheen

Bright color Slightly

1 1 Fading 1 Dull 1 Dull pink 1
dull sheen faded
Brightnes
Skin
s
Fading color 2 Faded 2 No color 2 Faded 2 Faded 2

Pale/brown
color 3
appearing

Translucent 0 Translucent 0 Translucent 0 Translucent 0 Translucent 0

Appeara Transpar
nce ency
Opaque 1 Opaque 1 Opaque 1 Opaque 1 Opaque 1

Firm and Firm and Firm and Firm and

Flesh Texture 0 0 Firm elastic 0 0 0
springy springy springy springy
 Atlantic salmon Barramundi Blue‐spotted Crimson snapper Saddletail
emperor snapper

Quality parameter Description Sco Description Sco Description Sco Description Sco Description Sco
re re re re re

Soft and Fingermark

1 1 Soft elastic 1 Soft 1 Soft 1
springy remains

Fingermark Fingermark
2 2 Mushy 2 Mushy 2
remains remains

Mushy 3

Bright red 0 Red 0 Bright red 0 Bright red 0 Bright red 0

Blood Dark brown 1 Orange 1 Brown 1 Orange 1 Orange 1

Brown 2 Brown 2 Brown 2

Fresh Fresh Fresh Fresh Fresh

seawater/se 0 seawater/se 0 seawater/se 0 seawater/se 0 seawater/se 0
aweed aweed aweed aweed aweed

Neutral 1 Neutral 1 Neutral 1 Neutral 1 Neutral 1

Odor
Fishy/metalli
2 Sour 2 Fishy 2 Acidic/fishy 2 Acidic/fishy 2
c

Sour 3 Acidic 3 Sour 3 Sour 3

None 0 No gaping 0 No gaping 0 None 0 None 0

Gaping Gaping Gaping

Slight gaping 1 Slight gaping 1 1 1 1
Gaping present present present

Significant Significant
2 2
gaping gaping

Quality
Index

Trimethylamine ranged between 0.19–0.37 mg/100 g at the beginning of the

experiments and reached 0.18–0.43 TMA mg/100 mg by Day five. In previous
studies, the average concentration for a freshly caught fish is 2 mg TMA‐N/100 g wet
weight; however, it can range between 1 and 4 mg/100 g (Oehlenschläger 1997). The
level of sensory rejection is variable among species; nevertheless it estimated that
between 10 and 15 mg TMA‐N/100 g in aerobic stored fish, and 30 mg TMA‐N/100 g
in packed cod (Dalgaard et al. 1993).

These results therefore agree with previous reports that suggest that the TMA
indicator is not suitable for the early stages of deterioration. Olafsdottir ( 1997)
discussed the fact that volatile amines such as TMA are present in fresh fish
(immediately after capture), but in very low levels and accumulate in the later stages
of conservation, depending on the species, temperature, and hygiene. Moreover,
TMAO can be first reduced to TMA by endogenous bacteria and later by the action
of gram‐negative bacteria as reported by Gram and Huss (1996). Therefore, the high
concentration of TMA has been considered mainly as an indicator of spoilage of fish
in an advanced state of deterioration.

A slow increase in TMA was also observed in the European sea bass (Kyrana and
Lougovois 2002), filleted trout (Chytiri et al. 2004), and Mediterranean fish
(Koutsoumanis and Nychas 1999), and the authors suggested that the specific
bacterial composition could have influenced the TMA production. Although
S. putrefaciens is a TMA producer, other spoilage organisms such as Pseudomonas
spp. do not produce TMA (Gram 1992). In addition, the potential presence of
Pseudomonas spp. can inhibit the growth of S. putrefaciens (Gram and Melchiorsen
1996) causing a reduction below the densities (8–9 log
10 cfu/g) required for TMA
production (Dalgaard et al. 1993; Gram and Huss 1996).

In this study, concentrations of TVB‐N in the fish fillet species showed a weak
correlation with concentrations of TMA (R 2 = 0.450). In general, the fish fillets
showed a moderate production of TVB‐N throughout the storage time, while TMA
levels always remained low. A similar trend has been observed during other fish
quality assessments. Low levels of TMA and an increase in TVB‐N up to
25 mg/100 g was observed in the European sea bass at the end of shelf life, after
19‐days storage in ice (Kyrana and Lougovois 2002). It has been suggested that the
increase detected by TVB‐N without the same impact on TMA can be due to
ammonia resulting from deamination of amino acids, rather than bacterial activity
(Stohr et al. 2001).

The pH value for the five fish species ranged from 6.2 to 6.5 at the beginning of the
experiments. The pH values are in agreement with the reported postmortem pH
range of 6.0–6.8 (Howgate 2009). There was a slight increase in pH over storage, but
changes in the pH values between the fish species during the entire storage trial
were not statistically significant (P > 0.05). Minimal changes in the pH values during
spoilage have also been observed for trout (Chytiri et al. 2004), in the first half of the
storage life of the European sea bass (Kyrana and Lougovois 2002), sardines (Erkan
and Özden 2008), haddock fillets (Olafsdottir et al. 2006), and MAP cod fillets (Dalgaard
et al. 1993).

It has been described that during spoilage the pH may increase because of
production of volatile bases (NH3 and TMA), due to SSO action(Fraser and Sumar
1998). However, because of variability in pH starting point among different species,

season, and other factors, pH level is not always a good predictor of spoilage
(Church 1998). For instance, Pacific coast fish has initial pH value slightly lower than
7, important microbial spoilage could be present with small rises in pH. In contrast
to some salmon species, in which the ultimate pH is 6.2, it has been found that
spoilage was present with different pH levels (Tarr 1954).
Sensorial analysis

Changes in the sensory attributes of the five finfish species fillets during storage at
4°C were evaluated using the QIA method summarized in Table 1. All fish fillets
showed an increase in QI demerit points throughout the period of storage as shown
in Figure 2A. At Day 1, the lowest QI score (0) was for Atlantic salmon and the
highest QI score (2.5) was for crimson snapper (Fig. 2A). After 5‐days storage,
Atlantic salmon showed the highest QI score (11) followed by barramundi with a QI
score of 7.5. The rest of the species presented a QI score between 5 and 6. This
demonstrates different rates of quality deterioration between the different fish
species as previously reported (Boulter et al. 2006; Chytiri et al. 2004; Huis In't Veld
1996).
Figure 2. Sensorial changes in vacuum‐packaged barramundi (♦), Atlantic salmon
(■), blue spotted emperor (▲), saddletail snapper (×), and crimson snapper (*) fillets
stored at 4°C.

The Torry scheme was used to evaluate the cooked fish and all fish fillets showed a
decrease in the Torry score with time as shown in Figure 2B. High negative
correlation (R 2 > 0.93) was observed between the two components evaluated (flavor
and odor) in fillets of barramundi, Atlantic salmon, and crimson snapper (R 2 = 0.94,
0.99 and 0.99; respectively) (Fig. 2B). By contrast, only a moderate correlation (R
2 = 0.51–0.58) between these parameters was observed in blue‐spotted emperor

and saddletail snapper. The accepted level for rejection (average of 5.5) was not
reached by the end of the 5‐days storage period for any of the fish species evaluated.
Atlantic salmon, had the lowest average score (6) at the end of storage period, which
is consistent with the results from the QIA.

Torry assessment as an indicator of freshness quality over time in general shows a

clear inverse correlation with the QIA tool: QI score = −1.127 × (days at
4°C) + 9.8595 (R 2 = 0.95). The results show that using the Torry assessment
criteria, all the samples were still suitable for consumption at Day 5.

Cooked fish flavor is considered to be an excellent indicator predictor of freshness

quality and shelf life (Lougovois et al. 2002). Our study showed differences in the
eating quality with storage time between species. This might have been due to a
variation in the size and age of fish used (Pedrosa‐Menabrito and Regenstein 1990)
or potential differences in the microbial flora. Further work would be required in this
area.

The Torry scheme is a recognized method to evaluate the quality of fish (spoilage
related to presence of off‐odors and off‐flavors), and it proved to be beneficial and
accurate in this study. In addition, it is a rapid and noninvasive method, which can
be performed, with adequate training, in situ. The raw fillet quality assessment
measures used during this study, once validated using the quality index established
methodologies (Boulter et al. 2006; Martinsdóttir et al. 2009) may be an alternative
quality measure as they were correlated well with Torry and microbiological
measures.

Correlation between microbiological, sensory, and chemical analysis

Sensory evaluation methods such as QIA and Torry Scheme have been used to
evaluate the shelf life of fish (Erkan and Özden 2008; Kyrana and Lougovois 2002;
Olafsdottir et al. 2006; Vaz‐Pires et al. 2008). They are highly accurate and efficient to
assess fish freshness, and closely related to consumers criteria. After 3‐days storage
at 4°C, fish fillet samples were above the reference limit (>106 cfu/g) and considered
unacceptable from the microbial point of view. However, they were still acceptable
by sensory panelist perception (Torry scores >5.5). In addition, although
microbiological counts were in the range of approximately 8–9 log10 CFU/g after
5 days, all fish fillets species were still acceptable by sensory analysis (QIA and
Torry scheme level >5.5). This contrast in acceptability based on different
assessment criteria (TVC and Torry) is shown in Figure 3. There was a significant
negative correlation between TVC and Torry scheme (r = −0.719 and P < 0.001)
Sensory acceptance of seafood above the microbiological reference limit (106 cfu/g)
have also been observed for other fish species (Kyrana and Lougovois 2002; Hozbor
et al. 2006; Olafsdottir et al. 2006).

Fig. 3. Torry Scheme and Total Viable Count relationship for vacuum‐packaged
barramundi (♦), Atlantic salmon (■), blue‐spotted emperor (▲), saddletail snapper
(×), and crimson snapper (*) fillets stored at 4°C.

The different microbiological media used for fish fillets assessment were plotted
against QIA scores. TVC provided good correlation with QI score for Atlantic salmon
(R 2 = 0.96), saddletail snapper (R 2 = 0.96), and crimson snapper (R 2 = 0.94)
(Fig. 4).
Figure 4. Quality Index Method and Total Viable Count relationship for vacuum‐
packaged (A) Atlantic salmon, (B) saddletail snapper, and (C) crimson snapper fillets
stored at 4°C.

Trimethylamine is a useful quality index for microbial spoilage of vacuum‐packed

cod fillets (Dalgaard et al. 1993). The fact that TMA did not increase during the overall
storage would suggest that TMA as an indicator is not suitable for the early stages
of deterioration in Atlantic salmon, barramundi, blue‐spotted emperor, saddletail
snapper, and crimson snapper fillets. Similarly, TMA has previously shown little
value during spoilage evaluation of cod and ocean perch fillets (Magnusson and
Martinsdottir 1995). In the case of TVB‐N, levels above the maximum limit
(35 mg/100 g) indicated by the EU were only observed for blue‐spotted emperor,
saddletail snapper, and crimson snapper after 5‐days storage at 4°C. However, it
was not find statistically significant differences in microbial counts among fish
species on Day 5 to justify the TVB‐N differences observed for saddletail snapper
and crimson snapper. The result of this study suggest that microbiological and
sensory assessment are better indicators of early deterioration of the five fish
species tested than chemical indexes such as TVB‐N, TMA, and pH. Freshness
deterioration is a complex event of biochemical and microbial interactions. For this
reason, further investigation of the bacterial and chemical composition of the
different fish fillets species would help to understand better the quality deterioration.

Conclusions

This study evaluated the effectiveness of quality measurements to determine

freshness in vacuum‐packed fillets of five finfish species. The sensory evaluation
has shown that all fish fillets in study were still acceptable at the end of the period of
study, although microbiological analysis (TVC, H2S‐producing organisms, and Total
psychrotrophic organisms) showed an exponential bacterial growth throughout all
chilled storage period reaching unacceptable levels by day 3. Only Atlantic salmon
showed higher microbial counts, whereas the other four varieties did not show
significant difference between growths. The observed microbial growth did not have
a significant impact on the changes in pH and TMA production. TVB‐N levels
increased differently for each fish species. To determine the quality and loss of
freshness, it is important to combine quality attributes and correlate them with
sensory evaluation, which it is considered the most important method to monitor the
effect of storage conditions on fish freshness.

Conflict of Interest

None declared.

Acknowledgments

The authors thank Mark Tamplin for assisting with manuscript review, and Felicity
Denham, Rachel Tonkin, and Gillian Parton for technical assistance. This
investigation was undertaken as part of the project Australian Seafood CRC Project
2009/709 “Improving the Supply Chain for Selected WA Seafood Products”.

References

1. ABARES . 2013. Australian Bureau of Agricultural and Resource Economics

and Sciences. Australian Fisheries statistics 2012.
2. Anon . 1994. Preparation of culture media, diluents and reagents, AS 1766.
Standards Australia.
3. AOAC . 1995. Official method 971.14. Trimethylamine nitrogen in seafood.
AOAC Official Methods of Analysis, Arlington, VA, p.7.
4. Baird‐Parker T. C. 2000. The production of microbiologically safe and stable
foods Pp. 3–18 in Lund B. M., editor; and Baird‐Parker T. C., editor. , eds.
The microbiological safety and quality of food. Aspen Publishers,
Gaithersburg, MD.
5. Baixas‐Nogueras S., Bover‐Cid S., Vidal‐Carou M. G., Veciana‐Nogués M.
T., and Mariné‐Font A.. 2001. Trimethylamine and total volatile basic nitrogen
determination by flow injection/gas diffusion in Mediterranean hake
(Merluccius merluccius). J. Agric. Food Chem. 49:1681–1686. [PubMed]
6. Boulter M., Poole S., and Bremner A.. 2006. Australian Quality Index Manual:
A Tool for Evaluating Changes in Chilled Seafood. Sydney Fish Market:
Fisheries Research and Development Corporation.
7. Buchbinder L., Baris Y., and Goldstein L.. 1953. Further studies on new milk‐
free media for the standard plate count of dairy products. Am. J. Public Health
43:869–872. [PMC free article] [PubMed]
8. Church N. 1998. Map fish and crustaceans sensory enhancement. Food Sci.
Technol. Today 12:73.
9. Chytiri S., Chouliara I., Savvaidis I. N., and Kontominas M. G.. 2004.
Microbiological, chemical and sensory assessment of iced whole and filleted
aquacultured rainbow trout. Food Microbiol. 21:157–165.
10. Dalgaard P. 2000. Fresh and lightly preserved seafood Pp. 110–139 in Man
C. M. D., editor; and Jones A. A., editor. , eds. Shelf‐life evaluation of foods.
Aspen, London.
11. Dalgaard P., Gram L., and Huss H. H.. 1993. Spoilage and shelf‐life of cod
fillets packed in vacuum or modified atmospheres. Int. J. Food Microbiol.
19:283–294. [PubMed]
12. Erkan N., and Özden O.. 2008. Quality assessment of whole and gutted
sardines (Sardina pilchardus) stored in ice. Int. J. Food Sci. Technol.
43:1549–1559.
13. Fraser O. P., and Sumar S.. 1998. Compositional changes and spoilage in
fish (part II) – microbiological induced deterioration. Nutri. Food Sci. 98:325–
329.
14. Ghaly A. E., Dave D., Budge S., and Brooks M. S.. 2010. Fish spoilage
mechanisms and preservation techniques: review. Am. J. Appl. Sci. 7:846–
864.
15. Gorczyca E. M., Angawati A., and Putro S.. 1984. Microbiological quality of
retail “Wet” fish in Jakarta Pp. 191–193 in Spoilage of tropical fish and product
development: proceedings of a Symposium Held in Conjunction with the Sixth
Session of the Indo‐Pacific Fishery Commission Working Party on Fish
Technology and Marketing: royal Melbourne Institute of Technology,
Melbourne Australia, 23–26 October 1984, Food and Agriculture Organization
of the United Nations.
16. Gram L. 1992. Evaluation of the bacteriological quality of seafood. Int. J. Food
Microbiol. 16:25–39. [PubMed]
17. Gram L., and Huss H. H.. 1996. Microbiological spoilage of fish and fish
products. Int. J. Food Microbiol. 33:121–137. [PubMed]
18. Gram L., and Melchiorsen J.. 1996. Interaction between fish spoilage bacteria
Pseudomonas sp. and Shewanella putrefaciens in fish extracts and on fish
tissue. J. Appl. Bacteriol. 80:589–595. [PubMed]
19. Gram L., Trolle G., and Huss H. H.. 1987. Detection of specific spoilage
bacteria from fish stored at low (0°C) and high (20°C) temperatures. Int. J.
Food Microbiol. 4:65–72.
20. Gram L., Ravn L., Rasch M., Bruhn J. B., Christensen A. B., and Givskov M..
2002. Food spoilage – Interactions between food spoilage bacteria. Int. J.
Food Microbiol. 78:79–97. [PubMed]
21. Gulsun O., Esmeray K., Serhat O., and Fatih O.. 2009. Sensory,
microbiological and chemical assessment of the freshness of red mullet
(Mullus barbatus) and goldband goatfish (Upeneus moluccensis) during
storage in ice. Food Chem. 114:505–510.
22. Howgate P. 2009. Chapter 2. Traditional methods Pp. 19–41 in Rehbein H.,
editor; and Oehlenschaläger J., editor. , eds. Fishery products, quality, safety
and autheticity. Wiley‐Blackwell, Oxford.
23. Hozbor M. C., Saiz A. I., Yeannes M. I., and Fritz R.. 2006. Microbiological
changes and its correlation with quality indices during aerobic iced storage of
sea salmon (Pseudopercis semifasciata). LWT ‐ Food Sci. Technol. 39:99–
104.
24. Huis In't Veld J. H. J. H. I.. 1996. Microbial and biochemical spoilage of foods:
an overview. Int. J. Food Microbiol. 33:1–18. [PubMed]
25. Huss H. H. 1995. Quality and quality changes in fresh fish. FAO Fisheries
Technical Paper, vol. 348. Food and Agriculture Organization of the United
Nations, Roma, Pp. 31–50.
26. ICMSF . 1986. International commission on microbiological specification for
foods. Microorganisms in food 2 sampling for microbiological analysis:
principles and specific applications. 2nd Univ. of Toronto Press, Bufalo, NY.
27. Jørgensen B. R., and Huss H. H.. 1989. Growth and activity of Shewanella
putrefaciens isolated from spoiling fish. Int. J. Food Microbiol. 9:51–62.
[PubMed]
28. Kimura K., and Kiamukura S.. 1934. Detection of the onset of decomposition
in fish meat as shown by the content of ammonia. Proc. Pacific Sci. Congress
5:3709.
29. Koutsoumanis K., and Nychas G. J. E.. 1999. Chemical and sensory changes
associated with microbial flora of mediterranean boque (Boops boops) stored
aerobically at 0, 3, 7, and 10°C. Appl. Environ. Microbiol. 65:698–706.
[PubMed]
30. Kyrana V. R., and Lougovois V. P.. 2002. Sensory, chemical and
microbiological assessment of farm‐raised European sea bass
(Dicentrarchus labrax) stored in melting ice. Int. J. Food Sci. Technol. 37:319–
328.
31. Lougovois V., Kyranas E., and Kyrana V.. 2002. Comparison of selected
methods of assessing freshness quality and remaining storage life of iced
gilthead sea bream (Sparus aurata). Food Res. Int. 36:55–560.
32. Madhusudana Rao B., and Imam Khasim D.. 2009. Hydrogen sulphide
producing bacteria as indicators of spoilage of freshwater rohu (Labeo rohita)
fish. J. Food Sci. Technol. 46:377–379.
33. Magnusson H., and Martinsdottir E.. 1995. Storage quality of fresh and
frozen‐thawed fish in ice. J. Food Sci. 60:273–278.
34. Martinsdóttir E., Schelvis R., Hylding G., and Sveinsdóttir K.. 2009. Chapter
20. Sensory evaluation of seafood Pp. 425–443 in Rehbein H., editor; and
Oehlenschaläger J., editor. , eds. Fishery products, quality, safety and
autheticity. Wiley‐Blackwell, Oxford.
35. Milne D., and Powell S. M.. 2014. Limited microbial growth in Atlantic salmon
packed in a modified atmosphere. Food Control 42:29–33.
36. Nordic Committee on Food Analysis . 2006. Aerobic count and specific
spoilage organisms in fish and fish products. NKML Newsletter 184:1–6.
37. Oehlenschläger J. 1997. Volatile amines as freshness/spoilage indicators Pp.
571–586 in Luten J. B., editor; , Børresen T., editor; and Oehlenschläger J.,
editor. , eds. Seafood from producer to consumer, integrated approach to
quality. Elservier, Amsterdam, the Netherlands.
38. Olafsdottir G. M. 1997. Methods to evaluate fish freshness in research and
industry. Trends Food Sci. Technol. 8:258–265.
39. Olafsdottir G., Lauzon H. L., Martinsdottir E., and Kristbergsson K.. 2006.
Influence of storage temperature on microbial spoilage characteristics of
haddock fillets (Melanogrammus aeglefinus) evaluated by multivariate quality
prediction. Int. J. Food Microbiol. 111:112–125. [PubMed]
40. Pedrosa‐Menabrito A., and Regenstein J. M.. 1990. Shelf‐life extension of
fresh fish‐ a review part III‐ fish quality and methods of assessment. J. Food
Qual. 13:209–223.
41. Stohr V., Joffraud J. J., Cardinal M., and Leroi F.. 2001. Spoilage potential
and sensory profile associated with bacteria isolated from cold‐smoked
salmon. Food Res. Int. 34:797–806.
42. Tarr H. L. A. 1954. Microbiological deterioration of fish post‐mortem, its
detection and control. Bacteriol. Rev. 18:1–15. [PubMed]
43. Van Spreekens K. J. A. 1974. The suitability of a modification of Long and
Hammer's medium for the enumeration of more fastidious bacteria from fresh
fishery products. Antonie Leeuwenhoek 25:213–219.
44. Vaz‐Pires P., Seixas P., Mota M., Lapa‐Guimarães J., Pickova J., Lindo A.,
et al. 2008. Sensory, microbiological, physical and chemical properties of
cuttlefish (Sepia officinalis) and broadtail shortfin squid (Illex coindetii) stored
in ice. LWT ‐ Food Sci. Technol. 41:1655–1664.
VIII. Odour Detection Methods: Olfactometry and
Chemical Sensors
Sensors (Basel). 2011; 11(5): 5290–5322.
Magda Brattoli,1 Gianluigi de Gennaro,1,* Valentina de Pinto,1 Annamaria Demarinis
Loiotile,1 Sara Lovascio,1 and Michele Penza2

Abstract

The complexity of the odours issue arises from the sensory nature of smell. From
the evolutionary point of view olfaction is one of the oldest senses, allowing for
seeking food, recognizing danger or communication: human olfaction is a protective
sense as it allows the detection of potential illnesses or infections by taking into
account the odour pleasantness/unpleasantness. Odours are mixtures of light and
small molecules that, coming in contact with various human sensory systems, also
at very low concentrations in the inhaled air, are able to stimulate an anatomical
response: the experienced perception is the odour. Odour assessment is a key point
in some industrial production processes (i.e., food, beverages, etc.) and it is
acquiring steady importance in unusual technological fields (i.e., indoor air quality);
this issue mainly concerns the environmental impact of various industrial activities
(i.e., tanneries, refineries, slaughterhouses, distilleries, civil and industrial
wastewater treatment plants, landfills and composting plants) as sources of olfactory
nuisances, the top air pollution complaint. Although the human olfactory system is
still regarded as the most important and effective “analytical instrument” for odour
evaluation, the demand for more objective analytical methods, along with the
discovery of materials with chemo-electronic properties, has boosted the
development of sensor-based machine olfaction potentially imitating the biological
system. This review examines the state of the art of both human and instrumental
sensing currently used for the detection of odours. The olfactometric techniques
employing a panel of trained experts are discussed and the strong and weak points
of odour assessment through human detection are highlighted. The main features
and the working principles of modern electronic noses (E-Noses) are then described,
focusing on their better performances for environmental analysis. Odour emission
monitoring carried out through both the techniques is finally reviewed in order to
show the complementary responses of human and instrumental sensing.

Keywords: odour detection, odour concentration, sensory methods, dynamic

olfactometry, electronic nose, sensors, sampling methods, GC-O

1. Introduction

In the last decade great attention has been paid to the issue of air quality as it directly
affects both the environmental and human health. Air pollution has mainly an
anthropogenic source: everyday industrial and commercial activities introduce an
enormous and various amount of chemicals into the ambient air. Currently, people’s
awareness of the effects of anthropic activities on the environment rises from the
sensorial perception: nowadays olfactory nuisances, coming from various livestock
buildings and industrial activities, are at the top of the list of air pollution complaints
[1–3].

An odour is a mixture of light and small molecules, also at very low concentrations
in the inhaled air, that, upon coming in contact with the human sensory system, is
able to stimulate an anatomical response: the experienced perception is the odour
[4]. Chemicals transported by the inhaled air are trapped and dissolved into the
olfactory epithelium, a small region of both nasal cavities where odorants stimulate
an electrical response of the olfactory nerves: the olfactory signal is thus transmitted
to the brain, where the final perceived odour results from a series of neural
computations. Odours are recognized thanks to the memory effect of previous
experienced smells, thus accounting for the high subjectivity of the odour perception
[5,6].

The human sense of smell has often been regarded as the least refined of all the
human senses and far inferior to that of other animals. In fact, Aristotle (384–322
BC) blames this lack of finesse on the ducts in the human nose and claims that
people who have noses with narrower ducts have a keener sense of smell, but he
cites no experimental evidence for this assertion (Aristotle in Problemata XXXIII, and
in De Sensu et Sensibili in Parva Naturalia). Moreover, the Roman philosopher
Lucretius (99–55 BC) focused on the shape of the particles as conveying the quality
of the odour and speculated on human olfaction by considering the nature and role
of the odorant particles (Lucretius in De Rerum Natura). Also, the sense of smell is
intimately linked with our emotions and aesthetics, but, despite the importance of
odour, there is a lack of a suitable vocabulary to describe odours with precision. This
is recognised by Plato in Timaeus: “the varieties of smell have no name, but they are
distinguished only as painful and pleasant”.

The sense of smell enables people to detect the presence of some chemicals in the
ambient air: in the worst cases an odour is associated with a risk perception [7,8];
anyway, generally, it is the marker for a specific situation or activity. Due to its nature,
olfaction is becoming a tool of straightforward importance in various fields, such as
food and beverage quality assessment [4,9,10] or illness detection [11]; in addition
odour is more and more often regarded as an environmental concern [12–17]: a
complaint arises just from the personal sense of smell [18–20]. The closer and closer
proximity of industrial plants and farms, very often source of bad odours, to
residential zones, really limits the acceptability of such activities and leads to citizen’s
complaints [1,3,21]. Furthermore, odours strongly affect people’s daily life and
health, as, although they do not represent a risk for human health, smells could
cause both physiological symptoms (respiratory problems, nausea, headache) and
psychological stress [22–24].

The growing concern for human and environmental well being, along with the
increasing air pollution complaints submitted to regulators and government bodies,
has promoted the necessity for effective odour impact assessment and consequent
odour emission regulation [21]. A careful investigation of the odours issue requires
odorous air measurement by applying standardized scientific methods [1,2,25,26].

Instrumental approaches to the characterization of odorants are based on the

evaluation of the odorous air chemical composition. First of all the odorous air needs
to be collected for subsequent analysis: the traditional VOCs sampling methods, like
adsorbers or metal canister and polymer bags, are taken into account. The sampling
procedures ensure the sample integrity, preserve the odour originally associated to
the sample, minimize losses and chemical-physical interaction between odorants
and the sampler medium [27,28].

Gas Chromatography coupled with Mass Spectrometry (GC/MS) has been widely
used to analyse air quality, in order to produce a list of substances involved and their
concentration [29,30], but the main limit of this technique relies on the complexity of
the odour: the perceived odour results from many volatile chemicals, often at
concentration lower than the instrumental detection limit, that interact synergistically
or additively according to unpredictable rules [1,2,4]. Furthermore GC/MS
instrumentation is expensive and does not give information about human perception,
thus not allowing a linear correlation between a quantified substance and an
olfactory stimulus [31]. Nevertheless, to overcome these limits, some efforts have
been done in order to study the behaviour of odourants in a mixture and the potential
masking phenomena that may occur [32,33], and to assess a relationship between
instrumental and olfactometric methods [34].

The most sensitive and broader range odour detector is undoubtedly the mammalian
olfactory system, whose high complexity and efficiency derive from millions of years
of evolutionary development. The limits of traditional instrumental techniques in the
matter of odours has led to growing attention to odour measurement procedures
relying on the use of the human nose as detector, in compliance with a scientific
method [4,5,35]. As occurring in the trade industry (i.e., food, beverages, perfumes,
etc.) for many years, the sensory evaluation of smells by means of panels of sensory
trained evaluators has been the main odour assessment and quantification tool: the
so-called dynamic olfactometry is the standardized method used for determining the
concentration of odours and evaluating odour complaints [36,37]. This methodology
is based on the use of a dilution instrument, called olfactometer, which presents the
odour sample diluted with odour-free air at precise ratios, to a panel of human
assessors. The examiners are selected in compliance with a standardized procedure
performed using reference gases; only assessors who meet predetermined
repeatability and accuracy criteria are selected as panelists. The odour
concentration, usually expressed in odour units (ou/m 3) is numerically equal to the
dilution factor necessary to reach the odour threshold, that is the minimum
concentration perceived by 50% of population [37,38]. According to European
standardization, 1 ou/m3 is defined as the amount of odourant that, when evaporated
into 1 m3 of gas air at standard conditions, causes a physiological response from a
panel (detection threshold) equivalent to that of n-butanol (reference gas)
evaporated into 1 m3 of neutral gas [37]. The perception of odours is a logarithmic
phenomenon [39]; for this reason, in this kind of measurements it is necessary taking
into account that odour concentration is associated to odour intensity though a
defined logarithmic relation. Using other sensorial methods, subjective parameters,
such as the hedonic tone or the perceived odour strength, could be assessed [37].

An improvement in odour determination consists of a GC-MS coupled with

olfactometric detection (GC-MS/O) [40]. The gas chromatographic separation of an
odorous air sample could be useful for identifying specific odorant components: GC-
MS/O, thus, allows a deeper comprehension of the odorant composition as concerns
the compounds’ identification and quantification, offering the advantage of a partial
correlation between the odorant chemical nature and the perceived smell [41,42].
This instrumental approach tries to solve the odour complexity issue, which is also
the main reason for the careful procedures required for the sampling of odorous air.
Anyway the odour detection remains linked to the human perception. Although the
careful choice of panel members and the use of standard procedures for odorous
sample collection and analysis allow one to obtain reliable and repeatable
olfactometric measures, thus overcoming the subjectivity due to the human olfaction
variability, increasing attention is being paid to the availability of more objective
odour evaluation methods.

The discovery of materials with chemo-electronic properties has provided the

opportunity for the development of artificial olfactory instruments mimicking the
biological system [4,9,43,44]. In the last decade a large field of scientific research
has been devoted to the development of electronic-noses (E-Noses), that are
sensor-based machines olfaction capable of discrimination between a variety of
simple and complex odours. Like human olfaction, E-Noses are based on “an array
of electronic-chemical sensors with partial specificity to a wide range of odorants and
an appropriate pattern recognition system” [45]. In contrast to the ideal gas sensors,
which are required to be highly specific to a single chemical species, sensors for E-
Nose need to give broadly tuned responses like the olfactory receptors in the human
nose: in both cases the odour quality information and recognition is ensured by the
entire pattern of responses across the sensors array, rather than the response of
any one particular sensor. Furthermore, mimicking the data processing in the
biological systems, the incoming chemo-electronic signals are processed through
the use of data reduction techniques (PCA); in both human and electronic noses,
the function of odour recognition is finally achieved by means of some form of
associative memory for the storage and recall of the previously encountered odours.
A wide variety of competing sensor technologies (conducting polymers, piezoelectric
devices, electrochemical cells, metal oxide sensors [MOX] and metal-insulator
semiconductor field effect transistors [MISFETs]) are currently available:
independently of the considered device, sensor elements have to show fast,
reproducible and reversible responses to odour samples [43,46].

This review focuses on the state of the art of both human and instrumental sensing
currently employed for odour assessment. The main features and the working
principles of dynamic olfactometry and modern E-Noses, as monitoring tools for
environmental analysis, are described. Papers comparing the performances of both
techniques are finally reviewed in order to show the complementary responses of
human and instrumental detection.

2. Sampling Methods for Odour Compounds

Sampling is a critical phase of the measurement procedure and requires particular

attention in order to avoid sample losses due to sorption on the container or line
surfaces and to minimize these interferences. Sample contaminations can easily
occur if unsuitable or unclean materials are used; furthermore samples inevitably
degrade or alter over the time: the choice of sample containers materials, the method
for collecting odour and the time allowed between sampling and analysis are the
main critical points of the sampling procedure [28,47].

Materials

Materials for odour containers and sampling lines must themselves be odourless,
undergo minimal physical or chemical reactions with the air sample and have low
permeability in order to minimize sample losses through diffusion and/or adsorption.
Stainless steel, polytetrafluoroethylene (PTFE), tetrafluoroethylene
hexafluoropropylene copolymer (Teflon™), polyvinylfluoride (Tedlar™),
polyterephtalic ester copolymer (Nalophan NA™) and glass are considered
appropriate materials for odour sampling [37,38]. Therefore, odorous air is usually
collected in stainless steel containers, called canisters, polymer bags or on
adsorbent materials [48].

Sampling Devices
Canisters are pre-cleaned evacuated cylinders useful for air sampling. Passivated
canisters represent suitable devices for volatile and apolar molecules [49], as
suggested by the most used standardized procedure [50]. The principal advantages
of their use are that the air sample is collected without any breakthrough and there
is no degradation of the trapping materials. Canisters need to be carefully
conditioned and pretreated to avoid contamination problems and require complex
sampling apparatus. Moreover the container volume is limited to a few liters, unless
greater amounts of air samples are collected by means of pressurization, and they
are more expensive than polymer bags [51,52]. Canister sampling does not work for
dynamic olfactometry; only polymer-based bags are suitable for this use.

Polymer bags are mostly used for the collection of odorous compounds. In particular,
sampling bags of materials such as Tedlar™ or Nalophan™ are considered
appropriate [37,38,53]. Several researchers have investigated the features of plastic
bags in order to verify the existence of background emissions. Keener et al. [54] and
Trabue et al. [55] have shown that Tedlar™ bags emit acetic acid and phenol, which
might bias air samples collected for olfactory analysis. Moreover, they have
demonstrated that recovery of malodorous compounds is dependent on the
residence time in the Tedlar™, bag with longer residence times leading to lower
recovery. Reported background values in commercially available bags without pre-
cleaning are in the range of 20–60 ou/m3 in Tedlar™ [56], 30–100 ou/m3 in
Nalophan™ [57] or 2–30 ou/m3 and 10–50 ou/m3 in Tedlar™ and Nalophan™,
respectively [58]. In these studies the authors have reported that flushing the bags
with non-odorous air and, in some cases coupled by heating, background levels are
reduced to about 10 ou/m3. Laor et al. [59] have tested the odour background from
new bags and the impact of sample storage in both Tedlar™ and Nalophan™ bags,
focusing on odours emitted from municipal sewage, aeration basins, sludge,
livestock manure and coffee. They have verified that the odour background from new
non-flushed Tedlar™ and Nalophan™ bags (in which fresh air have been stored for
24 h) is as high as 75–317 ou/m3 for Tedlar™ or 36–43 ou/m3 for Nalophan™. For
pre-flushed bags the background is reduced to 25–32 ou/m3 for Tedlar™ or 19–22
ou/m3 for Nalophan™. This suggests that although new modern measurement
systems allow us to detect very low odour concentrations, special caution is needed
before considering values in the range of several to low tens of ou/m 3.

Odour bags are filled using a depression pump that works on the basis of the “lung”
technique; the bag is placed inside a rigid container evacuated using a vacuum pump
[37,38,53]. This method avoids contamination because there is no direct contact
between the pump and the sample. In order to get representative and reproducible
results, it is necessary to adapt the sampling technique to the types of odour sources.
In general, when a gas sample is very concentrated and/or it is very hot and humid,
it is necessary to use a dilution device for avoiding condensation risks.

When sampling is performed by canisters or bags, the reactivity among the different
compounds could compromise air sample stability and cause artifacts. For this
reason, it is necessary that samples should be analyzed as soon as possible after
sampling in order to minimize sample losses, degradation or alteration.
Cheremisinoff [60] asserts that samples are still useful as long as 48 h after
collection. In most cases, efforts are made to assess samples within 24 h of
collection. The European Standard EN 13725/2003 states that odour samples must
be analyzed within 30 h from sampling [37].

Sampling on adsorbent materials, packed in an appropriate tube, represents a

handier sampling method than canisters and bags because it allows one to sample
a great volume of air reducing the analytes in a small cartridge. The critical point is
the choice of adsorbents (usually porous polymers or activated carbon, graphitized
carbon black and carbon molecular sieves) [51,61–63], that depends on the
chemical features of the compounds to be sampled [52]. A combination of different
adsorbents is preferred to sample a wide class of compounds without breakthrough
problems [62]. The sampling on adsorbent materials can be applied in “active” or
“passive” mode. In active sampling, a defined volume of sample air is pumped at a
controlled flow-rate. Passive or diffusive sampling occurs by direct exposure to the
atmosphere; the process is governed by the adsorption properties of sorbent and
diffusion processes [64–66]. The passive method does not require bulky and
expensive pumps, that must be regularly checked, hindering field sampling, and it
costs less than the active one. Moreover, particular care, on the choice of sampling
volume, has to be taken to avoid breakthrough problems [51,52]. However, the active
modality allows a greater and more accurate sampling volume. For both procedures
the compounds can be recovered through thermal desorption or liquid extraction
[65].

Sampling Auxiliary Devices

The sampling devices described in the previous section are used for odour
concentration monitoring in ambient air or for punctual emissions. In case of areal
emissions [67], auxiliary devices are employed, depending on source features. Areal
sources can be distinguished as active or passive. The first ones are characterized
by a measurable outward airflow (i.e., biofilters with forced aeration) while the latter
do not have a measurable airflow (i.e., landfills, cumulus, tanks, etc.). In the case of
areal sources, it is generally very difficult to cover the whole emission area during
sampling; for this reason, representative sampling sites have to be established and
it is necessary using particular auxiliary devices for collecting odorous samples [68].
The investigations are conducted using a hood or a wind tunnel, depending on the
measurement conditions. According to German VDI 3475 Bl. 1 [69] and VDI 3477
[70] a static hood should be used for sample collection on active areal sources,
selecting a portion of the area and convoying the odourous air into the stack placed
over the hood. For passive areal sources, a wind tunnel is positioned over the
emitting surface; a known neutral air flow is introduced into the device, simulating
the action of wind on the liquid or solid surface [71,72]. Different papers have focused
on the evaluation of the performance of the existing types of chambers, hoods and
tunnels used to collect volatile materials samples under different operative
conditions [73]. Hudson and Ayoko [28,72] have shown that estimates of odour
emission rates are strongly influenced by the selection of sampling device.
Comparison of emission rates derived from turbulent and essentially quiescent
sampling devices confirms that the concentrations and emission rates provided by
these devices are quite different. Moreover emission rates measured with these
devices are subject to external influences, including ambient wind speed and
direction and the permeability of the emitting surface [72]. For improving the
performance of these devices and optimizing efficiency parameters, special
sampling chamber extension and a sampling manifold with optimally distributed
sampling orifices have been developed for the wind-tunnel sampling system [74] and
a suitable sampling system has been designed for the simulation of specific odour
emission rates from liquid area sources without outward flow [75].

3. Sensory Methods

Sensory measurements employ the human nose as the odour detector, relating
directly to the properties of odours as experienced by humans. Sensory
measurement techniques can be divided into two categories:

1. Quantitative measurements which couple the nose with some

instrumentation;
2. Parametric measurements in which the nose is used without any other device.

3.1. Instrumental Sensory Measurement

Dynamic Olfactometry

Instrumental sensory measurements employ the human nose in conjunction with an

instrument, called olfactometer, which dilutes the odour sample with odour-free air,
according to precise ratios, in order to determine odour concentrations.

The variables which will affect olfactometric measurements [12] are:

 - olfactometer design;
 - test procedure;
 - differing sensitivity of observers;
 - data quality;
 - measurement uncertainty.

Olfactometer design. The materials used in olfactometer construction should not

cause sample contamination or alteration through adsorption/desorption. Low-
adsorbency materials such as stainless steel, Teflon, Tedlar™ or glass are used and
internal surface areas are minimized. Risks of contamination can be prevented also
supplying neutral air between the successive presentations.

Test procedure. In the choice of the order of sample presentation to the panel, it is
important to consider that a descending order can enhance the effects of
adsorption/desorption, and moreover it could provoke olfactory adaptation in
panelists, since a weak odour (highest dilution) is more difficult to detect after
exposure to a strong odour (lower dilution). Nevertheless, when dilutions occurr in a
stict order, this kind of presentation can affect the panel response, because panelists
expect subsequent samples to be weaker or stronger. Among these problems, the
effects due to the choice of a descending order are more relevant, so an ascending
order presentation is preferred [12].

There are two standardized methods for the presentation of odour sample to the
panel: forced choice and yes/no method [37,38,53]. In the forced choice method, two
or more sniffing ports are used; the odour sample is presented at one port, and
neutral air at the other port(s). In this case, the examiners have to compare the
different presentations and to choose the port from which odour exits. In the yes/no
method each examiner sniffs from a single port and communicates if an odour is
detected or not. Odour samples diluted with neutral air or only neutral air can exit
from the sniffing port.

Sampling odour mixtures at different dilutions are presented to a group of selected

panelists for sniffing and their responses are recorded. Generally, the first mixture
presented to an odour panel is diluted with a very large volume of air in order to be
undetectable by the human nose. In subsequent presentations, the volume of diluent
is decreased by a predetermined and constant factor. After having set the factor, it
is possible to create a geometric progression of dilutions (for example power factor
of two: 216, 215, 214,…) useful to describe the logarithmic relation between odour
intensity and concentration [39]. The process continues until each panelist positively
detects an odour in the diluted mixture; at this stage the panelist has reached the
detection threshold for that odour [37,38,53]. This threshold is calculated as the
geometric mean between the dilution of the last negative answer and the dilution of
the first positive answer. The geometric mean is preferred for taking into account the
logarithmic relation between odour intensity and concentration [39]. Different
measurement cycles are carried out and the final result is calculated as the
geometric mean of the values obtained for single series, as mentioned before [76].

The concentration is expressed as the dilution required for achieving panel detection
threshold. Mathematically, the concentration is expressed as [77]:

C=(V0+Vf )/V0 (1)

where C is the odour concentration, V0 the volume of odorous sample and Vf the
volume of odour-free air required to reach the threshold.

By analogy, for a dynamic olfactometer the concentration is given by:

C= (Q0+Qf ) /Q0 (2)

where Q0 is the flow of odorous sample and Qf the flow of odour-free air required to
reach the threshold.
The concentrations may be expressed as threshold odour numbers (TON) or dilution
to threshold (D/T) ratios. Although the concentrations are dimensionless, it is
common to consider them as physical concentrations, and to express them as odour
units per cubic meter (ou/m3) [77,78].

Sensitivity of observers: panel selection. Panelists are qualified examiners used as

sensors in olfactometric analysis and their olfactive response (odour threshold) is
the measured parameter for calculating odour concentrations. However, the
sensitivity to odours is variable among different individuals, so panelists could
indicate different odour concentrations for the same sample. This effect is minimized
because the examiners are selected according to a standardized procedure in order
to choose individuals with average olfactive sensitivity, who constitute a
representative sample of human population [37,38,53,79]. The screening is usually
performed using reference gases [37,38,53,79]. In particular, the most used
reference gas is n-butanol and only assessors who meet predetermined repeatability
and accuracy criteria for this gas are selected as panelists [37]:

 - average n-butanol odour threshold in a range of 20–80 ppb (40 ppb

represents the accepted odour threshold for n-butanol)
 - antilog standard deviation of individual responses less than 2.3.

Panelists must be continuously screened and trained and they must observe a
simple behaviour code [34,35,50], whose fundamental prescription is that panelists
impaired by illness caused by a cold or other indispositions are excluded from
measurements.

Olfactometric data quality. Olfactometric data quality can be estimated according to

two sources of uncertainty: the panel referability to a standard and the coherence of
panel responses. In order to ensure the referability, the laboratory performances are
evaluated by accuracy and precision measures. The assumption is that the
laboratory performance to the standard odourant can be transferred to all odours
tested by the laboratory. An example of criteria applied to verify the laboratory
performance is reported as follows [37]:

 - Aod ≤ 0.217, where Aod indicates the laboratory accuracy;

 - r ≤ 0.477 or 10r ≤ 3.0, where r indicates the laboratory precision, meaning
that the difference between the results from any two consecutive
measurements will not be larger than a factor three (3.0) for 95% of the cases.

The coherence of panel results can be estimated according to a validation procedure

that permits one to exclude panel members who give invalid responses. An example
of this type of procedures is represented by “retrospective screening” [37], based on
the valuation of ΔZ parameter, calculated for each individual panel response as the
ratio between the individual threshold value ZITE and the geometric mean of all
individual threshold values Z̄ ITE obtained during a measurement sequence:

If  ZITE ≥ ẔITE then  ΔZ=ZITE/ẔITE (3)

If ZITE< ẔITE then ΔZ= −ẔITE/ZITE (4)

This parameter must satisfy the following relation:

−5≤ΔZ≤5
(5)

If one or more individual threshold values do not satisfy this criterion, then all
responses given by the panel member with an inadequate ΔZ must be eliminated by
the final result and the procedure is repeated until all data provided by panel member
are consistent with the criterion. The ΔZ parameter indicates the coherence of panel
members’ responses and puts in evidence the gaps eventually present compared to
the mean. Moreover, so a measurement may be considered valid it is necessary that
each panel member does not commit mistakes of more than 20% for the detection
of neutral air [37].

In addition to these standardized procedures, different studies have focused on the

determination of the analytical characteristics of the olfactometric method (reliability
and robustness) with the purpose of determining the operating conditions influencing
the final uncertainty associated with the measurements. In this field additional
procedures for improving accuracy and repeatability of olfactometric measure, by
optimization of panel selection [80], or by editing a quality control protocol based on
interlaboratory comparison studies [81–83] have been evaluated. Moreover, panel
repeatability tests have also been performed by presenting to panelists the same
environmental odour sample or standard odorant multiple times during one test
[84,85]. During these experiments, it has been shown that the time exposure affects
panel response and that the optimal duration for the employment of analysts in a
measure session is equal to two hours. By applying statistical methods, such as
ANOVA, it has been demonstrated that olfactometric variance is mainly affected by
within group variance compared to between group variance [84,85].

Measure uncertainty. Different attempts have been carried out for estimating a total
uncertainty to assign to olfactometric measurements. As specified before, in this
evaluation it is necessary to take into account the fact that the relation between odour
intensity and odour concentration is logarithmic [39]. For this reason, the confidence
interval is not symmetric around the average value [83,84]. It is possible to calculate
an upper (UL) and a lower limit (LL) of the 95% confidence interval of the odour
threshold, according to the following relations [86]:

lg ZUL=M+t s/N−−√N (6)
lg ZLL=M+t s/N−−√N (7)

where:

t Student factor depending on f = L – W – 1

f number of variances
L total of measuring sequences
W number of measuring sequences for series of measurements
N number of panelists
M arithmetic mean
S standard deviation
Field Olfactometric Measurements

It would be ideal to carry out odour measurements directly at the odorous site,
allowing continuous sampling of the odour without the need for storage.
Unfortunately, this approach involves the need to isolate the panel of observers from
the surrounding environment and to maintain them in an odour-free environment to
prevent olfactory adaptation or fatigue. Usually in situ measurements can be
performed using mobile laboratories even if their provision is much expensive.
Instead of direct olfactometry, it is preferable to collect odour samples in situ and
transfer them to an off-site odour laboratory for the assessment.

In 1958 the U.S. Public Health Service sponsored the development of an instrument
and a procedure for field olfactometry (a technique only suitable for ambient odour
concentration measurements). The first field olfactometer, called scentometer, is a
hand-held device that allows one to evaluate odours on site. A field olfactometer
creates a series of dilutions by mixing the odorous ambient air with odour-free
(carbon-filtered) air. The U.S. Public Health Service method defines the dilution
factor as Dilution to Threshold, D/T. The Dilution-to-Threshold ratio is a measure of
the number of dilutions needed to make the odorous ambient air non detectable.

The advantages of scentometry are that it is economically attractive and readings

are taken on site. Disadvantages include odour fatigue, because it is difficult not to
expose the sniffer to the ambient environment (which is often odorous) before the
scentometer is used, lack of dilution options and inability to rate sniffers against their
ability to sense a known reference concentration. Because this test is conducted on
site, some concern has been expressed regarding the ability of the sniffers to remain
objective when they are seeing sources of odour emissions. These include rapid
saturation of olfactory senses by some odorants, individual variation in sensitivity to
different odours, fatigue as a result of adaptation, and changes in climatic variables
(temperature, humidity, and wind speed) when measuring odours under field
conditions, as well as effects of age, gender, health and personal habits on the sense
of smell of individual panelists [87,88].

Two commercially available field olfactometers include the original scentometer,

developed in the late 1950s, and the Nasal Ranger™, introduced to the market in
2002. These devices are used in studies regarding the evaluation of odour impact
and have been compared with dynamic olfactometry or electronic noses [88],
showing that Nasal Ranger field olfactometer is efficient at measuring livestock farm
odour, and can provide consistent and accurate measuring results.

Hybrid Instrumentation: Gas Chromatography-Olfactometry (GC-O)

The opportunity of using sensory perception for the development of conventional

instruments for chemical analysis has been investigated. Gas chromatography-
olfactometry (GC-O) technique couples the traditional gas chromatographic analysis
with sensory detection, in order to study complex mixtures of odorous compounds
[40]. The GC-olfactometer consists of a traditional GC where a split column
distributes the eluate between a conventional detector, such as a flame-ionization
detector (FID) or a mass spectrometer (MS), and a sniffing port where a properly
trained person or panel could detect the active odour species. All commercially
available olfactometric ports are glass or PTFE cones, fitting the nose shape; the
eluate is delivered through a dedicated transfer line which is heated to avoid the
condensation of semivolatile analytes. In order to prevent the nasal mucous
membrane drying, especially in long time analysis, auxiliary gas (humid air) is added
to the eluate [89,90]. The sensory responses are recorded in an olfactogram: the
eluate splitting occurs allowing the analytes to reach both human and instrumental
detectors simultaneously, in order to compare the chromatogram with the
olfactogram [89,91].

The combination of a mass spectrometer with an olfactometric detector is particularly

advantageous as it allows the identification of odour-active compounds. Anyway, to
avoid different retention times due to the different working pressure of the two
detectors (a mass spectrometer and an olfactometer work under vacuum and
atmospheric pressure conditions, respectively), particular attention is required for
device assembling and in the choice of carrier and auxiliary gas flows [92].

Several methods have been developed to perform both qualitative and quantitative
evaluation of the odour related to each analyte leaving the chromatographic column
[89,93]. Dilution analysis methods, such as Charm (Combined Hedonic Aroma
Response Measurement) Analysis [89,91,94,95] and AEDA (aroma extract dilution
analysis) [89,91,96], are based on stepwise sample dilution, usually by a factor of
two or three: each dilution is sniffed until no odour is detected, thus the highest
dilution factor (FD) still allowing the odour perception is the odorant FD value. In the
AEDA olfactogram each odorant is represented by a bar whose height is proportional
to the odorant FD. In the Charm Analysis the beginning and the end of each odour
perception is also taken into account, thus the olfactogram peaks combine the smell
duration with the odour concentration [89,91]. Detection frequency methods use a
group of assessors instead of one or two of them: the odour intensity of each
compound is measured by means of the number of evaluators simultaneously
detecting the odour at the sniffing port [97]. In direct intensity measurement methods,
the intensity of the odour of the eluting compound is measured by means of different
kinds of quantitative scales, thus single, time-averaged measurements,
measurement registered after the elution of the analyte (posterior intensity
evaluation method) or dynamic measurement (OSME, fingerspan method) are used
[89,90,98]. The GC-O technique indicates the relevance of some chemicals in an
odorant allowing the assessment of single compounds, but it does not provide
information on their behaviour in a mixture [89].

The GC-O technique is widely used for the evaluation of food aromas [41,89,99–
102], but its application in the environmental field is increasing, thus moving the
odour emission assessment, from the solely olfactometric evaluation to the
characterization of volatile components causing the odour nuisance. Odours emitted
from animal production facilities have been often investigated by the GC-O approach
in order to identify the compounds responsible of the primary odour impact and
produce a deep screening of VOCs emitted in such activities by applying the GC-
MS analysis [42,103–112]. It is often found that some compounds, due to their low
odour threshold, can generate a high olfactory stimulus also at very low
concentration; furthermore some odours are perceived at the olfactometric port also
when the odorant compound is below the instrumental detection limit. Anyway the
GC-O technique does not allow the evaluation of the additive and/or synergic effect
of the single odorants in the true odour mixture, it limits its use to quantify the overall
odour intensity [103–106].

Due to the high complexity of real odorous air samples, multidimensional GC is

revealing a more powerful tool to allow a better livestock air resolution [42,108–111].
MDGC-O has also been employed to investigate the VOCs-odour-particular matter
(PM) interactions, as suspended particulate is an important odour carrier [112].

3.2. Parametric Sensory Measurements

Parametric sensory measurements have the advantage of being quick to obtain at

relatively low cost, as no particular equipment is required. Particular care has to be
taken for interpretation of results due to the variation in odour perception, even for
well-trained personnel [77]. Parameters which may be subjectively measured
include odour character, odour intensity and hedonic tone.

 - Odour character, often called odour quality, is a nominal scale of

measurement. Odours can be characterized using a reference vocabulary
with a standard list of descriptor terms [113].
 - Perceived odour intensity is the relative strength of the odour above the
recognition threshold (suprathreshold). Odour intensity is measured using
several methods, including: descriptive category scales, magnitude
estimation, and reference scales. There are several scales that usually
employ 3–10 categories and panelists must assess the intensity of the sample
according to the specified scale. The most common applied scale counts six
categories [60,78,114] from no odour to very strong odour.

Systematic measurements on wastewater plants and waste treatment facilities and

landfills have demonstrated that the intensity level of 3 (in a scale of six categories,
it represents a distinct odour) is associated with an odorant concentration of
approximately 4 ou/m3 [76].

Magnitude estimation is a procedure that compares the intensity of one odour with
another odour. In this case, the assessor assigns an arbitrary value of intensity to
the first odorant perceived and then attributes another value to the second sample
on the basis of the first. This method is very difficult to apply to different types of
odours, and is best suited for comparing similar odours [113].

The American Society for Testing and Materials recommends an intra-modal factory
matching procedure with the use of an odour reference scale for the evaluation of
suprathreshold odour intensity [115]. This standard presents two methods for
referencing the intensity of ambient odours to a standard scale: dynamic-scale and
static-scale. For dynamic scale dynamic olfactometry procedure is used, for static
scale a test by a set of bottles with fixed dilutions of a standard odorant in a water
solution [113] is performed.

 - Hedonic tone defines the pleasantness and unpleasantness of an odorant.

A method for determination of hedonic odour tone has been standardized
[116]. Dilutions are presented through an olfactometer to the panelists. If the
panelist detects an odour, the hedonic odour tone of the perceived
concentration must be evaluated according to a category scale ranging from
−4 (“extremely unpleasant”) through zero (“neither pleasant nor unpleasant”)
to +4 (“extremely pleasant”) [76]. The influence of hedonic tone and intensity
as suitable parameters for valuating odour impact and odour annoyance for
residents living in the area surrounding industrial activities has been studied
in several scientific works and taken into account in some government
guidelines [17,21,117–119].

4. Electronic Noses and Olfaction Systems: Overview and Principles of

Operation

Despite the importance of our perception of odour and flavour, there are problems
in comparing different persons’ experience of smell and in quantifying odour. This
need has created a desire for a more analytical approach to the quantitative
measurement of odour. For this purpose the field of instrumental analyzers such as
Electronic Noses (E-Noses) and Olfaction Systems (Machine Olfaction) has been
developed in response to this desire [120,121].

The Electronic Nose is a device developed to reproduce the human olfactory system.
It consists of three main parts:

 - sampling system of odours to be analyzed;

 - sensor system based on array of multiple sensing elements, or chemical
sensors;
 - data analysis and signal processing unit for feature extraction and significant
information.
The response of the chemical sensors with partial selectivity is measured upon
exposure to the sampled odour or multicomponent gas-mixture. The pattern based
on the overall response of a sensor array defines a chemical fingerprint related to a
given sampled odour. The recorded data of the sensors array response towards
various odours can be usually processed by pattern recognition techniques (i.e.,
artificial neural networks, multivariate statistical analysis) for their classification in
order to identify odour and quantify the concentration. A proper set of features can
be extracted from the recorded dataset to enhance the classification of odours
without loss of significant information.

Despite the efforts to arrive at a universal device that can achieve fine discrimination
of flavours, perfumes, smells, odours, analytes, and eventually replace the human
nose, the E-Nose is not a chemical analyzer and thus must be trained for any specific
application. However, this technical limitation of the E-Nose is combined to the
potential ability of human odour sensing by increasing the number of performing
individual sensors. This ability of the E-Nose to operate as biomimetic mammalian
olfaction should be demonstrated yet. Nevertheless, there are strong drivers to apply
E-Noses in the field of olfaction because alternatives, e.g., human test panels, either
are not practicable or are too expensive and time-consuming/ In particular, E-Noses
offer the advantages of real-time, in situ and remote control for olfactometric controls
of air-emissions.

The term electronic nose first appeared in the literature in 1988 proposed by Gardner
[122], it was discussed in a workshop on chemosensory techniques [123], and finally
defined in 1994 [45]. Gopel et al. [124] in 1990 demonstrated the application of
multicomponent analysis in chemical sensing for gas and odour detection. Ryan et
al. [125] from NASA employed an E-Nose in the Space Shuttle to monitor air quality
in the cabin. D’Amico et al. [126] demonstrated the monitoring of biological odour
filtration in closed environments with olfactometry and electronic noses. Sberveglieri
et al. [127] proposed a comparison of the performance of different features in sensor
arrays for an E-Nose. Gardner et al. [128] proposed the development of a new
olfaction system, called electronic Mucosa (e-Mucosa), based on advanced pattern
recognition algorithms for space and time classification of odorants. Romain et al.
[129] recently reviewed the use of metal oxide gas sensors for E-Nose environmental
applications.

The detection of odours has been applied to many industrial applications. They
include indoor air quality, health care, safety, security, environmental monitoring,
quality control of beverage/food products and food processing, medical diagnosis,
psychoanalysis, agriculture, pharmaceuticals, biomedicine, military applications,
aerospace, detection of hazardous gases and chemical warfare agents.

Chemical Sensors for E-Noses: Materials and Transducers

Chemical sensors for E-Nose applications need to be responsive to molecules in the

gas phase. Many different types of gas sensors are available and some of them have
been used in E-Noses at one time or another; however, nowadays, commercial
instruments take into account two main types of gas sensors (metal oxide [MOX]
and conducting polymer [CP] resistive sensors). Recent studies are focused on the
evaluation of other types of solid-state gas sensors.

Chemical sensors comprise an appropriate and chemically-sensitive material

interfaced to a transducer, as shown in Figure 1. Hence, the solid-state sensors are
essentially constituted by a chemically sensitive interface (sensitive material) and a
transducer able to convert a chemical input (gas concentration or ions concentration)
and/or physical input (temperature, pressure, acceleration, etc.) into an output,
generally an electrical signal, by means of a conditioning and/or signal processing
electronics [122]. The input magnitudes or measurands include chemical and/or
biological magnitudes such as concentration and identity of unknown species in
gaseous or liquid phase, other than physical general magnitudes such as
temperature, pressure, speed, acceleration and force. A transduction process is
realized by converting the input-event or measurand into an output electrical signal
(analogue voltage or current, digital voltage) correlated to the measurand that
generates it. The output electrical signal is properly conditioned, processed and
stored for analysis.

Figure 1. Scheme of a solid-state chemical sensor with gas-sensitive material,

transducer and interface electronics.

Gas sensors, based on the chemical sensitivity of semiconducting metal oxides, are
readily available commercially and have been more widely used to make arrays for
odour measurement than any other single class of gas sensors. An in-depth
overview on sensor materials for odour detection can be found in literature [130,131].
The most common sensor materials for odour measurements are listed in Table 1.

Table 1. Most used gas-sensitive materials for chemical sensors.

Class of Materials Sensor Materials Technology

Thin-film metal SnO2, ZnO, WO3, In2O3, TiO2, MoO3, -Sputtering

oxides (MOX) etc. - Evaporation
Class of Materials Sensor Materials Technology

-Electrochemical
Conducting
Polypirroles, polytiophenes, etc. -Casting
polymers (CP)
- Spin-coating

Supramolecular Metal-porphyrins, phthalocyanines, -Electrochemical

materials etc. - Solvent casting

-High-temperature
SnO2, ZnO, WO3, In2O3, TiO2, MoO3,
Thick-films MOX material processing
etc.
- Sol-gel

Functional inorganic Metal catalysts (Pt, Pd, Au, Ag, Ru, Ti, -Sputtering
materials W, Ta, Mo, Cu, etc.), dopants, etc. - Evaporation

Cavitands, receptors, enzymes, -Casting

Molecular organic
antibodies, proteins, biomolecules, - Langmuir-Blodgett
materials
DNA, etc.

-Langmuir-Blodgett
Composites Fillers in host-matrix -Chemical routes
- PVD techniques

MOX nanostructures: nanowires,

nanotubes, nanorods, nanocrystals,
-CVD
nanoparticles, etc.
Nanomaterials -PVD
Carbon nanostructures : nanotubes,
- Chemical routes
nanowalls, nanofibers, nanoplatelets,
etc.

(PVD = Physical Vapor Deposition ; CVD= Chemical Vapor Deposition)

The classification of chemical sensors can be realized according to the transducer

used. The various categories of solid-state chemical sensors are differentiated by
the physical principle of the signal transduction by distinguishing the following
transducers: conductometric (resistive), optical, electrochemical,
mechanical/acoustic or ultrasonic, thermal and MOSFET. A detailed classification of
the solid-state chemical sensors is given in Table 2, showing the principle of
operation, the methods of sensor fabrication and some technical comments.
Additional definitions and principles of operation have been reported in literature
[132,133].

Table 2. Transducers used in chemical solid-state sensors.

 Methods of
Transducer Principle of operation Input/Output
fabrication

PVD
Electrical Conductivity:
Microfabrication Δc → Δσ → Δi →
Conductometric •Conducting Polymers
MEMS ΔV
• Metal Oxides
Screen printing

Absorption; Emission
Fluorescence Dip coating
Δc → Δn → ΔI →
Optical Chemiluminescence MEMS
Δi →ΔV
Evanescent Wave Microfabrication
Fiber Optics

Ionic Conductivity: Screen printing

•Amperometric Dip coating Δc → Δσ → Δi →
Electrochemical
•Potentiometric MEMS ΔV
• Voltammetric Microfabrication

Flow of thermal energy:

•Catalytic PVD Δc → ΔT → Δi →
Thermal
•Pyroelectric Microfabrication ΔV
• Calorimetric

Charge capacitive Δc → ΔΦ → Δi →
MOSFET Microfabrication
coupling ΔV

Piezoelectricity: PVD
Ultrasonic
•QCM Screen printing Δc → Δm → Δf
or Mechanical
•SAW Microfabrication Δc → Δm → Δf, Δϕ
or Acoustic
• TFBAR MEMS

MEMS = Micro Electro-Mechanical Systems; QCM = Quartz Crystal Microbalance;

SAW = Surface Acoustic Wave; TFBAR = Thin Film Bulk Acoustic Resonator;

Δc = variation of concentration; Δσ = variation of electrical conductivity; Δi = variation

of current;

ΔV = variation of voltage; Δn = variation of refractive index; ΔI = variation of light

intensity;

ΔT = variation of temperature; ΔΦ = variation of work function; Δm = variation of

mass;
Δf = variation of frequency; Δϕ = variation of phase of acoustic wave

The measurements of the odour concentration by solid-state sensors implemented

in the E-Nose should be standardized. Hence, the definition of the sensor
parameters is essentially in this issue [132,133].

The main sensor parameters are:

 - Sensitivity: is a measure of the magnitude of the output signal produced in

response to a given input magnitude (perturbation/stimulus), or the ratio
between two non-homogeneous magnitudes output signal/input measurand.
 - Response time: indicates the time that the sensor signal takes to pass from
10% to 90% of its excursion to reach a new steady state, during the response
dynamics.
 - Recovery time: indicates the time that the sensor signal takes to pass from
90% to 10% of its excursion to reach a new steady state, during the recovery
dynamics.
 - Resolution: is the measure of the minimal variation of the input magnitude
to which the sensor is able to response for a given signal-to-noise ratio, at a
fixed working point.
 - Limit of Detection (LOD): is the minimum gas concentration that a sensor is
able to detect for a given signal-to-noise ratio.
 - Selectivity: characterizes the capability of the sensor to distinguish a given
input magnitude from another measurand belonging to a different class.
 - Drift: is the attitude of sensor output signal caused not by an external input
but by intrinsic reasons (sensor material, electronics) of the sensor.
 - Stability: is the attitude of the sensor to keep constant in the time its
metrological characteristics; in other words, its response in the time.
 - Repeatability: is the attitude of the sensor output signal towards a given fixed
input measurand in different repeated measurements.

Applications of E-Noses for Environmental Analysis

The application sectors of E-Noses for odour monitoring are indicated as follows:

 - measurement of odours produced by factories causing a public nuisance

 - measurement and quantification of airborne odours from other sources:
sewage farms, waste sites, agricultural activities, cattles, cars, etc.
 - measurement of odours inside buildings that may arise from harmful building
materials, faulty heating, ventilation systems
 - measurement of odours in workplaces to preserve worker health.

Many multiparameter portable sensor-systems have been studied and exploited in

field measurements for air quality control of toxic pollutants (NO x, CO, SO2, H2S)
[134], greenhouse (CO2, CH4) [134,135], refrigerant gases [135], warfare agent
simulants [136] with wireless functionalities [137] in urban areas [138] by using
traditional (chemoresistive) [134,135,139,140] and innovative (SAW) [136,141]
transducers.

Moreover, practical portable devices [142–145] have been developed for odour
monitoring of landfill municipal sites and for odour quantification by a sensor array.
In particular, Persaud et al. [143–145] used a single-point E-Nose instrument for
continuous monitoring along the perimeter of a municipal landfill site by measuring
methane and carbon dioxide as main components in a biogas produced by waste
fermentation.

Additionally, the E-Nose has been applied in in situ measurements [146–153] for the
identification of malodours sources [149], to control odour concentration emitted
from a malodour agricultural site [147] and a compost hall [151], to monitor the odour
emission from construction materials [150] and for the classification of fruity odours
[153], including odour emissions from a biofiltration system in a pig farm building
[152].

The new trends in the odour detection are strongly driven by nanotechnologies and
nanomaterials [154–157]. Nanotechnology has attracted a lot of attention recently,
particularly in the research and industrial communities. It offers many opportunities
for advancing our ability to impact on day-life and environment. The ability to design,
synthesize and manipulate specific materials at nanoscale lies at the very heart of
the future promise of nanotechnology. Nanomaterials may have unique physical and
chemical properties not found in their bulk counterparts, such as unusually large
surface area to volume ratios or high interfacial reactivity. Such properties can be
useful to develop new chemical capabilities arising from exciting new classes of
nanomaterials (e.g., nanotubes, nanowires, nanocrystals, nanoparticles, etc.).
Several studies concerning the use of nanomaterials as gas sensor materials have
been reported in literature. Penza et al. [155] studied an array of four sensors based
on carbon nanotube layers functionalized with metal catalysts for landfill gas
monitoring applications. Lieber et al. [157] developed an individual silicon-nanowire
to implement a field effect transistor (FET) functionalized with DNA and proteins for
detection of biological and chemical species in the area of healthcare and life
sciences. This device may be called a nanosensor. However, these nanosensors
based on individual nanowires have been integrated by Cheng et al. [154] in an array
of multiple sensing elements to implement a nanoelectronic nose based on hybrid
nanowire/nanotubes and micromachining technology for sensitive gas
discrimination. This nanoelectronic nose has great potential to detect and
discriminate a wide variety of gases, including explosives, nerve agents and odours.

5. Olfactometry and E-Noses: Comparison and Integrated Approach

As concerning the different techniques applied to odour determination previously

discussed, whose characteristics are summarized in Table 3, it was shown that no
one of the described techniques is able alone to give exhaustive informations about
the odorous emissions from different kinds of human activities that may cause
olfactory nuisance. Therefore, a comparison and/or an integration of the olfactometry
methods with the technologies of sensorial analysis is necessary to completely
evaluate odour impact [158].

Table 3. Characteristics of odour measurement techniques.

Other sensorial Electronic GC-

Olfactometry methods Nose O

–
Objective measurement of + + +

odour concentration
–
Quantitative measurement of + + +

odour concentration

Measurement standardization + +/– – –

Continuous measurement – +/– + –

Single species determination – – – +

Temporal representativity of – +/– + –

measurement

Time of analysis +/– – + +/–

Costs + +/– – +

(+ = high;

+/– = medium;

– = low)

Several correlations can be observed between trends in the discrimination properties

of the electronic nose and the human olfactory system [159]. Since E-Noses are not
able to provide odour concentrations, many authors have focused their attention on
the research of a correlation between olfactometric and sensorial results in order to
realize a fast, portable and not very expensive device to carry out frequent odour
measurements in case of complaints from the public or in presence of unstable odour
compounds.
The dynamics of odour emissions from a pig barn have been investigated by
olfactometry and using an electronic odour sensor. The sensor signals showed a
good relation to the odour concentration and revealed a promising potential of
electronic odour sensors to detect the dynamic and the level of odour concentrations
[160].

On samples from pig and chicken slurry electronic nose measurements based on
polypyrrole sensors have been evaluated against odour concentration
measurements by the olfactometric technique; electronic nose sensitivity was found
to be lower than the olfactometry one, showing the need to develop sensors to
specific groups of compounds [161].

Thus, an electronic nose equipped with 14 gas sensors suitably selected for
measuring odorous components from livestock farms has been developed. The
responses of the sensors have been found to be in good agreement with the
perceived odour intensities of a portable field olfactometer [88], and both the data
sets, used to train an expert system for supporting odour management of livestock
and poultry farm, have made possible to forecast the effectiveness of odour control
efforts before those control means were applied [162].

An electronic nose based on conducting polymer sensors, has been widely applied
in the analysis of odour samples from swine manure sources coupled with a NH 3
and a H2S sensor [163,164] and as an alternative to sensory analysis for assessing
the effectiveness of biofilters, showing good correlation with odour concentrations
[165]; together with olfactometry and gas chromatography to analyze indoor air from
swine finishing facilities, instead, the correlation between GC/MS analyses and E-
Nose was found better than between E-Nose and olfactometry. This result
suggested that human panelist responses may be based on detection of compounds
that are not included in GC/MS quantification procedures and are not well detected
by this electronic nose [166].

An electronic nose was used in an experimental farm to quantify the odour level
inside the animal room and a good correlation was found with the olfactometric
results on the same samples. E-Nose results showed an evolution of the odour with
animal activities during the day and with their age [167].

Sohn et al. used an artificial neural network, trained by the data sets obtained with
an electronic nose and dynamic dilution olfactometry, to predict the pig farm odour
concentrations emanating from an effluent pond and to develop a confident, rapid,
and cost-effective technique for odour measurement [168]; in addition they
demonstrated the relationship between odour emission rates and pond loading rates
on pig farm effluent ponds and the increased magnitude of emissions from a heavily
loaded effluent pond [169].

As concerns livestock farms, they also employed olfactometry and electronic nose
results to demonstrate the odour monitoring capability of a non-specific conducting
polymer-based array for evaluating the performance of a biofiltration system installed
at a commercial pig farm [152] and to develop an odour prediction model using PLS
(Partial Least Squares) to investigate the relationship between odour concentrations
inside the poultry shed and factors such as weather, bird age, ventilation rates and
other variables associated with the broiler production cycle [170].

Agricultural sources can also be a source of complaints, so a device able to carry

out field measurements is required. After application of cattle slurry to grassland, two
olfactometers and two electronic noses were used, demonstrating the ability of both
E-Noses to respond to the odour concentrations arising from cattle slurry
applications at levels which would be similar to those from a range of agricultural
sources [171].

Applying PCA (Principal Component Analysis) and then PLS regression, a good
correlation between odour units and sensors data of E-Nose has been found in odour
measurements from a rendering plant bio-filter inlet and outlet [172], and in
investigations on the organic fraction of municipal solid waste [173], demonstrating
that a correctly calibrated E-Nose could replace olfactometry as a tool for odour
impact measurement.

On the other hand, studying samples from different sewage treatment works, a
comparison between results of an electronic nose and dynamic olfactometry showed
there is no universal relationship between the electronic nose responses and odour
concentrations for sewage odours from a range of locations within different treatment
works, but only for odour samples which are source/site specific [174,175]. The
same result was obtained also on wastewater samples from different treatment
works. [176].

Experimental studies have been carried out with an E-Nose to determine the
detection limits of the selected sensors, using olfactometric measurements of odour
detection threshold concentration, and the sensors capability of discriminating
different odours in waste treatment plants. The sensors characterized by low
detection limits for the considered odorants, also showed a good capability of
discriminating these odorants from each other [177].

Moreover the use of a chemosensor system, calibrated with olfactometric data on a

waste incineration plant, allowed continuous monitoring behind a charcoal filter and
thus the identification of filter breakthrough [178].

Among the human activities that may generate problems related to unpleasant odour
emissions, landfills represent one of the major causes of odour complaints [16]: they
are difficult to monitor as they are characterized by a great variety of substances that
may cause odour nuisance and then they require the use of more than one technique
for odour determination.

For a complete characterization of odours at a landfill, Capelli et al. collected

samples in different zones inside the plant, at the boundaries and at the receptors,
and analyzed them with different techniques: olfactometry enabled a quantification
of the landfill odour emissions, giving indicative values of sensory impacts; chemical
analyses with GC-MS were useful to analyze odour composition, and electronic
noses (two at the boundaries and one at the nearest receptor) were used as a
management tool in order to monitor site changes or operational failures. This study
has shown that even if the results of the three different odour characterization
techniques do not necessarily correlate, each one contributes to solve the complexity
of odour measurement in the environment [179].

Other comprehensive investigations on landfill areas used olfactometry with a

dispersion modelling, odour patrol monitoring and an E-Nose [180]; dynamic
olfactometry, field determination of odour perception points and electronic noses to
create a calibration curve that allowed the translation of the global E-Nose response
into odour concentration units that could be compared to a warning threshold
concentration [181]. Another approach was carried out using results of olfactometric
analysis as the input for a dispersion model and two electronic noses for continuous
monitoring to determine the landfill odour impact on a specific receptor, and a very
good correspondence of the electronic nose responses with the odour detections
reported by the people living at the receptor and with the result of the odour
dispersion modelling was found [182].

Some authors used data sets, obtained by evaluating odour samples with both an
olfactometer and an electronic nose, to train artificial neural networks (ANN) and
develop a function to convert the measurements of an electronic nose into odour
concentrations. The odour concentrations measured with the olfactometer have
been used as observed values, and the responses of the electronic nose as input
variables [183]. Using this technique on composting plants, it was possible to get
characteristic patterns only for different parts of the plant, but, for these parts a good
similarity between the samples was shown [184].

For the estimation of odour disturbances from the biofilters for the treatment of
emissions from a municipal solid waste organic fraction composting plant, dynamic
olfactometry has been used to determine odour intensity and to verify the standards
of odour disturbance in combination with an electronic nose. Once a correlation
between the two methods was established, it was possible to carry out frequent
quantitative determinations of the biofilter emissions by simply using the electronic
nose, with consequent lower costs than dynamic olfactometry analysis [158].

The possibility of monitoring the time evolution of the odour concentration has also
allowed the use of an electronic nose suitably calibrated by olfactory measurements
to supply a warning signal when the compost odour is identified and exceeds a given
threshold [151].

A problem that requires continuous monitoring, is the assessment of the presence

of odours at a particular receptor, like a house whose owners often complain about
the unpleasant odours originating from a nearby plant. For a composting plant the
electronic nose response has been correlated with the odour concentration values
measured by dynamic olfactometry in order to use the instrument for the continuous
odour concentration measurement. Two electronic noses have been installed in the
house and in the composting plant; in correspondence to the measurements during
which the electronic nose inside the house detected the presence of odours from the
composting plant, the olfactory classes recognized by both instruments coincided.
Moreover, the electronic nose at the house detected the presence of odours from
the composting plant at issue in correspondence of each odour perception of the
house occupants [185].

An E-Nose was trained to analyze different gas samples of known olfactory quality
at different odour concentration values, and then installed in two different periods at
two receptors of a composting plant. Applying an appropriate data analysis, a high
correlation index was found between true and predicted odour concentration values,
thus demonstrating that an opportunely trained electronic nose and suitable data
processing methods may represent a valid solution to the problem of having a
system for continuously monitoring odours of environmental interest [186].

6. Conclusions

The increasing attention of the population to olfactory nuisances and the need to
provide a reliable qualification and quantification of odours has led to the
development of different odour measurement techniques. In particular, instrumental
sensory methods and chemical sensors have been described, showing the
advantages and disadvantages of each technique.

Although dynamic olfactometry represents the standardized objective method for the
determination of odour concentration, it is affected by some limitations. First of all
dynamic olfactometry provides point odour concentration data, however, it is not
sufficient to evaluate completely a case of olfactory nuisance because it does not
allow one to perform continuous and field measurements, useful for monitoring the
industrial processes causing odour emissions. Moreover, dynamic olfactometry
considers the whole odour mixture and do not discriminate the single chemical
compounds and their contribution to the odour concentrations. Odour samples are
difficult to store, because of their instability, and, therefore, require rapid time of
analysis. Finally, as it is well-known, olfactometry is too time-consuming and quite
expensive and moreover frequency and duration of analysis are limited.

On the other hand, electronic noses present lower analysis costs and quick results
and they allow one to carry out continuous monitoring in the field nearby sources
and receptors. After a training step, electronic noses are able to preview the class of
an unknown sample and then to associate environmental odours to a specific
source.

Since each technique satisfies only a part of the problems of odour monitoring, many
authors have focused their attention on carrying out comparisons and integrations
between olfactometry and E-Nose results. These applications show the opportunity
of using more than one approach for describing and understanding olfactory
nuisance cases as completely as possible.
References and Notes

1. Yuwono AS, Lammers PS. Odor pollution in the environment and the detection
instrumentation. Int Agr Eng J. VI. 2004:1–33.
2. Vincent A, Hobson J. Odour Control. Terence Dalton Publishing Ltd.; London, UK:
1998.
3. Stuetz R, Frechen FB, editors. Odours in Wastewater Treatment: Measurement,
Modelling and Control. IWA Publishing; London, UK: 2001.
4. Craven MA, Gardner JW, Bartlett PN. Electronic noses—Development and future
prospects. Trends Anal. Chem. 1996;15:486–493.
5. Pearce TC. Computational parallel between the biological olfactory pathway and
its analogue “The electronic nose”: Part I. Biological olfaction. BioSystems.
1997;41:43–67. [PubMed]
6. Freeman JW. The physiology of perception. Sci. Am. 1991;264:78–85. [PubMed]
7. Rosenkranz HS, Cunningham AR. Environmental odor and health hazards. Sci.
Total Environ. 2003;313:15–24. [PubMed]
8. Dalton P. Upper airway irritation, odor perception and health risk due to airborne
chemicals. Toxicol Lett. 2003;140–141:239–248. [PubMed]
9. Peris M, Escuder-Gilabert L. A 21st century technique for food control: Electronic
noses. Anal. Chim. Acta. 2009;638:1–15. [PubMed]
10. Suffet IH, Mallevialle J, Kawczynski E, editors. Advances in Taste-and-Odor
Treatment and Control. American Water Works Association Research Foundation;
Denver, CO, USA: 1995.
11. Gardner JW, Shin HW, Hines EL. An electronic nose system to diagnose illness.
Sens. Actuat. B. 2000;70:19–24.
12. Gostelow P, Parson SA, Stuetz RM. Odour measurements for sewage treatment
works. Water Res. 2001;35:579–597. [PubMed]
13. Wolkoff P, Nielsen GD. Organic compounds in indoor air-Their relevance for
perceived indoor air quality? Atmos. Environ. 2001;35:4407–4417.
14. Burgess JE, Parsons SA, Stuetz RM. Developments in odour control and waste
gas treatment biotechnology: A review. Biotechnol. Adv. 2001;19:35–63. [PubMed]
15. Héroux M, Pagé T, Gélinas C, Guy C. Evaluating odour impacts from a landfilling
and composting site: Involving citizens in the monitoring. Water Sci. Technol.
2004;50:131–137. [PubMed]
16. Sironi S, Capelli L, Centola P, Del Rosso R, II, Grande M. Odour emission factors
for assessment and prediction of Italian MSW landfills odour impact. Atmos. Environ.
2005;39:5387–5394.
17. Miedema HME, Walpot JI, Vos H, Steunenberg CF. Exposure-annoyance
relationships for odour from industrial sources. Atmos. Environ. 2000;34:2927–2936.
18. Both R, Sucker K, Winneke G, Koch E. Odour intensity and hedonic tone—
Important parameters to describe odour annoyance to residents? Water Sci.
Technol. 2004;50:83–92. [PubMed]
19. van Harreveld AP. From odorant formation to odour nuisance: new definitions
for discussing a complex process. Water Sci. Technol. 2001;44:9–15. [PubMed]
20. Wise PM, Olsson JM, Cain WS. Quantification of odor quality. Chem. Senses.
2000;25:429–443. [PubMed]
21. Nicell JA. Assessment and regulation of odour impacts. Atmos. Environ.
2009;43:196–206.
22. Wilson GE, Huang YC, Schroepfer W. Atmospheric sublayer transport and odor
control. J. Environ. Eng. Div. 1980;106:389–401.
23. Brennan B. Odour nuisance. Water Waste Treat. 1993;36:30–33.
24. Schiffman SS. Livestock odors: Implications for human health and well-being. J.
Anim. Sci. 1998;76:1343–1355. [PubMed]
25. Pierson S, Buckland A, Barrie I. The Use of Odour Measurement in the Control
of Odours. Proceedings of the Second CIWEM National Conference on Odour
Control in Sewage Treatment; London, UK. 30 April 1998.
26. Gostelow P, Longhurst PJ, Parsons SA, Stuetz RM. Sampling for Measurement
of Odours. IWA Publishing; London, UK: 2003.
27. Jang J, Kaye R. Sampling techniques for odour measurement. In: Stuetz R,
Frechen FB, editors. Odours in Wastewater treatment: Measurement, Modelling and
Control. IWA Publishing; London, UK: 2001. pp. 98–117.
28. Hudson N, Ayoko GA. Odour sampling 1: Physical chemistry considerations.
Bioresource Technol. 2008;99:3982–3992. [PubMed]
29. Davoli E, Gangai ML, Morselli L, Tonelli D. Characterisation of odorants
emissions from landfills by SPME and GC/MS. Chemosphere. 2003;51:357–368.
[PubMed]
30. Dincer F, Odabasi M, Muezzinoglu A. Chemical characterization of odorous
gases at a landfill site by gas chromatography–mass spectrometry. J. Chromatogr.
A. 2006;1122:222–229. [PubMed]
31. Di Francesco F, Lazzerini B, Marcelloni F, Pioggia G. An electronic nose for
odour annoyance assessment. Atmos. Environ. 2001;35:1225–1234.
32. Kim K-H. Experimental demonstration of masking phenomenon between
competing odorants via an air dilution sensory test. Sensors. 2010;10:7287–7302.
[PMC free article] [PubMed]
33. Kim K-H. The averaging effect of odorant mixing via air dilution sensory test: A
case study on reduced sulfur compounds. Sensors. 2011;11:1405–1417. [PMC free
article] [PubMed]
34. Kim K-H, Park S-Y. A comparative analysis of malodor samples between direct
(olfactometry) and indirect (instrumental) methods. Atmos. Environ. 2008;42:5061–
5070.
35. Walker JC. The performance of the human nose in odour measurement. Water
Sci. Technol. 2001;44:1–7. [PubMed]
36. Schulz TJ, van Harreveld AP. International moves towards standardisation of
odour measurement using olfactometry. Water Sci. Technol. 1996;34:541–547.
37. 2003. EN13725: Air Quality—Determination of Odour Concentration by Dynamic
Olfactometry; Committee for European Normalization (CEN), Brussels, Belgium.
38. 2004. ASTM E679-04: Standard Practice for Determination of Odor and Taste
Thresholds by a Forced-Choice Ascending Concentration Series Method of Limits;
American Society for Testing and Materials, Philadelphia, PA, USA.
39. Stevens SS. The psychophysics of sensory functions. Am. Sci. 1960;48:226–
253.
40. Leland JV, Scheiberle P, Buettner A. In: Gas Chromatography–Olfactometry The
State of the Art. Acree TE, editor. American Chemical Society; Washington, DC,
USA: 2001.
41. Friedrich JE, Acree TE. Gas Chromatography Olfactometry (GC/O) of dairy
product. Int. Dairy J. 1998;8:235–241.
42. Lo YCM, Koziel JA, Cai L, Hoff SJ, Jenks WS, Xin H. Simultaneous chemical
and sensory characterization of volatile organic compounds and semi-volatile
organic compounds emitted from swine manure using solid phase microextraction
and multidimensional Gas Chromatography–Mass Spectrometry–Olfactometry. J.
Environ. Qual. 2008;37:521–534. [PubMed]
43. Pearce TC. Computational parallel between the biological olfactory pathway and
its analogue “The electronic nose”: Part II. Sensor-based machine olfaction.
BioSystems. 1997;41:69–90. [PubMed]
44. Snopok BA, Kruglenko IV. Multisensor systems for chemical analysis: State-of-
the-art in Electronic Nose technology and new trends in machine olfaction. Thin Solid
Films. 2002;418:21–41.
45. Gardner JW, Bartlett PN. A brief history of electronic noses. Sens Actuat B.
1994;18–19:211–220.
46. Ampuero S, Bosset JO. The electronic nose applied to dairy products: A review.
Sens. Actuat. B. 2003;94:1–12.
47. Duffee RA, Cha SS. Consideration of physical factors in dynamic olfactometry.
J. Air Pollut. Control Assoc. 1980;30:1294–1295.
48. Krol S, Zabiega a B, Namiesnik J. Monitoring VOCs in atmospheric air II. Sample
collection and preparation. Trends Anal. Chem. 2010;29:1101–1112.
49. Wang DKW, Austin CC. Determination of complex mixtures of volatile organic
compounds in ambient air: Canister methodology. Anal Bioanal Chem.
2006;386:1099–1120. [PubMed]
50. Compendium of Methods for the Determination of Toxic Organic Compounds in
Ambient air. 2nd ed. US Environmental Protection Agency; Cincinnati, OH, USA:
1999. EPA Compendium method TO-14A: Determination of volatile organic
compounds (VOCs) in ambient air using specially prepared canisters with
subsequent analysis by gas chromatography.
51. Camel V, Caude M. Trace enrichment methods for the determination of organic
pollutants in ambient air. J. Chromatogr. A. 1995;710:3–19.
52. Kumar A, Víden I. Volatile organic compounds: sampling methods and their
worldwide profile in ambient air. Environ. Monit. Assess. 2007;131:301–321.
[PubMed]
53. Australian/New Zealand Standard™ Stationary Source Emissions Part 3:
Determination of Odour Concentration by Dynamic Olfactometry. Standards
Australia International Ltd.; Sydney, Australia: 2001. AS/NZS 4323.3:2001.
54. Keener KM, Zhang J, Bottcher RW, Munilla RD. Quantification of odorants from
Tedlar bags. Proceedings of ASABE Annual Meeting; Chicago, IL, USA. 6 April
2010; p. 024165.
55. Trabue SL, Anhalt JC, Zahn JA. Bias of Tedlar bags in the measurement of
agricultural odorants. J. Environ. Qual. 2006;35:1668–1677. [PubMed]
56. Parker DB, Rhoades MB, Koziel J, Spinhirne J. Background Odors in Tedlar®
Bags Used for Cafo Odor Sampling. Proceedings of ASABE Annual Meeting; Las
Vegas, NV, USA. July 29 2003; p. 034144.
57. Juarez-Galan JM, Martinez JV, Amo A, Valor I. Background odour from sampling
bags. Influence in the analysis of the odour concentration. Chem. Eng. Trans.
2008;15:87–94.
58. Miller RM, McGinley MA. Evaluation of background odour in Tedlar and
Nalophan sample bags. Proceedings of Water Environment Federation,
WEF/A&WMA Odors and Air Emissions; Phoenix, AZ, USA. 6–8 April 2008; pp. 590–
604.
59. Laor Y, Ozer Y, Ravid U, Hanan A, Orenstein P. Methodological aspects of
sample collection for dynamic olfactometry. Chem. Eng. Trans. 2010;23:55–60.
60. Cheremisinoff PN. Industrial Odour Control. Butterworth-Heinemann Ltd; Oxford,
UK: 1988.
61. Gawrys M, Fastyn P, Gawlowski J, Gierczak T, Niedzielski J. Prevention of water
vapour adsorption by carbon molecular sieves in sampling humid gases. J.
Chromatogr. A. 2001;933:107–116. [PubMed]
62. Harper M. Review: Sorbent trapping of volatile organic compounds from air. J.
Chromatogr. A. 2000;885:129–151. [PubMed]
63. Matisová E, Škrabáková S. Carbon sorbents and their utilization for the
preconcentration of organic pollutants in environmental samples. J. Chromatogr. A.
1995;707:145–179.
64. Gorecki T, Namiesnik J. Passive sampling. Trends Anal. Chem. 2002;21:276–
291.
65. Bruno P, Caselli M, de Gennaro G, Solito M, Tutino M. Monitoring of odor
compounds produced by solid waste treatment plants with diffusive samplers. Waste
Manage. 2007;27:539–544. [PubMed]
66. Seethapathy S, Gorecki T, Li X. Passive sampling in environmental analysis. J.
Chromatogr. A. 2008;1184:234–253. [PubMed]
67. Blatt 2: Umweltmeteorologie, Emissionen von Gasen, Gerühen und Stäuben aus
diffusen Quellen, Deponien (Environmental meteorology, emission of gases, odour
and dust from diffuse sources, landfills) Beuth Verlag; Berlin, Germany: 2000. VDI-
Richtlinie 3790.
68. Bockreis A, Steinberg I. Measurement of odour with focus on sampling
techniques. Waste Manage. 2005;25:859–863. [PubMed]
69. Blatt 1 Entwurf: Emissionsminderung, Biologische Abfallbehandlungsanlagen,
Kompostierung und Vergärung, Anlagenkapazität P0.75 Mg/h (P6750 Mg/a)
(Emission reduction, biological waste treatment plants, composting and digestion,
plant size P0.75 Mg/h (P6750 Mg/a)) Beuth Verlag; Berlin, Germany: 2000. VDI-
Richtlinie 3475.
70. Biologische Abgasreinigung: Biofilter (Biological waste gas purification:
Biofilters) Ausgabe: Beuth Verlag; Berlin, Germany: 2004. VDI-Richtlinie 3477.
71. Frechen FB, Frey M, Wett M, Löser C. Aerodynamic performance of a low-speed
wind tunnel. Water Sci. Technol. 2004;50:57–64. [PubMed]
72. Hudson N, Ayoko GA. Odour sampling. 2. Comparison of physical and
aerodynamic characteristics of sampling devices: a review. Bioresour. Technol.
2008;99:3993–4007. [PubMed]
73. Jiang J, Kaye R. Sampling techniques for odour measurement. In: Stuetz R,
Frechen FB, editors. Odours in Wastewater Treatment: Measurement, Modelling
and Control. IWA Publishing; London, UK: 2008. pp. 95–119.
74. Wang X, Jiang J, Kaye R. Improvement of a wind-tunnel sampling system for
odour and VOCs. Water Sci. Technol. 2001;44:71–77. [PubMed]
75. Capelli L, Sironi S, del Rosso R, Céntola P. Design and validation of a wind
tunnel system for odour sampling on liquid area sources. Water Sci. Technol.
2009;59:1611–1620. [PubMed]
76. Frechen FB. Odour measurement and odour policy in Germany. Water Sci.
Technol. 2000;41:17–24.
77. Koe LCC. Sewage odors quantification. In: Cheremisinoff PN, editor.
Encyclopedia of Environmental Control Technology, Wastewater Treatment
Technology. Gulf Publishing Company; Houston, TX, USA: 1989. pp. 423–446.
78. Frechen FB. Odour emissions of wastewater treatment plants: Recent German
experiences. Water Sci. Technol. 1994;30:35–46.
79. The Offensive odor control law Latest Amendment by Law. Ministry of the
Environment Government of Japan; Chiyoda-ku, Tokyo, Japan: 1995. Available
online: http://www.env.go.jp/en/laws/air/odor/cm.html (accessed on 20 February
2011).
80. Capelli L, Sironi S, Del Rosso R, Centola P, Bonati S. Improvement of
olfactometric measurement accuracy and repeatability by optimization of panel
selection procedures. Water Sci. Technol. 2010;61:1267–1278. [PubMed]
81. van Harreveld AP. A review of 20 years of standardization of odor concentration
measurement by dynamic olfactometry in Europe. J. Air Waste Manage. Assoc.
1999;49:705–715.
82. Maxeiner B. Olfactometric Interlaboratory Comparison Test 2005. Proceedings
of the A&WMA/WEF 2006 Odors and Air Emissions; Hartford, CT, USA. 9–12 April,
2006; pp. 688–699.
83. Higuchi T. Estimation of uncertainty in olfactometry. Water Sci. Technol.
2009;59:1649–1655. [PubMed]
84. McGinley MA, McGinley CM. Precision of Olfactometry and Odor Testing
Results. Water Environment Federation/Air & Waste Management Association
Specialty; Proceedings of Odors and Air Emissions; Hartford, CT, USA. 9–12 April
2006.
85. Bruno P, Caselli M, Brattoli M, de Gennaro G, de Gennaro L, De Leonibus MA,
Parenza AE. Analytical characteristics of odour concentration measure by dynamic
olfactometry: Preliminary results. Chem. Eng. Trans. 2008;15:129–133.
86. Olfactometry-Odour Threshold Determination, Part 1: Fundamentals. Vol. 1
Verein Deutsche Ingenieure Verlag; Düsseldorf, Germany: 1986. VDI-guideline
3881, VDI-Handbuch, Reinhaltung der Luft.
87. Rappert S, Muller R. Odor compounds in waste gas emissions from agricultural
operations and food industries. Waste Manage. 2005;25:887–907. [PubMed]
88. Pan L, Yang SX, de Bruyn J. Factor analysis of downwind odours from livestock
farms. Biosyst. Eng. 2007;96:387–397.
89. van Ruth SM. Methods for gas chromatography-olfactometry: A review. Biomol.
Eng. 2001;17:121–128. [PubMed]
90. Plutowska B, Wardencki W. Application of gas chromatography–olfactometry
(GC–O) in analysis and quality assessment of alcoholic beverages—A review. Food.
Chem. 2008;107:449–463.
91. Benzo M, Gilardoni G, Gandini C, Caccialanza G, Finzi PV, Vidari G, Abdo S,
Layedra P. Determination of the threshold odor concentration of main odorants in
essential oils using gas chromatography-olfactometry incremental dilution
technique. J. Chromatogr. A. 2007;1150:131–135. [PubMed]
92. Hochereau C, Bruchet A. Design and application of a GCSNIFF/MS system for
solving taste and odour episodes in drinking water. Water Sci. Technol. 2004;49:81–
87. [PubMed]
93. Acree TE, Barnard J. In: Trends in Flavour Research. Maarse H, van der Heij
DG, editors. Elsevier; New York, NY, USA: 1994. pp. 211–220.
94. Acree TE, Barnard J, Cummingham D. A procedure for the sensory analysis of
gas chromatographic effluents. Food. Chem. 1984;14:273–286.
95. Debonneville C, Orsier B, Flament I, Chaintreau A. Improved hardware and
software for quick gas chromatography–olfactometry using CHARM and GC-‘SNIF’
analysis. Anal. Chem. 2002;74:2345–2351. [PubMed]
96. Grosch W. Detection of potent odorants in foods by aroma extract dilution
analysis. Trends Food Sci. Tech. 1993;4:68–73.
97. van Ruth SM. Evaluation of two gas chromatography-olfactometry methods: The
detection frequency and perceived intensity method. J. Chromatogr. A.
2004;1054:33–37. [PubMed]
98. Ferreira V, Pet’ka J, Aznar M, Cacho J. Quantitative gas chromatography-
olfactometry. Analytical characteristics of a panel of judges using a simple
quantitative scale as gas chromatography detector. J. Chromatogr. A.
2003;1002:169–178. [PubMed]
99. Curioni PMG, Bosset JO. Key odorants in various cheese types as determined
by gas chromatography-olfactometry. Int. Diary J. 2002;12:959–984.
100. Hallier A, Courcoux P, Sérot T, Prost C. New gas chromatography-olfactometric
investigative method, and its application to cooked Silurus glanis (European catfish)
odor characterization. J. Chromatogr. A. 2004;1056:201–208. [PubMed]
101. Machiels D, van Ruth SM, Posthumus MA, Istasse L. Gas chromatography-
olfactometry analysis of the volatile compounds of two commercial Irish beef meats.
Talanta. 2003;60:755–764. [PubMed]
102. Garruti DS, Franco MRB, da Silva MAAP, Janzantti NS, Alves GL. Assessment
of aroma impact compounds in a cashew apple-based alcoholic beverage by GC-
MS and GC-olfactometry. LWT. 2006;39:372–377.
103. Zahn JA, Hatfield JL, Do YS, DiSpirito AA, Laird DA, Pfeiffer RL.
Characterization of volatile organic emissions and wastes from a swine production
facility. J. Environ. Qual. 1997;26:1687–1696.
104. Schiffman SS, Bennett JL, Raymer JH. Quantification of odors and odorants
from swine operations in North Carolina. Agr. Forest Meteorol. 2001;108:213–240.
105. Rabaud NE, Ebeler SE, Ashbaugh LL, Flocchini RG. Characterization and
quantification of odorous and non-odorous volatile organic compounds near a
commercial dairy in California. Atmos. Environ. 2003;37:933–940.
106. Kai P, Schäfer A. Identification of key odour components in pig house air using
hyphenated gas chromatography olfactometry. Agr Eng Int: CIGR J Sci Res.
2004;VI:1–11.
107. Koziel JA, Cai L, Wright D, Hoff SJ. Solid-phase microextraction as a novel air
sampling technology for improved, GC-Olfactometry-based assessment of livestock
odors. J. Chromatogra. Sci. 2006;44:451–457. [PubMed]
108. Bulliner EA, Koziel JA, Cai L, Wright D. Characterization of Livestock odors
using steel plates, solid-phase microextraction, and multidimensional gas
chromatography-mass spectrometry-olfactometry. J. Air Waste Manage. Assoc.
2006;56:1391–1403. [PubMed]
109. Laor Y, Koziel JA, Cai L, Ravid U. Chemical-Sensory Characterization of Dairy
Manure Odor Using Headspace Solid-Phase Microextraction and Multidimensional
Gas Chromatography Mass Spectrometry-Olfactometry. J. Air Waste Manage.
Assoc. 2008;58:1187–1197. [PubMed]
110. Koziel JA, Lo YCM, Cai L, Wright DW. Simultaneous characterization of VOCs
and Livestock odors using solid-phase microextraction—multidimensional gas
chromatography- mass spectrometry-olfactometry. Chem. Eng. Trans. 2010;23:73–
78.
111. Zhang S, Cai L, Koziel JA, Hoff SJ, Schmidt DR, Clanton CJ, Jacobson LD,
Parker DB, Heber AJ. Field air sampling and simultaneous chemical and sensory
analysis of livestock odorants with sorbent tubes and GC–MS/olfactometry. Sens.
Actuat. B. 2010;146:427–432.
112. Cai L, Koziel JA, Lo YC, Hoff SJ. Characterization of volatile organic
compounds and odorants associated with swine barn particulate matter using solid-
phase microextraction and gas chromatography–mass spectrometry–olfactometry.
J. Chromatogr. A. 2006;1102:60–72. [PubMed]
113. St. Croix Sensory, Inc.; Lake Elmo, MN, USA: 2005. A Review of The Science
and Technology of Odor Measurement.
114. Olfactometry-Determination of Odour Intensity, Part 1. Vol. 1 VDI-guideline
3882/1, VDI-Handbuch Reinhaltung der Luft, Verein Deutsche Ingenieure Verlag;
Düsseldorf, Germany: 1992.
115. Standard Practice for Referencing Suprathreshold Odor Intensity. ASTM
International; Philadelphia, PA, USA: 2004. E544-99.
116. Olfactometry-Determination of Hedonic Odour Tone, Part 2. Vol. 1 VDI-
Handbuch Reinhaltung der Luft, Verein Deutsche Ingenieure Verlag; Düsseldorf,
Germany: 1994. VDI-guideline 3882/2.
117. Sarkara U, Hobbs SE. Odour from municipal solid waste (MSW) landfills: a
study on the analysis of perception. Environ. Int. 2002;27:655–662. [PubMed]
118. Sucker K, Both R, Bischoff M, Guski R, Winneke G. Odor frequency and odor
annoyance. Part I: Assessment of frequency, intensity and hedonic tone of
environmental odors in the field. Int. Arch. Occup. Environ. Health. 2008;81:671–
682. [PubMed]
119. Sucker K, Both R, Bischoff M, Guski R, Winneke G. Odor frequency and odor
annoyance Part II: dose–response associations and their modification by hedonic
tone. Int. Arch. Occup. Environ. Health. 2008;81:683–694. [PubMed]
120. Gardner JW, Bartlett PN. Electronic Noses-Principles and Applications. Oxford
University Press; Oxford, UK: 1999.
121. Rock F, Barsan N, Weimar U. Electronic Nose: Current status and future trends.
Chem. Rev. 2008;108:705–725. [PubMed]
122. Gardner JW. Pattern recognition in the Warwick electronic nose. Proceedings
of 8th International Congress of the European Chemoreception Research
Organisation; Coventry, UK. 18–22 July 1988; p. 9.
123. Gardner JW, Bartlett PN. Pattern recognition in gas sensing. In: Moseley P,
Norris J, Williams D, editors. Techniques and Mechanisms in Gas Sensing. Adam
Hilger; Bristol, UK: 1991. pp. 347–380.
124. Vaihinger S, Gopel W. Multicomponent analysis in chemical sensing. In: Gopel
W, Jones TA, Kleitz M, Lundstrom I, Seiyama T, editors. Sensors: A Comprehensive
Study: Chemical Sensors. 2/3. VCH; Weinheim, Germany: 1990. pp. 191–237.
125. Ryan MA, Zhou H, Buehler MG, Manatt KS, Mowrey VS, Jackson SP, Kisor
AK, Shevade AV, Honer ML. Monitoring space shuttle air quality using the jet
propulsion laboratory electronic nose. IEEE Sens. J. 2004;4:337–347. [PubMed]
126. Willers H, de Gijsel P, Ogink N, D’Amico A, Martinelli E, Di Natale C, van Ras
N, van der Waarde J. Monitoring of biological odour filtration in closed environments
with olfactometry and an electronic nose. Water Sci. Technol. 2004;50:93–100.
[PubMed]
127. Pardo M, Sberveglieri G. Comparing the performance of different features in
sensor arrays. Sens. Actuat. B. 2007;123:437–443.
128. Che Harun FK, Taylor JE, Covington JA, Gardner JW. An electronic nose
employing dual-channel odour separation columns with large chemosensor arrays
for advanced odour discrimination. Sens. Actuat. B. 2009;141:134–140.
129. Romain AC, Nicolas J. Long term stability of metal oxide-based gas sensors for
e-nose environmental applications: An overview. Sens. Actuat. B. 2010;146:502–
506.
130. Sberveglieri G, editor. Gas Sensors. Kluwer Academic; Dordrecht, The
Netherland: 1992.
131. Gopel W, Jones TA, Kleitz M, Lundstrom I, Seiyama T, editors. Sensors—A
Comprehensive Survey—Chemical and Biochemical Sensors, Part I-II. 2–3 Wiley-
VCH; Weinheim, Germany: 1992.
132. D’Amico A, Di Natale C. A contribution on some basic definitions of sensors
properties. IEEE Sens. J. 2001;1:183–190.
133. D’Amico A, Di Natale C, Taroni A. Sensors Parameters in Sensors for domestic
applications. In: D’Amico A, Sberveglieri G, editors. Proceedings of the First
European School on Sensors (ESS’94) World Scientific; Singapore: 1995.
134. Al-Ali AR, Zualkernan I, Aloul F. A mobile GPRS-sensors array for air pollution
monitoring. IEEE Sens. J. 2010;10:1666–1671.
135. Delpha C, Lumbreras M, Siadat M. Discrimination and identification of a
refrigerant gas in a humidity controlled atmosphere containing or not carbon dioxide:
Application to the electronic nose. Sens. Actuat. B. 2004;98:46–53.
136. Alizadeh T, Zeynali S. Electronic nose based on the polymer coated SAW
sensors array for the warfare agent simulants classification. Sens. Actuat. B.
2008;129:412–423.
137. Qu J, Chai Y, Yang SX. A real-time de-noising algorithm for E-Noses in a
wireless sensor network. Sensors. 2009;9:895–908. [PMC free article] [PubMed]
138. Ma Y, Richards M, Ghanem M, Guo Y, Hassard J. Air pollution monitoring and
mining based on sensor grid in London. Sensors. 2008;8:3601–3623.
139. Burgeois W, Romain AC, Nicolas J, Stuetz RM. The use of the sensor arrays
for environmental monitoring: interests and limitations. J. Environ. Monitor.
2003;5:852–860. [PubMed]
140. Pijolat C, Riviere B, Kamionka M, Viricelle JP, Breuil P. Tin dioxide gas sensor
as a tool for atmospheric pollution monitoring: Problems and possibilities for
improvements. J. Mater. Sci. 2003;38:4333–4346.
141. McGill RA, Nguyen VK, Chung R, Shaffer RE, DiLella D, Stepnowski JL, Mlsna
TE, Venezky DL, Dominguez D. The “NRL-SAWRHINO” a nose for toxic gases.
Sens. Actuat. B. 2000;65:10–13.
142. Micone PG, Guy C. Odour quantification by a sensor array: An application to
landfill gas odours from two different municipal waste treatment works. Sens. Actuat.
B. 2007;120:628–637.
143. Persaud KC, Woodyan NCP, Sneath RW. Development of a perimeter odor
monitoring system for landfill sites. Proceedings of IEEE Sensors 2008 Conference;
Lecce, Italy. 26–29 October 2008; pp. 1360–1363.
144. Persaud KC, Wareham P, Pisanelli AM, Scorsone E. Electronic Nose—New
condition monitoring devices for environmental applications. Chem. Senses.
2005;30:i252–i253. [PubMed]
145. Multisensor Systems Homepage. Available online:
http://www.multisensor.co.uk (accessed on 15 February 2011).
146. Koziel J. What is that smell? Nature. 2008;45:726–728.
147. Persaud KC, Khaffaf SM, Hobbs PJ, Sneath RW. Assessment of conducting
polymer odour sensors for agricultural malodour measurements. Chem. Senses.
1996;21:496–505. [PubMed]
148. Romain AC, Nicolas J. Monitoring malodours in the environment with an
electronic nose: requirements for the signal processing. In: Gutierrez A, Marco S,
editors. Biologically Inspired Signal Processing for Chemical Sensing. Vol. 188.
Springer; Berlin: 2009. pp. 121–134.
149. Romain AC, Andre P, Nicolas J. Three years experiment with the same tin oxide
sensor arrays for the identification of malodours sources in the environment. Sens.
Actuat. B. 2002;84:217–277.
150. Romain AC, Degrave C, Nicolas J, Lor M, Vause K, Dinne K, Maes F, Goelen
E. Olfactometry, chemical and e-nose measurements to characterize odor emission
of construct materials for the implementation of the European construction products
directive (CPD) on a Belgian level. Proceedings of ISOEN; Brescia, Italy. 15–17 April
2009; pp. 527–528.
151. Nicolas J, Romain AC, Ledent C. The electronic nose as a warning device of
the odour emergence in a compost hall. Sens. Actuat. B. 2006;116:95–99.
152. Sohn JH, Dunlop M, Hudson N, Kim TI, Yoo YH. Non-specific conducting
polymer-based array capable of monitoring odour emissions from a biofiltration
system in a piggery building. Sens. Actuat. B. 2009;135:435–464.
153. Tang KT, Chiu SW, Pan CH, Hsieh HY, Ling YS, Liu SC. Development of a
portable electronic nose systems for the detection and classification of fruity odors.
Sensors. 2010;10:9179–9193. [PMC free article] [PubMed]
154. Cheng PC, Ishikawa FN, Chang HK, Ryu K, Zhou C. A nanoelectronic nose: A
hybrid nanowire/carbon nanotube sensor array with integrated micromachined
hotplates for sensitive gas discrimination. Nanotechnology. 2009;20:125503.
[PubMed]
155. Penza M, Rossi R, Alvisi M, Serra E. Metal-modified and vertically-aligned
carbon nanotube sensors array for landfill gas monitoring applications.
Nanotechnology. 2010;21:105501. [PubMed]
156. Fryxell GE, Cao G, editors. Environmental Applications of Nanomaterials.
Imperial College Press; London, UK: 2007.
157. Patolsky F, Lieber CM. Nanowire nanosensors. Mater Today. 2005:20–28.
158. Littarru P. Environmental odours assessment from waste treatment plants:
Dynamic olfactometry in combination with sensorial analysers “electronic noses”
Waste Manag. 2007;27:302–309. [PubMed]
159. Doleman BJ, Lewis NS. Comparison of odor detection thresholds and odor
discriminablities of a conducting polymer composite electronic nose versus
mammalian olfaction. Sens. Actuat. B. 2001;72:41–50.
160. Brose G, Gallmann E, Hartung E, Jungbluth T. Detection of the dynamics of
odour emissions from pig farms using dynamic olfactometry and an electronic odour
sensor. Water Sci. Technol. 2001;44:59–64. [PubMed]
161. Hobbs PJ, Misselbrook TH, Pain BF. Assessment of odours from livestock
wastes by a photoionization detector, an electronic nose, olfactometry and gas
chromatography-mass spectrometry. J. Agr. Eng. Res. 1995;60:137–144.
162. Pan L, Yang SX. A new intelligent electronic nose system for measuring and
analysing livestock and poultry farm odours. Environ. Monitor. Assess.
2007;135:399–408. [PubMed]
163. Qu G, Omotoso MM, Gamal El-Din M, Feddes JJR. Development of an
integrated sensor to measure odors. Environ. Monitor. Assess. 2008;144:277–283.
[PubMed]
164. Omotoso MM, Qu G, Gamal El-Din M, Feddes JJR. Development of an
Integrated Electronic Nose to Reliably Measure Odors. Proceedings of the Seventh
International Symposium; Beijing, China. 18–20 May 2005.
165. Gutierrez-Osuna R, Schiffman SS, Nagle HT. Correlation of Sensory Analysis
with Electronic Nose Data for Swine Odor Remediation Assessment. Proceedings
of the 3rd European Congress on Odours, Metrology and Electronic Noses; Paris,
France. 19–21 June 2001.
166. Gralapp AK, Powers WJ, Bundy DS. Comparison of olfactometry, gas
chromatography, and electronic nose technology for measurement of indoor air from
swine facilities. Trans. ASABE. 2001;44:1283–1290.
167. Fuchs S, Strobel P, Siadat M, Lumbreras M. Evaluation of unpleasant odor with
a portable electronic nose. Mater. Sci. Eng. C. 2008;28:949–953.
168. Sohn JH, Smith R, Yoong E, Leis J, Galvin G. Quantification of odours from
piggery effluent ponds using an electronic nose and an artificial neural network.
Biosyst. Eng. 2003;86:399–410.
169. Sohn JH, Smith RJ, Yoong E. Process studies of odour emissions from effluent
ponds using machine-based odour measurement. Atmos. Environ. 2006;40:1230–
1241.
170. Sohn JH, Hudson N, Gallagher E, Dunlop M, Zeller L, Atzeni M. Implementation
of an electronic nose for continuous odour monitoring in a poultry shed. Sens. Actuat.
B. 2008;133:60–69.
171. Misselbrook TH, Hobbs PJ, Persaud KC. Use of an electronic nose to measure
odour concentration following application of cattle slurry to grassland. J. Agr. Eng.
Res. 1997;66:213–220.
172. Boholt K, Andreasen K, den Berg F, Hansen T. A new method for measuring
emission of odour from a rendering plant using the Danish Odour Sensor System
(DOSS) artificial nose. Sens. Actuat. B. 2005;106:170–176.
173. Orzi V, Cadena E, D’Imporzano G, Artola A, Davoli E, Crivelli M, Adani F.
Potential odour emission measurement in organic fraction of municipal solid waste
during anaerobic digestion: Relationship with process and biological stability
parameters. Bioresour. Tech. 2010;101:7330–7337. [PubMed]
174. Stuetz RM, Engin G, Fenner RA. Sewage odour measurements using a sensory
panel and an electronic nose. Water Sci. Technol. 1998;38:331–335.
175. Stuetz RM, Fenner RA, Engin G. Assessment of odours from sewage treatment
works by an electronic nose, H2S analysis and olfactometry. Water Res.
1999;33:453c–461c.
176. Fenner RA, Stuetz RM. The application of electronic nose technology to
environmental monitoring of water and wastewater treatment activities. Water
Environ. Res. 1999;71:282–289.
177. Capelli L, Sironi S, Cèntola P, del Rosso R. MOS sensors for the recognition of
environmental odours: evaluation of sensor sensitivity towards odour concentration
and humidity. Chem. Eng. Trans. 2008;15:307–314.
178. Boeker P, Haas T, Diekmann B, Schulze LP. Continuous online odour
measurements. Chem. Eng. Trans. 2008;15:255–260.
179. Capelli L, Sironi S, Del Rosso R, Cèntola P, Grande M. A comparative and
critical evaluation of odour assessment methods on a landfill site. Atmos. Environ.
2008;42:7050–7058.
180. Li X. East Asia Workshop on Odour Measurement and Control Review, Ministry
of the Environment. Odour impact and control at a landfill site in Hong Kong; pp. 8–
86. Japan, 13–14 October 2003.
181. Romain AC, Delva J, Nicolas J. Complementary approaches to measure
environmental odours emitted by landfill areas. Sens. Actuat. B. 2008;131:18–23.
182. Snidar R, Culòs B, Trovarelli A, Soldati A, Sironi S, Capelli L. Evaluation of
odour emissions from a landfill through dynamic olfactometry, dispersion modelling
and electronic noses. Chem. Eng. Trans. 2008;15:315–322.
183. Qu G, Feddes JJR, Armstrong WW, Coleman RN, Leonard JJ. Measuring odor
concentration with an electronic nose. Trans. ASABE. 2001;44:1807–1812.
184. Bockreis A, Jager J. Odour monitoring by the combination of sensors and neural
networks. Environ. Model. Softw. 1999;14:421–426.
185. Sironi S, Capelli L, Cèntola P, del Rosso R, Grande M. Continuous monitoring
of odours from a composting plant using electronic noses. Waste Manag.
2007;27:389–397. [PubMed]
186. Sironi S, Capelli L, Cèntola P, Del Rosso R. Development of a system for the
continuous monitoring of odours from a composting plant: Focus on training, data
processing and results validation methods. Sens. Actuat. B. 2007;124:336–346.

IX. Chemical and Sensory Changes Associated with

Microbial Flora of Mediterranean Boque (Boops boops)
Stored Aerobically at 0, 3, 7, and 10°C
Appl Environ Microbiol. 1999 Feb; 65(2): 698–706.
Konstantinos Koutsoumanis and George-John E. Nychas*

ABSTRACT

The development of a microbial population and changes in the physicochemical and

sensorial characteristics of Mediterranean boque (Boops boops), called gopa in
Greece, stored aerobically at 0, 3, 7, and 10°C were studied. Pseudomonads and
Shewanella putrefaciens were the dominant bacteria at the end of the storage
period, regardless of the temperature tested. Enterobacteria and Brochothrix
thermosphacta also grew, but their population density was always 2 to 3 log 10 CFU
g−1 less than that of pseudomonads. The concentration of potential indicators of
spoilage, glucose and lactic acid, decreased while that of the α-amino groups
increased during storage. The concentrations of these carbon sources also
decreased on sterile fish blocks inoculated with strains isolated from fish microbial
flora. The organic acid profile of sterile fish blocks inoculated with the above-
mentioned bacteria and that of naturally spoiled fish differed significantly. An
excellent correlation (r = −0.96) between log10 counts of S. putrefaciens or
Pseudomonas bacteria with freshness was observed in this study.

It is well known from studies with meat that although low in comparison with those
of protein and lipids, the concentrations of compounds such as glycogen, glucose,
and lactate are all sufficient to support massive microbial growth. These compounds
can affect the type (e.g., saccharolytic, proteolytic) and rate of spoilage and,
moreover, seem to be the principal precursors of those microbial metabolites that
we perceive as spoilage (59). The concentrations of, e.g., glucose and lactate in fish
are similar to those reported for meat (43, 60, 64), and their importance has been
underestimated (31), although they have been used alternatively for sensory and
microbiological analysis to determine fish freshness (68). Among the disadvantages
of sensory analysis, which is probably the most appropriate method, and
microbiological analysis is that the reliability of the former depends on highly trained
panels to minimize subjectivity, which makes it costly and unattractive for routine
analysis. For the latter, at least in traditional form (microbial counts, etc.), we get
retrospective information and presuppose that the specific spoilage organisms are
known and detectable by a chosen technique.

For this reason, both microbiological and sensory analyses can be replaced by
biochemical methods (e.g., based on nucleotide catabolism, production of amines,
trimethylamine [TMA], and sulfur compounds) or physical methods (Torymeter, K
value) which associate findings with microbial growth on fish (29, 30). With respect
to spoilage indicators or microbial metabolites, there is a lack of relevant information
about fish compared with that about meat. The basis of these methods is that as
bacteria grow on fish, they utilize nutrients and produce by-products. Determination
of the quantities of these metabolites could provide us with information about the
degree of spoilage.

However, fish spoilage depends on specific spoilage organisms, and moreover,

these are not the same in every case and are dependent on the climatic and storage
conditions, the type of fish, and even the place in which the fish was harvested (19,
31). For example, Shewanella putrefaciens is the only specific spoilage bacterium of
marine cold-water fish stored in ice, and the number of S. putrefaciens bacteria is
inversely related to remaining shelf life (31). On the other hand, Pseudomonas sp.
and S. putrefaciens are the specific spoilage bacteria of marine and freshwater
tropical fish stored in ice (31). No report of similar studies with fish from the
Mediterranean Sea—temperate waters—is available in the literature.

The aim of this work was to provide information on (i) the microbial attributes of the
Mediterranean boque (Boops boops), a marine fish very popular with consumers
with emphasis on specific spoilage organisms and (ii) the role of microbial
metabolites (e.g., lactate, formic acetate, free amino acids, etc.) as indicators of
spoilage in fish stored aerobically at 0, 3, 7, and 10°C.

MATERIALS AND METHODS

Preparation of fish fillets.

Fresh, gutted boque (B. boops) stored in ice after capture was bought from a local
fishery shop within 6 to 8 h after being caught. The fish were transported in ice to
the laboratory within 30 to 45 min of their purchase. On arrival at the laboratory, they
were divided into quarters, and the portions were kept at 0, 3, 7, and 10°C. Three
independent storage experiments were conducted, and in each experiment there
were 12 sampling times per temperature tested. On each sampling occasion, two
fish were analyzed.
Sample preparation.

Fish (25 g) was transferred aseptically to a stomacher bag (Seward Medical, London,
United Kingdom), 225 ml of 0.1% peptone water with salt (NaCl, 0.85%, wt/vol) was
added, and the mixture was homogenized for 60 s with a stomacher (Lab Blender
400; Seward Medical).

Microbiological media and enumeration.

Samples (0.1 ml) of serial dilutions of treated (either inoculated with the isolates or
naturally inoculated) fish homogenates were spread on the surface of the
appropriate media in petri dishes for determination of the total viable count on
modified Long-and-Hammer agar (66) and incubated at 10°C for 7 days. The
medium was composed of the following (grams per liter of distilled water): Proteose
Peptone (P 0431; Sigma), 20; gelatin (4070; Merck), 40; K 2HPO4, 1; NaCl, 10; agar
(L11; Oxoid), 15; ammonium ferric citrate, 0.25. Pseudomonads were determined on
cetrimide fusidin cephaloridine agar (Oxoid code CM 559, supplemented with
selective supplement SR 103) after incubation at 20°C for 2 days (52). (iii)
Brochothrix thermosphacta was determined on streptomycin sulfate-thallous
acetate-cycloheximide (actidione) agar (Oxoid code CM 881, supplemented with
selective supplement SR 151) after incubation at 20°C for 3 days (22). For members
of the family Enterobacteriaceae and hydrogen sulfide-producing bacteria, 1.0 ml
was inoculated into 10 ml of molten (45°C) violet red bile glucose agar (Oxoid code
CM 485) and iron agar (IA; Oxoid code CM 867), respectively. After setting, a 10-ml
overlay of molten medium was added. For the former, incubation was at 30°C for 24
h. The large colonies with purple haloes were counted (54). IA plates were incubated
at 20°C for 4 days (34). Black colonies formed by production of H2S were
enumerated after 2 to 3 days (23).

Detection of Photobacterium phosphoreum was done by using the direct microscopy

method. Thirty colonies were selected randomly from modified Long-and-Hammer
medium (66) containing 100 to 200 colonies. Large, gram-negative coccobacilli
similar to those described by Dalgaard (8) were counted as P. phosphoreum (11).

Three replicates of at least three appropriate dilutions (1) were enumerated. All
plates were examined visually for typical colony types and morphological
characteristics associated with each growth medium. In addition, the selectivity of
each medium was checked routinely by Gram staining and microscopic examination
of smears prepared from randomly selected colonies from all of the media.

Bacterial strains.

Pure cultures of Pseudomonas sp., S. putrefaciens, P. phosphoreum, B.

thermosphacta, and lactic acid bacteria were screened in sterile boque fish muscle
blocks for spoilage potential, i.e., the ability to produce chemical changes typical of
the spoiling product substrate used in inoculation studies of sterile gopa fish tissue.
With the exception of P. phosphoreum, all organisms (B. thermosphacta,
Shewanella sp., Pseudomonas sp., and lactic acid bacteria) were selected at the
latest stage of aerobic storage of boque fish at 3°C. Colonies were isolated from
streptomycin sulfate-thallus acetate-cycloheximide (actidione) agar (22), IA,
Pseudomonas agar supplemented with cetrimide fusidin cephaloridine agar (52),
and MRS medium and characterized further as B. thermosphacta, S. putrefaciens,
Pseudomonas sp., and Lactobacillus sp., respectively. The catalase test was used
for further confirmation of B. thermosphacta. The isolates from Mead-and-Adams
(52) and IA media were examined for colony shape and pigmentation (44), Gram
reaction (36), cell morphology (phase-contrast microscopy), flagellar arrangement
(51), oxidase reaction (46), and aerobic and anaerobic breakdown of glucose (39).
Moreover, the API 20NE System (BioMérieux) was applied for the following tests:
nitrite reduction, indole production, arginine dihydrolase, urease, esculin hydrolysis,
gelatin hydrolysis, β-galactosidase, and assimilation of the energy sources glucose,
arabinose, mannose, mannitol, N-acetylglucosamine, maltose, gluconate, caprate,
adipate, malate, citrate, and phenyl acetate. Gram-negative, motile rods with positive
catalase and oxidase reactions, oxidative glucose metabolism, and arginine
dihydrolase activity were identified as Pseudomonas sp. if they did not reduce
trimethylamine oxide (TMAO) or produce H2S. The identity of Shewanella sp. was
further confirmed by salmon pink pigment on nutrient agar and copious H 2S
production in test tubes with soft agar composed of the following (grams per liter of
distilled water): tryptone, 10; sodium chloride, 5; Na2S2O3 · 5H2O, 0.5; l-cysteine,
0.6; ferric citrate, 0.3; agar, 4.5 (pH 7.4 ± 0.2). It was also inoculated by stabbing and
incubated at 20 to 25°C for 7 days. Copious production of H 2S was noted by
blackening of the medium after 24 h. For the characterization of Lactobacillus sp.,
production of CO2 and growth at different temperatures (3, 15, 30, and 45°C), salt
concentrations (8 and 10%), and pHs (3, 3.7, and 8.5) were used in combination with
the API 50CH tests.

The inocula were prepared as follows. Each bacterial strain was maintained on
slopes of the appropriate medium agar at 4°C. A loopful of a fresh working subculture
(ca. 106 CFU) was used for inoculation of the corresponding broth (100 ml in a 250-
ml conical flask). The flask was incubated aerobically without agitation at 25°C for
18 h. Cells were harvested, washed by centrifugation, and washed with sterile saline,
and an appropriate dilution in saline (NaCl, 0.85%, wt/vol) was used for inoculation.

For P. phosphoreum (kindly provided by P. Dalgaard, Danish Ministry of Fisheries,

Lyngby, Denmark), growth medium broth (13) was used, and it was precultured and
incubated at 15°C.

Growth on sterile fish blocks.

Fish tissue blocks were prepared as described by Gram and Melchiorsen (32). Gopa
were killed, bled, and iced immediately after being caught and then brought to the
laboratory within 5 to 6 h. The skin surface was cleaned with 100% alcohol to sanitize
the surface. The skin of the back muscle was removed aseptically, and the sterile
tissue below was excised and cut into 15- to 20-g pieces. The pieces were inoculated
immediately with the bacterial isolates. To prepare fish extract, fillets were blended
with distilled water (1:1.5) for 60 s. The homogenate was heated at 80°C for 3 min,
and then it was passed through cheese cloth. The resulting stock was subsequently
filtered with 0.45- and 0.22-μm (pore size) microfilters (Millipore) to ensure sterile
fish juice. Bacterial isolates were grown at 3 and 10°C in fish extract for 3 to 5 days,
and the cultures were diluted in 3 and 10°C sterile physiological saline. Appropriate
dilutions were prepared at the above-mentioned temperatures, and then the tissue
pieces were dipped in the bacterial suspension for 3 min in chilled, sterile, 17-cm-
diameter plastic petri dishes. The inoculated tissue pieces were placed in sterile, 17-
cm-diameter plastic petri dishes. Control tissue was dipped in sterile diluent. All
inoculations were done in triplicate, and two independent trials were carried out. One
2-g piece was removed from each sample for microbiological and physicochemical
analysis.

Curve fitting.

The growth data (12 sampling times) from the enumeration of different groups of
microbial association were fitted to estimate the maximum specific growth rate (μmax).
The Gompertz function was used (24). The modified Gompertz curve is defined by
the equation log10 N(t) = A + C exp{−exp[−B(t − M)]}, where N(t) is the bacterial count
at time t, B is the relative maximum growth rate (log10 CFU hour−1), M is the time
(hours) at which the absolute growth rate is at a maximum, A is the lowest asymptote
of the curve, and C is the upper asymptote of the curve. After fitting of the curves,
the parameters B, M, C, and A were obtained by using nonlinear regression with the
Fig.P version 2.5 software (2). The parameters were used to calculate μmax with the
equation μmax = BC/e.

pH.

The pH value was recorded by a pH meter (Metrohm 691), and the glass electrode
was applied directly to the flesh.

Chemical analysis.

Fish flesh (10 g) was transferred to a beaker, 100 ml of perchloric acid (HClO 4, 1 M)
was added, and the contents were homogenized for 1 min with a blender. The
homogenate was centrifuged (15 min at 4,000 × g), and then the supernatant was
decanted and stored at −80°C. After thawing, a portion of the deproteinized
supernatant was neutralized with KOH (5 M) and diluted with water to 1:20 to give a
solution for analysis. l-Lactic acid and d-glucose were assayed enzymatically by the
methods described by Noll (58) and Kunst et al. (47), respectively, in which 10 g was
reduced to a fine suspension with 100 ml of cold water (3 to 5°C) in an Omni mixer
(Waring, New Hartford, United Kingdom). The suspension was centrifuged (5 min at
4,000 × g at 3°C) and filtered, and the clear filtrate was used to determine water-
soluble protein and α-amino groups by using the spectrophotometric assays
described by Church et al. (6). TMA was quantitatively measured by a colorimetric
method (3). TMA content was expressed as milligrams of TMA-N per 100 g of fish.

HPLC analysis of organic acids.

After the microbiological examination, a portion (20 ml) was filtered with Whatman
no. 1 paper. The clear filtrate, after addition of trifluoroacetic acid (1% [vol/vol] final
concentration) and sodium azide (final concentration, 0.1%), was centrifuged (10
min at 3,000 × g), filtered with a Millipore 0.22-μm-pore-size filter, and stored at
−80°C before high-performance liquid chromatography (HPLC) analysis. The
profiles of the organic acids of fish were analyzed by HPLC (Spectra Physics P2000
two-pump system with a UV/VIS detector using low-inertia scanning technology—
similar to a photodiode array—and software from Spectra Physics, San Jose, Calif.)
using a Rheodyne 7125 injector and an Aminex HPX-87H 5-μm column (300 by 7.8
mm; Bio-Rad Laboratories, Richmond, Calif.). The compounds were separated
isocratically with 0.009 N H2SO4 in distilled water (flow rate, 0.7 ml min−1). The peak
width was 12, the peak threshold was 600, and the attenuation was 32. The whole
spectra (190 to 330 nm) of the chromatograms were analyzed. The solvents were
HPLC grade, and for identification of peaks, solutions of reference substances (citric,
lactic, acetic, tartaric, malic, succinic, formic, and propionic acids) were analyzed by
using the same program, and the retention times (RT) and spectra were compared.
The contribution of each identified compound was expressed as the peak area eluted
in each chromatograph. The precision of the results was always better than ±5%.

Sensory evaluation.

The dorsal half of each fillet was heated (80°C, 15 min) in an unsealed plastic bag,
and the quality was assessed by 8 to 10 trained panelists. A scoring scale with three
categories was used. Class 1 corresponded to high-quality fillets without any off odor
or off flavor, class 2 corresponded to fillets that had slight off odors or flavors but
were still acceptable, and class three corresponded to fillets of unacceptable quality.
The shelf life limit was defined as the point when 50% of the panelists rejected the
fillets (11).

Prediction of remaining shelf life.

At each temperature, the relationship between the remaining shelf life, estimated by
sensory methods, and the log of the total viable counts of Pseudomonas and H2S-
producing bacteria was calculated by using the following simple linear regression:
remaining shelf life (days) = intercept (β0) + slope (β1) × log10(microbial numbers/g).
The coefficient of determination (R2) was calculated as a measure of the percentage
of the total variability in the remaining shelf life data explained by the model. Fig.P
version 2.5 software (2) was used to fit all models and calculate the parameters.

RESULTS
Microbiological analysis.

The changes in the microbial flora of boque fish during storage under aerobic
conditions at 0, 3, 7, and 10°C are shown in Table Table1.1. For practical reasons,
Table Table11 shows only those microbiological changes (7 out of 12 to 15)
occurring in samples taken at the same time for all storage conditions, while for the
calculation of μmax, all observations were taken into account (Table (Table2).2). Total
viable counts reached ca. 9 log10 CFU g−1 by the end of the storage period under
aerobic conditions, regardless of the storage temperature used. It needs to be noted
that there was no statistically significant difference between bacterial numbers on
Long-and-Hammer medium and those on IA when all colonies were counted in the
latter medium, although in a most cases, IA and Long-and-Hammer medium gave
higher counts than plate count agar (results not shown). Pseudomonads were
dominant at that time in fish stored at all of the temperatures tested, followed by S.
putrefaciens (Table (Table1).1). However, it should be noted that the μmax of S.
putrefaciens was found to be higher than that of Pseudomonas sp. (Table (Table2)2)
and, thus, ended up close to the Pseudomonas value, even though the initial counts
were lower (1.5-log difference) (Table (Table1).1). B. thermosphacta, a bacterium
more common in meat products, was found to be the third main member of the
microbial association of boque fish (Table (Table1)1) and outgrew the
Enterobacteriaceae.

TABLE 1. Changes in total viable countsa of pseudomonads, B.

thermosphacta, S. putrefaciens, Enterobacteriaceae, and P. phosphoreumb on
Mediterranean boque stored aerobically at 0, 3, 7, or 10°C

Temp Mean total viable count ± SD

(°C)
and
storag B. S.
e time Overall Pseudomona thermosphac putrefacien Enterobacteriace
(h) ds ae
ta s

5.55 ± 0.6
NA,c 0 4.37 ± 0.45 2.77 ± 0.24 3.3 ± 0.32 2.34 ± 0.52
7

5.93 ± 0.6
52 4.69 ± 0.37 3.00 ± 0.32 3.60 ± 0.21 2.95 ± 0.64
9

6.75 ± 0.2
99 5.53 ± 0.35 3.90 ± 0.41 4.80 ± 0.12 3.00 ± 0.69
8
Temp Mean total viable count ± SD
(°C)
and
storag B. S.
e time Overall Pseudomona thermosphac putrefacien Enterobacteriace
(h) ds ae
ta s

7.62 ± 0.4
147 6.33 ± 0.32 4.70 ± 0.12 5.90 ± 0.05 3.47 ± 0.36
6

8.05 ± 1.0
172 7.02 ± 0.25 5.00 ± 0.21 6.60 ± 1.03 4.00 ± 0.45
2

8.46 ± 1.0
220 8.18 ± 0.67 5.30 ± 0.11 7.60 ± 0.84 4.69 ± 0.23
8

8.45 ± 0.9
248 8.36 ± 1.23 5.25 ± 0.05 7.47 ± 0.75 4.60 ± 0.35
8

5.93 ± 0.3
52 4.44 ± 0.32 3.07 ± 0.11 4.36 ± 0.23 2.77 ± 0.51
6

7.14 ± 0.4
99 6.60 ± 0.17 4.38 ± 0.23 6.78 ± 0.26 2.97 ± 0.36
1

7.97 ± 0.1
147 7.26 ± 0.30 5.10 ± 0.32 6.95 ± 0.13 4.35 ± 0.45
1

8.60 ± 0.8
172 8.12 ± 0.27 5.40 ± 0.41 7.64 ± 0.74 4.47 ± 0.23
5

8.95 ± 1.0
220 8.40 ± 0.89 5.69 ± 0.21 7.83 ± 0.56 5.30 ± 0.42
4

8.69 ± 1.1
248 8.58 ± 0.90 5.77 ± 0.15 7.95 ± 0.99 5.69 ± 0.38
0

6.38 ± 0.3
20 5.48 ± 0.32 3.32 ± 0.14 4.84 ± 0.12 3.38 ± 0.63
6

7.17 ± 0.5
30 6.16 ± 0.25 4.43 ± 0.32 5.39 ± 0.32 3.84 ± 0.53
3
Temp Mean total viable count ± SD
(°C)
and
storag B. S.
e time Overall Pseudomona thermosphac putrefacien Enterobacteriace
(h) ds ae
ta s

7.58 ± 0.3
47 7.06 ± 0.25 5.04 ± 0.37 6.53 ± 0.38 4.61 ± 0.42
4

8.06 ± 0.6
54 7.33 ± 0.30 5.59 ± 0.40 6.74 ± 0.54 4.60 ± 0.36
5

8.98 ± 0.6
71 8.13 ± 0.67 5.77 ± 0.65 7.87 ± 0.87 5.38 ± 0.36
6

8.76 ± 1.0
97 8.43 ± 0.74 6.30 ± 0.62 7.99 ± 0.80 6.00 ± 0.54
4

10

6.56 ± 0.6
20 5.65 ± 0.45 3.54 ± 0.23 5.14 ± 0.26 3.56 ± 0.48
5

7.33 ± 0.5
30 6.07 ± 0.63 4.60 ± 0.34 5.71 ± 0.35 4.06 ± 0.39
4

8.55 ± 0.3
47 7.00 ± 0.68 5.71 ± 0.37 7.39 ± 0.37 5.11 ± 0.40
5

8.72 ± 0.3
54 7.77 ± 0.32 6.00 ± 0.37 7.79 ± 0.30 5.77 ± 0.30
6

9.29 ± 0.3
71 8.39 ± 1.05 6.00 ± 0.30 7.90 ± 0.42 6.07 ± 0.30
7

9.26 ± 0.3
97 8.84 ± 0.68 6.47 ± 0.40 8.10 ± 0.64 6.25 ± 0.32
0
aLog
10 CFU gram−1.
bP. phosphoreum was less than 1% of the total viable count.
cNA, not applicable.

TABLE 2. Overall μmax hour−1 and those of pseudomonads, B. thermosphacta,

S. putrefaciens, and Enterobacteriaceae grown on naturally spoiled boque fish
and on sterile fish blocks inoculated with bacteria and stored aerobically at 0,
3, 7, and 10°C

Mean μmax h−1 ± SD

Fish and bacterial
groupa

0°C 3°C 7°C 10°C

All 0.019 ± 0.002 0.031 ± 0.001 0.064 ± 0.003 0.070 ± 0.002

Pseudomonads 0.023 ± 0.001 0.037 ± 0.002 0.066 ± 0.001 0.080 ± 0.002

B. thermosphacta 0.019 ± 0.003 0.031 ± 0.003 0.060 ± 0.001 0.079 ± 0.003

S. putrefaciens 0.030 ± 0.001 0.045 ± 0.003 0.076 ± 0.002 0.102 ± 0.003

Enterobacteriaceae 0.012 ± 0.001 0.022 ± 0.001 0.056 ± 0.002 0.075 ± 0.002

Pseudomonads 0.025 ± 0.005 NDb ND 0.110 ± 0.014

B. thermosphacta 0.030 ± 0.006 ND ND 0.109 ± 0.012

S. putrefaciens 0.032 ± 0.004 ND ND 0.173 ± 0.023

Lactic acid bacteria 0.012 ± 0.001 ND ND 0.105 ± 0.010

aA, naturally spoiled boque; B, sterile boque inoculated with bacteria.
bND, not determined.

P. phosphoreum was detected only in the initial storage period, and its contribution
was less than 1% of the total microflora. In general, the contribution of P.
phosphoreum was extremely small and rather unimportant for boque fish stored
aerobically (results not shown). At the time of rejection, none of the selected colonies
from Long-and-Hammer medium was counted as P. phosphoreum. It needs to be
noted that this bacterium did not grow in sterile fish flesh inoculated with strains
tested in this study. Their μmax values of the other bacteria used in this study were
found to be lower than the μmax of those strains grown individually in sterile boque
flesh (Table (Table22).

Remaining shelf life.

The shelf life of fresh boque fish stored at 0, 3, 7, and 10°C was determined as 174
± 10, 103 ± 8, 60 ± 5, and 44 ± 7 h, respectively, based on the scoring scale
described in Materials and Methods. Excellent correlations (r, >0.96) were observed
between the counts (log10 CFU gram−1) of pseudomonads and H2S-producing
bacteria with the remaining shelf life at all temperatures (Table (Table3).3). The log
total viable counts showed less of a correlation with the remaining shelf life than did
log numbers of Pseudomonas sp. and H2S-producing bacteria. In addition, total
viable count data only ranged from approximately 5.5 to 8.5 log units (Table
(Table1),1), which limited their usefulness for evaluation of fish freshness.

TABLE 3. Parameter values and statisticsa fitted to the remaining shelf life data
of Mediterranean boque stored aerobically at 0, 3, 7, and 10°C

Bacteria and No. of Intercept % of total

Slope (β1) ± Correlation
storage temp data (β0) ± 95% variability
95% CI coefficient
(°C) points CIb explained

All

0 23 23.7 ± 3.23 −3.02 ± 0.23 85 −0.923

3 23 19.0 ± 3.08 −2.42 ± 0.41 81 −0.901

7 19 8.6 ± 1.32 −0.99 ± 0.12 80 −0.897

10 17 7.1 ± 1.19 −0.88 ± 0.14 80 −0.895

Pseudomonads

0 23 16.2 ± 1.52 −2.29 ± 0.21 95 −0.978

3 23 14.0 ± 1.91 −1.94 ± 0.24 92 −0.959

7 19 7.1 ± 0.67 −0.89 ± 0.13 95 −0.974

10 17 5.6 ± 0.70 −0.76 ± 0.11 93 −0.965

H2S producers

0 23 13.9 ± 0.7 −2.13 ± 0.10 98 −0.992

3 23 12.6 ± 0.4 −1.91 ± 0.11 99 −0.990

7 19 5.8 ± 0.4 −0.75 ± 0.14 96 −0.983

10 17 4.6 ± 0.6 −0.66 ± 0.12 91 −0.957

aDetermined as described in Materials and Methods.
bCI, confidence interval.
Changes in physicochemical characteristics.

The reduction of glucose in boque fillets and the rise in α-amino groups and pH
values are shown in Fig. Fig.1.1. This decrease was more pronounced in boque fish
samples stored at 7 and 10°C than in samples stored at 0 and 3°C (Fig. (Fig.1).1).
The glucose decrease was followed by an increase in pH values (Fig. (Fig.1c),1c),
as well as by an increase in proteolysis and an increase in free amino acids (Fig.
(Fig.1c).1c). l-Lactate was also found to decrease (Table (Table4).4). Thus, the
increase in pH (Fig. (Fig.1b)1b) can be due to the l-lactic acid decrease (Table
(Table4)4) and, further, to deamination of amino acids. In this study, the rate of
lactate decrease in fresh boque fish was affected by the storage temperature (Table
(Table4).4). In this study, no TMA was produced during the storage of boque fish
under the conditions studied (results not shown).

Changes in glucose (a), α-amino groups (equivalent of glycine) (b), and pH (c) of
naturally spoiled boque fish during storage at 0°C (□), 3°C (○), 7°C (▵), and 10°C (⋆).
Each point is the mean of two samples taken from different experiments (coefficient
of variation of the mean of samples taken from different experiments, <5.5%). Each
sample was analyzed in duplicate (coefficient of variation of samples from the same
experiment, <0.55%).
FIG. 1. Changes in glucose (a), α-amino groups (equivalent of glycine) (b), and pH
(c) of naturally spoiled boque fish during storage at 0°C (□), 3°C (○), 7°C (▵), and
10°C (⋆). Each point ...

TABLE 4. Changes in the areas under lactic, formic, and acetic acid peaks and
unknown peaks with RT of 14.8 and 17.6 min during storage of boque fish at
0, 3, 7, and 10°C

Area under peak (103)a at 210 nm

Peak and temp (°C)

0h 30 h 47 h 52 h 71 h 99 h 172 h 248 h

Lactic acid

0 865 750 616 549 345

3 NAb 683 NA 220

7 809 461 230

10 714 388 124

Formic acid

0 13 NDc 69 67 87

3 NA 57 NA 176

7 93 220 332

10 130 168 195

Acetic acid

0 9 ND 7 11 10

3 NA 12 NA 38

7 58 35 81

10 40 45 91

Bd
 Area under peak (103)a at 210 nm

Peak and temp (°C)

0h 30 h 47 h 52 h 71 h 99 h 172 h 248 h

0 22 21 44 60 347

3 NA 199 NA 348

7 81 217 217

10 294 393 393

De

0 5,622 5,666 6,137 5,064 1,822

3 NA 4,322 NA 521

7 3,229 1,158 3,229

10 316 458 318

aEach value is the mean of two samples taken from different experiments (coefficient
of variation of the mean of samples taken from different experiments, <5%). Each
sample was analyzed in duplicate (coefficient of variation of samples from the same
experiment, <0.65%).
bND, not detected.
cNA, not analyzed.
dRT, 14.8 min.
eRT, 17.6 min.

When the physicochemical changes of sterile boque fish were examined, it was
found that glucose was used by all of the bacteria tested (Fig. (Fig.2).2). Lactate
utilization was more pronounced in sterile blocks inoculated with the test bacteria
and stored at 10°C than in samples stored at 0°C. At 0°C, lactate decreased only in
sterile blocks inoculated with Pseudomonas sp. and S. putrefaciens (Tables
(Tables55 and and6),6), while it remained at the same level in samples inoculated
with B. thermosphacta and lactic acid bacteria. In general, we can say that the
organic acid profile observed in sterile fish blocks differed with the bacteria used in
this study (Tables (Tables55 and and6).6). Indeed, for example, the increase of the
areas under the formic acid peak and unidentified peak C (Tables (Tables55 and
and6)6) was more pronounced in samples inoculated with S. putrefaciens than in
samples inoculated with Pseudomonas sp. or B. thermosphacta. Unidentified peak
A (Table (Table55 and and6)6) was a characteristic only of samples inoculated with
Pseudomonas sp. Surprisingly, these two unidentified peaks were not present in the
organic acid profile derived from naturally spoiled boque fish (Table (Table4).4). It
should be noted that all of the above-mentioned organic acid changes were only
evident in natural and deliberately inoculated samples, and thus, it can be concluded
that the autolytic enzymes do not contribute to these changes (Tables (Tables55 and
and6,6, uninoculated samples).

FIG. 2. Changes in the concentration of glucose in uninoculated sterile samples (a) and in
samples inoculated with Pseudomonas sp. (□) or S. putrefaciens (▵) (b), or with B.
thermosphacta (○) or lactic acid bacteria (◊) (c) and stored at 0°C (open symbols) or 10°C
(closed symbols). Each point is the mean of two samples taken from different experiments
(coefficient of variation of the mean of samples taken from different experiments, <4.5%).
Each sample was analyzed in duplicate (coefficient of variation of samples from the same
experiment, <0.60%).

TABLE 5. Changes in the areas under lactic acid, formic acid, and acetic acid
peaks and four unknown peaks with RT of 14.05, 14.8, 16.25, and 17.6 min
during storage at 0°C of sterile boque fish left uninoculated or inoculated with
different microorganisms
 Area under peak (103)a at 210 nm

Peak and organism

0h 72 h 216 h 408 h 552 h

Lactic acid

Pseudomonas sp. 1,150 NAb 1,115 93 NDc

S. putrefaciens 1,297 1,296 928 240

B. thermosphacta 1,250 1,194 1,172 1,239

Lactobacillus sp. 1,400 1,542 1,522 1,470

None (uninoculated) 1,150 1,145 1,156 1,139

Formic acid

Pseudomonas sp. 3.5 NA ND 135 288

S. putrefaciens 104 400 7,781 7,955

B. thermosphacta 17 46 60 92

Lactobacillus sp. 33 9 13 ND

None (uninoculated) 3.7 3.6 3.3 3.4

Acetic acid

Pseudomonas sp. 1.1 NA ND ND ND

S. putrefaciens 1.2 25 118 185

B. thermosphacta 0.9 82 66 80

Lactobacillus sp. 1.0 2 4.5 4.5

None (uninoculated) 1.1 1.2 1.0 1.3

A, Pseudomonas sp. ND ND ND 379 1,335

Pseudomonas sp. 41 NA 34 69 198

S. putrefaciens 34 47 145 205

 Area under peak (103)a at 210 nm

Peak and organism

0h 72 h 216 h 408 h 552 h

B. thermosphacta 38 121 61 167

Lactobacillus sp. 35 93 131 1,028

None (uninoculated) 42 43 40 40

C, S. putrefaciens ND ND 33 533 717

Pseudomonas sp. 4,500 NA 11,139 6,162 4,930

S. putrefaciens 4,069 4,117 2,444 4,760

B. thermosphacta 4,261 4,398 3,968 4,389

Lactobacillus sp. 7,330 4,106 3,572 4,100

None (uninoculated) 4,534 4,527 4,478 4,502

aEach value is the mean of two samples taken from different experiments (coefficient
of variation of the mean of samples taken from different experiments, <5%). Each
sample was analyzed in duplicate (coefficient of variation of samples from the same
experiment, <0.7%).
bNA, not analyzed.
cND, not detected.

TABLE 6. Changes in the areas under lactic acid, formic acid, and acetic acid
peaks and four unknown peaks with RT of 14.05, 14.8, 16.25, and 17.6 min
during storage at 10°C of sterile boque fish left uninoculated or inoculated with
different microorganisms

Area under peak (103)a at 210 nm

Peak and organism

0h 36 h 72 h 108 h 144 h

Lactic acid
 Area under peak (103)a at 210 nm

Peak and organism

0h 36 h 72 h 108 h 144 h

Pseudomonas sp. 1,150 1,252 927 288 116

S. putrefaciens 1,116 NAb 878 564

B. thermosphacta 1,219 1,181 378 772

Lactobacillus sp. 1,265 1,082 NA 759

None (uninoculated) 1,160 1,145 1,140 1,155

Formic acid

Pseudomonas sp. 3.5 121 NDc 48 306

S. putrefaciens 1,552 NA 1,955 2,706

B. thermosphacta 17 46 60 92

Lactobacillus sp. 5 5.1 NA 22

None (uninoculated) 3.8 3.5 3.3 4

Acetic acid

Pseudomonas sp. 1.1 0.9 0.9 0.9 111

S. putrefaciens 82 NA 142 128

B. thermosphacta 23 125 149 157

Lactobacillus sp. 8 13 NA 160

None (uninoculated) 1.2 1.0 1.3 1.2

A, Pseudomonas sp. ND ND 109 305 912

B, Pseudomonas sp. 41 44 62 38 410

S. putrefaciens 154 NA 126 467

B. thermosphacta 31 34 30 64

Lactobacillus sp. 48 153 NA 170

 Area under peak (103)a at 210 nm

Peak and organism

0h 36 h 72 h 108 h 144 h

None (uninoculated) 39 40 39 42

C, S. putrefaciens ND 544 NA 875 945

Pseudomonas sp. 4,500 5,900 4,612 2,377 2,400

S. putrefaciens 4,323 NA 4,601 3,609

B. thermosphacta 4,439 4,490 5,514 3,013

Lactobacillus sp. 3,487 4,557 NA 3,900

None (uninoculated) 4,400 4,510 4,520 4,490

aEach value is the mean of two samples taken from different experiments (coefficient
of variation of the mean of samples taken from different experiments, <5%). Each
sample was analyzed in duplicate (coefficient of variation of samples from the same
experiment, <0.7%).
bNA, not analyzed.
cND, not detected.

DISCUSSION

The initial and final microbial associations of fresh boque fish were found to be similar
to those reported in the literature (9, 12, 19, 31, 45, 57, 65). Of P. phosphoreum and
B. thermosphacta, two bacteria whose contribution to fish spoilage has been
reported only recently (9, 12, 18, 19, 20), the latter was found to be a member of the
final microbial association, while the former did not grow in boque fish during aerobic
storage (Table (Table11).

The remaining bacterial groups examined in the present study, primarily

Pseudomonas sp. and S. putrefaciens, a late spoiler in the temperature range of
−1.4 to 15°C, have been reported to be the specific spoilage bacteria in temperate
and tropical waters (31, 33, 35, 49). That the presence of these two bacteria
correlated better with remaining shelf life than did the total viable count (Table
(Table3)3) could be explained by the fact that spoilage is more often a result of the
production of off odors and flavors caused by specific spoilage organisms, which are
only a fraction of the total microflora (40). On the basis of numerous data reported in
the literature, it is concluded that spoilage of fresh fish is due to the activity of more
than one specific spoilage organism. Similar results have been reported for other
fresh fish stored aerobically (5, 18, 19, 23, 34, 38, 42). However, it should be noted
that although a number of data concerning the correlation between H 2S-producing
bacteria and freshness have been collected, Pseudomonas sp. has not received the
appropriate attention for the effect of microbial interaction on spoilage (42). This can
be important in understanding spoilage, as it was found that there is an interaction
between the above-mentioned bacteria. Indeed, it was reported that Pseudomonas
sp. can inhibit the growth of S. putrefaciens due to the ability of the former to produce
siderophores, and this interaction can be the major factor governing the development
of spoilage flora (32). On the other hand, studies with fresh meat have shown that
pseudomonads predominated over the other meat bacteria because of their faster
growth rates, their greater affinity for oxygen, and, as a consequence, their greater
catabolism of glucose and lactate (26, 27). In general, the microbe-microbe
interaction can also be influenced by factors such as oxygen and substrate limitation
(14), and thus, microbial competition (21, 55) not only affects the development of
microbial association in a fish ecosystem but could also influence the chemical
changes which are essentially an expression of the development of such an
ecosystem. Indeed, this can be seen in our findings. Both of the bacteria mentioned
above were present in spoiled boque fish and were also studied individually in sterile
blocks under comparable aerobic conditions. There was no similarity among the
organic acid profiles of Pseudomonas sp.-inoculated, S. putrefaciens-inoculated,
and naturally spoiled samples, regardless of the storage temperature (Tables
(Tables4,4, ,5,5, and and6).6). This can possibly be attributed to the fact that in the
latter case, the decreasing availability of the carbon and energy substrate may cause
competition between bacteria. This competition usually forces the microorganisms
to regulate their enzymatic and nutrient uptake systems appropriately. Bacteria
belonging to different genera seem to differ in substrate or oxygen affinity (25), since
Drosinos (15) has reported a minor contribution of glucose and oxygen affinity to the
domination of P. fragi over the pseudomonads P. fluorescens and P. lundensis on
chilled meat. Thus, it was suggested that other key physiological properties may
contribute to the succession of Pseudomonas spp. in meat ecosystems (15). Among
these, the metabolic versatility of pseudomonads that allows them to grow on a wide
range of substrates can be considered a competitive advantage over more
specialized species with greater substrate specificity (14). Thus, organisms
belonging to this genus grew both in the protein-free fraction of fish press juice
containing soluble components and in the protein fraction devoid of soluble
compounds (67), although their growth was faster in the former case. In a medium
consisting of a mixture of both of the fractions, prolific growth of the organisms was
noticed, which was accompanied by protein breakdown (67). An increase in the
concentration of α-amino groups (Fig. (Fig.1)1) in naturally spoiled fresh fish was
evident in this study.

As far as substrate specificity is concerned, several studies with meat and fish have
shown that bacteria can grow on flesh at the expense of one or more of the low-
molecular-weight soluble components such as glucose, lactic acid, certain amino
acids, nucleotides, TMAO, and water-soluble proteins (18, 20, 50). Glucose and
lactate seem to be the substrates which are attacked first by the various groups of
spoilage bacteria under aerobic and anaerobic conditions (25). Although these
compounds are also present in fish muscle (18, 20, 43, 50, 53, 60, 68), their role as
intrinsic determinants of spoilage has been discussed in detail only in fresh-meat
ecosystems (59). These two substrates initially affect the composition of microflora
developing during storage, the metabolic products produced by the flora (switch from
saccharolytic to amino acid-degrading metabolism), and the cell density attained at
the onset of spoilage (16, 17, 18, 20, 25, 48). Glucose and lactate have been
proposed as potential spoilage indicators for fish and meat (4, 56, 63, 68).

Since none of the bacteria ceased growing because of substrate exhaustion under
aerobic conditions, oxygen availability was suggested to be the limiting factor mainly
due to massive growth of bacteria (61). Under such conditions, those bacteria which
can switch to using readily available TMAO as an electron acceptor must have a
great ecological advantage over bacteria lacking this ability. This could be the case
for S. putrefaciens, a nonfermentative bacterium (62), growing on the exposed areas
of the fish, where it gets energy for growth by oxidation of easily diffusible extractives
(such as glucose, lactate, and free amino acids). By the time a significant number of
bacteria can penetrate the muscle due to proteolysis (28, 37, 61, 67), the fish is
largely unacceptable for human consumption due to the formation of off odors.
Indeed, S. putrefaciens produces large amounts of TMA (11), while Pseudomonas
cannot use TMAO to produce TMA (35). We should mention that TMA was not found
during the storage period in this study. This is in agreement with results related to
other Mediterranean fish (12, 18, 19), where no TMA was produced or extremely
small quantities of TMA were produced. Of course, the absence of TMA in our study
can be due either to the fact that it is not detectable because the S. putrefaciens
count never reached 108 CFU g−1 (10, 11) or to the low concentration or absence of
TMAO in boque fish. The former can be of great importance because of the potential
use of TMA as a spoilage indicator. Reasons include the varying content of the
precursor TMAO in different fish species and seasonal variations within species (7).

Other microbial metabolites found in this study (e.g., formic and acetic acids; Tables
Tables4,4, ,5,5, and and6)6) were produced either by the microbial association or
as a result of microbial interaction with the environment and showed good
correlations with the microbial numbers (results not shown), but there is a lack of
information about their effect on the sensory characteristics of fish. It is generally
accepted that the use of microbial metabolites as potential indicators should meet,
among others, the following criteria (41): (i) the compound should be absent or at
least present at low levels initially, (ii) it should increase with storage, and (iii) it
should be produced by the dominant microbial flora and have a good correlation with
sensory characteristics. Determination of the quantities of metabolites could provide
us with information about the degree of spoilage. The identification of the ideal
metabolite that can be used for spoilage assessment has proved a difficult task for
the following reasons. (i) Most metabolites are specific to certain organisms (e.g.,
gluconate is specific to pseudomonads), and the absence of these organisms or their
inhibition naturally or due to food ecology measures imposed by humans provides
incorrect spoilage information. (ii) The metabolites are the result of the consumption
of a specific substrate, but the absence of a given substrate or its presence in small
quantities does not preclude spoilage. (iii) The rate of microbial metabolite
production and the metabolic pathways of these bacteria are affected by the
environmental conditions imposed (e.g., pH, oxygen tension, temperature, etc.). (iv)
Accurate detection and measurement require sophisticated procedures, highly
educated personnel, time, and equipment. (v) Retrospective information about many
of these metabolites is not satisfactory. More research is needed in this field.

ACKNOWLEDGMENTS

This study was a part of two research projects on fish safety and quality (EPET II
and fair-1090) funded by the Greek Ministry of Development (General Secretariat of
Research and Technology) and the European Union (DGXIV), respectively.

We thank C. Genigeorgis for reviewing the manuscript.

REFERENCES

1. Anonymous. Microorganisms in foods. 1. Their significance and methods of

enumeration. 2nd ed. Toronto, Ontario, Canada: University of Toronto Press; 1978.
p. 45.
2. Anonymous. Fig.P for windows. User’s manual, version 2.5. Cambridge, United
Kingdom: Fig.P Biosoft Software; 1995.
3. Association of Official Analytical Chemists. Trimethylamine nitrogen in seafood.
In: Sidney W, editor. Official methods of analysis. 15th ed. Arlington, Va: Association
of Official Analytical Chemists; 1990. p. 123.
4. Boers R H, Dijkmann K E, Wijngaards G. Shelf-life of vacuum-packaged wild boar
meat in relation to that of vacuum-packaged pork: relevance of intrinsic factors. Meat
Sci. 1994;37:91–102. [PubMed]
5. Chai T, Chen C, Rossen A, Levin R F. Detection and incidence of specific species
of spoilage bacteria on fish. II. Relative incidence of Pseudomonas putrefaciens and
fluorescent pseudomonads on haddock fillets. Appl Microbiol. 1968;16:1738–1741.
[PMC free article] [PubMed]
6. Church C F, Swaisgood H E, Porter D H, Catignani L G. Spectrophotometric assay
using o-phthaldialdehyde on determination of proteolysis in milk and isolated milk
proteins. J Dairy Sci. 1983;66:1219–1227.
7. Dainty R H. Chemical/biochemical detection of spoilage. Int J Food Microbiol.
1996;33:19–34. [PubMed]
8. Dalgaard, P. 1997. Personal communication.
9. Dalgaard P. Qualitative and quantitative characterization of spoilage bacteria from
packed fish. Int J Food Microbiol. 1995;26:319–333. [PubMed]
10. Dalgaard P. The effect of anaerobic conditions and carbon dioxide. FAO Fish
Tech Pap. 1995;348:68–76.
11. Dalgaard P, Gram L, Huss H H. Spoilage and shelf life of cod fillets packed in
vacuum or modified atmospheres. Int J Food Microbiol. 1993;19:283–294. [PubMed]
12. Dalgaard P, Mejlholm O, Christiansen T J, Huss H H. Importance of
Photobacterium phosphoreum in relation to spoilage of modified atmosphere-
packaged fish products. Lett Appl Microbiol. 1997;24:373–378.
13. Dalgaard P, Ross T, Kamperman L, Neumeyer K, McMeekin T A. Estimation of
bacterial growth rates from turbidimetric and viable count data. Int J Food Microbiol.
1994;23:391–404. [PubMed]
14. Dolfing J, Gottschal J C. Microbe-microbe interactions. In: Mackie R I, White B
A, Isaacson R E, editors. Gastrointestinal microbiology. 2. Gastrointestinal microbes
and host interactions. New York, N.Y: Chapman & Hall; 1996. pp. 373–433.
15. Drosinos E H. Microbial associations of minced lamb and their ecophysiological
attributes. Ph.D. thesis. Bath, United Kingdom: University of Bath; 1994.
16. Drosinos E H, Board R G. A survey of minced lamb packaged in modified
atmospheres. Fleischwirtschaft. 1995;75:281–284.
17. Drosinos E H, Board R G. Attributes of microbial associations of meat growing
as xenic batch cultures in a meat juice at 4°C. Int J Food Microbiol. 1995;26:279–
293. [PubMed]
18. Drosinos E H, Lambropoulou K, Mitre E, Nychas G-J E. Attributes of fresh gilt-
head seabream (Sparus aurata) fillets treated with potassium sorbate, sodium
gluconate and stored under a modified atmosphere at 0 ± 1°C. J Appl Microbiol.
1997;83:569–575.
19. Drosinos E H, Nychas G-J E. Brochothrix thermosphacta, a dominant organism
in Mediterranean fresh fish (Sparus aurata) stored under modified atmosphere. Ital
J Food Sci. 1996;4:323–329.
20. Drosinos E H, Nychas G-J E. Production of acetic acid in relation to the content
of glucose during modified atmosphere storage of gilt-head seabream (Sparus
aurata) at 0 ± 1°C. Food Res Int. 1998;9:711–717.
21. Fredrickson A G, Stephanopoulos G. Microbial competition. Science.
1981;213:972–979. [PubMed]
22. Gardner G A. A selective medium for the enumeration of Microbacterium
thermosphactum in meat and meat products. J Appl Bacteriol. 1966;29:455–460.
[PubMed]
23. Gennari M, Campanini R. Isolamento e caratterizzazione di Shewanella
putrefaciens da pesce fresco e alterato, carni fresche e alterate, prodotti lattiero-
caseari, acqua e suolo. Ind Aliment. 1991;30:965–976. , 988.
24. Gibson A M, Bratchell N, Roberts T A. Predicting microbial growth: growth
responses of salmonellae in a laboratory medium as affected by pH, sodium chloride
and storage temperature. Int J Food Microbiol. 1988;6:155–178. [PubMed]
25. Gill C O. The control of microbial spoilage in fresh meats. In: Pearson A M,
Dutson T R, editors. Advances in meat research: meat and poultry microbiology. T.
L. New York, N.Y: Macmillan; 1986. pp. 49–88.
26. Gill C O, Molin G. Modified atmospheres and vacuum packaging. In: Russell N
J, Gould G W, editors. Food preservatives. Glasgow, Scotland: Blackie; 1991. pp.
172–199.
27. Gill C O, Newton K G. The development of aerobic spoilage flora on meat stored
at chill temperatures. J Appl Bacteriol. 1977;43:189–195. [PubMed]
28. Gill C O, Penney N. Penetration of bacteria into meat. Appl Environ Microbiol.
1977;33:1284–1286. [PMC free article] [PubMed]
29. Gill T A. Chemical and biochemical indices in seafood quality. In: Huss H H,
Jacobsen M, Liston J, editors. Quality assurance in the fish industry. Amsterdam,
The Netherlands: Elsevier; 1992. pp. 377–387.
30. Gill T A. Advanced analytical tools in seafood science. In: Luten J B, Børresen
T, Oehlenschläger J, editors. Developments in food science. 38. Seafood from
producer to consumer. Integrated approach to quality. Amsterdam, The Netherlands:
Elsevier Science; 1997. pp. 479–490.
31. Gram L, Huss H H. Microbiological spoilage of fish and fish products. Int J Food
Microbiol. 1996;33:121–137. [PubMed]
32. Gram L, Melchiorsen J. Interaction between fish spoilage bacteria Pseudomonas
sp. and S. putrefaciens in fish extracts and on fish tissue. J Appl Bacteriol.
1996;80:589–595. [PubMed]
33. Gram L, Oundo J O, Bon J. Storage life of Nile perch (Lates niloticus) in relation
to temperature and initial bacterial load. Trop Sci. 1989;29:221–236.
34. Gram L, Trolle G, Huss H H. Detection of specific spoilage bacteria from fish
stored at low (0°C) and high (20°C) temperatures. Int J Food Microbiol. 1987;4:65–
72.
35. Gram L, Webell-Neergaard C, Huss H H. Bacteriology of fresh and spoiling Lake
Victorian Nile perch (Lates nilotocus) Int J Food Microbiol. 1990;10:303–316.
[PubMed]
36. Gregersen T. Rapid method for distinction of Gram-negative and Gram-positive
bacteria. Eur J Appl Microbiol. 1978;5:123–127.
37. Gupta L K, Nagamohini Y. Penetration of poultry meat by Pseudomonas and
Lactobacillus spp. World J Microbiol Biotechnol. 1992;8:212–213. [PubMed]
38. Herbert R A, Hendrie M S, Gibson D M, Shewan J M. Bacteria active in the
spoilage of certain sea foods. J Appl Bacteriol. 1971;34:41–50. [PubMed]
39. Hugh R, Leifson E. The taxonomic significance of fermentative versus oxidative
metabolism of carbohydrates by various gram negative bacteria. J Bacteriol.
1953;66:24–26. [PMC free article] [PubMed]
40. Huss H H, Dalgaard P, Hansen L, Ladefoged H, Pedersen A, Zittan L. The
influence of hygiene in catch handling on the storage life of iced cod and plaice. J
Food Technol. 1974;9:213–221.
41. Jay J M. Microbial spoilage indicators and metabolites. In: Pierson M D, Sterm
N J, editors. Foodborne microorganisms and their toxins: developing methodology.
Basel, Switzerland: Marcel Dekker, Inc.; 1986. pp. 219–240.
42. Jøorgensen B R, Gibson D M, Huss H H. Microbiological quality and shelf life
prediction of chilled fish. Int J Food Microbiol. 1988;6:295–307. [PubMed]
43. Kakouri A, Drosinos E H, Nychas G-J E. Storage of Mediterranean fresh fish
(Boops boops and Sparus aurata) under modified atmospheres or vacuum at 3 and
10°C. In: Luten J B, Børresen T, Oehlenschläger J, editors. Developments in food
science. 38. Seafood from producer to consumer. Integrated approach to quality.
Amsterdam, The Netherlands: Elsevier Science; 1997. pp. 171–178.
44. King E D, Ward M K, Roney D E. Two simple media for the demonstration of
pyocyanin and fluorescein. J Lab Clin Med. 1954;44:301–307. [PubMed]
45. Koutsoumanis K, Taoukis P, Drosinos E H, Nychas G-J E. Proceedings of the
Final Meeting of the Concerted Action “Evaluation of Fish Freshness.”. Paris,
France: International Institute of Refrigeration; 1998. Lactic acid bacteria and
Brochothrix thermosphacta—the dominant spoilage microflora of Mediterranean
fresh sea fish stored under modified atmosphere packaging conditions; pp. 158–165.
46. Kovacs N. Identification of Pseudomonas pyocyanea by oxidase reaction.
Nature. 1956;178:703–707. [PubMed]
47. Kunst A, Draeger B, Ziegenhorn J. Colorimetric methods with glucose oxidase
and peroxidase. In: Bergmeyer H U, Bergmeyer J, Graßl M, editors. Methods of
enzymatic analysis. 3rd ed. 6. Metabolites 1: carbohydrates. Weinheim, Germany:
Verlag Chemie; 1984. pp. 178–185.
48. Lambropoulou K A, Drosinos E H, Nychas G-J E. The effect of glucose
supplementation on the spoilage microflora and chemical composition of minced
beef stored aerobically or under a modified atmosphere at 4°C. Int J Food Microbiol.
1996;30:281–291. [PubMed]
49. Lima dos Santos C A M. The storage life of tropical fish in ice—a review. Trop
Sci. 1981;23:97–127.
50. Liston J. Recent advances in the chemistry of iced fish spoilage. In: Martin R E,
Flicks G J, Bebard C E, Ward D E, editors. Chemistry and biochemistry of marine
products. Conn: AVI Publishing Co.; 1982. pp. 27–36.
51. Mayfield C I, Inniss W E. A rapid, simple method for staining bacterial flagella.
Can J Microbiol. 1977;23:1311–1313. [PubMed]
52. Mead G C, Adams B W. A selective medium for the rapid isolation of
Pseudomonas associated with poultry meat spoilage. Br Poultry Sci. 1977;18:661–
670. [PubMed]
53. Montero P, Mackie I M. Changes in intramuscular collagen of cod (Gadus
morhua) during post mortem storage in ice. J Sci Food Agric. 1992;59:89–96.
54. Mossel D A A, Eelderink L, Koopmans M, Rossem F V. Influence of carbon
source, bile salts and incubation temperature on recovery of Enterobacteriaceae
from foods using MacConkey-type agars. J Food Prot. 1979;42:470–475.
55. Mossel D A A, Ingram M. The physiology of the microbial spoilage of foods. J
Appl Bacteriol. 1955;18:232–268.
56. Nassos P S, King A D, Jr, Stafford A E. Lactic acid concentration and microbial
spoilage in anaerobically and aerobically stored ground beef. J Food Sci.
1985;50:710–712. , 715.
57. Nickelson R, Finne G, Hanna M O, Vandezant C. Minced fish flesh from
nontraditional Gulf of Mexico finfish species: bacteriology. J Food Sci.
1980;45:1321–1326.
58. Noll F. l-(+)-Lactate. In: Bergmeyer H U, Bergmeyer J, Graßl M, editors. Methods
of enzymatic analysis. 3rd ed. 6. Metabolites 1: carbohydrates. Weinheim, Germany:
Verlag Chemie; 1984. pp. 582–588.
59. Nychas G-J E, Drosinos E H, Board R G. Chemical changes in meat. In: Board
R G, Davies A, editors. Microbiology of meat and poultry. London, England: Blackie
Academic & Professional; 1998. pp. 288–326.
60. Parisi G, Geri G, Mecati M, Guidotti P, Franci O. Proceedings of the Conference
of IIR Commission C2, Preservation Techniques Applied to Biodiversity
Management in Fish Farming. Paris, France: International Institute of Refrigeration;
1996. Changes in some chemical and rheological parameters and in the indices of
rigor in the muscle of rainbow trout and sea bass in the 24 hour post mortem period;
pp. 159–166.
61. Ringø E, Stenberg E, Strøm A R. Amino acids and lactate catabolism in
trimethylamine oxide respiration of Alteromonas putrefaciens NCMB 1735. Appl
Environ Microbiol. 1984;47:1084–1089. [PMC free article] [PubMed]
62. Scott J H, Nealson K H. A biochemical study of the intermediary carbon
metabolism of Shewanella putrefaciens. J Bacteriol. 1994;176:3408–3411. [PMC
free article] [PubMed]
63. Seymour I J, Cole M B, Coote P J. A substrate-mediated assay of bacterial proton
effux/influx to predict the degree of spoilage of beef mince stored at chill
temperatures. J Appl Bacteriol. 1994;76:608–615. [PubMed]
64. Tarr H L A. Post mortem changes in glycogen, nucleotides sugar phosphates
and sugars in fish muscles—a review. J Food Sci. 1966;31:846–855.
65. Tassou C C, Drosinos E H, Nychas G-J E. Inhibition of resident microbial flora
and pathogen inocula on cold fresh fish fillets in olive oil, oregano and lemon juice
stored under a modified atmosphere or air. J Food Prot. 1996;59:31–34.
66. Van Spreekens K J A. The suitability of a modification of Long and Hammer’s
medium for the enumeration of more fastidious bacteria from fresh fishery products.
Arch Lebensm Hyg. 1974;25:213–219.
67. Venugopal V. Extracellular proteases of contaminant bacteria in fish spoilage: a
review. J Food Prot. 1990;53:341–350.
68. Watanabe E, Endo H, Ikeda Y, Shibamoto N, Toyama K. Determination of
glucose in fish muscle and serum with an enzyme sensor. Bull Jpn Soc Sci Fish.
1986;52:711–717.
X. Effect of Porcine Collagen Peptides on the
Rheological and Sensory Properties of Ice Cream
Korean J Food Sci Anim Resour. 2015; 35(2): 156–163.
Liying Li, Jae-Hyeong Kim, Yeon-Ji Jo, Sang-Gi Min,* and Ji-Yeon Chun*

Abstract

The effects of low molecular-weight collagen peptides derived from porcine skin
were investigated on the physicochemical and sensorial properties of chocolate ice
cream. Collagen peptides less than 1 kDa in weight were obtained by sub-critical
water hydrolysis at a temperature of 300℃ and a pressure of 80 bar. Ice cream was
then prepared with gelatin powder and porcine skin hydrolysate (PSH) stabilizers
mixed at seven different ratios (for a total of 0.5 wt%). There was no significant
difference in color between the resulting ice cream mixtures. The increase in
apparent viscosity and shear thinning of the ice cream was more moderate with PSH
added than with gelatin. Moreover, the samples containing more than 0.2 wt% PSH
had enhanced melting resistance, while the mixture with 0.2 wt% PSH had the lowest
storage modulus at -20℃ and the second highest loss modulus at 10℃, indicating
that this combination of hydrocolloids leads to relatively softer and creamier
chocolate ice cream. Among the seven types of ice creams tested, the mixture with
0.2 wt% PSH and 0.3 wt% gelatin had the best physicochemical properties.
However, in sensory evaluations, the samples containing PSH had lower chocolate
flavor scores and higher off-flavor scores than the sample prepared with just 0.5 wt%
gelatin due to the strong off-flavor of PSH.

Keywords: porcine skin, low molecular-weight collagen peptides, sub-critical water

hydrolysis, ice cream, stabilizers

Introduction

Ice cream is a frozen product consisting of air cells, ice crystals, and fat droplets
dispersed in a serum phase. During processing, ice cream is frozen rapidly by
removing heat from the mix while stirring to incorporate air (Marshall et al., 2003).
Stabilizers, predominantly hydrocolloids, are used during preparation to provide a
smooth texture, for shape retention during melting, and to increase the viscosity of
the mix (Goff and Hartel, 2004; Marshall et al., 2003). Most importantly, these
cryoprotective hydrocolloids retard the growth of ice crystals during storage,
especially during temperature fluctuations, but they neither change the
thermodynamics of the ice cream mix nor alter ice nucleation kinetics (Marshall et
al., 2003; Muhr et al., 1986).

Gelatin, derived from collagen obtained from various animal by-products, is one of
the principal materials used as an ice cream stabilizer. It disperses easily and does
not cause wheying-off or foaming. However, gelatin is a relatively expensive
stabilizer only effective at high concentrations and tends to offer poor protection
against the effects of heat shock. It also requires prolonged aging (Wang and
Damodaran, 2009). With these drawbacks in mind, low molecular-weight (600-2,700
Da) collagen peptides derived from Alcalase hydrolysis of bovine gelatin have been
found to inhibit ice recrystallization in a supercooled ice cream mix (Wang and
Damodaran, 2009). These can therefore be considered as new stabilizers for ice
cream production.

Other than their use as stabilizers, collagen peptides have also attracted attention
for their bioactive properties. An antihypertensive peptide has been obtained from
porcine skin collagen (Ichimura et al., 2009). Numerous studies have investigated
the antioxidant properties of peptides obtained from collagenous sources; these may
protect living cells against free radical mediated oxidative damage (Kim et al., 2001).

Subcritical water, viz. pressurized water heated above its boiling point but below the
critical point of 374℃ and 221 MPa, has unique properties (Lee et al., 2013).
Compared with other hydrolysis methods, the “green” approach of using water as a
solvent in place of hazardous substances also expedites processing (Alargov et al.,
2002; Lee et al., 2013). Sub-critical water hydrolysis has been employed to efficiently
extract amino acids from collagen (Zhu et al., 2011). Furthermore, the effects of high
pressure/high temperature (sub-critical water) treatment on porcine placenta have
also been reported (Lee et al., 2013).
In this study therefore, low molecular-weight collagen derived from porcine skin (<
1kDa, porcine skin hydrolysate, PSH) was extracted using sub-critical water (300℃
and 80 bar). Chocolate ice cream was then prepared using PSH which contained 5
wt% collagen peptides as a stabilizer and cocoa powder because of its strong off-
flavor. The effects of different PSH contents on the physicochemical and sensorial
properties of chocolate ice cream were analyzed.

Materials and Methods

Preparation of porcine skin hydrolysates

Porcine skin hydrolysates (PSH) were produced by subcritical water hydrolysis. The
treatment conditions were optimized in a previous study (Kim et al., 2014). Porcine
skin was bought from a local butcher’s shop (Daeho Chooksan, Korea). The skin
was immersed in 70℃ water for 2 h to remove fat and other residues and then sliced
into 0.5 cm × 0.5 cm pieces. These were homogenized first for 3 min using a four-
wing blade blender (CNHR-26, Bosch, China) and then at 25,000 rpm for 5 min using
an Ultra-Turrax® blender (T25, IKA Labotechnik, Germany). The sub-critical water
system (300℃ and 80 bar) was composed of a control box, vessel (reactor), water
bath, heater, temperature controller, and pressure controller (SFE SYSTEM, CS-
1000, REXO, Korea). Porcine skin mixed with distilled water (1:2, w/w) was then
hydrolyzed and cooled directly in this system whose temperature and pressure were
reduced down to 40-45℃ and 0.1 bar, respectively in a 4℃ water bath with coolant
circulation. All processing steps were completed within 1 h.

Characterization of the porcine skin hydrolysates

All characterizations were performed as described by Kim et al. (2014). The samples
were centrifuged at 10,000 g for 5 min and the molecular weights of the peptides in
the supernatant were determined by gel permeation chromatography (GPC, YL
9100, Younglin Instrument Co. Ltd., Korea). The amino acid profile was determined
at the Animal Resources Research Center in Konkuk University (S4300 Amino Acid
Reaction Module, Systeme & Komponenten analytischer Meßtechnik, Germany).
The free amino group content was determined using the method of Benjakul and
Morrissey (1997), and expressed in terms of L-leucine (Nagarajan et al., 2012). The
pH and color of the filtered PSH solution were respectively determined using a pH
meter (Model S220, Mettler Toledo GmbH, Switzerland) and a colorimeter (CR-210
Chroma-meter, Minolta, Japan) calibrated using a white standard (CIE L* = +97.83,
CIE a*= −0.43, CIE b*= +1.96).

Ice cream preparation

Different ice cream samples were prepared with the stabilizer type and content
varied as follows: 0.5 wt% gelatin (Hayashi Pure Chemical Industries Ltd., Japan),
0.5 wt% PSH, PSH 0.1 (0.1 wt% PSH+0.4 wt% gelatin), PSH 0.2 (0.2 wt% PSH+0.3
wt% gelatin), PSH 0.25 (0.25 wt% PSH+0.25 wt% gelatin), PSH 0.3 (0.3 wt% PSH+
0.2 wt% gelatin), and PSH 0.4 (0.4 wt% PSH+0.1 wt% gelatin). The ice cream used
in this study consisted of 11 wt% milk fat and 59.3 wt% water (provided as milk and
whipped cream, Maeil Dairies Industry Co., Ltd., Korea), 10 wt% non-fat milk solids
(skimmed milk powder, Maeil Dairies Industry Co., Ltd., Korea), 12 wt% sucrose (CJ
CheilJedang Corp., Korea), 4 wt% corn syrup solids (Dextrose Equivalent 42, Grain
Processing Corp., USA), 0.2 wt% glyceryl monosearate (Daejung Chemicals &
Metals Co., Ltd., Korea), and 3 wt% cocoa powder (ADM Cocoa Pte. Ltd.,
Singapore) (Table 1).

Table 1. Composition of the chocolate ice cream samples used in the present study

Corn
Milk
Gelati PSH MSN syrup Glyceryl Cocoa Wate
Sampl fat Sucros
n (wt% F solid monosearat powde r
e (wt% e (wt%)
(wt%) ) (wt%) s e (wt%) r (wt%) (wt%)
)
(wt%)

Control 0.5 0

PSH
0.4 0.1
0.1

PSH
0.3 0.2
0.2

PSH
0.25 0.25 11 10 12 4 0.2 3 59.3
0.25

PSH
0.2 0.3
0.3

PSH
0.1 0.4
0.4

PSH
0 0.5
0.5

All ingredients were mixed, pasteurized at 70℃ for 30 min, and then homogenized
using a high-speed homogenizer (Ingenieurbüro CAT M. Zipperer GmbH, Germany)
for 30 min at 11,000 rpm. The ice cream mix was then cooled down to 4℃ and aged
overnight. The aged ice cream mix was frozen in a freezer (Sani-serv 407 1E, USA)
at a drawer temperature of −5℃, and packed into 30 mm × 5 mm containers and 120
mL specimen cups (SPL Lifesciences Co., Ltd., Korea). The ice creams were
hardened and stored at −20℃ for 24 h in a freezer (CRFD-0621, Samsung
Electronics Co., Ltd., Korea). The overall experimental procedure was repeated
three times giving a total of 21 formulations.
Rheological properties

Ice cream viscosity

A rheometer (Anton Paar, MCR 302, Austria) with a concentric cylinder (CC27) was
used to determine the viscosity of the ice cream mixes after aging. The
measurements were carried out for shear rates ranging from 1 s–1 to 100 s–1 at 4℃.
The apparent viscosity (Kokini viscosity) of the ice cream mixes, representing the
feeling of low-viscosity foods in the mouth, was calculated at a shear rate of 50 s–1
(Akhtar et al., 2006; Kokini 1987; Stanley and Taylor, 1993). The Ostwald-de Waele
model was used to relate the shear rate and the shear stress, through which the
consistency coefficient (K) and flow behavior index (n) can be calculated as follows:

σ=Kγn

where σ is the shear stress (Pa), K is the consistency coefficient (Pa·s n ), γ is the
shear rate (s–1), and n is the flow behavior index (dimensionless).

Oscillatory rheometry

Oscillatory thermo-rheometry measurements were carried out (Wildmoser et al.,

2004) using a rheometer (Anton Paar, MCR 302, Austria) with a plate-plate geometry
(PP25-S, plate diameter: 25 mm) to determine the dynamic rheological properties of
the ice cream mixtures. A movable hood covering the plate-plate setup reduced
temperature fluctuations with the environment. The test was performed at a constant
strain γ = 0.2% and angular frequency ω = 10 s–1. A 2 mm gap was maintained
between the plates. The temperature was gradually increased from 20℃ to 10℃ at
a constant heating rate of 2℃/min. The storage modulus G' and the loss modulus G"
were obtained from measurements, which respectively characterize the elastic and
viscous behaviors of the ice cream.

Melting test

Each time one sample was placed on a stainless mesh on top of a plastic funnel
attached to a stand, and a plastic sample dish was placed on an electronic balance
(BL-220H, Shimadzu Corp., Japan). This ice-cream melting system was placed in
an incubator (SW-90F, Sangwoo, Korea) maintained at 25℃. The balance was
connected to a computer and the weight of the ice cream melt was automatically
calculated every minute (Fig. 1).
Fig. 1. Schematic diagram of the ice cream melting system.

Color differences

The difference in color of the ice cream after hardening was measured using a
Chroma-meter (CR-400, Konica Minolta Optics, Inc., Japan). Results were
expressed using the L*a*b* scale, for lightness, redness, and yellowness,
respectively. The measurements were calibrated to a white standard (CIE
L*=+94.50, CIE a*=–0.53, CIE b*= +3.14) and the total color difference (∆E*) was
calculated using the following equation:

ΔE∗=(L∗2−L∗1)2+(a∗2−a∗1)2+(b∗2−b∗1)2−−−−−−−−−−−−−−−−−−−−−−−−−−−−√

where (L∗1, a∗1, b∗1) is the reference color and (L∗2, a∗2, b∗2) is the target color.

Sensory properties

Sensory characteristics including the appearance, chocolate flavor, off-flavor,

bitterness, creaminess, iciness, hardness, sweetness, smoothness, and coarseness
of the ice cream were evaluated by trained panelists. In these sessions, the attributes
and assessment techniques were defined and a practical sample evaluation was
performed. Sensory evaluations were carried out using a 5-point scale, with 1
representing “non-existent or imperceptible” and 5 representing “very intense”
properties.

Statistical analysis

All assays were carried out in triplicate. The data were expressed as means with
standard deviations and statistical analysis was performed using the Origin Pro 8
(OriginLab, USA) and SPSS 20.0 (SPSS Institute, USA) software. Means were
subjected to analysis of variance by Duncan’s multiple range test and Least-
significant difference at p<0.05.

Results and Discussion

Physical properties of porcine skin hydrolysates

In this study, PSH was produced through an optimized sub-critical water treatment
(300℃ and 80 bar) and the physical properties of the resulting PSH were similar to
those described in our previous report (Kim et al., 2014). Representative
characteristics of the PSH used here to prepare ice cream are shown in Table 2 and
Fig. 2. The relatively high pH, 9.22, is due to the increased self-ionization of water
during sub-critical water treatments (Brunner, 2009, 2014; Penninger et al., 2000;
Ravber et al., 2015; Watchararuji et al., 2008). The molecular weight peaks obtained
following the sub-critical water treatment of PSH are distributed between 434 Da and
626 Da. The free amino group content was 57.18 mM, with the concentrations of Gly
(28%) and Ala (13.1%) being the highest. In terms of color, the lightness, redness
and yellowness of the PSH were 33.6, 0.59, and 11.4, respectively.

Table 2. Physical properties of porcine skin collagen extract

Average
Free amino
pH molecular weight L*-value a*-value b*-value
acid (mM)
(Da)

PSH 9.22±0.01 500 57.18±0.69 33.6±0.05 0.59±0.06 11.4±0.06

Fig. 2. Gel permeation chromatography graph obtained for PSH with crosses
indicating the molecular weights of constituents.

Viscosity of the ice cream mixes

Viscosity, one of the most important rheological properties of ice cream mix, is
influenced by many factors notably its composition and the stabilizer type and
content. By forming a three-dimensional network of hydrated molecules,
hydrocolloids enhance the water-binding capacity of the ice cream mix (Kus et al.,
2005). Table 3 shows the values measured for the Kokini viscosity, consistency
coefficient, and flow behavior index of the ice cream mixes prepared with different
PSH-to-gelatin ratios. The flow behavior indexes are all less than one, characteristic
of the shear thinning behavior reported previously (Kaya and Tekin, 2001). Ice cream
mixes with higher consistency coefficients are considered more viscous. Soukoulis
et al. (2009) related enhanced apparent viscosity by increased Kokini viscosities and
consistency coefficients, and increased shear thinning by lower flow behavior
indexes. Here, the Kokini viscosity and consistency coefficient increases and the
flow behavior index decreases for lower PSH (higher gelatin) concentrations,
indicating that PSH leads to a more moderate increase in ice cream viscosity than
gelatin. This may be due to the low peptide content of collagen, only 5 wt% in PSH.
There is a significant (p<0.05) difference between the Kokini viscosities of the
control, PSH 0.1, and PSH 0.2 ice cream mixes. The control sample has the highest
Kokini viscosity (0.43±0.06 Pa·s), followed by PSH 0.1 (0.28±0.07 Pa·s) and PSH
0.2 (0.19±0.03 Pa·s). However, although the other mixes had lower Kokini
viscosities, these differences were not significant (p>0.05).

Table 3. Effect of hydrocolloids on the Kokini viscosity, consistency coefficient, and

flow behavior index of different chocolate ice cream mixes

Mean±SD values were calculated based on at least 3 repeat measurements.

a-dDifferent letters in the same column indicate significantly different values (p<0.05).

Kokini viscosity η50 Consistency coefficient K Flow behavior

Treatments n
(Pa·s) (Pa·s ) index n

Control 0.43±0.06a 2.65±0.31a 0.54±0.04d

PSH 0.1 0.28±0.07b 0.58±0.23b 0.83±0.04c

PSH 0.2 0.19±0.03c 0.32±0.09bc 0.87±0.03b

PSH 0.25 0.15±0.02cd 0.24±0.03c 0.88±0.01ab

PSH 0.3 0.15±0.04cd 0.24±0.08c 0.89±0.02ab

PSH 0.4 0.11±0.01d 0.14±0.01c 0.93±0.01a

PSH 0.5 0.10±0.00d 0.13±0.00c 0.92±0.01ab

Oscillatory rheometry
Oscillatory tests are used to determine the viscoelastic properties of complex food
systems (Dolz et al., 2006). Temperature has an important impact on the structure
of ice cream, with ice crystal microstructures dominating the rheological behavior of
frozen ice cream in the low temperature range, from –20℃ to –10℃. Thus, the
storage modulus G' and loss modulus G" below –10℃ report on the ice crystal size
and rigidity of the ice cream. As ice is melted away under constant heating from 0℃
to 10℃, the loss modulus G”, which characterizes the viscosity of the ice cream, can
be correlated with creaminess (Wildmoser et al., 2004).

The influence of hydrocolloids on the ice cream storage modulus at –20℃ and its
loss modulus at 10℃ are presented in Figs. 3 and and4,4, respectively. The PSH
0.2 and PSH 0.4 samples have the lowest storage modulus at –20℃ (0.45±0.03 MPa
and 0.48±0.03 MPa, respectively). These two combinations of stabilizers are
therefore the most effective at inhibiting ice crystal growth during ice cream
preparation. The control ice cream mix with the highest Kokini viscosity is also the
creamiest, with the highest loss modulus at 10℃ (34.01±2.14 Pa, p<0.05) among
the different samples, but also with the largest ice crystals as revealed by the highest
storage modulus at –20℃ (1.11± 0.03 MPa, p<0.05). The second and third most
viscous mixes (PSH 0.1 and PSH 0.2) are also creamy with no significant difference
(p>0.05) between their loss modulus at 10℃ (21.27±1.25 Pa and 20.63±3.11 Pa,
respectively).

Fig. 3. The influence of hydrocolloids on the ice cream storage modulus at –

20℃. Control, 0.5 wt% gelatin; PSH 0.1, 0.1 wt% PSH+0.4 wt% gelatin; PSH 0.2,
0.2 wt% PSH+ 0.3 wt% gelatin; PSH 0.25, 0.25 wt% PSH+0.25 wt% gelatin; PSH
0.3, 0.3 wt% PSH+0.2 wt% gelatin; PSH 0.4, 0.4 wt% PSH+0.1 wt% gelatin; PSH
0.5, 0.5 wt% PSH. Mean±SD values were calculated based on at least 3 repeat
measurements. a-dDifferent letters indicate significantly different (p<0.05)
values.

Fig. 4. The influence of hydrocolloids on the ice cream loss modulus at 10℃.
Control, 0.5 wt% gelatin; PSH 0.1, 0.1 wt% PSH+0.4 wt% gelatin; PSH 0.2, 0.2
wt% PSH+0.3 wt% gelatin; PSH 0.25, 0.25 wt% PSH+0.25 wt% gelatin; PSH 0.3,
0.3 wt% PSH+0.2 wt% gelatin; PSH 0.4, 0.4 wt% PSH+0.1 wt% gelatin; PSH 0.5:
0.5 wt% PSH. Mean±SD values were calculated based on at least 3 repeat
measurements. a-dDifferent letters indicate significantly different (p<0.05)
values.

Melting resistance of the ice cream

Melting resistance is an important parameter used to evaluate the physical stability

of frozen ice cream. According to Marshall et al. (2003), hydrocolloids improve the
resistance to melting due to the enhancement of water-holding and microviscosity
they provide. However, as shown in Table 4, although the ice creams with higher
viscosities (control and PSH 0.1) have longer first-dropping times (35.67±2.08 min
and 23.33±2.52 min, respectively), their melting rates (6.35±0.78 g/min and 4.43±
0.16 g/min, respectively) are significantly higher than those of the other five types of
ice cream (p<0.05). This result is in agreement with Kilara and Chandan (2008), and
proves that gelatin is not effective in preventing the effects of heat shock. The ice
cream with the third-highest viscosity (PSH 0.2) has a first dropping time and melting
rate that are not significantly different (p>0.05) from the values measured for the four
remaining samples, indicating that melting resistance is similarly enhanced for PSH
concentrations above 0.2 wt% and gelatin concentrations below 0.3 wt%.

Table 4. Effect of hydrocolloids on the melting quality of different chocolate ice cream
mixes
Treatments First dropping time (min) Melting rate (g/min)

Control 35.67±2.08a 6.35±0.78a

PSH 0.1 23.33±2.52b 4.43±0.16b

PSH 0.2 17.00±3.61c 4.02±0.19bc

PSH 0.25 15.33±0.58c 3.66±0.09c

PSH 0.3 17.67±1.53c 3.55±0.19c

PSH 0.4 16.33±1.15c 3.71±0.16c

PSH 0.5 15.33±1.15c 3.69±0.11c

Mean±SD values were calculated based on at least 3 repeat measurements.

a-cDifferent letters in the same column indicate significantly different (p<0.05) values.

Color differences

In this study, since the PSH are dark brown, the lightness, redness, and yellowness
of the ice cream mixtures were measured to evaluate the effects of hydrocolloids on
the color of hardened chocolate ice cream. Note that color differences in fresh ice
cream have seldom been investigated. As Table 5 shows, there is no significant
difference in color (p>0.05) between the ice creams prepared with different PSH to
gelatin ratios, with an average lightness, redness, and yellowness of 42.3, 10.5, and
17.1, respectively. Cocoa powder is similar in color to PSH but darker, therefore
varying the PSH concentration does not affect the color of the ice cream. In previous
tests of the effect of PSH on vanilla ice cream, its color quality was deemed not
acceptable because of stains and darkening.

Table 5. Effect of hydrocolloids on lightness, redness, yellowness and total color

difference of chocolate ice cream

Treatment L*-value a*-value b*-value ΔE

Control 42.53±0.41 10.71±0.23 16.84±0.23 54.91±0.40

PSH 0.1 43.03±0.17 10.74±0.18 16.94±0.22 54.46±0.13

PSH 0.2 41.64±0.85 10.32±0.11 16.93±0.44 55.69±0.93

PSH 0.25 42.39±0.32 10.37±0.30 17.25±0.27 55.07±0.32

Treatment L*-value a*-value b*-value ΔE

PSH 0.3 41.67±1.68 10.31±0.32 17.38±0.30 55.78±1.62

PSH 0.4 42.34±0.25 10.65±0.19 17.31±0.19 55.19±0.27

PSH 0.5 42.32±0.26 10.32±0.13 17.12±0.26 55.10±0.33

Mean±SD values were obtained from at least 3 repeat measurements.

None of the values are significantly different (p>0.05).

Sensory evaluation

The perceived texture and flavor of ice cream are considered to be the most
important factors influencing ice cream consumption. In addition to their other
effects, hydrocolloids influence the sensory properties of ice cream (Donhowe et al.,
1991; Rincon et al., 2006; Soukoulis et al., 2008). Comparing the physicochemical
properties of the different fresh samples, the PSH 0.2 composition has a relatively
high Kokini viscosity, enhanced resistance to melting, the lowest storage modulus at
–20℃, and the second highest loss modulus at 10℃. This mixture was therefore
chosen for sensory evaluations along with the control and PSH 0.5 samples.

As shown in Table 6, there was no significant difference (p>0.05) between the three
samples in terms of appearance, bitterness, creaminess, iciness, sweetness,
smoothness, or coarseness. Morris (1995) reported that increased shear thinning of
hydrocolloids may be associated with better flavor perception. Indeed, in the present
study, the sample with 0.5 wt% gelatin, which exhibited the most shear thinning,
gained a significantly higher chocolate flavor score (p<0.05). As expected since PSH
is known to cause strong off-flavors, the PSH 0.2 and PSH 0.5 mixtures have
significantly higher off-flavor scores (p<0.05) compared with the control ice cream.
The ice cream produced with 0.5 wt% gelatin has a significantly higher hardness
score (p<0.05) than the other two, which is in agreement with the oscillatory
rheometry tests that ice crystal growth is not inhibited effectively in the 0.5 wt%
gelatin sample.

Table 6. Effect of hydrocolloids on the sensory attributes of different chocolate ice

cream mixtures
 Treatments

Sensory parameters

Control PSH 0.5 PSH 0.2

Appearance 3.60±1.34a 4.20±0.84a 3.40±0.89a

Chocolate flavor 4.20±1.30a 2.60±0.89b 2.60±0.55b

Off-flavor 1.40±0.55b 3.40±1.34a 3.60±1.52a

Bitterness 3.60±0.55a 3.80±1.30a 3.00±0.71a

Creaminess 2.80±1.10a 3.00±1.00a 3.80±1.30a

Iciness 2.00±1.00a 2.00±0.71a 1.80±0.84a

Hardness 3.60±0.55a 2.20±0.84b 2.20±0.84b

Sweetness 3.20±1.10a 2.60±0.89a 3.60±1.14a

Smoothness 3.00±1.00a 3.40±0.89a 3.20±1.48a

Coarseness 3.80±0.84a 2.60±1.14a 2.60±0.89a

Mean±SD values were calculated based on at least 5 repeat measurements.

a,bDifferent letters in the same row indicate significantly different values (p<0.05).

Conclusion

This study analysed the impact of different PSH-to-gelatin ratios on the

physicochemical and sensorial properties of chocolate ice cream. In terms of the
physicochemical properties, the ice cream prepared with 0.2 wt% PSH and 0.3 wt%
gelatin was optimal among the seven compositions tested, with a relatively high
Kokini viscosity, enhanced melting resistance, the lowest storage modulus at –20℃,
and the second highest loss modulus at 10℃. However, in the sensory evaluations,
the control ice cream (0.5 wt% gelatin) was perceived to have the highest chocolate
flavor and the lowest off-flavor, possibly due to the strong off-flavor of PSH. The
addition of low molecular-weight collagen peptides derived from porcine skin was
therefore effective in improving the physical and rheological properties of the ice
cream, but its sensory properties were degraded by the off-flavor of PSH. Masking
the off-flavor of PSH is therefore an important goal for further studies in order to
manufacture ideal ice cream products.

Acknowledgments
Financial support for this study was obtained from the Korean Institute of Planning
and Evaluation for Technology in Food, Agriculture, Forest, and Fisheries,
Korea(iPET Project No. 311029-3) and this paper was supported by the KU
Research Professor Program of Konkuk University.

References

1. Akhtar M., Murray B. S., Dickinson E. Perception of creaminess of model oil-in-

water dairy emulsions: Influence of the shear-thinning nature of a viscosity-
controlling hydrocolloid. Food Hydrocolloid. (2006);20:839–847. doi:
10.1016/j.foodhyd.2005.08.006. [Cross Ref]
2. Alargov D., Deguchi S., Tsujii K., Horikoshi K. Reaction behaviors of glycine under
super- and subcritical water conditions. Orig. Life Evol. Biosph. (2002);32:1–12. doi:
10.1023/A:1013906319253. [PubMed] [Cross Ref]
3. Benjakul S., Morrissey M. T. Protein hydrolysates from pacific whiting solid
wastes. J. Agric. Food Chem. (1997);45:3423–3430. doi: 10.1021/jf970294g. [Cross
Ref]
4. Brunner G. Near critical and supercritical water. Part I. Hydrolytic and
hydrothermal processes. J. Supercritical Fluid. (2009);47:373–381. doi:
10.1016/j.supflu.2008.09.002. [Cross Ref]
5. Brunner G. Supercritical fluid science and technology In: Brunner G., editor.
Properties of pure water. Vol. 5. Elsevier; (2014). pp. 9–93.
6. Dolz M., Hernández M., Delegido J. Oscillatory measurements for salad dressings
stabilized with modified starch, xanthan gum, and locust bean gum. J. Appl. Polym.
Sci. (2006);102:897–903. doi: 10.1002/app.24125. [Cross Ref]
7. Donhowe D. P., Hartel R. W., Bradley R. L. Determination of ice crystal size
distributions in frozen desserts. J. Dairy Sci. (1991);74:3334–3344. doi:
10.3168/jds.S0022-0302(91)78521-4. [Cross Ref]
8. Goff H. D., Hartel R. W. Ice cream and frozen desserts In: Hui Y. H., editor.
Handbook of Frozen Foods. Marcel Dekker Inc; NY: (2004). pp. 499–570.
9. Ichimura T., Yamanaka A., Otsuka T., Yamashita E., Maruyama S.
Antihypertensive effect of enzymatic hydrolysate of collagen and Gly-Pro in
spontaneously hypertensive rats. Biosci. Biotechnol. Biochem. (2009);73:2317–
2319. doi: 10.1271/bbb.90197. [PubMed] [Cross Ref]
10. Kaya S., Tekin A. R. The effect of salep content on the rheological characteristics
of a typical ice cream mix. J. Food Eng. (2001);47:59–62. doi: 10.1016/S0260-
8774(00)00093-5. [Cross Ref]
11. Kilara A., Chandan R. C. Ice cream and frozen desserts In: Chandan R.C., editor.
Dairy Processing & Quality Assurance. Wiley-Blackwell; Oxford: (2008). pp. 364–
365.
12. Kim J. H., Jeong K. H., Shim J. B., Jeong Y. K., Jang M.Y., Jo Y. J., Min S. G.,
Chun J. Y. Effect of super- or sub-critical water treatment on physicochemical
properties of pig skin. P 13-033; 2014 Annual Meeting of Korean Society of food Sci.
and Technol.; Gwangju, Korea. (2014). pp. 164–165.
13. Kim S. K., Kim Y. T., Byun H. G., Park P. J., Ito H. Purification and
characterization of antioxidative peptides from bovine skin. J. Biochem. Mol. Biol.
(2001);34:214–219.
14. Kokini J. L. The physical basis of liquid food texture and texture-taste
interactions. J. Food Eng. (1987);6:51–81. doi: 10.1016/0260-8774(87)90021-5.
[Cross Ref]
15. Kus S., Altan A., Kaya A. Rheological behavior and time-dependent
characterization of ice cream mix with different salep content. J. Texture Stud.
(2005);36:273–288. doi: 10.1111/j.1745-4603.2005.00016.x. [Cross Ref]
16. Lee M. Y., Choi Y. C., Chun J. Y., Min S. G., Hong G.P. Effects of high
pressure/high temperature processing on the recovery and characteristics of porcine
placenta hydrolysayes. Korean J. Food Sci. An. (2013);33:474–480. doi:
10.5851/kosfa.2013.33.4.474. [Cross Ref]
17. Marshall R. T., Goff H. D., Hartel R. W. Ice cream. 6th ed. Kluwer
Academic/Plenum Publishers; NY: (2003). pp. 1–34.
18. Morris E. R. Polysaccharide rheology and in-mouth perception In: Stephen M.
A., editor. Food polysaccharides and their applications. Marcel Dekker Inc; NY:
(1995). pp. 515–546.
19. Muhr A. H., Blanshard J. M. V., Sheard S. J. Effects of polysaccharide stabilizers
on the nucleation of ice. J. Food Technol. (1986);21:587–603.
20. Nagarajan M., Benjakul S., Prodpran T., Songtipya P., Kishimura H.
Characteristics and functional properties of gelatin from splendid squid (loligo
formosana) skin as affected by extraction temperatures. Food Hydrocolloid.
(2012);29:389–397. doi: 10.1016/j.foodhyd.2012.04.001. [Cross Ref]
21. Penninger J. M. L., Kersten R. J. A., Baur H. C. L. Hydrolysis of diphenylether in
supercritical water: Effects of dissolved nacl. J. Supercrit. Fluid. (2000);17:215–226.
doi: 10.1016/S0896-8446(00)00046-2. [Cross Ref]
22. Ravber M., Knez Ž, Škerget M. Simultaneous extraction of oil- and water-soluble
phase from sunflower seeds with subcritical water. Food Chem. (2015);166:316–
323. doi: 10.1016/j.foodchem.2014.06.025. [PubMed] [Cross Ref]
23. Rincon F., Leon de Pinto G., Beltran O. Behavior of a mixture of Acacia
glomerosa, Enterolobium cyclocarpum and Hymenaea courbaryl gums in ice cream
preparation. Food Sci. Technol. Int. (2006);12:13–17. doi:
10.1177/1082013206060445. [Cross Ref]
24. Soukoulis C., Chandrinos I., Tzia C. Study of the functionality of selected
hydrocolloids and their blends with κ-carrageenan on storage quality of vanilla ice
cream. Food Sci. Technol. (2008);41:1816–1827.
25. Soukoulis C., Lebesi D., Tzia C. Enrichment of ice cream with dietary fibre:
effects on rheological properties, ice crystallisation and glass transition phenomena.
Food Chem. (2009);115:665–671. doi: 10.1016/j.foodchem.2008.12.070. [Cross
Ref]
26. Stanley N. I., Taylor L. J. Rheological basis of oral characteristics of fluid and
semi-solid foods: A review. Acta. Psychol. (1993);84:79–92. doi: 10.1016/0001-
6918(93)90074-2. [PubMed] [Cross Ref]
27. Wang S. Y., Damodaran S. Ice-structuring peptides derived from bovine
collagen. J. Agric. Food Chem. (2009);57:5501–5509. doi: 10.1021/jf900524y.
[PubMed] [Cross Ref]
28. Watchararuji K., Goto M., Sasaki M., Shotiprunk A. Value-added subcritical water
hydrolysate from rice bran and soybean meal. Bioresour. Technol. (2008);99:6207–
6213. doi: 10.1016/j.biortech.2007.12.021. [PubMed] [Cross Ref]
29. Wildmoser H., Scheiwiller J., Windhab E. J. Impact of disperse microstructure on
rheology and quality aspects of ice cream. LWT-Food Sci. Technol. (2004);37:881–
891. doi: 10.1016/j.lwt.2004.04.006. [Cross Ref]
30. Zhu G. Y., Zhu X., Fan Q., Wan X. L. Raman spectra of amino acids and their
aqueous solutions. Spectrochim. Acta, Part A. (2011);78:1187–1195. doi:
10.1016/j.saa.2010.12.079. [PubMed] [Cross Ref]
XI. Studies on Physical and Sensory Properties of
Premium Vanilla Ice Cream Distributed in Korean
Market
Korean J Food Sci Anim Resour. 2014; 34(6): 757–762.
Mi-Jung Choi1 and Kwang-Soon Shin*

Abstract

The object of this study was to investigate the difference in physical and sensory
properties of various premium ice creams. The physical properties of the various ice
creams were compared by manufacturing brand. The water contents of the samples
differed, with BR having the highest value at 60.5%, followed by NT and CS at 57.8%
and 56.9%, respectively. The higher the water content, the lower Brix and milk fat
contents in all samples. The density of the samples showed almost similar values in
all samples (p>0.05). The viscosity of each ice cream had no effect on the water
content in any of the brands. Before melting of the ice cream, the total color
difference was dependent on the lightness, especially in the vanilla ice cream, owing
to the reflection of light on the surface of the ice crystals. The CS product melted the
fastest. In the sensory test, CS obtained a significantly higher sweetness intensity
score but a lower score for color intensity, probably due to the smaller difference in
total color, by which consumers might consider the color of CS as less intense. From
this study, the cold chain system for ice cream distribution might be important to
decide the physical properties although the concentration of milk fat is key factor in
premium ice cream.

Keywords: physical properties, premium ice cream, sensory test

Introduction

Ice cream can be defined as a smooth, sweet, cold dessert food prepared from a
frozen mixture of milk products and flavorings, containing a minimum of 10% milk fat
(Karaman et al., 2014). The mixture is homogenized after pasteurization and aged
to improve the physical properties before the freezing process. Ice cream is a
representative frozen dairy product enjoyed by people of all ages due to its cooling
effect in the mouth. Nowadays, manufacturers continue to develop formulations of
ice cream mixtures according to consumer demands, resulting in the creation of
various brand names (Sun-Waterhouse et al., 2013).

In the Korean ice cream market, there are a number of ice cream names depending
on the ice cream company, each of which has their formulations as well. Many
studies are present in the journal literature about the creation of new ice cream
formulations (Dervisoglu, 2006; Dervisoglu and Yazici, 2006; Karaman and
Kayacier, 2012). Nevertheless, vanilla ice cream has been the most representative
of the consumer’s preference for a long time (Bodyfelt et al., 1988; Buck et al., 1986).

The grade of ice cream depends on the amount of milk fat content. Generally, ice
creams containing more than 12% are recognized as premium level. Milk fat is of
importance in qualifying ice cream because it is closely related with the flavor and
texture of the ice cream (Li et al., 1997). Milk fat acts as a thermal insulator which
affects the melting process of ice cream. In general, ice cream containing high milk
fat content could take a longer time to melt, so it has potential advantages in handling
during high storage temperature.

Ice crystals also play a key role in the sensorial properties of ice cream. Ice creams
with small ice crystals have softer texture, also minimizing the recrystallization during
frozen storage (Arbuckle, 1966; Flores and Goff, 1999). The formation of ice crystals
is dependent not only on the compositions of the ice cream, but also on the freezing
temperature (Arbuckle, 1977; Trgo et al., 1999).

In order to evaluate the quality of frozen food, the quality is evaluated after the
thawing and cooking process to analyze the physical properties. For that reason, it
is hard to evaluate the actual quality of food in the frozen state. In general, ice cream,
which contains milk fat, MSNF (milk solids not fat), sugars, emulsifier and stabilizer,
is consumed in the frozen state, and the quality is strongly related to the physical
properties (Arbuckle, 1986). According to Glickman (1991), hydrophilic gums as a
stabilizer bind water molecules directly or entrap them inside the gel matrix, which
reduces the amount of free water. Therefore, they provide soft texture and resistance
against heat shock, and contribute to maintaining the body while delaying melting.

Owing to technological development, many premium level ice creams are now being
commercially produced. Physical and rheological properties are crucial factors for
consumers to evaluate the quality of ice cream. Therefore, this study was conducted
to evaluate the physical and sensory properties of premium vanilla ice creams
produced by specific brands in Korea, and to provide baseline data for improving the
crucial factors of the ice cream related to physical and rheological properties.

Materials and Methods

Materials

In this study, premium level vanilla ice cream was purchased from CS, BR, and NT
companies, respectively. Each selected ice cream was manufactured on different
dates. The order of preparation was CS, NT, and BR. Each sample was formed for
experimental purposes and kept at -70℃. The experimental analysis was performed
on the same day for all the samples although the manufacturing dates were different.
All experiments were repeated on the 3 different days for determination (n=3).

Methods

Water content

The water content was determined by the method of AOAC (1995). Two gram
aliquots of the samples were put into a bottle and dried for 24 h at 105℃. Water
content was calculated as the difference of water weight before and after drying.

Brix

Brix of molten ice cream was determined via refractometer (Portable refractometer
C-2 REF-104, SPECTROLAB, England) at room temperature.

Milk fat content

The milk fat content was determined by the method of Soxhlet extract (Soxhlet
extractor Ser148, VELP, Italy).

Density

In order to estimate the overrun, the weight of 30 mL of molten ice cream was
determined at 25℃ under atmospheric pressure and the density of milk was
calculated.
Viscosity

The viscosity of molten ice cream was determined by a rotational viscometer (Visco
star-L, J. P. SELECTA, Spain) using Spindle L2. The temperature of the sample was
fixed at 4℃ and spindle 1 rotated at 200 rpm.

Color

The color of the frozen ice cream was measured using a spectrocolorimeter (CR
400, Minolta Co., Japan) calibrated with a white plate and light trap supplied by the
manufacturer. Color was expressed using the CIE L*, a*, b* color system (CIE,
1976). The total color difference was calculated using the equation below.

Total color difference = ΔL*2+Δa*2+Δb*2−−−−−−−−−−−−−−−−−√

Melting time

The heat shock resistance of the ice cream was determined by using a self-
manufactured system (Fig. 1). The melting time of one hundred grams of ice cream
kept at 20℃ to melt completely in the chamber was measured. Here, the initial
temperature of the ice cream was set to be -70℃. The onset melting time of ice
cream was initiated at the point of weight change. During the melting process, the
weight of the ice cream was determined every 1 min.

Fig. 1. The lab-manufactured apparatus for the determination of melting time

of ice cream.
Sensory test

The samples were served to 10 experienced panel members. The determination was
carried out in duplicate by the sensory panelists. The color, aroma, ice crystal size,
retention of air bubbles, hardness and sweetness (1=extremely undesirable,
5=extremely desirable) were evaluated using a 5-point descriptive scale. The
panelists were required to cleanse their palate between samples with water.

Statistical analysis

Analysis of variance was performed on all the variables measured using the general
linear model (GLM) procedure of the SAS statistical package (1999). The Duncan’s
multiple range test (p<0.05) was used to determine differences between treatment
means.

Results and Discussion

Water content

The water content of each sample is shown in Fig. 2. The water contents of the ice
cream from CS, BR, and NT were 56.9, 60.5, and 57.9%, respectively. The ice
creams contained water contents approximately around 55-60%. The water content
of the BR ice cream was the highest among the samples (p<0.05). In general, ice
cream mix containing a low amount of total solids (high water content) has
proportionately more water to freeze than that containing a higher amount of total
solids (low water content) when hardened at the same storage temperature (El Owni
and Zeinab, 2009).
Fig. 2. Water contents of various premium vanilla ice creams. Different letters
indicate significant differences between samples.

Brix

The Brix of each sample is shown in Fig. 3. The highest content of sugar was found
in the CS company brand, at 36.6%. The Brix of NT and BR ice cream was 36.2%
and 34.7%, respectively (p<0.05). However, there were no significant differences
between the CS and NT ice creams (p>0.05). It was supposed that Brix decreased
as the water contents increased. According to the authors, a greater extent of total
solid contents increases the resistance to flow of the serum phase as ice melted,
which leads to slower meltdown. On the other hand, ice creams with low levels of
total solid content (up to 30%) melted quickly (Silva Junior and Silva Lannes, 2011).

Fig. 3. Brix of various premium vanilla ice creams. Different letters indicate
significant differences between samples.

Milk fat content

The milk fat content is presented in Fig. 4. The milk fat content of CS, BR, and NT
ice cream was 13.6, 12.7 and 13.54%, respectively. The three samples contained
approximately 12-14% milk fat. Among the samples, BR contained the lowest value
(p<0.05). From these results, it could be seen that as the water content rose, the
milk fat content lowered.
Fig. 4. Milk fat contents of various premium vanilla ice creams. Different letters
indicate significant differences between samples.

Density

The densities of the samples are shown in Fig. 5. The densities of CS, BR, and NT
ice creams were 0.846 g/ mL, 0.884 g/mL, and 0.917 g/mL, respectively. The density
of CR ice cream was significantly lower than the other samples (p<0.05). However,
there were no significant differences between BR and NT (p>0.05). Therefore, it was
assumed that the density of ice cream depends on the water content. In addition,
the overrun time of CS was thought to be higher than the others, causing higher
contents of air cells by air injection time. Normally, the dispersed air cells in the ice
cream significantly affect the qualities of ice cream such as the soft mouthfeel of the
product (Park et al., 2006).
Fig. 5. Density of various premium vanilla ice creams. Different letters indicate
significant differences between samples.

Viscosity

The viscosities of the samples are displayed at Fig. 6. The viscosities of CS and BR
ice cream were 198 cP and 170 cP. Exceptionally, the viscosity of NT ice cream was
26.3 cP. These values presented significantly differences among the samples
(p<0.05). As considered in the previous results, it was supposed that the viscosity of
ice cream could be dependent not on water content, but on the addition of a
thickening agent or other stabilizer.
Fig. 6. Viscosity of various premium vanilla ice creams. Different letters
indicate significant differences between samples.

Total color difference

Based on the measurement of lightness, redness and yellowness, the total color
differences were calculated and shown in Fig. 7. The total color difference of CS ice
cream was 81.4 at the highest value before melting, followed by BR and NT ice
cream at 77.3 and 65.2. Similarly, the total color differences of CS, BR, and NT
samples were 66.7, 68.4 and 58.5 after melting. The main factor determining the
color properties of vanilla ice cream is the lightness value, as compared to redness
or yellowness. Furthermore, the value of lightness presented the highest value since
light reflects on the surface of ice crystals before melting of the ice cream. In general,
the color of the ice cream increased in whiteness as the fat content increased
(Roland et al., 1999), namely, by the different brand names. However, a relationship
between the fat content and whiteness intensity in the ice cream was not observed
herein due to the different water contents.

Fig. 7. Total color difference of various premium vanilla ice creams. Different
letters within the same bar indicate significant differences between samples.

Melting time

The melting ratio as a function of time is shown in Fig. 8, and the parameters
estimated in the melting curve are depicted in Fig. 9. The onset point of melting was
found at 67.3 min for BR, which was significantly longer than the 58.8 min for CS
and 57.7 min for NT (p<0.05). Furthermore, the end point of BR melting was
estimated at 121.5 min, which was also significantly later compared to the 104.9 min
of CS and 111.3 min of NT (p<0.05). The CS product had a significantly shorter
overall melting time than the BR and NT products.
Fig. 8. Melting ratio of various premium vanilla ice creams as a function of time;
(y: Molten ice cream weight (g) per total ice cream weight (g), x: melting time
(min)).

Fig. 9. Comparisons of melting profiles of premium vanilla ice creams.

Different letters within the same bar indicate significant differences between
samples.

Based on the statistical analysis, the overall melting time of BR and NT did not differ
from each other, though BR took more time to begin melting comparing to the NT
product. Consequently, melting was delayed in the order of BS, NT and then CS
products. This order of melting was reversely proportional to the contents of milk fat.
A possible explanation was suggested in that more time was required to begin
melting when the ice cream contained more water content. Based on the fact that
the melting point of milk fat is relatively low, it is possible that NT and CS, which
contained higher amounts of milk fat, began melting more quickly than BR. Another
explanation is that the NT product had a relatively low viscosity, which is evidence
that it likely contained lower amounts of stabilizer. This would result in the short
starting point of melting. However, while NT began to melt earlier than CS, it had a
similar end point, reflecting that the water content of CS was lower than NT.
Consequently, these factors would lead to the quick melt of the CS product.

In general, the melting properties of ice cream are known to be influenced by the fat
content. According to Roland et al. (1999), increase of the fat content of ice cream
from 7 to 10% caused an increase in the half-life of the ice cream. They noted that
the melting results corresponded to the hardness determinations, i.e., while there
were no changes for the 0.1, 3, or 7% samples, significant difference occurred when
the percentage of fat was increased from 7 to 10% milk fat. Although the melting
time and sample hardness were not significantly different among the 0.1, 3, and 7%
fat samples of ice cream in that study, the characteristics of the lower fat ice creams
differed from those of the 10% fat samples. In our study, although there were no
significant differences in the melting time due to fat contents, the onset temperature
of melting of the BR ice cream, containing relatively lower fat content, appeared
earlier than the others. Based on these analyses, the BR ice cream contained higher
water content and lower Brix, resulting in the earlier starting time of ice cream
melting. It appears from these results that the onset of melting of BR ice cream,
which contains lower fat, was later than the other ice creams. This seems to be
opposite to the results obtained by Roland et al. (1999). In general, the quality of ice
cream is characterized by hardness, melting properties, air entrapment, and ice
content (Roland et al., 1999). However, no significant relationship between fat
content and the physical properties of ice cream was found in this study.

Sensorial properties of premium ice cream

The sensorial properties of the premium ice creams are given in Fig. 10. For color
intensity, both BR and NT measured 3.3 and 3.8, which was significantly higher
intensity compared to the 2.4 of CS (p<0.05). The vanilla flavor intensities of all
products were from 2.7 to 3.4, whereas the intensities did not significantly differ with
one another. Ice crystal content tended to be high in NT and low in CS, although the
difference was not significant. Overrun intensities (texture) and hardnesses were
similar for all products. For sweetness, CS scored higher than BR and NT (p<0.05),
while no difference in sweetness between BR and NT was obtained.
Fig. 10. Sensory evaluation of various premium vanilla ice creams (5-point
scoring method). Different letters within the same bar indicate significant
differences between samples.

In the present study, CS obtained a significantly higher sweetness intensity score

but a lower score for color intensity, probably due to the smaller difference in total
color, by which consumers might consider the color of CS as less intense. The main
factor deciding the mouthfeel taste of ice cream is the content of milk fat.
Unfortunately, a significant relationship between the ice cream quality measured by
sensory test and the fat content among the samples of different brands of ice cream
could not be measured. However, based on the analysis of physical properties, the
quality of ice cream could be different because of the storage conditions, although
we did not trace the cold chain system of each company. For further research, the
effect of storage temperature on the recrystallization of different commercial ice
creams should be performed to evaluate the mouthfeel taste.

Acknowledgments

This work was supported by Kyonggi University Research Grant 2012.

References

1. AOAC. Official methods of analysis. 16th ed. Association of Official Analytical

Chemists; Washington, USA: (1995). p. 1141.
2. Arbuckle W. S. Relation of freezing and gardening to the body and texture of ice
cream. Ice Cream Field Trade J. (1966);148:34.
3. Arbuckle W. S. AVI Publ. Co.; Westport, CT.: (1977). Ice cream.
4. Arbuckle W. S. Van Nostrand Reinhole; New York: (1986). Ice cream.
5. Bodyfelt F. W., Tobias J., Trout G. M. Sensory evaluation of ice cream and related
products In: Bodyfelt F. W., Tobias J., Trout G. M., editors. The sensory evaluation
of dairy products. Van Nostrand Reinhold; NY: (1988). pp. 166–226.
6. Buck J. S., Walker C. E., Poerce M. M. Evaluation of sucrose esters in ice cream.
J. Food Sci. (1986);51:489. doi: 10.1111/j.1365-2621.1986.tb11163.x. [Cross Ref]
7. CIE. Official recommendations on uniform color spaces color difference equations
and metric color terms, Suppl. No. 2. CIE (Commission International de l’Éclairage)
Publication No. 15 Colorimetry. Paris, France: (1976).
8. Dervisoglu M. Influence of hazelnut flour and skin addition on the physical,
chemical and sensory properties of vanilla ice cream. Int. J. Food Sci. Technol.
(2006);41:657–661. doi: 10.1111/j.1365-2621.2005.01127.x. [Cross Ref]
9. Dervisoglu M., Yazici F. The effect of citrus fibre on the physical, chemical and
sensory properties of ice cream. Food Sci. Technol. Int. (2006);12:159–164. doi:
10.1177/1082013206064005. [Cross Ref]
10. El Owni O. A. O., Zeinab K. O. K. Chemical composition of ice cream produced
in Khartoum state, Sudan. Pakistan J. Nutr. (2009);8:158–160. doi:
10.3923/pjn.2009.158.160. [Cross Ref]
11. Flores A. A., Goff H. D. Ice crystal size distributions in dynamically frozen model
solutions and ice as affected by stabilizers. J. Dairy Sci. (1999);82:1399–1407. doi:
10.3168/jds.S0022-0302(99)75366-X. [Cross Ref]
12. Everington D. W. “Food Freezing: Today and Tomorrow” The special problems
of freezing ice cream In: Bald W.B., editor. Springer-Verlag; London. Berlin.
Heidelberg, NY, USA: (1991).
13. Karaman S., Kayacier A. Rheology of ice cream mix flavored with black tea or
herbal teas and effect of flavoring on the sensory properties of ice cream. Food
Bioprocess Technol. (2012);5:3159–3169. doi: 10.1007/s11947-011-0713-5. [Cross
Ref]
14. Karaman S., Toker Ö. S., Yüksel F., Çam M., Kayacier A., Dogan M.
Physicochemical, bioactive, and sensory properties of persimmon-based ice cream:
Technique for order preference by similarity to ideal solution to determine optimum
concentration. J. Dairy Sci. (2014);97:97–110. doi: 10.3168/jds.2013-7111.
[PubMed] [Cross Ref]
15. Li Z., Marshall R. H., Fernando H. L. Effect of milk fat content on flavor perception
of vanilla ice cream. J. Dairy Sci. (1997);80:3133–3141. doi: 10.3168/jds.S0022-
0302(97)76284-2. [PubMed] [Cross Ref]
16. Park S. H., Hong G. P., Kim J. Y., Choi M. J., Min S. G. The influence of food
hydrocolloids on changes in the physical properties of ice cream. Food Sci.
Biotechnol. (2006);15:721–727.
17. Roland A. M., Phillips L. G., Boor K. J. Effects of fat content on the sensory
properties, melting, color and hardness of ice cream. J. Dairy Sci. (1999);82:32–38.
doi: 10.3168/jds.S0022-0302(99)75205-7. [Cross Ref]
18. SAS. SAS/STAT Software for PC. Release 6.11. SAS Institute; Cary, NC, USA:
(1999).
19. Silva E., Jr., Silva Lannes S. Effect of different sweetener blends and fat types
on ice cream properties. Ciênc. Technol. Aliment. (2011);31:217–220. doi:
10.1590/S0101-20612011000100033. [Cross Ref]
20. Sun-Waterhouse D., Edmonds L., Wadhwa S. S., Wibisono R. Producing ice
cream using a substantial amount of juice from kiwifruit with green, gold or red flesh.
Food Res. Int. (2013);50:647–656. doi: 10.1016/j.foodres.2011.05.030. [Cross Ref]
21. Trgo C., Koxholt M., Kessler H. G. Effect of freezing point and texture regulating
parameters on the initial ice crystal growth in ice cream. J. Dairy Sci. (1999);82:460–
465. doi: 10.3168/jds.S0022-0302(99)75254-9. [Cross Ref]
XII. The effect of a novel low temperature-short time (LTST)
process to extend the shelf-life of fluid milk
Springerplus. 2016; 5(1): 660.
Phillip R. Myer, Kyle R. Parker, Andrew T. Kanach, Tengliang Zhu, Mark T. Morgan,
and Bruce M. Applegate

Abstract

Pasteurization has long been the standard method to extend the shelf-life of dairy
products, as well as a means to reduce microbial load and the risk of food-borne
pathogens. However, the process has limitations, which include cost effectiveness,
high energy input, and reduction of product quality/organoleptic characteristics. In
an effort to reduce these limitations and extend shelf-life, this study examined a novel
low temperature, short time (LTST) method in which dispersed milk in the form of
droplets was treated with low heat/pressure variation over a short treatment time, in
conjunction with pasteurization. Lactobacillus fermentum and Pseudomonas
fluorescens Migula were exposed to conventional pasteurization treatments with and
without LTST. Using these organisms, the LTST addition was able to reduce
microbial load below detection limits; 1.0 × 101 cfu/mL, from approximately 1.2 × 108
and 1.0 × 107 cfu/mL for L. fermentum and P. fluorescens Migula, respectively. In
addition, the shelf-life of the treated, raw, and uninoculated product was prolonged
from 14 to 35 days, compared with standard pasteurization, to as long as 63 days
with the LTST amendment. Sensory analysis of samples also demonstrated equal
or greater preference for LTST + pasteurization treated milk when compared to
pasteurization alone (α = 0.05). Conventional pasteurization was effective at
reducing the above mentioned microorganisms by as much as 5.0 log10 cfu/mL.
However, LTST was able to achieve 7.0–8.0 log10 cfu/mL reduction of the same
microorganisms. In addition, BActerial Rapid Detection using Optical scattering
Technology detected and identified microorganisms isolated both pre- and post-
treatment, of which the only organisms surviving LTST were Bacillus spp. Increased
lethality, improved shelf-life, and equal or better organoleptic characteristics without
increased energy consumption demonstrate the effectiveness of the incorporation of
LTST. The improved shelf-life may potentially have major impacts in the dairy
industry in terms of shipping and overall sustainability.

Electronic supplementary material

The online version of this article (doi:10.1186/s40064-016-2250-1) contains
supplementary material, which is available to authorized users.

Keywords: Low temperature, LTST, MST, Pasteurization, BARDOT

Background

The manufacturing and distribution dynamics of the fluid milk industry are constantly
impacted by the primary concern of prolonging shelf-life and improving safety. The
process of heating milk for a predetermined time at a predetermined temperature
(i.e. pasteurization) is aimed at reducing microbial load and addressing the previous
issues (Nada et al. 2012). Unfortunately, many characteristics of the pasteurization
process are neither effectively robust nor cost effective. Currently, the shelf-life of
unopened, pasteurized milk is 8–14 days depending on the intensity of the treatment
(Niamsuwan et al. 2011), and the process itself consumes a significant amount of
energy. Together, with rising fuel costs, the industrial sector must find solutions to
minimize energy and fuel consumption in order to cut expenditures and reduce their
carbon footprint.

Thermal processing, such as pasteurization, has been the standard method to

extend the shelf-life of dairy products. These high temperatures (70–120 °C) can
damage and/or cause structural modifications to proteins within the product, leading
to noticeable changes in sensorial characteristics (Siciliano et al. 2000). Lowering
the thermal threshold, while still achieving adequate microorganism reduction, would
be an advantageous scenario; reducing energy use/costs and increasing the overall
quality of the product. Such treatments exist, such as low temperature long-time
(LTLT) pasteurization, or Holder pasteurization. However, these treatments have
been shown to degrade and damage many biochemical components, such as
vitamins: C, folacin, and B6 (Van Zoeren-Grobben et al. 1987; Moltó-Puigmartí et al.
2011). High-pressure processing (HPP) is utilized as an alternative to thermal
treatment, but proteins, enzymes, polysaccharides and nucleic acids have also been
shown to be adversely affected (Balci and Wilbey 1999).

A viable process would reduce costs and energy input, improve the quality of the
product, and maintain or improve microorganism reduction potential. Such benefits
would have a tremendous impact on the industry, including: increased cost
effectiveness, higher quality product, and increased distribution distances. Increased
shelf-life would also allow for a decrease in distribution locations; distribution from a
few locations, rather than hundreds.

This study examined a novel, low temperature, short time (LTST) amendment for
pasteurization, in which low heat and variable pressure were utilized to aid in the
pasteurization of a fluid milk product. This method, utilizing an Millisecond
Technologies (MST) chamber unit and previously described (Arofikin 2010), was
characterized by examining microbial reduction, shelf-life, and sensory evaluation in
order to assess the efficacy of the treatment process. In addition, detection and
identification of microorganisms isolated both pre- and post-treatment were
conducted via BActerial Rapid Detection using Optical scattering Technology
(BARDOT; Banada et al. 2009; Singh et al. 2014) and 16S rRNA gene sequencing,
in order to determine the identity and characteristics of survivor organisms at
differing steps in the process, allowing insight into the effect on potential spoilage
organisms. We hypothesize that the LTST amendment will increase microbial
reduction and shelf-life, while maintaining the quality and organoleptic
characteristics of the product.

Results and discussion

LTST bacterial reduction

The LTST method depends on a mechanism in which low heat and low pressure are
utilized to pasteurize a fluid milk product (Arofikin 2010). In this process, milk is
dispersed in the form of droplets into a process chamber and heated for 0.02 s at or
below pasteurization temperatures (≤72.7 °C). The MST chamber schematic and
image are depicted in Fig. 1.
Fig. 1 MST unit. a Schematic of the MST Process (LTST) or (MST). b Image of the
MST unit (Millisecond Technologies, New York, NY)

In order to assess the efficacy of the LTST process, the logarithmic reduction of
bacterial species was examined utilizing model Gram-positive and -negative
organisms; Lactobacillus fermentum and Pseudomonas fluorescens Migula,
respectively. Tolerance to heat varies between organisms and thus, the extent of the
pasteurization treatment is determined by the thermal characteristics of its model
organisms, and time–temperature relationships must be examined. Additional file 1:
Fig. S1 shows the D- and Z-values for P. fluorescens Migula, while Additional file 1:
Fig. S2 shows similar information for L. fermentum. The values demonstrate the
greater thermal resistance of L. fermentum compared to that of P. fluorescens
Migula. With a Z-value 9.00, compared to 7.09, L. fermentum is able to withstand an
increase in thermal treatment greater than P. fluorescens Migula. This is further
demonstrated by the associated D-values: D60 = 4.92 and 44.29 s, for P. fluorescens
Migula and L. fermentum respectively. The results also demonstrate the capability
to assess the LTST method using these organisms as a model, due to the variance
in their thermal characteristics.

Utilizing P. fluorescens Migula and L. fermentum, the LTST amendment was applied
to assess the microbial reduction potential of the system. The sampling model is
depicted in Additional file 1: Fig. S3. The first two trials assessed the system, using
process temperatures at or near that of pasteurization temperatures (Table 1). The
results show that in most runs, significant populations (P < 0.05) of the inoculated
microorganisms survived pasteurization conditions (Table 1), but did not survive the
MST unit treatment which followed. For Trial 1, Run 1, the natural microbiota in the
raw milk was approximately 104 cfu/ml and the pasteurization conditions were
enough to fully eliminate these organisms. However, for Runs 2 and 3 in the first
trial, L. fermentum survived standard pasteurization. This was especially evident
when compared to that of the natural microbiota, as only 1–2 log10 (90–99 %)
cfu/mL of the microorganisms were eliminated due to similar pasteurization
conditions. The organisms that survived pasteurization were reduced by the MST
unit which followed, when operated under the conditions listed in Table 1B.

Table 1. LTST microbial reduction

(A) Micro counts (cfu/mL)

Indicator After
Run Before processing After MST
Trial organism pasteurizer
(cfu/mL) (cfu/mL)
(cfu/mL)1

1 None 5.7 × 104 a NSb NSb

1 2 L. fermentum 1.2 × 108 a 5.0 × 106 b NS

3 L. fermentum 1.2 × 108 a 1.2 × 107 b NS

P. fluorescens
1 6.4 × 106 a NSb NSb
Migula

P. fluorescens
2 2 9.0 × 106 a 3.4 × 105 b NSc
Migula

P. fluorescens
3 1.2 × 107 a 1.1 × 106 b NSc
Migula
(A) Micro counts (cfu/mL)
Indicator After
Run Before processing After MST
Trial organism pasteurizer
(cfu/mL) (cfu/mL)
(cfu/mL)1

(B) Trial 12 Trial 22

Run 1 2 3 1 2 3

MST Temp
61.0 61.5 61.6 60.1 55.8 51.3
In3 (°C)

MST
Chamber 73.7 73.7 75.7 74.1 71.2 68.3
(°C)

MST Temp
74.6 74.3 76.2 69.7 66.7 62.9
Out (°C)

A) The reduction of organisms inculcated into raw, prior-treated, milk before

processing, after pasteurization, and after pasteurization + MST treatment. B)
Process temperatures

Means among the groups were compared using ANOVA and the Tukey’s Test.
a,b,c

Within a row, means with a different superscript were different (P < 0.05)
1NS = No significant microorganisms recovered; populations were below the
detection limit (101 cfu/mL)
2Process temperature (°C) means are of three readings (beginning, middle, and end)
of each 30 min total run
3MST Temp In is approximately the same as pasteurizer temperature out

For Trial 2, the conditions were less severe, as LTST temperatures were reduced,
including the product entering the MST unit and the MST chamber (Table 1), for
each successive run. Only Run 1 was effective in reducing the inoculated population
of P. fluorescens Migula below the detection limit (1.0 × 101 cfu/mL; P < 0.05). The
lower LTST thermal conditions for Runs 2 and 3 resulted in only 1 log10 reduction.
However, in each run for Trial 2, the conditions of the MST unit that followed
pasteurization allowed for the significant reduction of all measurable, surviving
model microorganisms (P < 0.05). Based on these data, the LTST treatment
demonstrated significant reduction of microorganisms present in inoculated milk
using optimal operating conditions. Both L. fermentum and P. fluorescens Migula
added to the raw milk at high concentrations (106–108 cfu/mL) were below detection
limits after LTST treatment.

Shelf-life evaluation of LTST treated fluid milk

To determine the shelf stability of LTST treated milk, samples at differing process
temperatures were collected after the pasteurizer and the MST unit. These samples
were subsequently plated and counted for up to 63 days to determine the shelf-life
at 4 °C. Trial 3 microbial count data is shown in Table 2. The process temperatures
are also shown in Table 2. The table shows the microbial counts in samples that
were run through the pasteurizer (using FDA-required operating conditions) followed
by processing through the MST unit. Trial 3 was treated more vigorously with regard
to temperature, in that process temperatures were greater than previous trials. The
results indicated that, on average, samples taken after the pasteurizer had significant
microbial growth after 50 days of storage at 4 °C while all three samples taken after
the MST unit had no significant growth after 50 days. Two of the three samples taken
after the MST unit continued to have no significant growth after 63 days, at the end
of the testing period. The third sample showed growth, but only after 57 days.
However, this third sample was treated with only a 1 °C temperature increase within
the MST chamber, while samples from runs 1 and 2 were treated with 10 and 5 °C
increases, respectively, further demonstrating the effectiveness of the LTST method
to prolong shelf-life beyond that of pasteurization.

Table 2. Recoverable microorganisms during refrigerated storage at 4 °C

(A)

Microbial growth in samples over time

Duration (days) 0 21 28 36 43 50 57 63

After pasteurization (cfu/mL)

Run 1 57a 68a 9a 11a 15a 5 × 102 b 5 × 103 c 5 × 107 d

Run 2 17a 9a <1a 3a 14a 22a 3 × 105 b 3 × 106 c

Run 3 <1a 54a 21a 3a 8a 3 × 103 b 3 × 106 c 3 × 107 d

After MST unit (cfu/mL)

Run 1 2a 3a 8a 5a 7a 7a 3a <1a

Run 2 <1a 8a 4a 5a 7a 4a 4a <1a

Run 3 7a 6a 6a 7a 5a 57a 3.02 × 102 b 5 × 107 c

(B)

Microbial growth in samples over time

Duration (days) 0 7 14 21 28 35

After pasteurization (cfu/mL)

<1 <1 1.59 × 106

Run 1 <1a a a 1.67 × 104 b 6.70 × 105 c d

<1 <1 7.10 × 105

Run 2 1.7a a a 4.33 × 103 b 5.30 × 104 c d

After MST unit (cfu/mL)

<1 <1
Run 1 <1a a a 7.40 × 103 b 3.57 × 105 c 2.24 × 105 c

<1 <1
Run 2 <1a a a 7.67 × 103 b 1.63 × 105 c 3.80 × 105 c

(C)
Trial 31 Trial 41

Run
1 2 3 1 2

MST Temp In2 (°C) 73.0 73.0 73.1 57.2 54.3

MST Chamber (°C) 83.8 78.5 74.4 67.6 64.8

MST Temp Out (°C) 77.5 75.2 70.9 69.0 66.3

a,b,c, dMeans
among the groups were compared using ANOVA and the Tukey’s Test.
Within a row, means with a different subscript were different (P < 0.05)
1Process temperature (°C) means are of three readings (beginning, middle, and end)
of each 30 min total run
2MST Temp In is approximately the same as pasteurizer temperature out

Trial 4 was conducted with lower MST processing temperatures to possibly

determine a threshold for suitable effectiveness (Table 2B). The data in the table
demonstrate that at lower processing temperatures, shelf-life was less stable than
that of Trial 3.
The post-pasteurization addition of the LTST method was able to prolong the shelf-
life of the product beyond 14 days, referenced to that of conventional pasteurization,
to as much as 57 days (Marsili 2000). These data indicated that the LTST method
was effective at prolonging shelf-life, utilizing the temperatures listed for Trial 3.
Notably, although traditionally pasteurized, the MST process temperatures after
were greater in Trial 3, and were still below that of typical pasteurization
temperatures and contact times in comparison (Fromm and Boor 2004). An added
benefit to this process is that the residual energy from the traditional pasteurizer is
utilized for the LTST process, allowing for greater shelf-life without additional energy
inputs (Fig. 1).

Overall, shelf-life examination of the pasteurized + MST processed milk samples

held at 4 °C showed that no significant microbial growth occurred in samples for up
to 57 days when treated at typical, low-end pasteurization temperatures (Fromm and
Boor 2004). Additionally, two out of three samples taken after the MST unit continued
to have no growth at the end of the testing (63 days). Importantly, Trial 4
demonstrated that at lower temperatures, MST treatment was able to maintain the
same effectiveness as pasteurization. Even though shelf-life was not prolonged,
lower temperatures from MST treatment were at least as effective as pasteurization.
This could prove to be of great benefit when examining process energy inputs.

Sensory evaluation of LTST pasteurized fluid milk

After determining the effectiveness of the LTST method at reducing microbial load
while prolonging shelf-life, it was necessary to determine whether the novel
processing technology had an effect on the sensorial characteristics of the milk
product. Fifty, untrained panelists were given a chance to comment on likes and
dislikes of milk products. Sensory panelists examined color, aroma, taste, aftertaste,
and ranking (preference between samples). Table 3A summarizes the sensory
comparisons between pasteurized milk and pasteurized + MST treated milk
processed during Trial 3. Differences in color, aroma, taste, and aftertaste, were
detectable between the samples by panelists as designated in italic (P < 0.05). On
three separate occasions, panelists showed no preference between samples
produced by either treatment. Sensory evaluations required the raw milk to be
pasteurized during Trial 3 to meet FDA regulations, in addition to the MST
processing. For this reason, pasteurized samples were compared to
pasteurized + MST processed samples. Even under these conditions, sensory
panelists either favored pasteurized + MST processed samples or were unable to
detect a significant difference in taste and aftertaste compared to traditionally
pasteurized milk, with the exception of Run 1. These results were attributed to the
process conditions of the run, which had the greatest MST operating temperatures
(Table 2C).

Table 3. Sensory evaluation of pasteurized and pasteurized + MST processed

milk
(A) Trial 3 Sensory characteristics1

Color Aroma Taste Aftertaste Preference

Days after 2
Run MS
processing P MST P MST P P MST
T

5.6 5.3
1 7.49 7.42 6.36 5.83 4.60 4.21 P
2 8

4.7 4.0
21 2 7.16 7.28 5.12 5.30 4.02 3.54 –
0 8

5.0 4.5
3 7.28 7.14 5.44 5.40 5.14 5.06 –
4 4

5.3 4.9
1 7.10 6.14 5.41 4.98 4.69 4.45 P
7 8

5.0 4.6
28 2 7.24 7.14 5.38 5.72 5.72 4.96 MST
2 4

4.9 4.7
3 6.90 7.02 5.32 5.40 5.34 5.18 MST
0 4

5.2 4.7
1 7.14 6.40 5.48 5.28 4.44 3.94 P
8 4

5.1 4.7
36 2 7.12 6.86 5.32 5.48 5.58 5.02 –
8 6

5.3 4.8
3 6.73 6.86 5.26 5.24 5.54 4.92 MST
8 2

Sensory characteristics1
(B)
Color Aroma Taste Aftertaste
Tria Preference3
l 5 Run MS MS MS
P P P P MST
T T T

1& 7.2 6.0

7.26 6.08 6.86 7.05 6.15 6.38 MST
2 3 8
(C) Trial 5 processing temperatures

Run 14 Run 24

MST Temp In5 (°C) N/A 61.7

MST Chamber (°C) N/A 72.4

MST Temp Out (°C) N/A 74.2

Pasteurization temp 73.8 73.8

1Italic values were different (P < 0.05). Greater values indicate greater preference
2Preference; Indicates whether panelists preferred pasteurized sample (P) or
pasteurized + MST sample
3Preference;Indicates whether panelists preferred pasteurized sample (P) or
MST + pasteurized sample
4Process temperature (°C) means are of three readings (beginning, middle, and end)
of each 30 min total run
5MST Temp in is approximately the same as pasteurizer temperature out

Trial 5 was completed using slightly increased processing temperatures (Table 3C)
compared to Trial 4 (Table 2C), but was performed using a different process
configuration, allowing sensory evaluation of the product with MST + pasteurization.
One hundred panelists were used for sensory evaluation. The results in Table 3B
show that there were no significant differences in color, aroma, taste, or aftertaste
among the untrained panelists. In addition, preference of each treatment was
significantly greater for that of MST + pasteurized fluid milk. These results represent
similar trends seen with the sensory data from previous trials, and demonstrate that
reduction in microbial load using the MST unit, greater than that of traditional
pasteurization, has no effect on the sensorial quality of the milk seen with traditional
pasteurization.

Isolation and identification of survivor microorganisms

The effectiveness of the LTST method is dependent upon the physical and thermal
mechanisms involved in the process (Arofikin 2010), as well as the microorganisms
interrogated. The former has been the focus of the previous pasteurization research,
but the latter must also take precedence, which can be accomplished by examining
the microorganisms that survive each step in the treatment.
Light scattering technology can be used to differentiate colonies on plated agar that
normally would not be differentiated by simple, common, visual identification
(Banada et al. 2009). A label-free method called BActerial Rapid Detection using
Optical scattering Technology (BARDOT) can be used to aid in the identification of
bacterial species by their respective forward scattering patterns (Bae et al. 2011). In
this method, the biophysical characteristics of cultured bacteria result in differences
of light scattering from a transmitted laser beam, subsequently producing unique
scatter images (Bae et al. 2011). Using this automated system, researchers are able
to cost-effectively identify unique scatter images, representing the identification and
differentiation at the serovar level (Rajwa et al. 2010).

Using BARDOT, differing colonies from plated samples within Trial 4 were selected
based on their unique scatter images. Each colony, representing potentially different
bacterial species, was then identified using 16S rRNA gene sequencing. The
bacterial load of the raw milk was of low concentration (~10 2 cfu/mL) due to its
freshness and expedient delivery from the Purdue University Dairy Research and
Education Center. The identities and scatter images from BARDOT of the 21 isolated
colonies are listed in Fig. 2. All microorganisms were present at the start of
processing, but those surviving pasteurization and MST treatment were noted
(Table 4) and also ordered as such in Fig. 2. A phylogenetic analysis of organisms
present in the pasteurized sample and those surviving pasteurization + MST
treatment are depicted in Fig. 3a, b, respectively. Interestingly, many
Pseudomonads were identified in the raw sample and after pasteurization, which is
typical for dairy products (Cousin et al. 2001). Additionally, an Achromobacter
species was identified, which is also found in dairy (Poffé and Mertens 1988).
However, as predicted, many species that survived traditional pasteurization were
from the Bacillus or Paenibacillus genus, supporting the selection of L. fermentum
as a model Gram-positive organism for the LTST pasteurization process. Yet, the
identification of these spore-forming organisms suggests that assessment of the
LTST method needs to be more stringent, by possibly examining the process
treatment of heat/stress tolerant Bacillus species. Use of such organisms would
better serve the evaluation of the LTST method and ultimately allow for easier
approval of use over traditional pasteurization methods.
Fig. 2. Identification of pasteurization + LTST survivors. Representative scatter images of
bacterial species differentiated by BARDOT and identified by 16S rRNA gene sequencing.
Survivors were samples from before treatment, after pasteurization, and after LTST
treatment
Table 4. LTST method and pasteurization survivors designated by sampling
location

Organism Sample location within treatment

Paracoccus spp. S3 Post-pasteurization

Pseudomonas stutzeri strain M16-9-4 Post-pasteurization

Pseudomonas fragi Post-pasteurization

Pseudomonas fluorescens strain TCA3 Post-pasteurization

Pseudomonas spp. MC1 Post-pasteurization

Pseudomonas brassicacearum Post-pasteurization

Pseudomonas spp. 11BF10 Post-pasteurization

Pseudomonas spp. T7A Post-pasteurization

Microbacterium oxydans strain N3 Post-pasteurization

Bacillus subtilis strain 10010 Post-pasteurization

Klebsiella pneumoniae strain RDS7 Post-pasteurization

Achromobacter xylosoxidans strain PRE8 Post-pasteurization

Pseudomonas fluorescens strain V7c10 Post-pasteurization

Bacillus pumilus strain J2RP2 Post-pasteurization + MST

Bacillus pumilus strain L17 Post-pasteurization + MST

Bacilllus pumilus strain Znu-110 Post-pasteurization + MST

Bacillus spp. A111 Post-pasteurization + MST

Bacillus stratosphericus strain S6 Post-pasteurization + MST

Bacillus altitudinis strain KtMA2-6 Post-pasteurization + MST

Paenibacillus favisporus Post-pasteurization + MST

Paenibacillus spp. JSC-N3-214-3 Post-pasteurization + MST

Fig. 3. Phylogenetic analysis of microorganisms. Phylogenetic trees displaying the
relationships among species identified in the a traditionally pasteurized treatment and b
species surviving the pasteurization + MST treatment. Tree data was determined by the
analysis of 16S rRNA gene sequences. The scale bar represents substitutions per site.
Bootstrap values are shown at the nodes (based on 500 re-samplings). Survivors were
sampled from before treatment, after pasteurization, and after pasteurization + MST
treatment
Conclusions

The LTST addition demonstrated reduction in microbial load, prolonged shelf-life,

and minimal-to-no loss of sensorial/organoleptic properties tested. Still, several
parameters of the process need to be examined. Using more thermally robust model
microorganisms (e.g. Bacillus spp.) will help to better assess process efficacy.
Systematic means to better limit product contamination, beyond technological
parameters, such as the aforementioned test organisms, would facilitate the
application of LTST amendment beyond the laboratory to potential industry
applications.

The use of LTST as a method of pasteurization is additionally promising with regard

to its source of energy. Currently, FDA regulations require that raw milk be
pasteurized. To that end, the MST unit can be connected in-line with a standard
pasteurizer to enhance product shelf-life via greater log reduction of spoilage
organisms. Current traditional methods of pasteurization have been effective at
reducing microbial load by as much as 5.0 log10 cfu/mL (Guan et al. 2005).
However, LTST has been able to achieve 7.0–8.0 log10 reduction of
microorganisms. Additionally, its in-line connection to a traditional pasteurization
tube provides enough energy to run the MST unit, without addition of exogenous
heat energy. Thus, the energy-saving characteristics and the previous results serve
to demonstrate the effectiveness of the LTST process.

Methods

Organisms and growth conditions

Pseudomonas fluorescens strain Migula (ATCC Number: 27663) was grown in two,
1L flasks of Luria broth (LB) (pH 7.0–10 % tryptone, 5 % yeast extract, 10 % NaCl),
overnight (24 h), with constant aeration at 100RPM, and incubated at room
temperature (20–22 °C). The overnight cultures were then used to inoculate ten 16 L
carboys containing LB broth (pH 7.0). The inoculated carboys were mixed with
magnetic stir bars at room temperature, and allowed to grow for 3 days to ensure a
high cellular concentration. Growth was monitored via optical density (OD600)
measurement with an Eppendorf Biophotometer. Cells were centrifuged at 9803×g
for 8 min at 4 °C (Beckman Coulter Avanti J-25i) and resuspended in phosphate
buffered saline (PBS) (pH 7.0) in order to concentrate the cells from 160 to 2 L and
obtain a cellular concentration of 4 × 109 cfu/mL. The concentrated cells were then
stored at 4 °C until use the following day. The P. fluorescens strain Migula species
identity was confirmed by phage typing using Phi-S1.

Lactobacillus fermentum was grown in two 1L flasks of Lactobacilli MRS broth (BD
Diagnostics, Franklin Lakes, New Jersey) (pH 6.5) overnight, with constant shaking
at 100RPM, and incubated at 37 °C. The overnight cultures were then used to
inoculate ten 16L carboys containing MRS broth (pH 6.5). The inoculated carboys
were mixed using magnetic stir bars at 37 °C, and allowed to grow for 3 days to
ensure a high cellular concentration. Cells were centrifuged at 9803×g for 8 min at
4 °C and resuspended in phosphate buffered saline (PBS) (pH 7.0) in order to
concentrate the cells from 160 to 2 L and obtain a cellular concentration of
1.2 × 1010 cfu/mL. The concentrated cells were then stored at 4 °C until use the
following day.

Decimal reduction and thermal death time determination

Decimal reduction and thermal death time were conducted in a water bath, heated
by an Isotemp Immersion Circulator (model 730, ThermoFisher Scientific, Waltham,
MA), over a magnetic stir plate (Lab-Line Multi-Magnestir, No. 1278, ThemoFisher
Scientific, Waltham, MA). Temperature was monitored with American Society for
Testing and Materials (ASTM) thermometers accurate to ±0.2 °C. P. fluorescens
strain Migula and L. fermentum were grown overnight and incubated in LB broth
(26 °C), and MRS broth (37 °C), respectively, with constant shaking at 100RPM.
Experiments were conducted in triplicate. Three 250 mL dual-neck round bottom
flasks were filled with 100 mL Grade A ultra-high-temperature (UHT) whole milk
(Parmalat, Grand Rapids, MI) and allowed to acclimate to the respective temperature
for 10 min; 54.4, 57.2, and 60 °C for P. fluorescens strain Migula and 60, 62.8, and
65.5 °C for L. fermentum. Flasks were then inoculated with 1 mL of appropriate
culture and treated thermally for 5, 10, 20, 40, 80, and 160 s. At the designated time,
1 mL of the sample was extracted from the flask and placed in 9 mL of ice-water-
chilled PBS. Samples were allowed to recover for 10 min in an ice water bath.
Thermally treated samples were serially diluted in PBS, plated on their respective
media agar (LB, MRS), and incubated at their appropriate temperature (26 °C,
37 °C) overnight. Colonies were counted 1 day later and averaged to obtain log
reduction values.

The log numbers of the survivors at each time were used to determine D-values. The
D-value was determined from the negative reciprocal of the slopes of the regression
lines (log10 cfu/mL vs. time of exposure to the thermal treatment, at constant
temperature; Mazzola et al. 2003). A linear regression was used from log D-values
versus temperature, and the Z-value was obtained from the absolute value of the
inverse of the slope.

MST processing

The specifics of the MST technology are described in detail in US patent 7,708,941
(Arofikin, 2010). Briefly, the system consists of a balance tank, product pumps,
magnetic flow meters, temperature transmitters, level transmitters, tubular heat
exchanger with raw regenerator, heater, hold tube, MST chamber, pasteurized
regenerator, and cooler sections. The system is started up on water, then sanitized
by circulating water at 98 °C for 20 min. After sanitization, the system is cooled down
and controlled at the desired operating temperatures for the experiment. The system
is equipped with: regenerator bypass on the raw regenerator to allow for control of
the temperature out of the regenerator and cascade hot water to heater product out
temperature control to control the feed temperature to the MST inlet. The product is
then fed to the inlet nozzle or nozzles of the MST chamber at a controlled pressure
(pressure = 800,000 Pa). Milk droplets are sprayed into the unit, at which time the
milk is then heated to temperature within 0.02 s, which establishes the desired effect.
The internal temperature of the MST unit is controlled by the cascade temperature
control of hot water to internal temperature. This temperature is controlled at a value
10 °C higher than the inlet temperature. The lower jacket is cooled to prevent further
heat treatment of the droplets collecting on the bottom of the MST Chamber. The
discharge pump from the MST chamber controls a minimum level in the discharge
leg of the MST chamber, and acts as the motive force to push through the
downstream pasteurized regenerator and cooling sections of the process, and finally
discharged to the destination. The system is controlled for circulation of water, water
to product change-over, product discharge, product to water change-over, and
clean-in-place (CIP) cycles. The milk is then cooled to 4 °C and expelled from the
processor.

MST-pasteurization milk bacterial reduction

Each trial consisted of three separate runs of homogenized milk from the Purdue
University Dairy Research and Education Center (W. Lafayette, IN). Three hundred
gallons (~1136L) per run were processed using the tubular pasteurizer followed by
the MST chamber (Arofikin 2010). Milk was pumped into a mixing vat and inoculated
with 1L of the stored, concentrated P. fluorescens Migula or L. fermentum (4 × 109
and 1.2 × 1010 cfu/mL, respectively). Raw milk bacterial load was 5.7 × 104
(±1.5 × 103) cfu/mL. The inoculated milk was then pumped into the holding basin
within the pasteurization unit and processed. The duration of each run was
approximately 30 min. Samples of milk were collected at 3 time points (10 min
intervals) throughout the run, and at locations before processing, after the
pasteurizer, and after the MST chamber. Before extraction, sterile sample
diaphragms were attached to collecting ports, and were additionally treated with
70 % ethanol. Milk was extracted and collected via sterile 250 mL QMI sampling
assembly bags (QMI, Oakdale, MN), with an 18 gauge needle and three feet of
tubing. After the sampling bag was filled, the needles were removed from the port
and capped with a Luer lock fitting and cap. Collected milk was immediately stored
in an ice water bath and collecting ports were again sterilized with 70 % ethanol.

Samples from the processed runs were aseptically drawn, diluted in phosphate
buffered saline (PBS, pH 7), plated, and incubated on LB agar (17 %, 26 °C) and
Lactobacilli MRS agar (17 %, 37 °C) to determine the bacterial reduction of P.
fluorescens Migula and L.fermentum, respectively. Samples were stored at 4 °C
between plating periods. Samples were plated weekly for 5 weeks (35 days) to track
the growth of the inoculated runs. Plates were counted after 2 days of growth.

Shelf-life determination

Experimental runs were completed without bacterial inoculation in order to assess

the shelf-life of the milk product and to perform sensory evaluation. Processing was
conducted similarly to the previous method; albeit without the bacterial mixing vat.
Extraction and collection of the processed milk were performed similarly to the
previous procedures (conditions are previously noted). In addition, milk was
collected after pasteurization alone and pasteurization + MST and into sterile, 2 L,
brown glass bottles for sensory evaluation. The bottles were housed in a laminar
flow hood in order to prevent post-processing contamination. After extraction, the
bottles were stored at 7 °C until sensory evaluation.

Samples from the raw milk processed runs were aseptically drawn, diluted in
phosphate buffered saline (PBS, pH 7), plated on plate count agar (PCA, HiMedia
Laboratories, Mumbai, India), and incubated at both 26 and 7 °C to assess
psychrophile survivability and growth. Samples were stored at 4 °C between plating
periods. Samples were plated weekly for 5–9 weeks (35–63 days), dependent upon
the study, to determine the shelf-life of the processed raw milk. Plates incubated at
26 °C were counted after 2 days of growth, while plates incubated at 7 °C were
counted after 5 days of growth.

Sensory evaluation

Panelists participated voluntarily. This study was approved by the local research
ethics committee (IRB Protocol Number 1209012647). Sensory evaluation using a
paired comparison test of 50 untrained panelists was performed. Samples from
pasteurized, pasteurized + MST, and MST + pasteurized treated milk were
compared for preference/acceptability. Testing was performed 21, 28, and 36 days
after processing. The products were kept in refrigeration when not in use (7 °C). Prior
to testing, the milk was poured into a pitcher and a hand blender was used to further
homogenize the milk. Each panelist was given a 2 sample in a 5 oz. drinking cup
and was first asked hedonic and ranking questions. The panelists were given a
chance to comment on likes and dislikes. Parameters examined were color, aroma,
taste, aftertaste, and ranking.

Multiple comparison tests were conducted. Tukey’s HSD was performed to control
for maximum experiment-wise error rate and can be used without F protection.
According to standard practice, LSD and Duncan’s were only considered if the
ANOVA P value was deemed acceptable to control for experiment-wise error rates
(under the complete null hypothesis). If Duncan’s Multiple Range Test was used,
only the largest critical range was reported. If automatic significance was selected,
an available significance level was chosen for the multiple comparison test based on
the observed P value.

Application of BARDOT to detect and identify bacterial species

Growth on plates from representatives of the raw processed runs were initially
differentiated by BARDOT (Banada et al. 2009), selected, serially diluted in PBS and
plated on PCA to obtain isolated colonies, then incubated at 26 °C for 24 h or until
colony size reached 1.3 ± 0.2 mm. Colony size was used as a fixed parameter
because growth rates are variable among species. Scatter images of colonies were
acquired and analyzed using the BARDOT system (Advanced Bioimaging Systems,
W. Lafayette, IN). Unique scatter images were selected for identification by 16S
rRNA gene sequencing.

Bacterial identification by 16S rRNA gene sequencing

Cultures were identified by 16S rRNA gene sequencing of PCR-amplified products

(Lane 1991; Marchesi et al. 1998). 16S rRNA-specific primer pairs, 27F and 1492R
were used to amplify the target gene 1465 bp in length. PCR conditions include: an
initial denaturation at 94 °C for 4 min, followed by 30 cycles consisting of 94 °C for
55 s, 46 °C for 55 s, and 72 °C for 4 min, and final extension at 72 °C for 9 min. PCR
products were sequenced, quality checked, and cleaned by the Purdue Genomics
Core Facility (Purdue University, W. Lafayette, IN), and the 16S reads were
classified using the NCBI nucleotide collection database. Phylogenetic analysis was
conducted from the resultant sequences and analyzed via alignment, curation,
phylogeny, and tree rendering programs from Dereeper et al. (2008).

Authors’ contributions

PRM aided in sample acquisition, carried out all sample manipulations, nucleic acid
extraction, amplification, and preparation for sequencing; participated in all statistical
analyses; generated all figures; participated in design of the study; contributed to
conclusions; and drafted the manuscript. KRP, ATK, and TZ aided in sample
acquisition, instrumental in the oversight of and design of the study, and contributed
to conclusions. MTM oversaw and participated in all statistical analyses, food
process engineering, oversaw sensory testing, aided in the development of the
model, contributed to conclusions; and participated in design of the study. BMA
oversaw and participated in all statistical analyses, participated in design of the
study, contributed to conclusions, oversaw the food microbiology component, aided
in sample acquisition and use of BARDOT. All authors read and approved the final
manuscript.

Acknowledgements

We thank Carla Rosenfield, Eileen Duarte Gomez, and Amy Fleishman Littlejohn for
technical assistance. This research was partially supported through a cooperative
agreement with the Agricultural Research Service of the US Department of
Agriculture Project Number 1935-42000-035 and the Center for Food Safety
Engineering at Purdue University, and Millisecond Technologies.

Competing interests

The authors declare that they have no competing interests.

Abbreviations

BARDOT BActerial Rapid Detection using Optical scattering Technology

HPP high-pressure processing

LTLT low temperature-long-time

LTST low temperature-short time

MST Millisecond Technologies

UHT ultra-high-temperature

Additional file

10.1186/s40064-016-2250-1 D- and Z-values for P. fluorescens Migula. Figure S2.

D- and Z-values for L. fermentum. Figure S3. Sampling model.(153K, doc)

Notes

This paper was supported by the following grant(s):

Agricultural Research Service 1935-42000-035 to Bruce M. Applegate.

Millisecond Technologies.

Contributor Information

Phillip R. Myer, Email: ude.ktu@reymp.

Kyle R. Parker, Email: moc.liamg@rekrapyk.

Andrew T. Kanach, Email: ude.eudrup@hcanaka.

Tengliang Zhu, Email: moc.liamtoh@20028002ltz.

Mark T. Morgan, Email: [email protected].

Bruce M. Applegate, Phone: (765) 496-7920, Email: ude.eudrup@etagelppa.

References

 Arofikin NV (2010) Liquid product pressure treatment method and device. US

Patent 7,708,941, 4 May 2010
 Bae E, Aroonnual A, Bhunia AK, Hirleman ED. On the sensitivity of forward
scattering patterns from bacterial colonies to media composition. J
Biophotonics. 2011;4:236–243. doi: 10.1002/jbio.201000051. [PubMed]
[Cross Ref]
 Balci AT, Wilbey RA. High pressure processing of milk-the first 100 years in
the development of a new technology. Int J Dairy Technol. 1999;52:149–155.
doi: 10.1111/j.1471-0307.1999.tb02858.x. [Cross Ref]
 Banada PP, Huff K, Ba E, Rajwa B, Aroonnual A, Bayraktar B, Adil A,
Robinson JP, Hirleman ED, Bhunia AK. Label-free detection of multiple
bacterial pathogens using light-scattering sensor. Biosens Bioelectron.
2009;24:1685–1692. doi: 10.1016/j.bios.2008.08.053. [PubMed] [Cross Ref]
 Cousin MA, Jay JM, Vasavada PC. Psychrotrophic microorganisms. In: Ito K,
Downes FP, editors. Compendium of methods for the microbiological
examination of foods. Washington, DC: American Public Health Association;
2001.
 Dereeper A, Guignon V, Blanc G, Audic S, Buffet S, Chevenet F, Dufayard
JF, Guindon S, Lefort V, Lescot M, Claverie JM, Gascuel O. Phylogeny. fr:
robust phylogenetic analysis for the non-specialist. Nucleic Acids Res.
2008;36:W465–W469. doi: 10.1093/nar/gkn180. [PMC free article] [PubMed]
[Cross Ref]
 Fromm HI, Boor KJ. Characterization of pasteurized fluid milk shelf-life
attributes. J Food Sci. 2004;69:M207–M214. doi: 10.1111/j.1365-
2621.2004.tb09889.x. [Cross Ref]
 Guan D, Chen H, Hoover DG. Inactivation of Salmonella typhimurium DT 104
in UHT whole milk by high hydrostatic pressure. Int J Food Microbiol.
2005;104:145–153. doi: 10.1016/j.ijfoodmicro.2005.01.014. [PubMed] [Cross
Ref]
 Lane DJ. 16S/23S rRNA sequencing. Nucleic acid techniques in bacterial
systematics. New York: Wiley; 1991. pp. 115–175.
 Marchesi JR, Sato T, Weightman AJ, Martin TA, Fry JC, Hiom SJ, Wade WG.
Design and evaluation of useful bacterium-specific PCR primers that amplify
genes coding for bacterial 16S rRNA. Appl Environ Microbiol. 1998;64:795–
799. [PMC free article] [PubMed]
 Marsili RT. Shelf-life prediction of processed milk by solid-phase
microextraction, mass spectrometry, and multivariate analysis. J Agric Food
Chem. 2000;48:3470–3475. doi: 10.1021/jf000177c. [PubMed] [Cross Ref]
 Mazzola PG, Penna TC, da S Martins AM. Determination of decimal reduction
time (D value) of chemical agents used in hospitals for disinfection purposes.
BMC Infect Dis. 2003;3:24. doi: 10.1186/1471-2334-3-24. [PMC free article]
[PubMed] [Cross Ref]
 Moltó-Puigmartí C, Permanyer M, Castellote AI, López-Sabater MC. Effects
of pasteurisation and high-pressure processing on vitamin C, tocopherols and
fatty acids in mature human milk. Food Chem. 2011;124:697–702. doi:
10.1016/j.foodchem.2010.05.079. [Cross Ref]
 Nada S, Ilija D, Igor T, Jelena M, Ruzica G. Implication of food safety
measures on microbiological quality of raw and pasteurized milk. Food
Control. 2012;25:728–731. doi: 10.1016/j.foodcont.2011.12.022. [Cross Ref]
 NCBI nucleotide collection database (2013) National Center for
Biotechnology Information, National Library of Medicine, Bethesda MD.
http://blast.ncbi.nlm.nih.gov/. Accessed 18 Dec 2013
 Niamsuwan S, Kittisupakorn P, Mujtaba IM (2011) Optimization approach to
minimize energy consumption in pasteurized milk process. International
conference on chemistry and chemical process. Bangkok, Thailand
 Poffé R, Mertens W. Heat-stable proteases of psychrotrophic bacteria
isolated from cooled raw milk. Appl Microbiol Biotechnol. 1988;27:437–442.
doi: 10.1007/BF00451609. [Cross Ref]
 Rajwa B, Dundar MM, Akova F, Bettasso A, Patsekin V, Hirleman DE, Bhunia
AK, Robinson JP. Discovering the unknown: detection of emerging pathogens
using a label-free light-scattering system. Cytom A. 2010;77:1103–1112. doi:
10.1002/cyto.a.20978. [PMC free article] [PubMed] [Cross Ref]
 Siciliano R, Rega B, Amoresano A, Pucci P. Modern mass spectrometric
methodologies in monitoring milk quality. Anal Chem. 2000;72:408–415. doi:
10.1021/ac990590n. [PubMed] [Cross Ref]
 Singh AK, Bettasso AM, Bae E, Rajwa B, Dundar MM, Forstere MD, Liu L,
Barrette B, Lovchike J, Robinson JP, Hirleman ED, Bhunia AK. Laser optical
sensor, a label-free on-plate Salmonella enterica serovar colony-detection
tool. mBio. 2014;5(1):e01019-13. doi: 10.1128/mBio.01019-13. [PMC free
article] [PubMed] [Cross Ref]
 Van Zoeren-Grobben D, Schrijver J, Van den Berg H, Berger HM. Human
milk vitamin content after pasteurisation, storage, or tube feeding. Arch Dis
Child. 1987;62:161–165. doi: 10.1136/adc.62.2.161. [PMC free article]
[PubMed] [Cross Ref]
XIII. Expectation or Sensorial Reality? An Empirical
Investigation of the Biodynamic Calendar for Wine
Drinkers
PLoS One. 2017; 12(1): e0169257.
Wendy V. Parr,1,* Dominique Valentin,2 Phil Reedman,3 Claire Grose,4 and James
A. Green5
Andrey E. Ryabinin, Editor

Abstract

The study’s aim was to investigate a central tenet of biodynamic philosophy as

applied to wine tasting, namely that wines taste different in systematic ways on days
determined by the lunar cycle. Nineteen New Zealand wine professionals tasted
blind 12 Pinot noir wines at times determined within the biodynamic calendar for wine
drinkers as being favourable (Fruit day) and unfavourable (Root day) for wine tasting.
Tasters rated each wine four times, twice on a Fruit day and twice on a Root day,
using 20 experimenter-provided descriptors. Wine descriptors spanned a range of
varietal-relevant aroma, taste, and mouthfeel characteristics, and were selected with
the aim of elucidating both qualitative and quantitative aspects of each wine’s
perceived aromatic, taste, and structural aspects including overall wine quality and
liking. A post-experimental questionnaire was completed by each participant to
determine their degree of knowledge about the purpose of the study, and their
awareness of the existence of the biodynamic wine drinkers’ calendar. Basic wine
physico-chemical parameters were determined for the wines tasted on each of a
Fruit day and a Root day. Results demonstrated that the wines were judged
differentially on all attributes measured although type of day as determined by the
biodynamic calendar for wine drinkers did not influence systematically any of the
wine characteristics evaluated. The findings highlight the importance of testing
experimentally practices that are based on anecdotal evidence but that lend
themselves to empirical investigation.

Introduction

Influence of extrinsic factors on perceived liking and quality of foods and beverages
is now well established, with research demonstrating effects from a variety of stimuli
including ambient music, consumption location (e.g., a laboratory vs a restaurant)
and wine bottle shape [1]. The present study investigated a contentious factor,
namely the impact of the lunar cycle on perception of Pinot noir wine qualities by
experienced wine professionals.

Over recent decades the wider agricultural philosophy of biodynamics, founded by

Austrian philosopher Rudolf Steiner in the 1920s [2], has systematically increased in
influence within the international wine industry. This is exemplified by recent articles
in media such as The Drinks Business [3] and The Independent [4]. Conceivably, a
reason for the current interest in biodynamic practice relates to a desire by wine
producers to farm more sustainably, with focus on environmental issues such as soil
health in keeping with general societal trends. Recently there has been extension of
biodynamic philosophy from viticultural practice and wine production to include wine
tasting [5]. The aim of the present study was to investigate a central tenet of
biodynamic philosophy as applied to wine tasting, namely that wines taste different
in systematic ways on days determined by the lunar cycle.

Biodynamic practices have their basis in a series of lectures titled Spiritual

foundations for a renewal of agriculture [2]. Biodynamic agriculture follows the
central tenets of organic production, namely exclusion of synthetic fertilisers and
chemical herbicides that are common in conventional agriculture, and inclusion of
practices such as mulching, manure and composts, but adds further elements.
These include focus on lunar rhythms, self-sufficiency, and the use of specially
prepared sprays and composts that involve cow manure. There is much scepticism
amongst scientists about biodynamics as a form of agriculture, this lack of scientific
respectability presumably, at least in part, due to the relative lack of sound empirical
data on the topic. The dearth of sound data in turn is understandable in that many of
the beliefs and practices of biodynamics do not lend themselves easily to empirical
testing. Further, it is virtually impossible to control the myriad of potentially
confounding factors when making comparisons of viticultural and/or oenological
practice as a function of farming type (conventional; organic; biodynamic). Despite
these difficulties, a longitudinal study over 21 years that compared biodynamic,
bioorganic, and conventional farming practices in Switzerland [6] reported enhanced
soil fertility and increased biodiversity in their organic plots relative to the
conventionally farmed plots. However, in a recent viticultural study [7], fruit quality
was reported as not influenced by the organic and biodynamic management systems
investigated. In terms of wine production, one of the few studies comparing
biodynamic and organic viticulture that has been published in a peer-reviewed
journal reported no consistent differences in any of the physical, chemical, and
biological parameters measured [8]. To our knowledge there are no wine sensory
data in support of advantages of biodynamic practices over conventional farming
practices in terms of enhanced sensory experience although a recent investigation
reported on this topic indirectly. When over 70,000 wine assessments (wine ratings)
given to wines in three publications, Wine Advocate, Wine Enthusiast, and Wine
Spectator, were analysed [9], the authors concluded that organic or eco-certification,
the latter including organic and biodynamically-produced wines, was statistically
associated with increased wine quality ratings in red wines, but not in white wines.

The biodynamic calendar for wine drinkers was first produced in German
approximately fifty years’ ago by Maria Thun and is now published by her son
Matthias Thun [5]. Since 2010, the calendar has been published annually in the
English language for the United Kingdom time zone, namely for GMT/British
Summer time, by Floris Books in Edinburgh and is available also in ‘phone App form.
The iPhone App called ‘Wine Tonight’ not only provides UK consumers with
information as to whether it is considered a favourable day or unfavourable day to
drink their wine but as well offers conversion to other time zones including those of
southern hemisphere countries.

The wine-drinking advice in the calendar is based on biodynamic farming principles,

a key tenet being that many agricultural practices are timed according to the moon’s
cycle. The calendar advises wine drinkers accordingly; that is, the calendar provides
‘days’ when the moon’s rhythms suggest that a wine will taste its best. The ‘days’
are seldom 24 hour periods but are temporal intervals categorised according to star
constellations and the movement of the moon in terms of ascending and descending
cycles. Four types of days have been identified [5] based on when the moon moves
through the twelve constellations. When the moon moves into Sagittarius, Aries, and
Leo a Fruit day takes over; when in Libra, Aquarius, and Gemini, a Flower day
occurs; Leaf days involve the moon moving through Scorpio, Pisces, and Cancer
while Root days take over when the moon moves into Virgo, Capricorn and Taurus.
Fruit and Flower days are considered favourable for wine tasting while Leaf and Root
days are best avoided.

To our knowledge, there are no published, sound empirical data to support the notion
that wines taste different according to where the moon is in its lunar cycle. None-
the-less, the notion that the moon’s rhythms exert influence on the taste of a wine in
systematic ways does lend itself to empirical investigation. Anecdotal evidence in
the form of wine industry media [10] suggests that some professionals in the wine
industry, in particular wine producers and retail outlet and wine distribution company
staff, appear to accept that the moon may exert some sort of influence over how a
beverage tastes on a particular day, despite the lack of scientific evidence. For
example, in the United Kingdom, several major supermarkets and wine retail outlets
such as Tesco and Marks & Spencer have been reported as organising their wine
tasting sessions around "good" days (Flower days; Fruit days) and "bad" days (Leaf
days; Root days) as dictated by the lunar calendar [10] [11] [5]. Published anecdotal
reports provide some details as to precisely how wines are expected to change in
terms of being ‘better’ or ‘worse’ as a function of the lunar cycle such as becoming
more tannic or bitter on Leaf days and Root days, the days that are argued as not
favourable for wine tasting, while expressing better their freshness and aromatic
qualities on Fruit days and Flower days, days argued as best for wine tasting [12]
[13].

It is conceivable that respect within the wine industry for the notion that the moon’s
rhythms influence a wine’s taste has its basis in attempts to understand why some
wines do appear to taste differently across different days. Anecdotal evidence
suggests that it is not unusual for wine professionals and knowledgeable wine
consumers to report that a wine, including a bottled wine, appears to taste “different”
when tasted at varying time periods such as on consecutive days or weeks. For
example, a wine may be perceived as tasting different across two successive
tastings of the same wine, or “not showing well” on a particular day [14]. There are
many reasons that could underlie such perceived differences including wine
composition factors, weather and atmospheric pressure, and human perception
factors including memory and mood of the taster. Hence, the potential factor
investigated in the present study is one possibility only, but one that does lend itself
to empirical investigation.

One convoluted notion to justify why the moon may influence the taste of wine is
based on the known impact of the lunar cycle on the tides. This has been interpreted
and extended by followers of biodynamic philosophies to argue that the moon
therefore also affects the water in plants, the water in a wine, and the water in the
human body. The latter notion includes the argument that human behaviour (e.g.,
moods) is affected by the lunar cycle such that human sensory experience including
the tasting of wine is influenced by the moon. Hence, there are broadly two
possibilities: a wine may change (e.g., in some aspects of composition such as
bonding of chemical compounds) according to the lunar cycle, or the taster may
perceive the wine to change. In the present study we separate these two aspects by
providing sensory data addressing the question “are the wines reported by
experienced wine professionals as tasting better on Fruit days in comparison with
Root days?”, as well as providing wine composition data where data analysis was
conducted on the same wines employed in the sensory study on the same day on
which they were tasted (i.e., on both a Fruit day and on a Root day).

To test the notion that wines taste different in systematic ways on days determined
by the biodynamic wine drinkers’ calendar [5], several aspects of methodology
required consideration. First, it was essential that all wine tasters were blind as to
the purpose of the study and in terms of following the biodynamic wine drinkers’
calendar. To determine this, each wine professional completed a questionnaire after
all tasters had participated in the study, the questionnaire seeking the relevant
information [see S1 Questionnaire]. A second and very important factor was to effect
a correct time-zone change to ensure that our selected Fruit days and Root days
were valid in terms of the ascending and descending moon cycle. This was achieved
by consultation with several relevant people and their publications (e.g., the Astro-
Calendar produced by Brian Keats in Tasmania) [15], and confirming that the
conversions offered for New Zealand Summer Time matched the conversion
available via the iPhone App associated with the calendar for the United Kingdom
time zone produced by Floris Books (www.florisbooks.co.uk) [5]. A third important
factor was to operationalise different in terms of its application to how the wines
would be expected to taste on a good day (Fruit day) and on a less favourable day
(Root day) according to the calendar. Anecdotal evidence from online, published
reports [12], along with interview data from UK wine-industry professionals
purporting to use the wine drinkers’ calendar, was employed to determine the wine
type best suited to testing our hypothesis, along with the wine characteristics most
likely to serve as appropriate descriptors. Pinot noir was selected as the wine of
choice. This was due to its varietal characteristics, namely an aromatic profile that
could be more-or-less expressive, and a tannin profile that could range between
silkiness and harshness [16], these aspects of wine sensory experience involving
the predominant attributes reported within the anecdotal evidence as changing
according to the lunar cycle. Third, we selected relatively young wines for the study
given that according to the biodynamic calendar [5] a wine greater than five years of
age may be advantaged by being tasted on a Leaf day rather than on a Fruit or
Flower day. Fourth, we selected wines from biodynamic, organic and conventional
wine producers to comprise the sample set for the study as the lunar cycles’
influences are argued as relevant irrespective of wine-production mode [5]. Finally,
since several media references made to tasting according to biodynamic philosophy
[17] have suggested that weather patterns (e.g., atmospheric pressure) may also
influence how a wine tastes, we collected relevant meteorological data. During the
time that the experimental tastings were conducted, we recorded mean per hour
measures of moisture (rain in mm), sunshine (minutes), wind speed, and
atmospheric pressure.

Summary and hypothesis

In the current study our experimental hypothesis was that Pinot noir wines would be
reported as tasting different in systematic ways on days determined by the
biodynamic calendar for wine drinkers [5]. More specifically, we predicted that the
wines would be perceived as more aromatic, fruity, concentrated, and overall
flavoursome on Fruit days than on Root days. Conversely, wines were predicted to
be perceived on a Root day as less balanced, more aggressive in terms of tannin
influences, and with any green or leafy characteristics, over-oaking, or faults (e.g.,
reductive phenomena) becoming dominant.

Methods and Materials

Participants

Nineteen New Zealand (NZ) wine professionals participated in the study. All
participants were experienced with production and tasting of Pinot noir wines. They
were members of a panel of wine tasters who regularly participate in wine sensory
research tastings and are involved in various types of wine production methods
including conventional, organic and biodynamic. Mean age of the participants was
41.5 years (age range = 29–60 years), and there were 5 females and 14 males. The
majority of participants were oenologists, winemakers and wine producers (N = 16),
one reported her major occupation as viticulturist, and two participants were wine
science educators. Seven of the participants were also formally designated wine
judges. Two participants only reported that they were smokers. Mean number of
years of wine industry experience was 18.2 years (range = 8–32 years). The
experiment was performed in keeping with ethical requirements of the Lincoln
University Human Ethics Committee, NZ, with informed written consent obtained
prior to participation. Participants were ‘blind’ as to the purpose of the tasting, with
experimenter-provided instructions advising them only that the wine varietal under
evaluation was Pinot noir, and that they would need to attend two separate tasting
sessions.

Materials

Twelve Pinot noir wines from NZ’s major Pinot noir producing areas were selected
for the experiment (Table 1). Eight wines were from the 2012 vintage and 4 were
from the 2013 vintage. The wines were from conventional, organic, and biodynamic
producers. Wines were sourced directly from their producers, and all were sealed
with screw-cap closure to ensure consistency between bottles and between tasting
days. The wines, listed in Table 1, comprised four wines from Marlborough, three
from Central Otago, two from each of Martinborough and Nelson, and one wine from
Canterbury (Waipara). All wines were 100% Pinot noir and ranged in price between
NZ$30 - $50. The wines were selected by senior researchers and wine professionals
on the basis of three criteria. These criteria were that each wine was judged by its
producers as (i) exhibiting Pinot noir varietal fruity characters; (ii) comprising a
phenolic (tannin) composition that provided perceived substance in terms of wine in-
mouth structure, and (iii) being relatively youthful (younger than three years of age).
The wines were stored at 140 C until 24 hours before a session at which time they
were slowly brought up to ambient temperature (220 C, + or– 1).

Table 1. Pinot noir wines employed in the study.

Wine Year Alcohol v/v Region of NZ Production method

LDHR 2013 13.6 Marlborough Conventional

ASB 2012 13.0 Marlborough Conventional

CHPN 2013 13.7 Marlborough Organic

HPN 2012 13.3 Marlborough Organic

MDPN 2012 13.4 Central Otago Conventional

APN 2012 14.1 Central Otago Organic

Wine Year Alcohol v/v Region of NZ Production method

QRPN 2012 14.0 Central Otago Biodynamic

WWW 2013 14.1 Waipara Conventional

PPPN 2012 12.7 Martinborough Conventional

MVTT 2013 13.8 Martinborough Conventional

WNPN 2012 13.1 Nelson Organic

NMPN 2012 13.1 Nelson Conventional

Procedure

Sensory study

The study was conducted at the sensory facilities of the Marlborough Wine Research
Centre (MWRC) in Blenheim, NZ. The specialised sensory facilities at MWRC
permitted the important variables such as ambient temperature, sound, ambient
odours and between-participant communication to be controlled as advised for
sensory experimentation [18]. Each taster participated in two sessions separated by
approximately one week, one session on a Fruit day and a second session on a Root
day, these ‘days’ determined by the biodynamic calendar for wine tasting in 2014 [5].
It was not possible to control strictly the temporal gap between the two tastings per
person but the time between each person’s two sessions ranged between 7–9 days.
Four to nine people participated at any particular time, and all tastings were
conducted between the hours of 1 pm and 7 pm on a tasting day. Twelve participants
undertook their first session on a Fruit day and their second session on a Root day.
The remaining seven participants tasted the wines in the reverse order; that is, their
first session was conducted on a Root day and their second session on a Fruit day.
Each session lasted approximately two hours.

The wines for evaluation comprised 25-mL samples that were served at ambient
temperature in standardised tasting glasses [19]. A new bottle of each wine was
opened each day that the experiment was conducted and the wines were first
checked for faults by two experienced wine professionals. For the within-session,
replicate data collection, the wines were re-poured between flight 1 and flight 2 for
each participant. The glasses were coded with 3-digit numbers and were covered
with plastic Petri dishes. Each participant evaluated the 12 wines within the sample
set four times, twice in each session with the two within-session tastings separated
by a 20-minte break. Order of the 12 wines within a flight was varied between-subject
but remained constant within-subject. That is, the wine samples were presented in
a different order specific to each participant according to a Williams Latin square
arrangement generated by FIZZ software (Biosystemes, Courtenon, France). On the
other hand, the wine order specific to any particular participant remained the same
for each of the participant’s four tastings of the 12 wines to eliminate any possibility
of wine order as a confounding effect across Fruit and Root day tastings. Water was
available throughout each session.

Participants were seated in separate booths or at separate tables where their 12

wines were positioned. They were advised that they were to undertake two tasting
tasks within the session, and that they could proceed with the tasks at their own
pace. Participants were also advised that they were welcome to take a break at any
time should they choose such, but that they must take a 20-minute break between
the two descriptive rating tasks within the session. Specific instructions to
participants prior to each task included that they were to evaluate each wine, in the
order presented, via 20 experimenter-provided descriptors. They were also informed
that the wine evaluations were to be undertaken by global perception, that is, by full
tasting involving olfaction, taste, and trigeminal stimulation. They were further
advised that all wine was to be expectorated (i.e., not swallowed). The 20 descriptors
(see Table 2) were presented in the same order for every participant and for all four
descriptive rating tasks performed by each participant. The descriptors spanned a
range of varietal-relevant aroma, taste, and mouthfeel characteristics, and were
selected with the aim of elucidating both qualitative and quantitative aspects of each
wine’s aromatic (e.g., fruity) and more structural aspects (e.g., balance; length in
mouth; harshness of tannins). There were eight Intensity Descriptors (aromatic
intensity; fruit notes; green notes; reductive notes; concentration in mouth;
bitterness; astringency; sweetness), six Quality Evaluation descriptors (overall
quality; oak integration; acid/flavour balance; harmony of components; overall
structure; length in mouth), four Qualitative descriptors (expressiveness; fruit
ripeness; tannins; colour) and finally two Overall Appreciation descriptors (Pinot noir
typicality; liking). Each descriptor was rated via a 100 mm, horizontal visual analogue
scale with the scale anchors as in Table 2.

Table 2. The 20 descriptors employed in the experiment in the order presented.

Descriptors Scale anchors

Intensity descriptors

Aromatic intensity Low—intense

Fruit notes Low—intense

Green notes Low—intense

Reductive notes Low—Intense

Concentration in mouth Low—intense

Bitterness Low - Intense

Descriptors Scale anchors

Astringency Low - Intense

Sweetness Low - Intense

Quality evaluation

Overall quality Poor—Outstanding

Oak integration Poor—Outstanding

Acid/Flavour balance Poor—Outstanding

Harmony of components Poor—Outstanding

Overall structure Poor—Outstanding

Length in mouth Poor–Outstanding

Qualitative descriptors

Expressiveness Closed—Expressive

Fruit ripeness Unripe–Raisined

Tannins Harsh–Soft

Colour Light—Dark

Overall appreciation

Pinot noir typicality Atypical—Typical

Liking Dislike—Like

Post-experiment questionnaire

After all 19 participants had completed both sessions of the experiment, each
participant was sent a Questionnaire [S1 Questionnaire] to determine their degree
of knowledge about the purpose of the study, and their awareness of the existence
of the wine drinkers’ calendar [5].

Wine basic parameters

At the time of the sensory sessions on both a Fruit day and a Root day, wine samples
were taken for physico-chemical analysis of standard wine parameters. The wine
parameters were determined by InfraRed spectrometry using Fourier
Transformation (IRFT) with a WinescanTM FT2 (FOSS) that was calibrated with wine
samples analysed in accordance with official OIV practices. Samples were analysed
in duplicate and parameters were quantified using a high-input calibration file.
Relative standard deviations were exclusively lower than 10%.

Collection of meteorological information

Hourly measures of air pressure (hPa) for Blenheim were recorded from the NZ
Meteorological Service website (www.metservice.com) during each tasting session,
along with basic meteorological data including rainfall, sunshine minutes, relative
humidity, and wine speed.

Data Analysis

Due to the controversial nature of the topic under investigation, the data analysis
was performed ‘blind’. That is, the data were analysed by a research colleague who
was unaware of the purpose of the study, not involved in any aspects of planning or
implementation of the experiment, and received the dated coded so that the study’s
variables were not apparent (e.g., Fruit day data were coded as Variable X, and Root
day data were coded as Variable Y).

Sensory data

To consider the effect of Fruit versus Root days, mixed effects ANOVAs were
computed for each descriptor with condition (Fruit day; Root day) and replicate as
fixed factors, and with wine and subject as random factors, along with an interaction
term between condition and replicate. A significance level of 0.1 was adopted for all
analyses to ensure that we did not miss any potential effects, even though this is a
more liberal criterion than the usual 0.05. To evaluate the strength of each effect
partial eta squared values were computed for all ANOVA factors. This index reflects
the proportion of variance attributable to each factor.

The formulation of the biodynamic calendar for wine drinkers’ hypothesis [5]
proposes that the changes in sensory experience between Fruit and Root days affect
the sensory experience of wines in general. To evaluate whether there was any
interaction between specific wines and the biodynamic tasting days (Fruit vs Root
day) we carried out a second series of mixed effects with wine as a fixed factor.

Finally, to specifically explore whether there was a difference for the one wine in the
sample set that was produced by biodynamic methods, we also calculated ANOVA
for only that wine, with condition and replication as fixed factors, subject as random
factor, and the condition by replicate interaction.
Link between sensory data and wine basic physico-chemical parameters

To obtain a more synthetic picture of the sensory data and evaluate their link with
basic physico-chemical parameters, we carried out a principal component analysis
(PCA). The sensory descriptors were entered into the analysis as active variables.
Each wine was represented by four rows corresponding to the four tastings of each
wine (two tastings on each tasting day). The physico-chemical parameters were
entered as supplementary, continuous variables. The type of tasting day (Fruit vs
Root) and wine origin were entered as supplementary nominal variables to evaluate
whether these two factors influenced the global description of the wines.

Results

Post-experiment questionnaire data

Eighteen of the 19 study participants returned completed questionnaires. Ten of the

18 tasters reported knowledge of the existence of the biodynamic wine drinkers’
calendar, while eight had no knowledge of it. Of the 10 tasters aware of the
calendar’s existence, three reported ever having tasted wine according to the
instructions of the calendar, and one person only reported previously tasting wines
according to the lunar cycle. Interestingly, two of the three who reported previously
having tasted wines according to the calendar also reported in a prior question that
they had never tasted wines “according to the lunar cycle (the stars and the moon)”.
Hence, an inconsistency occurs, suggesting some ignorance or ambiguity for these
individuals regarding the basis for the instructions in the calendar. Finally, when
asked in the final question about the current tasting of Pinot noir wines, not one
person commented that they had tasted the wines according to either the lunar cycle
or the calendar.

Sensory data

Mean ratings to the 12 wines for each descriptor on Fruit days and on Root days are
presented in Fig 1. The ANOVA results, with wine as random factor, are summarised
in Table 3. There was a significant effect (p < 0.10) for three descriptors, bitterness,
oak integration, and concentration. However, one only of these was consistent with
the biodynamic wine tasting calendar prediction, with concentration in mouth being
higher on a Fruit than a Root day. For the other two significant effects, oak integration
was higher on a Root day and bitterness was higher on a Fruit day, these two effects
being counter to the biodynamic calendar prediction and anecdotal evidence that the
wines would taste “better” on a Fruit than Root day. For many of the other wine
characteristics that could be expected to be rated higher on a Fruit day (left of Fig
1), responses were in fact slightly higher on Root days. Similarly, descriptors that
could be expected to be rated lower on a Fruit day were rated higher on a Fruit day
(right of Fig 1). Further supporting the lack of evidence for superior perceived wine
quality on a Fruit day in comparison with a Root day, partial eta squared values did
not exceed 0.005, meaning the differences in ratings between a Fruit and a Root day
tasting accounted for less than half a percent of variation.

Mean ratings (+/- SE) to the twelve wines for each descriptor on Fruit days and on
Root days.

Fig 1 Mean ratings by Fruit/Root day.

Table 3. Summary of ANOVA analysis for each descriptor.

Descriptor Effect Variance explained

Conditio Conditio
Conditio Replicat n Conditi Replica n Win
Wine
na eb *Replica on te *Replica e
te te

F(11,87
F(1,877) F(1,877) F(1,877) (%) (%) (%) (%)
7)

Intensity descriptors

Aromatic
1.4 1.7 0.1 13.4*** 0 0 0 14
intensity

Fruit notes 2.6 0 0.2 18.2*** 0 0 0 19

Green notes 2.2 5.3* 1.5 4.0*** 0 1 0 5

Reductive
0.3 0.6 0.1 10.5*** 0 0 0 12
notes

Concentratio
3.6† 0.3 5.1* 7.4*** 0 0 1 9
n in mouth

Bitterness 3.1† 5.6* 0.2 4.0*** 0 1 0 5

Astringency 0.3 5.7* 0.4 20.3*** 0 1 0 20

Sweetness 0.1 18.9*** 1.6 6.9*** 0 2 0 8

Quality evaluation

Overall
0 0.4 1 12.5*** 0 0 0 14
quality

Oak
2.9† 3.7† 0.5 13.2*** 0 0 0 7
integration

Acid/flavour
0 2.2 0.1 6.3*** 0 0 0 7
balance

Harmony of
0.3 3.1† 0 11.4*** 0 0 0 13
components

Overall
0.6 3.1† 0.6 9.2*** 0 0 0 8
structure
Descriptor Effect Variance explained

Conditio Conditio
Conditio Replicat n Conditi Replica n Win
Wine
na eb *Replica on te *Replica e
te te

F(11,87
F(1,877) F(1,877) F(1,877) (%) (%) (%) (%)
7)

Length in
1.4 1.2 3.1† 19.8*** 0 0 0 4
mouth

Qualitative descriptors

Expressiven
1.7 2.9† 0.1 8.6*** 0 0 0 10
ess

Fruit
1.7 0 2.4 6.0*** 0 0 0 7
ripeness

Tannins 1.4 10.7*** 2.2 18.2*** 0 1 0 19

Colour 1.3 0.2 1.6 58.0*** 0 0 0 40

Overall appreciation

Pinot noir
0 0 6.9** 5.3*** 0 0 1 6
typicality

Liking 0 1.3 0.4 10.2*** 0 0 0 11

Mean
(F/Effect 1.245 3.345 1.405 13.17 0 0.3 0.1 11.9
Size)

Note

† p < 0.10.

* p < 0.05.

** p < 0.01.

*** p < 0.001.

a Condition = Fruit vs Root Day.
b Replicate = within-session tastings.

In contrast with the lack of descriptor-rating differences between Fruit and Root days,
there were several significant differences between replicate ratings within a session
(Table 3; Fig 2). Differences were observed for green notes, bitterness, astringency,
sweetness, tannins, oak integration, harmony of components, overall structure and
expressiveness. The characteristics that could be considered not desirable in Pinot
noir wines [16], namely green notes, bitterness and astringency, were rated higher
in the second tasting within a session. Conversely, several qualities desirable in
Pinot noir wines including tannin softness, harmony of components, and wine overall
expressiveness were judged less positively in the replicate tasting within a session.
However, despite these differences being statistically significant, they typically
accounted for a small 1% of total variance observed. There were also two statistically
significant interactions between replicate and condition, one for Pinot noir typicality
and the other for concentration in mouth. However these interactions are not in
keeping with the biodynamic wine tasting hypothesis.
 Mean ratings (+/- SE) to the twenty wine descriptors as a function of within-session
replicate: Replicate 1 = first assessment of the wines; Replicate 2 = repeat
assessment of the wines.

Fig 2. Mean ratings for each sensory descriptor by Replicate 1/Replicate 2.

In contrast to the small differences attributable to condition of Fruit day versus Root
day and to within-session replicate tasting, a significant effect of Wine, accounting
on average for 12% of the variance, was observed for all descriptors [see S1–S3
Figs]. That is, the Pinot noir wines were judged as differing in the varietal and quality
wine characteristics measured, including in perceived overall quality and in liking.
These results are in agreement with previously published work involving NZ Pinot
noir wines [16].

The ANOVA carried out with wine as fixed factor showed no significant
wine*condition effects, p > .26. [See S1 Fig for detailed results and ratings of each
wine for each descriptor by Fruit and Root days].

Analyses for the sole biodynamic wine, wine QRPN, also showed no evidence of
differences between Fruit and Root days (p > .20). Finally, the data presented in S1
Fig show that type of wine production management (see Table 1), namely
conventional, organic or biodynamic, was not a factor in quality ratings of the Pinot
noir wines. This last comment should be qualified in that the small sample set of
wines in this study included one biodynamic wine only.

Association between sensory data and wine basic physico-chemical

parameters

Mean results of the WinescanTM FT2 (FOSS) analysis on each wine on a Fruit day
and a Root day are presented in Table 4. The variability across Fruit and Root day
measures, and the negative values for malic acid and glucose, can be interpreted in
terms of calibration and measurement error as reported in Foss Application Notes
7a, 9, 11, 13, & 14a [20]. Of importance, there is no evidence that wine composition
factors that have potential to influence perception of wine quality (e.g., pH; volatile
acidity) changed systematically in the direction predicted by the biodynamic wine
drinkers’ calendar.

Table 4. Mean data for each wine from Winescan analysis of basic physico-chemical
parameters on each of a Fruit Day and a Root Day.

Reducing Total Tartaric Malic Lactic Ethanol Volatile

Wine Day pH CO2 Glucose Fructose
sugar Acidity Acid Acid Acid V/V Acidity

LDHR Fruit 1.0 3.60 3.7 2.2 -0.5 1.4 13.58 366.79 0.7 -0.9 -0.5
 Reducing Total Tartaric Malic Lactic Ethanol Volatile
Wine Day pH CO2 Glucose Fructose
sugar Acidity Acid Acid Acid V/V Acidity

Root 1.1 3.59 3.7 2.2 -0.5 1.5 13.73 386.93 0.7 -0.8 -0.4

ASB Fruit 0.4 3.53 3.6 2.7 -0.3 1.6 12.99 363.61 0.6 -0.6 -1.0

Root 0.4 3.53 3.7 2.7 -0.2 1.7 13.15 381.96 0.6 -0.5 -1.0

CHPN Fruit 0.1 3.53 3.4 2.4 -0.3 1.4 13.68 421.64 0.5 -0.9 -0.6

Root 0.1 3.52 3.4 2.4 -0.2 1.4 13.89 438.34 0.5 -0.8 -0.6

HPN Fruit 0.2 3.74 3.5 1.6 -0.4 2.6 13.30 362.05 0.6 -1.8 -0.7

Root 0.2 3.74 3.5 1.6 -0.3 2.6 13.49 376.88 0.6 -1.7 -0.7

MDPN Fruit -0.2 3.66 3.5 1.9 -0.3 1.9 13.44 339.47 0.6 -1.7 -1.0

Root -0.1 3.67 3.5 1.9 -0.3 1.9 13.61 359.89 0.7 -1.5 -1.0

APN Fruit 0.1 3.56 3.9 2.2 -0.3 2.0 14.05 416.58 0.7 -0.8 -0.8

Root 0.1 3.57 3.9 2.2 -0.2 2.0 14.26 429.71 0.7 -0.7 -0.8

QRPN Fruit -0.1 3.61 3.4 1.8 -0.2 1.7 13.99 466.95 0.5 -1.1 -0.8

Root -0.1 3.61 3.3 1.8 -0.1 1.7 14.21 502.30 0.5 -1.0 -0.7

WWW Fruit 0.9 3.69 3.7 2.3 -0.2 1.7 14.07 389.12 0.7 -0.4 -0.4

Root 0.8 3.69 3.7 2.3 -0.2 1.6 14.25 410.90 0.7 -0.4 -0.4

PPPN Fruit 1.2 3.55 3.6 2.4 -0.5 1.7 12.72 387.64 0.5 -0.1 -0.5

Root 1.1 3.54 3.6 2.3 -0.4 1.7 12.99 419.57 0.5 -0.2 -0.5

MVTT Fruit 0.5 3.61 3.4 2.4 -0.2 1.3 13.82 447.40 0.6 -0.6 -0.6

Root 0.4 3.61 3.4 2.4 -0.1 1.4 14.03 464.70 0.6 -0.5 -0.7

WNPN Fruit 0.6 3.62 3.3 1.6 -0.3 1.9 13.07 436.04 0.5 -0.7 -0.7

Root 0.4 3.63 3.4 1.6 -0.2 1.9 13.28 460.76 0.5 -0.7 -0.7

NMPN Fruit 0.5 3.64 3.6 1.9 -0.2 1.7 13.12 364.19 0.6 -0.7 -0.7
 Reducing Total Tartaric Malic Lactic Ethanol Volatile
Wine Day pH CO2 Glucose Fructose
sugar Acidity Acid Acid Acid V/V Acidity

Root 0.4 3.65 3.5 1.8 -0.1 1.7 13.37 386.93 0.6 -0.7 -0.6

PCA results are reported in Fig 3. The first principal component (51% of variance)
opposes the wines that were the most liked, these exhibiting ripe fruits, intense
aroma, harmony, and expressiveness, with wines more green, astringent, and bitter
(Fig 3A). These latter two sensory characteristics were associated with several wine
composition parameters, namely Folin C Index (a measure of wine total phenolics),
malic acid, ethanol, and CO2. On the other hand, more positive wine attributes such
as fruit ripeness, concentration, expressiveness, good structure, acid-flavour
balance, harmony, sweetness, and soft tannins were associated with wine
composition factors related to sugars and acids, including volatile acidity.
PCA of sensory data: (A) Loadings of sensory variables (black) and basic physico-
chemical variables (green) plotted in PCA space; (B) Projections of each wine by
Fruit/Root day and replicate 1 and 2 in PCA space, along with centroids for region
(Central Otago, Marlborough, Martinborough, Nelson, Wairarapa) and centroids for
Fruit/Root days.

Fig 3. PCA of sensory data (active variable), wine basic parameters

(supplementary continuous variables), tasting days and wine origin
(supplementary nominal variable).

According to our experimental hypothesis based on the lunar calendar [5], wines
tasted on Fruit days should have high loading on the positive side of this component
and wine tasted on Root days should load on the negative side. Fig 3B shows that
there is no evidence for this as there are as many wines tasted on the Root as on
the Fruit days projecting onto both sides of the component. The second principal
component (17% of variance) opposes one wine to all other wines independently of
the type of day on which it was tasted. Globally, the position of the barycentres of
the nominal variables on Fig 3B shows no effect of tasting day (the Fruit and Root
day barycentres project very close to one another in the centre of the map) but shows
an effect of wine origin (the barycentres of origins are spread out along the first
component). This last result indicates that the absence of an effect of tasting day
cannot be attributed to a lack of sensitivity of the participants given that their
descriptions were sensitive enough to detect differences in terms of wine origin. This
is in agreement with the ANOVA results that show that the differences between
repeat tastings within a tasting day were of the same magnitude as the differences
between tasting days.

Meteorological information

Table 5 shows meteorological data and Table 6 reports measures of air pressure
(hPa) for Blenheim, the location of the study, during each tasting session. Although
there were differences in some measures, there is no systematic difference across
Fruit and Root days that could conceivably be responsible for the minor differences
in either sensory or physico-chemical data reported on the 12 wines.

Table 5. Meteorological data collected at Blenheim Weather Station, Grovetown

Park campus of the Marlborough Research Centre, Blenheim, New Zealand,
between the hours of 12 md and 7 pm during which time the experimental wine
tastings took place.
 Mean
Mean dry bulb Mean Total Mean wind
relative
Date Day temperature sunshine Rainfall Speed
humidity
°C (minutes) (mm) (Km/hour)
(%)

25/11/14 FRUIT 21.54 31 0 16.55 64.47

26/11/14 FRUIT 24.95 480 0 28.39 29.08

27/11/14 ROOT 21.95 411 0 26.87 44.3

4/12/14 FRUIT 22.39 81 0 18.69 50.01

5/12/14 ROOT 14.71 0 0 8.93 78.7

Table 6. Blenheim hourly air pressure readings taken from the NZ Meteorological
web site during conduction of the tastings.

Date Time Air pressure hPa Date Time Air pressure hPa

25/11/14 1pm 1009 4/12/14 1pm 1014

FRUIT DAY 2pm 1010 FRUIT DAY 2pm 1014

3pm 1008 3pm 1014

4pm 1007 4pm 1013

5pm 1006 5pm 1014

6pm 1005 6pm 1014

7pm 1005 7pm 1014

26/11/14 1pm 1003 5/12/14 1pm 1022

FRUIT DAY 2pm 1003 ROOT DAY 2pm 1022

3pm 1003 3pm 1021

4pm 1003 4pm 1022

5pm 1003 5pm 1022

6pm 1003 6pm 1022

7pm 1003 7pm 1022

Date Time Air pressure hPa Date Time Air pressure hPa

27/11/14 1pm 1003

ROOT DAY 2pm 1002

3pm 1001

4pm 1001

5pm 1001

6pm 1000

7pm 1001

Discussion

The aim of the present study was to test an extension of biodynamic agriculture
which argues that wines will taste different in systematic ways on days designated
by the biodynamic calendar for wine drinkers [5] as favourable (Fruit days) or
unfavourable (Root days) for wine tasting. The outcome of the study is clear; the
results demonstrate that judgments to the twelve Pinot noir wines were little
influenced by tasting day. That is, there was little variability in descriptor ratings by
experienced wine professionals between tastings of the wines on Fruit days versus
Root days, despite our hypothesis testing adopting a lenient criterion for rejection of
the null hypothesis. In fact, there was more variability within session, where wines
were tasted twice on either a Fruit day or a Root day, than between sessions (Fruit
vs Root day). As well, there were significant differences amongst the Pinot noir wines
reported for every wine characteristic on which the wines were evaluated. These
latter effects argue that failure to support our experimental hypothesis was not due
to lack of discrimination by our experienced, wine-professional tasters who clearly
found and reported significant differences amongst the twelve wines including in
overall quality.

To address the topic under investigation, we implemented various essential

methodological requirements, the majority of which presumably are lacking when
wine industry tastings are conducted according to the biodynamic wine drinkers’
calendar [12] and when positive results are reported in wine industry media. The
most important of these was to minimise confounding variables (e.g., by retaining
the same order for the 12 wines for each taster across their four evaluations of the
wines) and to ensure that the study’s tasters were blind as to the purpose of the
study. That is, we needed to know that the tasters did not apply the biodynamic wine
drinkers’ calendar [5] guidelines to their tasting experience, this likely influencing
their expectations and tasting behaviour during their evaluation of the Pinot noir
wines in the current study. That the tasters were blind in terms of not considering the
lunar cycle predictions while undertaking the experimental tastings was validated by
the post-experiment, questionnaire data. The questionnaire data further informed us
as to the validity of our assumption that conducting the study in the Southern
Hemisphere would minimise the likelihood that our tasters were current advocates
of biodynamic wine tasting practice. Three tasters only reported ever having tasted
wine according to the instructions of the calendar, and almost half of the study’s
participants were blind to the existence of the wine drinkers’ calendar.

We considered the topic of lunar influence on wine tasting important to investigate

for several reasons. First, it is now widely established that contextual factors, both
intrinsic and extrinsic, may influence sensorial assessment of wine quality [21],
presumably as a result of cognitive influence based on expectations. Suggestions of
a lunar effect, widely dismissed by many as pseudo-science or “absolute rubbish”
[14], deserved to be tested empirically. Second, there appears increasing reference
to wine tasting driven by the lunar calendar in wine industry media [3] [10] [14] and
more recently the development of an iPhone App produced by Floris Books. Again,
testing the underlying basis for a lunar-effect notion appeared a responsible
undertaking given apparent interest in the phenomenon. Third, and related to the
second point, a scientific basis for this empirically testable aspect of biodynamic
practice could aid scientific respectability for advocates and practitioners of
biodynamic philosophy of which there are many in the international wine industry.

Interestingly, our data show that the type of wine production management, namely
conventional, organic or biodynamic, was not a factor in determining the influence of
Fruit and Root days on how a wine was evaluated. This is in keeping with information
provided in the biodynamic wine drinkers’ calendar [5]. Further, and somewhat as a
side issue, type of wine production was not a factor in overall quality ratings of the
Pinot noir wines. In fact, several of the wines judged highest in overall quality were
wines produced by conventional production practices. These data however must be
treated with caution due to the low and unequal numbers of wines from each wine-
production category, notably from biodynamic production. A future study, aimed
specifically at testing the interaction between type of wine production and tasting
day, is required before firm conclusions can be drawn on this point.

While failing to support the major tenet of the biodynamic wine drinkers’ calendar
that wines are perceived as tasting better or worse according to the lunar cycle, the
question remains as to why some wines can appear to taste better on some days in
comparison with others [3]. In the present study we were not able to measure all
possible factors pertaining to the taster (e.g., mood or stress level of the taster;
influence of ovarian hormones). We did however record data regarding objective
measures pertaining to the tasting location that have been put forward by some
authors [13] [17] as possibilities for influencing how a wine tastes, namely
meteorological and air pressure data. These data provide no evidence of conditions
that could lead to systematic influence on how the wines were perceived.

Although the biodynamic wine drinkers’ calendar does not comment on whether the
proposed source of change in a wine across Fruit and Root days involves perceived
differences or differences in wine composition, we conducted basic physico-
chemical analysis on the wines in the sample set on each type of tasting day to
consider both possibilities. As well, we sourced all the wines employed in the study
from their producers rather than retailers to minimise any bottle differences, and
stored the wines in the same location prior to the experiment being conducted.
Information from the developers of Foss Winescan methodology demonstrates that
any differences in our data demonstrate variability within a range expected due to
calibration and measurement error [20].

Conclusion

In conclusion, the findings reported in the present study provide no evidence in

support of the notion that how a wine tastes is associated with the lunar cycle. The
Pinot noir wines in the sample set were judged by experienced wine professionals
as varying significantly in a range of characteristics. However, the day on which there
were tasted did not influence these judgments. It is conceivable that the anecdotal
reports of sensory effects that have been described in wine-industry media could be
due to expectation effects rather than actual differences in the wines. Consumers
expecting a wine to be more expressive and aromatic on Fruit days might actually
perceive them as such through top down cognitive effects [22]. Such top down
effects involving a range of factors have been reported previously. For example,
Rose Pangborn and colleagues found that a white wine colored pink to give it the
appearance of Rosé wine was perceived by wine professionals as sweeter than a
non-coloured wine sample [23]. Likewise, researchers in Bordeaux reported that
colouring a white wine with odourless anthocyanin to make it red led wine experts to
describe the wine’s flavour as that of a red wine [24]. These results highlight the
importance of testing, where possible, anecdotally-based notions and practices in
the food and beverage industries. Further work, replicating this study and
manipulating the lunar calendar information provided to the tasters, may help in
validating the hypothesis pertaining to expectation-driven effects.

Supporting Information

S1 Fig
Mean descriptor rating for each descriptor and each wine by condition (fruit
day, root day).

(DOCX)

Click here for additional data file.(63K, docx)

S2 Fig
Mean descriptor rating for each descriptor and each wine by replicate within
fruit day session.

(DOCX)
Click here for additional data file.(64K, docx)
S3 Fig
Mean descriptor rating for each descriptor and each wine by replicate within
root day session.

(DOCX)

Click here for additional data file.(64K, docx)

S1 Questionnaire
Questionnaire completed by each participant at the end of the experiment.

(DOCX)

Click here for additional data file.(14K, docx)

Acknowledgments

We thank Lily Stuart for assistance with conducting the experiment, Rob Agnew for
providing meteorological data, and Abby Albright for Winescan data analysis. Finally,
we thank the Marlborough wine professionals who served as study participants, and
the New Zealand wine producers who generously sponsored the study by providing
Pinot noir wines.

Funding Statement

This paper was supported by the following grant(s):

the Cresswell Jackson Wine Competition Trust of New Zealand to Wendy Parr.

We thank the Cresswell Jackson Wine Competition Trust of New Zealand for a
funding award to Wendy Parr and Phil Reedman. This was the first award made by
the Trust; http://www.spiegelauiwc.co.nz/assets/cjnzwt-review-february-2016.pdf.
The funders had no role in study design, data collection and analysis, decision to
publish, or preparation of the manuscript.

Data Availability

All data files are available on the Open Source Framework (https://osf.io/bcjez/).

References

1. Spence C, Piqueras-Fiszman B. The perfect meal: The multi-sensory science of

food and dining. Oxford: Wiley-Blackwell; 2014.
2. Steiner R, Gardner M. Spiritual Foundations for the Renewal of Agriculture: a
Course of Lectures Held at Koberwitz, Silesia, June 7 to June 16, 1924. Available:
http://www.rudolfsteineraudio.com/agriculture/agriculture.html. Accessed 22 July
2016.
3. Kennedy B. Tasting biodynamically. Db Reader, March 10, 2014. Available:
https://www.thedrinksbusiness.com/2014/03/tasting-biodynamically/
4. Gallagher P. Wine growers are toasting biodynamic methods. The Independent.
14 May 2015. Available: www.independent.co.uk/life-style/food-and-
drink/features/wine-growers-are-toasting-biodynamic-methods-10248274.html
Accessed 22 July 2016.
5. Thun M, Thun M. When wine tastes best 2014: A biodynamic calendar for wine
drinkers. Floris Books: UK; 2013.
6. Mäder P, Fliessbach A, Dubois D, Gunst L, Fried P, Niggli U. Soil fertility and
biodiversity in organic farming. Science. 2002. May 31;296 (5573):1694–7. doi:
10.1126/science.1071148 [PubMed]
7. Döring J, Frisch M, Tittmann S, Stoll M, Kauer R. Growth, Yield and Fruit Quality
of Grapevines under Organic and Biodynamic Management. PLoS ONE. 2105;
10(10): e0138445 doi: 10.1371/journal.pone.0138445 [PMC free article] [PubMed]
8. Reeve J, Carpenter-Boggs L, Reganold J, York A, McGourty G, McCloskey L. Soil
and wine grape quality in biodynamically and organically managed vineyards. Am.
J. Enol. Vitic. 2005; 56: 367–376.
9. Delmas MA, Gergaud O, Lim J. Does organic wine taste better? An analysis of
experts’ ratings? American Association of Wine Economists AAWE Paper No. 190,
Economics. 2016. Available: http://www.wine-economics.org/aawe/wp-
content/uploads/2016/01/AAWE_WP190.pdf Accessed 22 July 2016
10. Booth R. Tesco and supermarket rivals go for wine tasting by moonlight. The
Guardian, 18 April, 2009. Available:
https://www.theguardian.com/business/2009/apr/18/wine-lunar-calendar-tesco-
supermarkets Accessed: 22 July 2016
11. BBC. Can lunar cycles affect the taste of wine? BBC News. 20 April, 2009.
Available: http://news.bbc.co.uk/2/hi/uk_news/magazine/8008167.stm Accessed:
22 July 2016.
12. Blackadder JH. Enjoying wine with the Biodynamic calendar. April 9, 2014.
Available: http://www.wine.co.za/news/news.aspx?NEWSID=24680 Accessed: 22
July 2016
13. Blasberg M. Werewolf in Napa: How the moon affects wine. Menuism Wine Blog,
April 19, 2014. Available: https://www.menuism.com/blog/biodynamic-wine/
Accessed: 22 July 2016
14. Robinson J. Moody burgundy and other (fairy?) stories. 4 March, 2006. Available:
http://www.jancisrobinson.com/articles/moody-burgundy-and-other-fairy-stories.
Accessed: 22 July 2016.
15. Keats B. 2014 Antipodean AstroCalendar. Brian Keats Publishing: Tasmania,
Australia; 2014.
16. Valentin D, Parr WV, Peyron D, Grose C, Ballester J. Colour as a driver of Pinot
noir wine quality judgments: An investigation involving French and New Zealand
wine professionals. Food Qual Prefer. 2016; 48: 251–261.
17. Goode J. Jamie Goode’s wine anorak. Part 10: an interview with Monty Waldin.
2009. Available: www.wineanorak.com/biodynamic10.htm Accessed: 22 July 2016.
18. ASTM; Physical Requirement Guidelines for Sensory Evaluation Laboratories,
ASTM STP 913. Philadelphia: ASTM publications; 1986.
19. ISO. Sensory analysis–apparatus–wine-tasting glass. Geneva, Switzerland:
ISO; 1977.
20. Foss. Application Notes 7a (Ethanol), 9 (CO2), 11 (Folin C Index), 13 (Glucose),
& 14a (Malic acid) 2010: www.foss.dk
21. Piqueras-Fiszman B, Spence C. The weight of the bottle as a possible extrinsic
cue with which to estimate the price (and quality) of the wine? Observed correlations.
Food Qual Prefer. 2012; 25: 41–45.
22. Parr WV. Making sense of wine: Cognitive psychology’s contribution to
understanding wine tasting and wine tasters. Proceedings of Oeno2015: 10th
International Wine Symposium, Bordeaux, France, June 29 –July 1, 2015.
23. Pangborn RM, Berg HW, Hansen B. The influence of color on discrimination of
sweetness in dry table wine. Am J Psychol. 1963; 76: 492–495.
24. Morrot G, Brochet F, Dubourdieu D. The color of odors. Brain Lang. 2001; 79:
309–320. doi: 10.1006/brln.2001.2493 [PubMed]
XIV. Physicochemical and Sensory Evaluation of Whole Soybean Curd
Supplemented with Pine Needle Powder

Prev Nutr Food Sci. 2015 Jun; 20(2): 148–152.

Jun Ho Lee

Abstract

To develop functionally and nutritionally improved whole soybean curd (WSC), the
effects of partial (0 ~ 4%) replacement with pine needle powder (PNP) on the quality
characteristics of WSC were investigated. The moisture content and pH ranged from
76.96~77.80% (wet basis) and 6.69~6.74, respectively, with no considerable
differences. Lightness significantly decreased with higher PNP content in the
formulation (P<0.05), as indicated by visual observation that the color of WSC
became darker. Redness and yellowness also significantly decreased (P<0.05). The
texture profile analysis indicated that WSC containing PNP was softer and less
cohesive than control WSC. 2,2-Diphenyl-1-picrylhydrazyl and 2,2′-azino-bis(3-
ethylbenzthiazoline-6-sulphonic acid) radical scavenging activities significantly
increased (P<0.05) with higher substitution of PNP, and they were well correlated.
Results from the consumer test revealed that WSC with 1% PLP received the most
favorable acceptance scores for sensory attributes, including overall acceptability.
The present study indicated that the characteristics of PNP may play a role in
improving WSC quality in terms of physicochemical, sensorial, and antioxidant
characteristics.

Keywords: whole soybean curd, pine needle powder, physicochemical, sensory,

antioxidant

INTRODUCTION
Soybean curd (tofu), a traditional soybean product composed mainly of protein and
fat, is a prime component of Asian food culture (1). The consumption and popularity
of tofu continues to grow due to its status as a healthy vegetable protein. On the
other hand, soybean curd residue (SCR), a byproduct of soybean food processing,
is discarded as agro-industrial waste and has raised concerns over severe
environmental pollution (2). Nevertheless, SCR is rich in fat, starch, protein, and
sugar and has been recognized as a valuable ingredient due to its plentiful dietary
fiber content and various bioactive compounds (3). Whole soy bean curd (WSC) is
made from soymilk after removal of SCR, and WSC includes SCR along with a
curdling agent added to the coagulation. WSC can be made in a relatively short
period of time compared to soybean curd, and losses of soluble and insoluble
proteins, dietary fibers, and other nutritional components can be lessened (4).

Korean pine (Pinus koraiensis), a large conifer with important economic value, is
found across Korea, Japan, and the Northeastern part of China (5). P. koraiensis
has long been used in Oriental fork medicine due to its anti-fatigue, anti-aging, and
anti-inflammatory effects as well as a food supplement (6). Its leaves (called pine
needles) contain large amounts of protein, vitamins, and minerals (7) and are utilized
extensively as an important food ingredient (8). Therefore, the development of a
novel functional food containing pine needles would be beneficial owing to its
physiological activities and therapeutic effects. Pine needle powder (PNP) has been
successfully used in noodles (9), meat (10), rice cupcakes (11), rice madeleine (12),
and tofu (13).

Rapidly growing concerns about healthy diets and increased demand for the use of
novel food ingredients, especially soybean products, have led to investigations of
WSC with value-added ingredients, such as PNP. To the best of our knowledge, little
to no information is available on the effect of PNP on the physicochemical and
sensory properties of WSC. Pine needles contain antioxidants and other nutritional
components, and if added to foods as a supplement, it can provide beneficial effects
on nutrition and health. Therefore, it would be beneficial to develop a novel
formulation of WSC products with PNP.

In this study, PNP was supplemented to soybean milk in order to improve its
functional and nutritional qualities. The aim of this research was to determine the
effects of PNP addition (0~4%) on the physicochemical and sensory characteristics
of WSC. The antioxidant properties of WSC were also determined.

MATERIALS AND METHODS

Materials

PNP was procured from Garunara (Seoul, Korea), and soybeans were purchased
from a local market. Microbial transglutaminase (TGase) and silicon defoamer were
purchased from Ajinomoto Co. (Tokyo, Japan) and Dow Corning Korea Ltd. (LS-303,
Seoul, Korea), respectively.
WSC preparation

TGase-treated WSC was produced according to the method used by Jin and Lee
(14) with slight modifications. Briefly, an appropriate amount of yellow soybeans
(57.6~60.0 g), PNP (0.0~2.4 g), distilled water (240 mL), and defoamer (1.2 mL)
were mixed and homogenized (ULTRA-TURRAX, IKA, Staufen, Germany) at 10,000
rpm for 2 min. The mixture was partially heated to 80°C for 10 min to produce the
soybean milk, which was then cooled down to 50°C at room temperature and
homogenized again at 10,000 rpm for 5 min after adding TGase (3 g). Subsequently,
the soybean milk was incubated at 50°C for 1 h, followed by heat treatment at 80°C
for 20 min. WSC (0~4% depending on the amount of replacement) was then stored
in a 4°C refrigerator until use.

Physicochemical analyses

A sample (5 g) of WSC was mixed with 45 mL of distilled water and homogenized

for 10 min. The mixture was held at ambient temperature for 3 h to separate the solid
and liquid phases. The pH of the supernatant was measured using a pH meter
(pH/Ion 510, Oakton Instruments, Vernon Hills, IL, USA). Moisture content was
obtained by drying a specific amount (5 g) of WSC sample to a constant weight at
50°C in an oven (FOL-2, Jeio Tech Co., Daejeon, Korea), and the results were
reported on a wet basis (w.b.).

Texture profile analysis (TPA) of WSC samples (3×3×3 cm) was performed using a
computer-controlled Advanced Universal Testing System (model LRXPlus, Lloyd
Instrument Limited, Fareham, Hampshire, UK) at room temperature. Each sample
was compressed twice to 30% of its original height using a cylindrical-shaped probe
(12.45 mm in diameter) at 3 mm/s speed, and the trigger was set at 0.4 kgf. Hardness
is expressed as the peak height of the force on the first bite, whereas cohesiveness
quantifies the internal resistance of food structure (9). Color determination was
measured using a Chroma-meter (model CM-600d, Minolta Co., Osaka, Japan) set
for CIELAB color space, and reported as CIE L*-value (lightness), a*-value
(redness), and b*-value (yellowness).

Determination of free radical scavenging activity

2,2-Diphenyl-1-picrylhydrazyl (DPPH) radical scavenging activities of the samples

were measured based on their hydrogen-donating or radical scavenging activity
using a stable DPPH radical. The assay was performed as previously described by
Blois (10) with some modifications. Briefly, 0.2 mM solution of DPPH radical in
ethanol was prepared, after which 5 mL of this solution was added to 1 mL of sample
solution in ethanol at different concentrations and then shaken and left to stand for
10 min. Decolorization of DPPH-donated protons was determined by measuring the
absorbance at 517 nm using a spectrophotometer (Optizen 2020 UV Plus, Mecasys
Co., Ltd., Daejeon, Korea). The scavenging activity of DPPH radical was calculated
using the following equation:
Radical scavenging activity (%)=Absorbancecontrol−AbsorbancesampleAbsorbanc
econtrol×100

The spectrophotometric analysis of 2,2′-azino-bis (3-ethylbenzothiazoline-6-

sulphonic acid) (ABTS) radical scavenging activity was measured according to the
method used by Re et al. (11) with slight modifications. The ABTS+· was produced
by the reaction between 7.4 mM ABTS in H2O and 2.6 mM potassium persulfate
during storage in the dark at room temperature for 12~16 h. Before use, ABTS +·
solution was diluted with methanol to obtain an absorbance of 1.1±0.02 at 734 nm.
Subsequently, 3 mL of ABTS+· solution was added to 0.1 mL of sample. After 10
min, the percent inhibition at 734 nm was calculated for each concentration relative
to the blank absorbance.

Sensory evaluation

The Hedonic test was used to determine the degree of overall preference scores for
WSC. For this study, 30 untrained volunteer consumers were recruited from the
University and informed that they would be evaluating WSC. Five samples (3×3×3
cm) were presented in random order, and panelists were asked to evaluate the
acceptability of color, flavor, taste, chewiness, and overall acceptability. Consumer
participants were asked to evaluate preference levels using a seven-point hedonic
scale (7=like extremely, 6=like moderately, 5=like slightly, 4=neither like nor dislike,
3=dislike slightly, 2=dislike moderately, and 1=dislike extremely). Panelists received
a tray containing the samples at room temperature (randomly coded using a three-
digit number), a glass of water, and an evaluation sheet. Participants were instructed
on how to evaluate the samples and were not required to expectorate or consume
the entire volume served. Overall acceptance was evaluated first, and another
session was held to evaluate the rest of the attributes. There was an inter-stimulus
interval of 30 s imposed between samples, to allow time to recover from adaptation.
Participants were advised to rinse their palates between samples. Enough space
was given to handle the samples and questionnaire, and evaluation time was not
constrained. No specific compensation was given to the participants.

Statistical analysis

Each measurement was conducted five times, except for hardness (n=15),
antioxidant activities (n=3), and sensory evaluation (n=30). The experimental data
were subjected to analysis of variance (ANOVA) using the general linear models
(GLM) procedure. Mean values were compared using Duncan’s multiple range test.
The significance was defined at the 5% level.

RESULTS AND DISCUSSION

Physicochemical characteristics
Table 1 describes the physicochemical characterization of WSC supplemented with
different levels of PNP. Moisture of the samples ranged from 76.96 to 77.80% (w.b.),
and the control samples exhibited the highest moisture content. The pH values of
WSC ranged from 6.69 to 6.74, and decreased upon addition of increasing amounts
of PNP but the changes were little. Thus, it seems that PNP supplementation
resulted in WSC of slightly lower moisture content and pH.

Table 1. Physicochemical properties of WSC supplemented with different

levels of PNP

PNP level (%)

Properties

0 1 2 3 4

Moisture content 77.80±0.32a 77.40±0.14a 77.32±0.23a 77.64±0.26ab 76.96±0.62b

(%)

pH 6.74±0.01a 6.72±0.01b 6.72±0.00b 6.71±0.01b 6.69±0.01c

Color

L* 81.81±0.18a 78.60±0.12b 74.39±0.18c 73.70±0.15d 71.79±0.11e

a* 1.87±0.07a 1.19±0.09b 1.16±0.07b 1.16±0.08b 1.10±0.07c

b* 20.79±0.43a 18.29±0.29b 17.29±0.17c 17.00±0.32c 16.98±0.20c

Texture

Hardness (N) 2.42±0.08a 2.38±0.07a 2.22±0.11b 1.74±0.07c 1.22±0.06c

Cohesiveness 0.34±0.04a 0.30±0.03b 0.31±0.01b 0.29±0.02b 0.24±0.04c

(–)

Means with different letters (a–e) in the same row are significantly different according
to Duncan’s multiple range test (P<0.05).

Surface color, together with texture, is a very important element for consumers’ initial
acceptability (12). WSCs made with PNP were significantly different from the control
(P<0.05). The L*-value for the control (81.81±0.18) was higher than any of the values
seen for WSCs containing PNP. This was also true for a*- and b*-values. L*-values
decreased significantly upon addition of increased amounts of PNP (P<0.05). a*-
and b*-values showed similar trends but no significant differences were found among
the 1~3% samples in a*-values and the 2~4% samples in b*-values. Similar color
changes were reported for cookies (13) and tofu (14) supplemented with PNP.
The TPA results presented in Table 1 show that addition of an increasing amount of
PNP from 0 to 4% resulted in a reduction of WSC hardness from 2.42 to 1.22 N as
well as a reduction of cohesiveness values from 0.24 to 0.34. Cohesiveness is a
dimensionless value used to define the internal resistance of WSC structure, and it
an indicator of the ability of a material to stick together (9,15). These results show
that the WSC containing PNP is softer and less cohesive when compared to the
control WSC. Woo et al. (16) also reported a reduced hardness of tofu with added
roasted sorghum powder.

Free radical scavenging activities

Antioxidant activities were investigated based on DPPH radical-scavenging activity

and ABTS radical cation assay. The usage of PNP in the WSC formulation enhanced
the levels of antioxidant activity (Fig. 1). This can be attributed to the inherent rich
antioxidant capacities of PNP as compared to soybean powder. There were
significant increases in electron donating ability (EDA) values of bound phenolic
extracts in the samples with PNP as compared to those of the control sample (P<
0.05). EDA values increased significantly with increased levels of PNP (P<0.05).
ABTS values also significantly increased after addition of 3% PNP compared to the
control samples (P<0.05), whereas no significant differences were found in ABTS
values among the control, 1, and 2% samples (P>0.05). Samples enriched with 4%
PNP showed the highest antioxidant potential. Data also showed a positive
correlation between the antioxidant capacities of the various samples. EDA
increased as ABTS increased. A similar trend was reported by Kilci and Gocmen
(17) for tarhana (Turkish fermented cereal food) supplemented with oat flour. The
antioxidant capacities of various extracts are mainly due to the total phenolic content
and total phenolic acids (18).

Fig. 1. DPPH and ABTS radical scavenging activities of WSC supplemented with
different levels of PNP. Means within the same activity without a common letter (a–
e) are significantly different (P<0.05).
Sensory findings

Color, flavor, taste, chewiness, and overall acceptance of control WSC and PNP-
supplemented WSC were evaluated, and the results are presented in a radar plot
(Fig. 2). The highest flavor, taste, and overall acceptance scores were obtained for
samples supplemented with 1% PNP (5.19, 5.16, and 5.87, respectively). Color
values obtained by the consumer acceptability test of PNP-supplemented WSC
varied between 3.63 and 6.47. The control received the highest score, and the
values decreased with increased levels of PNP. Excessive addition of PNP seemed
to have a detrimental effect on this property. Similarly, the highest chewiness value
of 4.61 was obtained in the control samples, but it was not significantly different from
that of the 1% samples (P>0.05). Overall acceptability scores of WSC containing
PNP were the highest for the 1% samples as mentioned above. The results of the
sensory analysis showed that supplementation WSC with 1% PNP improved most
of the sensory attributes evaluated. However, further increasing the levels of PNP
seemed to have some undesirable sensorial effects.

Fig. 2. Radar plot of sensory results of WSC supplemented with different levels of
PNP. Means within the same attribute without a common letter (a–d) are significantly
different (P<0.05).

In conclusion, the present study established the feasibility of using PNP as a

supplement for enhancing the nutritional quality of WSC. The results of this study
showed that nutritious WSC can be prepared by supplementing 1% PNP, with
increased physicochemical, sensorial, and antioxidant characteristics.

Footnotes

AUTHOR DISCLOSURE STATEMENT

The authors declare no conflict of interest.

REFERENCES

1. Cui H, Zhao C, Lin L. The specific antibacterial activity of liposome-encapsulated

Clove oil and its application in tofu. Food Control. 2015;56:128–134. doi:
10.1016/j.foodcont.2015.03.026. [Cross Ref]
2. Shi M, Yang Y, Hu X, Zhang Z. Effect of ultrasonic extraction conditions on
antioxidative and immunomodulatory activities of a Ganoderma lucidum
polysaccharide originated from fermented soybean curd residue. Food Chem.
2014;155:50–56. doi: 10.1016/j.foodchem.2014.01.037. [PubMed] [Cross Ref]
3. Joo SI, Kim JE, Lee SP. Physicochemical properties of whole soybean curd
prepared by microbial transglutaminase. Food Sci Biotechnol. 2011;20:437–444.
doi: 10.1007/s10068-011-0061-2. [Cross Ref]
4. Lee HJ, Shin HC, Lee YS, Kim JY, Moon YH, Park KH, Moon JH. Comparison of
quality characteristics of soybean curd and whole soybean curd. Korean J Food Sci
Technol. 41:117–121.
5. Choi HD, Koh YJ, Choi IW, Kim YS, Park YK. Anticariogenic activity and
glucosyltransferase inhibitory effects of extracts from pine needle and twig. Korean
J Food Sci Technol. 2007;39:336–341.
6. Watanabe K, Momose F, Handa H, Nagata K. Interaction between influenza virus
proteins and pine cone anti-tumor substance that inhibits the virus multiplication.
Biochem Biophys Res Commun. 1995;214:318–323. doi: 10.1006/bbrc.1995.2290.
[PubMed] [Cross Ref]
7. Zhang JM, Shi XF, Fan B. Chemical composition and pharmacological activities
of Cedrus deodara. Chinese Trad Plant Med. 2009;31:928–933.
8. Chen YH, Hsieh PC, Mau JL, Sheu SC. Antioxidant properties and mutagenicity
of Pinus morrisonicola and its vinegar preparation. LWT-Food Sci Technol.
2011;44:1477–1481. doi: 10.1016/j.lwt.2011.01.016. [Cross Ref]
9. Lu TM, Lee CC, Mau JL, Lin SD. Quality and anti-oxidant property of green tea
sponge cake. Food Chem. 2010;119:1090–1095. doi:
10.1016/j.foodchem.2009.08.015. [Cross Ref]
10. Blois MS. Antioxidant determinations by the use of a stable free radical. Nature.
1958;181:1199–1200. doi: 10.1038/1811199a0. [Cross Ref]
11. Re R, Pellegrini N, Proteggente A, Pannala A, Yang M, Rice-Evans C.
Antioxidant activity applying an improved ABTS radical cation decolorization assay.
Free Radic Biol Med. 1999;26:1231–1237. doi: 10.1016/S0891-5849(98)00315-3.
[PubMed] [Cross Ref]
12. Zucco F, Borsuk Y, Arntfield SD. Physical and nutritional evaluation of wheat
cookies supplemented with pulse flours of different particle sizes. LWT–Food Sci
Technol. 2011;44:2070–2076. doi: 10.1016/j.lwt.2011.06.007. [Cross Ref]
13. Choi HY. Antioxidant activity and quality characteristics of pine needle cookies.
J Korean Soc Food Sci Nutr. 2009;38:1414–1421. doi:
10.3746/jkfn.2009.38.10.1414. [Cross Ref]
14. Son BG, Kim HE, Lee JH. Physicochemical and consumer preference
characteristics of tofu incorporated with pine needle powder. J Korean Soc Food Sci
Nutr. 2015;44:296–301. doi: 10.3746/jkfn.2015.44.2.296. [Cross Ref]
15. Ulziijargal E, Yang JH, Lin LY, Chen CP, Mau JL. Quality of bread supplemented
with mushroom mycelia. Food Chem. 2013;138:70–76. doi:
10.1016/j.foodchem.2012.10.051. [PubMed] [Cross Ref]
16. Woo KS, Ko JY, Seo MC, Song SB, Oh BG, Lee JS, Kang JR, Nam MH.
Physicochemical characteristics of the tofu (soybean curd) added sorghum
(Sorghum bicolor L. Moench) powder. J Korean Soc Food Sci Nutr. 2009;38:1746–
1752. doi: 10.3746/jkfn.2009.38.12.1746. [Cross Ref]
17. Kilci A, Gocmen D. Phenolic acid composition, antioxidant activity and phenolic
content of tarhana supplemented with oat flour. Food Chem. 2014;151:547–553. doi:
10.1016/j.foodchem.2013.11.038. [PubMed] [Cross Ref]
18. Fang Z, Hu Y, Liu D, Chen J, Ye X. Changes of phenolic acids and antioxidant
activities during potherb mustard (Brassica juncea, Coss.) pickling. Food Chem.
2008;108:811–817. doi: 10.1016/j.foodchem.2007.11.033. [PubMed] [Cross Ref]
XV. Chemical characterization of commercial liquid smoke
products
Food Sci Nutr. 2013 Jan; 1(1): 102–115.
Naim Montazeri,1 Alexandra CM Oliveira,1 Brian H Himelbloom,1 Mary Beth Leigh,2
and Charles A Crapo1

Abstract

The objective of this study was to determine important chemical characteristics of a

full-strength liquid smoke, Code 10-Poly, and three refined liquid smoke products
(AM-3, AM-10 and 1291) commercially available (Kerry Ingredients and Flavors,
Monterey, TN). The pH of the products were significantly different (P < 0.05) and
ranged from 2.3 (Code 10-Poly) to 5.7 (1291). The pH was inversely correlated with
titratable acidity (R2 = 0.87), which was significantly different (P < 0.05)
among products ranging from 10.3% acetic acid (Code 10-Poly) to 0.7% acetic acid
(1291). Total phenol content was quantified using the Gibbs reaction; the only liquid
smoke containing appreciable level of phenolic compounds was Code 10-Poly at
3.22 mg mL−1. Gas chromatography-mass spectrometry (GC-MS) analysis of liquid
smoke dichloromethane extracts revealed that carbonyl-containing compounds were
major constituents of all products, in which 1-hydroxy-2-butanone, 2(5H)-furanone,
propanal and cyclopentenone predominated. Organic acids were detected by GC-
MS in all extracts and correlated positively (R2 = 0.98) with titratable acidity. The
GC-MS data showed that phenolic compounds constituted a major portion of Code
10-Poly, and were detected only in trace quantities in 1291. The refined liquid
smokes had lighter color, lower acidity, and reduced level of carbonyl-containing
compounds and organic acids. Our study revealed major differences in pH, titratable
acidity, total phenol content, color and chemical make-up of the full-strength and
refined liquid smokes. The three refined liquid smoke products studied have less
flavor and color active compounds, when compared with the full-strength product.
Furthermore, the three refined products studied have unique chemical
characteristics and will impart specific sensorial properties to food systems.
Understanding the chemical composition of liquid smokes, be these refined or full-
strength products, is an important step to establish their functions and appropriate
use in food systems.

Keywords: GC-MS, Gibbs reaction, liquid smoke, phenol content

Introduction

Liquid smokes have been used extensively in food systems to impart flavor
characteristics that are similar to smoked food products (Varlet et al. 2010). These
may be used to preserve quality and ensure safety of foods (Schubring 2008; Martin
et al. 2010). Liquid smokes are usually obtained from the condensation of wood
smoke produced by smoldering wood chips or sawdust under limited oxygen.
Commercial full-strength liquid smokes are commonly fractionated, purified and
concentrated to yield aqueous, oil or dry powder products. Through the refining
process, undesirable polycyclic aromatic hydrocarbons (PAH) are removed, and the
intensity of flavor and color in the resulting refined liquid smoke is adjusted
(Cadwallader 2007; Varlet et al. 2010). Refined liquid smokes generally offer more
flexible applications to particular food systems when compared with full-strength
liquid smoke products.

The chemical composition of liquid smokes depends primarily on the wood type and
moisture content of wood, the latter influences the pyrolysis temperature and the
duration of smoke generation (Guillen and Ibargoitia 1999; Cadwallader 2007). A
majority of the dry mass of wood is composed of cellulose, hemicelluloses and lignin.
Thermal decomposition of cellulose produces anhydroglucose, carbonyl-containing
compounds and furans. Hemicellulose decomposition is similar to cellulose
decomposition, but yields acetic acid and carbon dioxide. Partial pyrolysis of lignin
produces various types of phenolic compounds (Miler and Sikorski 1990). Therefore,
the thermal degradation of wood results in a complex mixture of compounds, which
characterize the overall organoleptic, antioxidative and antibacterial properties of
full-strength liquid smokes (Guillen and Manzanos 1999a; Milly et al. 2005; Wei et al.
2010).

Baltes et al. (1981) found the major proportion of commercial full-strength liquid
smoke to be composed of water (11–92%), tar (1–17%), acids (2.8–9.5%), carbonyl-
containing compounds (2.6–4.6%) and phenol derivatives (0.2–2.9%). However, in
the manufacturing of liquid smokes, a variety of ingredients may be used, such as
salts, fatty acids, fatty esters and carriers like saccharides (Guillen and Manzanos
1997; Varlet et al. 2010). Phenolic compounds contribute to smoke flavor and color of

liquid smokes, and also have antibacterial and antioxidant properties (Clifford et al.
1980; Maga 1987; Varlet et al. 2010). Carbonyl-containing compounds impart sweet or
burnt-sweet aroma and tend to soften the heavy smoky aroma associated with
phenolic compounds with some “typical smoke-cured” aroma and flavors (Fujimaki
et al. 1974; Kim et al. 1974; Kostyra and Barylko-Pikielna 2006). Furthermore,
carbonyl-containing compounds are involved in textural changes in smoked food
caused by interaction with proteins, and contribute the golden-brown color of smoked
products due to reaction with amino acids, and the formation of Maillard reaction
products (Varlet et al. 2007).

Liquid smokes, either full-strength or refined, offer several advantages over wood
smoke, such as better control of PAH content, wider diversity of applications to food
systems, superior product homogeneity, easiness to store and less environmental
pollution (Varlet et al. 2010). A disadvantage of full-strength liquid smokes is their
high content of flavor and color active compounds, which limit their application when
used for specific purposes such as antimicrobial agents (Vitt et al. 2001). As stated,
full-strength liquid smokes can be refined for various applications. The desired
aroma and color characteristics of the final product are achieved by adjusting the
content of phenol derivatives, carbonyl-containing compounds and organic acids
(Kim et al. 1974; Underwood 1992; Toledo 2007). It has been demonstrated that
commercial liquid smoke preparations are effective against various types of spoilage
and pathogenic microorganisms (Milly et al. 2005). However, at times it may be
challenging to ensure that the liquid smoke used for a particular application will
deliver the desired final sensorial properties of a food product. For instance, liquid
smokes with high content of carbonyl-containing compounds are generally applied
to food systems as browning agent (Underwood 1992), while refined liquid smokes
containing reduced levels of color and flavor active compounds are applied as color
preservatives in raw tuna and salmon (Schubring 2008) or as an antimicrobial additive
in frankfurters (Martin et al. 2010).

Commercial production and fractionation of liquid smokes involve proprietary or

patented information; thus, the specific chemical constituents of these commercial
products are generally not disclosed to buyers. This poses difficulty for food product
developers to determine appropriate levels and types of liquid smoke products to be
used for specific applications in food systems. Establishing the chemical composition
of commercial liquid smokes is the first step to better understand their interaction
with the chemical components of food systems. It may also be advantageous in
predicting the resulting sensorial properties of the final product. These are two key
points to define the functions and appropriate uses of liquid smokes in food systems.

The objective of this study was to determine important chemical characteristics of a

full-strength liquid smoke, Code 10-Poly, and three refined liquid smoke products
(AM-3, AM-10 and 1291) commercially available (Kerry Ingredients and Flavors,
Monterey, TN). The chemical volatile and semi-volatile constituents of these
products were identified using gas chromatography-mass spectrometry (GC-MS)
analysis. Color, pH, titratable acidity and total phenol content were also determined.

Material and Methods

Procurement and handling of liquid smokes

A quantity of 2 L each of full-strength liquid smoke (Code 10-Poly lot no.

0831038703) and of three refined liquid smoke fractions (AM-3 lot no. 09009038703,
AM-10 lot no. 0910038701 and 1291 lot no. 0316038820) were procured from Kerry
Ingredients and Flavors in June 2010. Manufacturer information for these
preparations indicated that the three refined liquid smokes with less color and smoky
flavors could be applied to foods as antibacterial and/or shelf-life-enhancing
additives with minimized undesirable sensorial impacts. These refined liquid smokes
are produced using different processes except AM-10 which is a refined product
from AM-3 (M. van der Bleek, Kerry Ingredients and Flavors, pers. comm.). Each
product was transferred immediately from the plastic containers to 1 L glass jars
covered with two layers of aluminum foil and stored at 4°C until analysis. Code 10-
Poly was analyzed along with the refined liquid smokes to compare the extent of
refining process on final product characteristics. Analyses were conducted within 2
weeks of sample procurement.

Color

The Gardner Delta Color Comparator (BYK Additives and Instruments, Columbia,
MD) was used to determine the color of duplicate liquid smoke samples. The
comparator is equipped with two wheels embedded with nine glass filters with colors
ranging from colorless through deep amber. The colors correspond to the Gardner
color values ranging from 1 to 18. According to the manufacturer's instruction, about
10 mL of a liquid sample was placed in a glass tube, which was positioned between
the two wheels. The wheels were rotated until the filter glass closest in color to the
liquid was in place and the filter notation was recorded.

pH and titratable acidity

An aliquot of 5 mL (ca. 5 g) was sampled from each liquid smoke, and pH at 25°C
was recorded with a SevenEasy pH meter (Mettler Toledo, Schwerzenbach,
Switzerland) equipped with a probe (model WD-35801-00; Oakton Instruments,
Vernon Hills, IL). For titratable acidity (TA) (as % acetic acid), the liquid smokes were
diluted (1:4 w/v) in deionized water and titrated to pH 8.3 using 0.1 N NaOH (J. T.
Baker Chemical Co., Phillipsburg, NJ) as described by Sadler and Murphy ( 2003). All
pH measurements were conducted in triplicate.

Extraction of volatile and semi-volatile compounds in liquid smokes

The commercial liquid smokes were subjected to liquid–liquid extraction as
described by Guillen and Manzanos (1999a) with minor modifications. A sample of 25
mL of each liquid smoke was extracted with 50 mL dichloromethane (Honeywell
Burdick & Jacksons, Muskegon, MI) in a 100 mL two-neck round-bottom flask
(Chemglass, Vineland, NJ) fitted with a thermometer. The upper neck of the flask
was fitted with a water-jacketed reflux condenser with an internal coil-type cold finger
(Kimble Chase, Vineland, NJ) connected to a recirculating cooler (Model F 12,
Julabo, Allentown, PA) kept at 6.5°C. Another end of the condenser was open to the
atmosphere. The flask was heated in an oil bath for 7 h at 40°C and stirred with a
magnetic stirrer bar. A 12.5 mL aliquot of the extract was transferred to a 20 mL
capped amber glass vial (Agilent Technologies, Wilmington, DE), and 4 g of
anhydrous Na2SO4 (Mallinckrodt Chemicals, Phillipsburg, NJ) was added. After 4
h, the extract aliquot was centrifuged (5 min, 3,000 rpm) and the liquid portion
carefully transferred, using a glass Pasteur pipette, to another 20 mL capped amber
glass vial. The extract volume was reduced at 25°C using a gentle stream of
research-grade nitrogen gas. The final volume for each of the three refined liquid
smoke extracts (AM-3, AM-10 and 1291) was 1 mL, a tenth of the volume
suggested by Guillen and Manzanos (1999a). In the case of Code 10-Poly, it was not
possible to reduce its dichloromethane extract to 1 mL because constant volume
was achieved at 4.5 mL; therefore, this was its final volume.

Separation, identification and qualification of volatile and semi-volatile

compounds in liquid smokes

A gas chromatograph (GC, Model 6890; Agilent Technologies) interfaced with a

mass spectrometer (MS, Model 5973; Agilent Technologies) equipped with a DB-
5HT capillary column (30 m × 0.25 mm × 0.1 μm, Model 122–5731;
Agilent Technologies) was used. Helium was used as a carrier gas at 1.0 mL min−1
of flow and average velocity of 36 cm s−1. The GC was operated in constant flow
mode. The nominal inlet pressure and temperature were 7.67 psi and 150°C,
respectively. The transfer line temperature was kept at 280°C. The injection volume
was 1 μL at 1:100 inlet split ratio. The Code 10-Poly extract was exactly diluted
with dichloromethane to 1:10 (v/v) because gas chromatography of a 1 μL aliquot
of the dichloromethane extract that had been reduced to a constant volume of 4.5
mL, injected at 1:100 inlet split ratio, yielded misshaped chromatographic peaks
displaying significant front-tailing, poor symmetry and extensive overlaps of
contiguous peaks. Differently, the three refined liquid smoke fractions (AM-3, AM-10
and 1291) were analyzed using 1 μL aliquots of their dichloromethane extracts that
had been reduced to a final volume of 1 mL, corresponding to one tenth of the
volume suggested by Guillen and Manzanos (1999a). When the refined liquid smoke
extracts were reduced to a final volume of 10 mL (Guillen and Manzanos 1999a),
fewer peaks were observed at much lower abundances. Initial oven temperature was
50°C and held for 0.5 min, followed by an increment of 2°C min−1 until 250°C, and
a final hold time of 15 min. The total run time was 115.5 min. The MS was
operated in electron impact ionization mode at 70 eV and acquired data after a
solvent delay of 2 min. The mass range for acquisition was 50–300 amu at a scan
rate of 5.02 scans s−1. The temperature of the MSD transfer line, source and
quadrupole were 280°C, 230°C and 150°C, respectively. Compounds were identified
using the National Institute of Standards and Technology'98 (NIST'98) library
(Agilent Technologies). When the MS spectra quality (MSQ) was >90%, the
compound was considered “likely known,” when between 80% and 90% MSQ as
“tentatively identified” and when MSQ <80% as “unknown.” Peak areas were
recorded and reported as total ion count (TIC) and as peak percentage (% TIC),
while retention times were recorded and reported in minutes.

Total phenol content

Nicholson (1984) method was used to measure total phenol content (TPC) in liquid
smokes. This method is a modification of Tucker (1942) procedure based on the Gibbs
(1927) method that quantifies the intensity of a colorimetric reaction of the Gibbs
reagent with phenolic compounds. The stock Gibbs reagent was prepared by
dissolving 0.25 g of 2,6-dichloroquinone-4-chloroimide (Sigma-Aldrich, St. Louis,
MO) in 30 mL of absolute alcohol (Spectrum, Gardena, CA). The stock solution is
stable for several months if kept in a cold and dark environment (Svobodova et al.
1978). A working solution, to be used immediately, was prepared by adding 1 mL of
the stock solution to 15 mL of deionized water (nanopure). A buffer was prepared
by mixing 125 mL of 0.4 M boric acid (J. T. Baker Chemical Co., Phillipsburg,
NJ), 125 mL of 0.4 M (Sigma-Aldrich, Steinheim, Germany) and 40 mL of 0.2
M NaOH (Fisher Scientific, Fair Lawn, NJ), and bringing the mixture volume to 1 L
using deionized water. The liquid smokes were diluted with deionized water to 1:40,
1:400 and 1:4,000 (v/v, liquid smoke: deionized water).

A sample of 5 mL of liquid smoke diluted in deionized water and five phenol

solutions with concentrations of 3, 6, 9, 12 and 15 μg mL−1 phenol (Sigma-Aldrich,
St. Louis, MO) were each mixed with 5 mL of pH 8.3 buffer. Deionized water was
used as blank. A quantity of 1 mL of 0.15 N NaOH was added followed by 1 mL
of the Gibbs working solution. The mixture was kept in the dark for 25 min at 25°C.
Indophenols result from the reaction as indicated by blue color development (Gibbs
1927). The absorbance was recorded at 635 nm using a Cary 50 UV–Visible
Spectrophotometer (Varian Inc., Walnut Creek, CA) and the TPC calculated using a
calibration curve linear to 15 μg mL−1 phenol (R2 = 0.996). The TPC was
determined in each of the liquid smoke solutions in triplicate (Nicholson 1984).

Statistical analyses

Linear regression and correlation were performed using Microsoft® Excel 2011 (Ver.
14.2.3: Microsoft Corporation, Santa Rosa, CA) to establish standard curve for TPC
measurement and to evaluate the statistical relationships among TA, pH and organic
acid content (α = 0.05).

Results and Discussion

Color
Freshly prepared liquid smokes were bright yellow; however, color changes quickly
take place because some of the smoke components condense or polymerize
rendering the mixture brown (Guillen and Manzanos 1997, 1999a, 1999b; Simko 2005). The
four liquid smokes studied showed no visible turbidity or precipitate formation during
2 months of storage at 4°C. The Code 10-Poly had the darkest Gardner color value
of 16, followed by AM-3, AM-10 and 1291 with values of 11, 5 and 2, respectively.
The Gardner Delta Color Comparator is a simple, inexpensive, and non-destructive
instrument that is used to measure color of vegetable oil (Cermak and Isbell 2002 and
fish oil (U.S. Food and Drug Administration 2003). Limited information is available in
the literature regarding the instrumental quantification of liquid smoke color. Our
results demonstrated that this instrument shows potential in an industry setting for
quick measurement of color in liquid smoke products.

pH and titratable acidity

Acids influence flavor (tartness), color, texture and microbial stability of food
(Hollenbeck 1977; Sadler and Murphy 2003; Rozum 2007). Acidity of liquid smokes
depends on the wood source, processing steps and refining parameters (Guillen and
Ibargoitia 1999; Rozum 2007; Sung et al. 2007; Toledo 2007). Liquid smokes are usually
acidic with a pH of 1.5–5.5 (Toth and Potthast 1984); however, an alkaline liquid
smoke (pH = 7.7) prepared from black tea leaves is reported by Sung et al.
(2007) These authors observed that the pH of bread flour liquid smoke was 4.2, and
when combined with black tea, the pH of resulting liquid smoke increased to 5.5
(Sung et al. 2007. The Code 10-Poly had the highest acidity with pH of 2.3 and TA
(% acetic acid) of 10.3 (mean ± 0.0 SD, n = 3). On the other hand, the refined
ones had a higher pH and lower TA. The pH and TA of AM-3, AM-10 and 1291 were
4.3, 4.2, and 5.7, and 2.2, 2.3, and 0.7 (% acetic acid), respectively (mean ± 0.0
SD, n = 3). In the liquid smoke manufacturing process, acids may be neutralized
to decrease the harshness of smoke flavor (Toledo 2007). This change in pH may
affect food texture, color development and flavor intensity (Maga 1987; Fiddler et al.
1970). Among the product tested, 1291 had the highest pH and lowest TA, which

indicates a probable neutralization step during manufacturing, and less acidic flavor
change to food.

The pH of the four products investigated were inversely correlated (R2 = 0.87; TA
= −2.99(pH)+16.17) to TA values and similar to the results reported by Milly et al.
(2005). However, a correlation between organic acid content and pH is not observed
for liquid smokes prepared from different woods (Chen and Maga 1993). Similar
results are observed in wine (Plane et al. 1980) and milk (Walstra et al. 2006). The
lack of correlation between the two parameters could be attributed to differences in
dissociation constants of the organic acids present in the products. A pH
measurement reflects the quantity of hydronium ions (H3O+) in the media; however,
not all organic acids may be dissociated (Sadler and Murphy 2003). Conversely, TA
is based on the ability of an alkaline titrant to dissociate weak organic acids (Sadler
and Murphy 2003). Based on our results, TA appeared to be a better predictor of
acidity in liquid smokes. With respect to organic acids impacts on flavor, a
combination of hydrogen ions, anions and/or protonated acid species dictates acid
taste (Plane et al. 1980; Sadler and Murphy 2003; Neta et al. 2007).

Volatile and semi-volatile compounds in liquid smokes

Dichloromethane is a widely used solvent to extract the chemical constituents of

liquid smokes (Edye and Richards 1991; Guillen et al. 1995, 2001; Sung et al. 2007).
Dichloromethane has a low boiling point (40°C) and extracts aromatic compounds
efficiently (Guillen and Manzanos 1999a). Some drawbacks of using a
dichloromethane extraction is the escape of low boiling point <40°C compounds
(Guillen et al. 2001) and partitioning of very polar compounds, such as short-chain
organic acids, between the liquid smoke aqueous phase and the dichloromethane
phase. Furthermore, there are non-volatiles and particulate matter that may
contribute to smoke flavor but do not extract with dichloromethane (Maga 1987).

The 1:100 inlet split ratio analysis provided better separation and peak resolution for
the majority of detected compounds, but was not sufficient to identify some
compounds present at low abundance. Therefore, all four extracts were also injected
using a 1:50 inlet split ratio to investigate the possible identity of certain peaks
(Tables 1 and and22 footnote). Selected chromatograms of Code 10-Poly and
1291 dichloromethane extract, as two extreme examples of a full-strength and a
refined liquid smoke, are represented in Figure 1. Phenolic compounds are an
important group of smoke constituents responsible for smoke flavor notes in liquid
smokes and in smoked food products. Phenols also have antibacterial and
antioxidant properties (Maga 1987). Phenolic compounds were only detected in the
Code 10-Poly extracts and in trace quantities in the 1291 extracts. Carbonyl-
containing compounds (aldehydes, ketones, furan and pyran derivatives) constituted
a major portion of all four extracts. This class of compounds provides odor and flavor
background notes described as sweet, caramel and magi (instant broth), which
rounds-up the smoky odor in liquid smokes (Kostyra and Barylko-Pikielna 2006),
contributes to the browning coloration in smoked foodstuff (Varlet et al. 2007), and
may display antibacterial properties (Milly et al. 2005).

Table 1. Volatile and semi-volatile compounds tentatively identified in

dichloromethane fractions of Code 10-Poly

tR Compounds (mass
MSQ1 TIC2 %TIC Other names3
(min) spectra data)

Phenolic compounds

6.38 Phenol 94 554.2 0.5

9.08 2-methylphenol 95 350.0 0.3 Ό-cresol

tR Compounds (mass
MSQ1 TIC2 %TIC Other names3
(min) spectra data)

92
10.02 4-methylphenol 601.5 0.6 ρ-cresol
(93)4

10.32 2-methoxyphenol 97 3,908.4 3.9 Guaiacol

13.51 Dimethylphenols 93 101.6 0.1

4-ethylphenol OR 3- 62-70
14.56 121.9 0.1
ethylphenol (80)4

93
14.78 2-methoxy-3-methylphenol 179.0 0.2
(94)4

2-methoxy-4-methylphenol
15.60 OR 2-methoxy-5- 95-95 2,457.1 2.4
methylphenol

16.79 1,2-benzenediol 91 3,319.3 3.3 Pyrocatechol; catechin

95
19.41 3-methoxy-1,2-benzenediol 1,262.3 1.2 3-methoxypyrocatechol
(97)4

20.15 3-methyl-1,2-benzenediol 95 846.5 0.8 3-methylcatechol

91
22.01 4-methyl-1,2-benzenediol 751.8 0.7 4-methylcatechol
(93)4

20.51 4-ethyl-2-methoxyphenol 90 1,281.3 1.3 4-ethylguaiacol

21.03 4-methoxy-3-methylphenol 83 184.2 0.2

Anisaldehyde;m-
3-hydroxybenzaldehyde
formylphenol;ρ-
OR 4- 96–
23.03 346.0 0.3 formylphenol;Ό-
hydroxybenzaldehyde OR 90–81
formylphenol
2-hydroxybenzaldehyde
(salicylaldehyde)

24.95 2,6-dimethoxyphenol 94 12,482.4 12.3 Syringol

25.17 Eugenol 98 206.3 0.2 ρ-allylguaiacol

2-hydroxy-3-
27.39 Vanillin 81 1,043.2 1.0
methoxybenzaldehyde
tR Compounds (mass
MSQ1 TIC2 %TIC Other names3
(min) spectra data)

4,5-dimethyl-1,3- 59
28.64 352.1 0.3
benzenediol (92)4

Cerulignol; 4-
31.06 2-methoxy-4-propylphenol 87 347.3 0.3
propylguaiacol

4-hydroxyacetyl-2-
38.23 80 187.0 0.2
methylphenol

2,6-dimethoxy-4-(2- 97
39.34 858.2 0.8 4-allylsyringol
propenyl) phenol (98)4

4-hydroxy-3- (4-hydroxy-3-
74
41.54 methoxybenzeneacetic 431.6 0.4 methoxyphenyl) acetic
(81)4
acid acid

4-hydroxy-3,5- 90
42.21 2,273.0 2.2 Syringaldehyde
dimethoxybenzaldehyde (93)4

Total phenolic compounds 34,445.9 34.0

Aldehydes and ketones

2.60 Propanal 87 1,960.8 1.9

2.44 1-hydroxy-2-butanone 87 2,710.9 2.7

2-methyl-2-cyclopenten-1-
4.28 94 398.2 0.4
one

87
4.70 1,2-cyclopentanedione 2,388.3 2.4
(91)4

4.80 2,5-hexanedione 87 315.5 0.3

4-methyl-2-
5.61 hydroxycyclopent-2-en-1- 90 133.4 0.1
one

86
5.81 1-(acetyloxy)-2-butanone 452.6 0.4
(91)4

3,4-dimethyl-2-
6.60 95 220.1 0.2
cyclopenten-1-one
tR Compounds (mass
MSQ1 TIC2 %TIC Other names3
(min) spectra data)

2-hydroxy-3-methyl-2-
7.85 94 5,459.4 5.4 Cyclotene
cyclopenten-1-one

2,3-dimethyl-2- 89
8.16 438.4 0.4
cyclopenten-1-one (94)4

3,4-dimethyl-2-
8.95 hydroxycyclopent-2-en-1- 95 416.7 0.4
one

3-ethyl-2-hydroxy-2-
11.87 95 1,261.5 1.2
cyclopenten-1-one

13.85 2,3-dihydroxybenzaldehyde 95 320.8 0.3

3,4- 83 1-(3,4-dihydroxyphenyl)
19.97 390.2 0.4
dihydroxyacetophenone (87)4 ethanone

1-(3-hydroxyphenyl)
28.08 93 252.3 0.2
ethanone

37.47 2-ethoxy-4-anisaldehyde 80 365.9 0.4

1-(4-hydroxy-3- Acetovanillone; 4-acetyl-

32.51 97 965.6 1.0
methoxyphenyl) ethanone 2-methoxyphenol

(2E)-3-(4-hydroxy-2-
4-hydroxy-2- 90
45.71 135.8 0.1 methoxyphenyl)-2-
methoxycinnamaldehyde (95)4
propenal

Acetosyringone; 4-
1-(4-hydroxy-3,5-
46.04 97 2,662.7 2.6 acetyl-2,6-
dimethoxyphenyl) ethanone
dimethoxyphenol

3,5-dimethoxy-4- 93
57.89 504.2 0.5
hydroxycinnamaldehyde (97)4

Total aldehydes and

21,726.3 21.4
ketones

Furans and pyrans

tR Compounds (mass
MSQ1 TIC2 %TIC Other names3
(min) spectra data)

91
2.88 Furan-3-carbaldehyde 121.3 0.1 3-furfural; 3-furaldehyde
(94)4

3.10 Furan-2-carbaldehyde 91 5,716.2 5.6 2-furfural; 2-furaldehyde

Tetrahydro-2,5-
3.82 90 610.1 0.6
dimethoxyfuran

4.38 1-(2-furanyl)-ethanone 94 387.6 0.4 Acetylfuran

4.42 Butyrolactone 90 513.1 0.5 2(3H)-furanone

2-butenolide; γ-
4.48 2(5H)-furanone 91 2,735.6 2.7
crotonolactone

87
5.06 5-methyl-2(5H)-furanone 175.6 0.2 β-Angelica lactone
(90)4

5-methyl-2-
5.68 94 605.5 0.6 5-methylfurfural
furancarboxaldehyde

6.05 Methyl furan-3-carboxylate 83 124.0 0.1 Methyl ester 3-furoic acid

6.41 3-methyl-2(5H)-furanone 91 426.7 0.4 2-methyl-2-butenolide

2,5-dihydro-3,5-dimethyl-2-
6.97 91 548.2 0.5
furanone

78
8.36 4-methyl-5H-furan-2-one 732.1 0.7 4-methyl-2(5H)-furanone
(87)4

90 3-hydroxy-2-methyl-4h-
11.51 Maltol 757.6 0.7
(93)4 pyran-4-one

5-(hydroxymethyl)-2-
18.01 94 1,020.2 1.0
furancarboxaldehyde

Total furans and pyrans 14,473.6 14.3

Organic acids

2.74 Butanoic acid 90 187.9 0.2 Butyric acid

3.25 2-butenoic acid 86 257.6 0.3 Crotonic acid

tR Compounds (mass
MSQ1 TIC2 %TIC Other names3
(min) spectra data)

3-hydroxy-4-
30.56 80 4,999.2 4.9 Isovanillic acid
methoxybenzoic acid

65.70 9-(E)-octadecanoic acid 99 460.8 0.5 Oleic acid

Total organic acids 5,905.4 5.8

Miscellaneous

1,4:3,6-dianhydro-α-d- 83
15.70 465.1 0.5
glucopyranose (89)4

17.89 2,3-anhydro-d-mannosan 90 521.9 0.5

1,2,3-trimethoxy-5- 90
28.32 330.2 0.3 3,4,5-trimethoxytoluene
methylbenzene (92)4

3-hydroxy-benzoic acid 83
29.84 317.6 0.3
methyl ester (87)4

1,4-dimethoxy-2- 83-80
34.19 methylbenzene OR 3- (87- 200.1 0.2
isopropylthiophenol 86)4

4-hydroxy-3-methoxy- 93
34.36 136.8 0.1 Vanillic acid methyl ester
benzoic acid methyl ester (94)4

90 6-hydroxy-2H-1-
36.04 6-hydroxycoumarin 154.4 0.2
(93)4 benzopyran-2-one

Total miscellaneous 2,126.0 2.1

Total area of known

78,677.3 77.6
compounds

Total area of unknown

9,593.7 9.5
compounds with >2 %TIC
1MSQ, MS spectrum quality according to NIST98 library.
2Code 10-Poly was injected at 1:10 dilution, the area of peaks are multiplied by 10,
and reported in 104 TIC.
3Extracted
from NIST Chemistry WebBook, NIST Standard Reference Database
Number 69 (http://webbook.nist.gov/chemistry/name-ser.htm, accessed November
2011).
4MSQ values in parentheses are for peaks when samples were injected at 1:50 inlet
split ratio.

Table 2. Volatile and semi-volatile compounds tentatively identified in

dichloromethane fractions of refined liquid smokes

AM-3 AM-10 1291

tR Compounds
MS %T MS %T MS %T Other
(mi (mass spectra 1 TIC2 TIC TIC
Q IC Q IC Q IC names3
n) data)

Phenolic compounds

81 3-
20. 3-methyl-1,2-
(90 9.8 0.1 methylcatech
15 benzenediol
)4 ol

Total phenolic
0.0 0.0 0.0 0.0 9.8 0.1
compounds

Aldehydes and ketones

80
2.6 1,741 1,409 693.
Propanal 83 5.7 (90 8.2 80 6.3
0 .3 4 .8 1
)

86
2.4 1-hydroxy-2- 2,342 1,920 11. 5,02 46.
(90) 7.6 86 87
4 butanone 4 .5 .9 2 0.5 0

64
3.0 4-hydroxy-2- 257.
(90) 493.8 1.6 93 267.8 1.6 83 2.4
7 pentanone 4 6

91
3.1 2-cyclopenten-1-
83 873.9 2.8 (93 166.7 1.0 93 98.4 0.9
5 one
)4

72
3.5 2,056
2-butanone 80 6.7 (80 947.8 5.5
6 .1
)
 AM-3 AM-10 1291

tR Compounds
MS %T MS %T MS %T Other
(mi (mass spectra 1 TIC2 TIC TIC
Q IC Q IC Q IC names3
n) data)

4.8
2,5-hexanedione 90 116.4 0.7
0

90
5.8 1-(acetyloxy)-2-
(91) 163.2 0.5 91 14.5 0.1
1 butanone 4

5.9 3-methyl-2-
91 326.8 1.1
6 cyclopenten-1-one

6.8 1,2-
93 30.7 0.1
3 cyclohexanedione

2-hydroxy-3- 94
7.8 1,004
methyl-2- (95) 3.3 Cyclotene
5 4 .8
cyclopenten-1-one

Total aldehydes 9,033 29. 4,829 28. 6,08 55.

and ketones .0 3 .4 2 4.1 7

Furans and pyrans

2- 90
2.3
methoxytetrahydro (91 965 6.4
5
furan )4

2.6 Tetrahydro-2- 103.

80 19.4 0.1 87 0.9
3 furanol 7

3.8 Tetrahydro-2,5- 100.

94 0.9
2 dimethoxyfuran 4

83 3-
3.9 Dihydro-2H-pyran-
86 392.5 1.3 (86 195.4 1.1 tetrahydropyr
2 3(4H)-one
)4 anone

3.9 Tetrahydro-2-
83 16.4 0.1
3 furanmethanol
 AM-3 AM-10 1291

tR Compounds
MS %T MS %T MS %T Other
(mi (mass spectra 1 TIC2 TIC TIC
Q IC Q IC Q IC names3
n) data)

3.9 Tetrahydro-2-
80 29.6 0.1
8 methyl-2-furanol

86
4.3 1-(2-furanyl)-
(87) 185.1 0.6 Acetylfuran
8 ethanone 4

4.4 437. 2(3H)-

Butyrolactone 91 4.0
2 4 furanone

2-butenolide;
94 90
4.4 7,361 23. 3,767 22. 243. γ-
2(5H)-furanone (95) (95 91 2.2
8 4 .2 9 .7 0 9 crotonolacto
)4
ne

5.0 5-methyl-2(5H)- β-Angelica

95 685.7 2.2 95 155.9 0.9
6 furanone lactone

91
6.4 3-methyl-2(5H)- 2-methyl-2-
91 569.3 1.8 (93 65.4 0.4
1 furanone butenolide
)4

6.2
2H-pyran-2-one 93 14.6 tr Coumalin
2

2,5-dihydro-3,5-
6.9
dimethyl-2- 91 270.2 0.9
7
furanone

8.2 3,4-dimethyl-2,5- Dimethylmal

93 11.9 0.1
2 furandione eic anhydride

4-methyl-
8.3 4-methyl-5H-furan-
91 985.3 3.2 2(5H)-
6 2-one
furanone

12. 5-acetyldihydro-
83 99.7 0.6 Solerone
29 2(3H)-furanone
 AM-3 AM-10 1291

tR Compounds
MS %T MS %T MS %T Other
(mi (mass spectra 1 TIC2 TIC TIC
Q IC Q IC Q IC names3
n) data)

5-
15.
hydroxymethyldihy 86 112.0 0.7
57
drofuran-2-one

5-(hydroxymethyl)-
18. 2- 3,277 10.
94
01 furancarboxaldehy .9 6
de

Total furans and 13,77 44. 4,431 25. 1,59 14.

pyrans 1.4 7 .9 8 1.9 6

Organic acids

2.2 175. Propionic

Propanoic acid 90 197.0 0.6 90 414.0 2.4 90 1.6
9 1 acid

3.0
3-butenoic acid 90 13.8 0.1
4

3.2
2-butenoic acid 93 20.2 0.1 64.0 0.6 Crotonic acid
5

Levulinic
11. 4-oxo-pentanoic
80 124.9 0.4 acid; 4-oxo-
42 acid
valeric acid

65. 9-(E)-
93 44.1 0.3 Oleic acid
70 octadecanoic acid

252.
Total organic acids 321.8 1.0 478.3 7.4 2.3
9

Miscellaneous

3.0 2-propenyl-
83 787.5 4.6
8 butanoate

11.
Glycocyanidine 80 265.3 1.5
91
 AM-3 AM-10 1291

tR Compounds
MS %T MS %T MS %T Other
(mi (mass spectra 1 TIC2 TIC TIC
Q IC Q IC Q IC names3
n) data)

1,4:3,6-dianhydro-
15. 1,248
.alpha.-d- 83 140.8 0.5 93 7.3
70 .5
glucopyranose

86
17. 2,3-anhydro-d- 1,408
93 4.6 (93 725.2 4.2
89 mannosan .9
)4

Total 1,549 3,026 17.

5.0
miscellaneous .6 .5 6

Total area of 24,67 80. 12,76 74. 7,93 72.

known compounds 5.8 1 6.1 4 8.7 7

Total area of
unknown
2,125 1,532 548.
compounds with 6.9 8.9 5.0
.9 .3 5
peak area >2
%TIC
1MSQ, MS spectrum quality according to NIST98 library.
2Area in 104 TIC.
3Extracted from NIST Chemistry WebBook, NIST Standard Reference Database
Number 69 (http://webbook.nist.gov/chemistry/name-ser.htm, accessed November
2011); tr, %TIC <0.1.
4MSQ values in the parentheses are for peaks when samples were injected at the
1:50 inlet split ratio. Note: The blank cells for TIC and %TIC indicate that the
chemicals were not detectable.
Figure 1. Selected chromatograms of Code 10-Poly (left) and 1291 (right) at 1:100
inlet split ratio, where abundance is reported as the total ion count (TIC) and time in
minutes (min).

Phenolic compounds

Several phenolic compounds were detected in the Code 10-Poly and accounted for
34% TIC (Table 1). Identification of phenol was confirmed by injecting a phenol
standard solution (Sigma-Aldrich) into the GC-MS along with the sample analysis
(data not shown).

Syringol (2,6-dimethoxyphenol) was the most abundant phenolic compound,

followed by guaiacol (2-methoxyphenol), and pyrocatechol (1,2-benzenediol). The
ratio of syringol derivatives to guaiacol derivatives was approximately 1.2:1 and
lower than 3.3:1, which is considered an indicator that hardwood was used as a
wood source (Baltes et al. 1981). When softwood is used to produce smoke, syringol
derivatives are barely detectable (Edye and Richards 1991; Guillen et al. 1995). Our
results indicated a mixture of hardwood and softwood might have been used to
produce Code 10-Poly. Guaiacol and pyrocatechol are antioxidants found in smoke
(Fujimaki et al. 1974; Guillen et al. 1995, 2001; Maga 1987). Along with 4-
methylguaiacol and syringol, guaiacol contributes to overall smoky flavor and odor
(Wasserman 1966). Syringol and its derivatives are not associated with odor notes
typically detected in freshly smoked food products (Kostyra and Barylko-Pikielna
2006). The relative ratio of the total content of guaiacol and its derivatives to that of

phenol, cresols and xylenols may impart a considerable effect on the variability in
the aroma of the phenolic fractions of wood smoke (Fujimaki et al. 1974). 4-
methylphenol (ρ-cresol) and 2-methylphenol (Ό-cresol) are present in Code 10-Poly
and are important contributors to the typical, palatable smoke-curing profile in liquid
smoke product (Kostyra and Barylko-Pikielna 2006). Pyrocatechol is another
important phenol in Code 10-Poly with antioxidant property that contributes to a
heavy burnt and phenolic odor and some sweetness in liquid smokes (Fujimaki et
al. 1974; Guillen et al. 1995, 2001). Among the refined liquid smokes, 3-methyl-1,2-
benzenediol was only detected in trace quantities in 1291.

Aldehydes and ketones

The refining process to produce AM-10 from AM-3 reduced the content of aldehydes
and ketones in AM-10 (Table 2). The most refined 1291 contained the highest
proportion of aldehydes and ketones at about 56% TIC. The most abundant
compound was 1-hydroxy-2-butanone, which was also the major constituent of the
other two refined liquid smokes. The 1-hydroxy-2-butanone is detectable in ether
extracts of refined liquid smoke prepared from the branches of walnut tree (Juglans
sp.) (Wei et al. 2010) and a commercial liquid smoke preparation (Charsol, Red
Arrow Products Corp., Milwaukee, WI) (Fiddler et al. 1970). Formation of 1-hydroxy-
2-butanone occurs during low temperature pyrolysis of wood hemicelluloses (Yi-min
et al. 2009). The aroma perception of 1-hydroxy-2-butanone has a sweet, musty,
coffee and grain-like odor (Kaseleht et al. 2011). Its isomer 3-hydroxy-2-butanone
contributes to the pleasant butter-like odor of cold-smoked Atlantic salmon (Salmo
salar) and as an indicator of spoilage (Jonsdottir et al. 2008).

Propanal was detected in all liquid smoke extracts, but was least abundant in 1291
(Tables 1 and and2).2). This aldehyde is detected in liquid smoke from grapevine
shoots (Vitis vinifera L) and beechwood (Fagus sylvatica L) (Guillen and Ibargoitia
1996), but is not detected in a commercial Spanish liquid smoke (Guillen and Ibargoitia
1998). The 2-cyclopenten-1-one and its derivatives are usually found in wood or

smoke in various amounts (Maga 1987; Guillen and Ibargoitia 1999) and have the
organoleptic characteristics described as bitter taste and odor with grassiness and
slight effect on smoke flavor (Kim et al. 1974). The 2-cyclopenten-1-one was absent
in Code 10-Poly, but most of the methyl and dimethyl derivatives were detected
(Table 1). Cyclotene (2-hydroxy-3-methyl-2-cyclopenten-1-one) was found in Code
10-Poly and AM-3 and detectable in AM-10 and 1291 only when samples were
analyzed using an inlet split ratio of 1:50 (data not shown). These compounds are
found usually in wood smoke or liquid smoke (Maga 1987; Guillen et al. 1995; Guillen
and Manzanos 1996; Guillen and Ibargoitia 1999). Their organoleptic characteristic is
described as bitter taste and grass odor notes (Kim et al. 1974). Similar to ρ-cresol
and Ό-cresol, cyclotene mainly contributes to liquid smoke odor intensity and is
characterized as a palatable taste typical of smoked foods (Kostyra and Barylko-
Pikielna 2006). This ketone is detected in Atlantic salmon treated with liquid smoke
and contributes to a potent “cooked/spicy” odor in the product (Serot et al. 2007).

Furans and pyrans

Furans and pyrans result from thermal breakdown of cellulose and hemicellulose
(Maga 1987) and contribute overall smoky odor to liquid smokes that soften partially
the heavy smoky aroma of phenolic compounds (Kim et al. 1974). Their formation is
attributed to Maillard reactions (Varlet et al. 2007). Furans are considered a
secondary group of odor-active compounds, after phenolic compounds, in smoked
Atlantic salmon extract (Serot et al. 2007). Pyrans impart green and roasty odor
notes to salmon treated with liquid smoke (Serot et al. 2007). Furans and pyrans
comprised about 26% and 45% TIC of AM-10 and AM-3, respectively (Table 2).
The predominant compound was 2(5H)-furanone (2-butenolide), and its isomeric
methyl derivatives were detected in three of the liquid smokes extracts studied but
not in the 1291 extract. Furfurals (furan-2-carbaldehyde and furan-3-carbaldehyde)
were detected only in Code 10-Poly and were in agreement with the chemical
composition of liquid smoke prepared from grapevine shoot and beechwood (Guillen
and Ibargoitia 1996), and walnut tree branches (Wei et al. 2010). Organoleptic
characteristics of furfural are described as sweet, bread-like and caramel-like
(Guillen and Ibargoitia 1996), and are linked with development of smoke odor in
smoked foods (Serot et al. 2007; Toledo 2007). Detected in Code 10-Poly and AM-3,
5-methyl-2-furancarboxaldehyde imparts sweet fragrant and floral odor notes
contributing to partial softening of the heavy smoky aroma of phenolic compounds
in liquid smokes (Kim et al. 1974). Maltol was found only in Code 10-Poly and is one
of the main furans detected in beechwood liquid smoke (Guillen and Ibargoitia 1999)
and present in considerable amounts in several other liquid smoke products (Guillen
and Ibargoitia 1996; Guillen and Manzanos 1997 1999a). Maltol is a chemical reference
to “sweet odor” for sensory evaluation and training of panelists and has higher flavor
intensity than furfurals (Ojeda et al. 2002). Other minor compounds in this group
were 2(3H)-furanone in 1291 and Code 10-Poly and 4-methyl-5H-furan-2-one in AM-
3 and Code 10-Poly (Tables 1 and and22).

Organic acids

Organic acids result from the partial pyrolysis of wood cellulose and hemicellulose
(Gilbert and Knowles 1975). During fish smoking, a range of organic acids may be
deposited on the product surface (Guillen and Errecalde 2002). Organic acids are
known for their impact on flavor (tartness), color, texture and microbial stability of
food (Hollenbeck 1977; Sadler and Murphy 2003; Rozum 2007). Among these, acetic and
lactic acids are added to food products for preservative (Baltes et al. 1981) and
antibacterial purposes (Doores 1993). Butyric, caproic, capric and enanthic acids are
mainly strong aroma carriers (Baltes et al. 1981). However, none of these organic
acids were detected in our samples and was probably due to their high polarity and
low affinity to dichloromethane. Similarly, acetic acid is not quantifiable in the
dichloromethane extract of liquid smoke obtained from thyme (Thymus vulgaris L.)
(Guillemet and Manzanos 1999a). However, acetic acid is found in liquid smoke
obtained from beechwood (Guillen and Ibargoitia 1999) and the most abundant
organic acid in liquid smoke prepared from a mixture of ponderosa pine (Pinus
ponderosa) and cottonwood (Populus trichocarpa) (Edye and Richards 1991) and
walnut tree branches (Wei et al. 2010).
Propanoic acid has a slightly pungent odor that contributes to the overall odor of
liquid smokes and has antibacterial activity against spore-forming bacteria and
molds in food (Doores 1993; Guillen and Manzanos, 1999a). This was the predominant
organic acid in AM-3, AM-10 and 1291 (Table 2). Propanoic acid was a component
of liquid smoke prepared from thyme (Guillen and Manzanos, 1999a). However, only
trace amounts are detectable in liquid smoke from a mixture of ponderosa pine and
cottonwood (Edye and Richards 1991). In Code 10-Poly, 3-hydroxy-4-
methoxybenzoic acid was the predominant organic acid and may have been formed
by secondary oxidation of aldehydes (Edye and Richards 1991).

The organic acid content in the liquid smoke products studied followed the values
determined for TA and pH, as found by others (Edye and Richards 1991; Guillen and
Ibargoitia 1996). The pH and TA of all products investigated were inversely correlated
(R2 = 0.87). A good linear correlation (R2 = 0.98; TA = 2 × 10−7[organic
acid content] + 1.17) was established between organic acids content and TA.
However, a low correlation (R2 = 0.77; pH = −4 × 10−8[organic acid
content] + 4.86) was observed for pH and organic acid content. Unlike pH,
titratable acidity correlates well with the content of acetic acid (Edye and Richards
1991). Results indicated that TA was a more reliable parameter than pH for

determining organic acid content of liquid smokes, and may these be full-strength or
refined.

Other compounds

Miscellaneous compounds such as sugars, benzene and esters were found in AM-
3 and AM-10 (Table 2). Through pyrolysis, cellulose hydrolyzes to glucose followed
by dehydration to 1,6-anhydroglucose (beta-glucosan) (Simon et al. 2005) and may
explain its presence in liquid smokes. A third fragmentation produces acetic acid and
its homologs (Simon et al. 2005) that may also be present in liquid smokes. The
1,2,3-trimethoxy-5-methyl-benzene was detected only in the Code 10-Poly and is
detected at low concentrations in smoked Atlantic salmon treated with liquid smoke
(Serot et al. 2007). Its characteristic odor descriptions are cooked and earthy notes
(Serot et al. 2007).

In addition to the advantage of 1:50 inlet split ratio injection in enhanced identification
of some compounds, it improved the detection of other compounds that were not
detected at 1:100 inlet split ratio injection. Among carbonyl-containing compounds,
4-methyl-5H-furan-2-one was detected in AM-10 with 93% MSQ, and 5-
hydroxymethyl-2-furancarboxaldehyde in 1291 with 94% MSQ. The ester 2-
propenyl-butanoate was detected in 1291 with 83% MSQ. Pyrazine derivatives were
not found in any samples analyzed, and these compounds contribute to a
background smoky aroma associated with hickory smoke of liquid smokes (Maga
and Chen 1985; Guillen and Manzanos, 1999a).

Total phenol content

The calibration curve of phenol standards was linear for 3–15 μg mL−1 (R2 =
0.997, absorbance = 0.077[phenol] + 0.009). The absorbance values for the
phenol standard solutions ranged from 0.21 to 1.22. Blue color developed only with
Code 10-Poly, which had a TPC of 3.22 ± 0.03 mg mL−1. This value was much
lower than the phenol content reported in an aqueous fraction of liquid smokes
prepared from mixed dried hardwood sawdust, which is in the range of 9.9–11.1
mg mL−1 (Ramakrishnan and Moeller 2002). The GC-MS analysis of the
dichloromethane extracts (Table 2) revealed that neither AM-3 or AM-10 contained
phenolic compounds, while 1291 contained only a trace amount of 3-methyl-1,2-
benzenediol (0.1% TIC) and was below the minimum phenol concentration of 0.5 ×
10−4 mg mL−1 required for blue color formation from indophenols (Gibbs 1927).

Conclusion

This study determined key product characteristics of three commercially refined

liquid smokes and one full-strength liquid smoke. The refining of liquid smokes
eliminated phenolic compounds and, to a lesser extent, reduced contents of
carbonyl-containing compounds and organic acids. As a result, the refined liquid
smoke fractions AM-3, AM-10 and 1291 had significantly lower acidity, lighter color,
and rather distinct chemical make-up when compared with the full-strength Code 10-
Poly. The observed differences demonstrate that knowledge of the chemical
characteristics of liquid smoke preparations is important to better understand their
interactions with food components. Lower levels of carbonyl-containing compounds,
as determined in the refined liquid smokes, may result in fewer changes to the
original color and texture of the food product. Concomitantly, removal of phenol
derivatives from liquid smokes may lower their impact in the final flavor of the food
product. Nevertheless, prediction of impacts of liquid smokes in the organoleptic
characteristic of food products is difficult because of the possibility of synergistic and
antagonistic effects liquid smoke components may have with specific constituents of
the food matrix. Overall, the results from this study are useful to food product
developers seeking to determine a suitable liquid smoke product, and its appropriate
level of use for application in a particular food system.

Acknowledgments

This publication is the result of research sponsored by USDA-CSREES (grant no.

2008-34385-19314) and Alaska Sea Grant with funds from the National Oceanic and
Atmospheric Administration Office of Sea Grant, Department of Commerce, under
grant no. NA10OAR4170097 (project no. E/142-01), and from the University of
Alaska Fairbanks with funds appropriated by the state. The authors thank Mark van
der Bleek (pers. comm.) for providing the liquid smoke samples and valuable
consultations.

Conflict of Interest

None declared.
Funding Information

This publication is the result of research sponsored by USDA-CSREES, Alaska Sea

Grant and the University of Alaska Fairbanks.

References

1. Baltes W, Wittkowski R, Sochtig I, Block H, Toth L. Ingredients of smoke and

smoke flavor preparations The quality of foods and beverages. In:
Charalambous G, Inglett G, editors. New York, NY: Academic Press; 1981.
pp. 1–19.
2. Cadwallader KR. Wood smoke flavor Handbook of meat, poultry and seafood
quality. In: Nollet LML, editor. Ames, IA: Blackwell Publishing; 2007. pp. 201–
210.
3. Cermak SC, Isbell TA. Pilot-plant distillation of meadowfoam fatty acids. Ind.
Crops Prod. 2002;15:145–154.
4. Chen J, Maga J. Wood smoke composition Food flavors, ingredients, and
composition. In: Charalambous G, editor. New York, NY: Elsevier; 1993. pp.
1001–1007.
5. Clifford MN, Tang SL, Eyo AA. Smoking of foods. Proc. Biochem. 1980;15:8–
26.
6. Doores S. Antimicrobials in foods. In: Davidson PM, Branen AL, editors.
Organic Acids. New York, NY: Marcel Dekker, Inc.,; 1993. pp. 95–136.
7. Edye LA, Richards GN. Analysis of condensates from wood smoke:
components derived from polysaccharides and lignins. Environ. Sci. Technol.
1991;25:1133–1137.
8. Fiddler W, Wasserman WAE, Doerr RC. A “smoke” flavor fraction of a liquid
smoke solution. J. Agric. Food Chem. 1970;18:934–936.
9. Fujimaki M, Kim K, Kurata T. Analysis and comparison of flavor constituents
in aqueous smoke condensates from various woods. Agric. Biol. Chem.
1974;38:45–52.
10. Gibbs HD. Phenol tests: III The indophenol test. J. Biol. Chem. 1927;72:649–
664.
11. Gilbert J, Knowles ME. The chemistry of smoked foods: a review. Int. J. Food
Sci. Technol. 1975;10:245–261.
12. Guillen MD, Errecalde MC. Volatile components of raw and smoked black
bream (Brama raii) and rainbow trout (Oncorhynchus mykiss) studied by
means of solid phase microextraction and gas chromatography/mass
spectrometry. J. Sci. Food Agric. 2002;82:945–952.
13. Guillen MD, Ibargoitia ML. Volatile components of aqueous liquid smokes
Vitis vinifera L shoots and Fagus sylvatica L wood. J. Sci. Food Agric.
1996;72:104–110.
14. Guillen MD, Ibargoitia ML. New components with potential antioxidant and
organoleptic properties, detected for the first time in liquid smoke flavoring
preparations. J. Agric. Food Chem. 1998;46:1276–1285.
15. Guillen MD, Ibargoitia ML. Influence of the moisture content on the
composition of the liquid smoke produced in the pyrolysis process of Fagus
sylvatica L. wood. J. Agric. Food Chem. 1999;47:4126–4136. [PubMed]
16. Guillen MD, Manzanos MJ. Study of the components of a solid smoke
flavouring preparation. Food Chem. 1996;55:251–257.
17. Guillen MD, Manzanos MJ. Characterization of the components of a salty
smoke flavouring preparation. Food Chem. 1997;58:97–102.
18. Guillen MD, Manzanos MJ. Smoke and liquid smoke. Study of an aqueous
smoke flavouring from the aromatic plant Thymus vulgaris L. J. Sci. Food
Agric. 1999a;79:1267–1274.
19. Guillen MD, Manzanos MJ. Extractable components of the aerial parts of
Salvia lavandulifolia and composition of the liquid smoke flavoring obtained
from them. J. Agric. Food Chem. 1999b;47:3016–3027. [PubMed]
20. Guillen MD, Manzanos MJ, Zabala L. Study of a commercial liquid smoke
flavoring by means of gas-chromatography mass-spectrometry. J. Agric.
Food Chem. 1995;43:463–468.
21. Guillen MD, Manzanos MJ, Ibargoitia ML. Carbohydrate and nitrogenated
compounds in liquid smoke flavorings. J. Agric. Food Chem. 2001;49:2395–
2403. [PubMed]
22. Hollenbeck CM. Novel concepts in technology and design of machinery for
production and application of smoke in the food industry. Pure Appl. Chem.
1977;49:1687–1702.
23. Jonsdottir R, Olafsdottir G, Chanie E, Haugen JE. Volatile compounds
suitable for rapid detection as quality indicators of cold smoked salmon
(Salmo salar) Food Chem. 2008;109:184–195. [PubMed]
24. Kaseleht K, Leitner E, Paalme T. Determining aroma-active compounds in
Kama flour using SPME-GC/MS and GC-olfactometry. Flavour Frag. J.
2011;26:122–128.
25. Kim K, Kurata T, Fujimaki M. Identification of flavor constituents in carbonyl,
non-carbonyl, neutral and basic fractions of aqueous smoke condensates.
Agrc. Biol. Chem. 1974;38:53–63.
26. Kostyra E, Barylko-Pikielna N. Volatiles composition and flavour profile
identity of smoke flavourings. Food Quality Pref. 2006;17:85–95.
27. Maga JA. The flavor chemistry of wood smoke. Food Rev. Int. 1987;3:139–
183.
28. Maga J, Chen Z. Pyrazine composition of wood smoke as influenced by wood
source and smoke generation variables. Flavour Frag J. 1985;1:37–42.
29. Martin EM, O'Bryan CA, Lary RY, Griffis CL, Vaughn KLS, Marcy JA, et al.
Spray application of liquid smoke to reduce or eliminate Listeria
monocytogenes surface inoculated on frankfurters. Meat Sci. 2010;85:640–
644. [PubMed]
30. Miler KBM, Sikorski ZE. Smoking Seafood: resources, nutritional
composition, and preservation. In: Sikorski ZE, editor. Boca Raton, FL: CRC
Press; 1990. pp. 163–180.
31. Milly PJ, Toledo RT, Ramakrishnan S. Determination of minimum inhibitory
concentrations of liquid smoke fractions. J. Food Sci. 2005;70:M12–M17.
32. Neta ERD, Johanningsmeier SD, Drake MA, McFeeters RF. A chemical basis
for sour taste perception of acid solutions and fresh-pack dill pickles. J. Food
Sci. 2007;72:S352–S359. [PubMed]
33. Nicholson MD. 1984. Tar-depleted liquid smoke treatment of food casings
U.S. Patent 4431033.
34. Ojeda M, Barcenas P, Perez-Elortondo FJ, Albisu M, Guillen MD. Chemical
references in sensory analysis of smoke flavourings. Food Chem.
2002;78:433–442.
35. Plane RA, Mattick LR, Weirs LD. An acidity index for the taste of wine. Am. J.
Enol. Vitic. 1980;3:265–268.
36. Ramakrishnan S, Moeller PW. Liquid smoke: product of hardwood pyrolysis.
Abst. Papers Ame. Chem. Soc. 2002;47:366–367.
37. Rozum JJ. Introduction to smoke condensates International smoked seafood
conference proceedings. In: Kramer DE, Brown L, editors. Anchorage, AK:
Alaska Sea Grant College Program; 2007. pp. 63–64.
38. Sadler GD, Murphy PA. pH and titratable acidity Food analysis. In: Nielsen
SS, editor. New York, NY: Springer; 2003. pp. 207–225.
39. Schubring R. Use of “filtered smoke” and carbon monoxide with fish. J. Cons.
Prot. Food Safety. 2008;3:31–44.
40. Serot T, Varlet V, Cardinal M, Knockaert C, Prost C. Olfactometric
determination of the most potent odor-active compounds in salmon muscle
(Salmo salar) smoked by using four smoke generation techniques. J. Agric.
Food Chem. 2007;55:4518–4525. [PubMed]
41. Simko P. Factors affecting elimination of polycyclic aromatic hydrocarbons
from smoked meat foods and liquid smoke flavorings. Mol. Nutr. Food Res.
2005;49:637–647. [PubMed]
42. Simon R, de la Calle B, Palme S, Meier D, Anklam E. Composition and
analysis of liquid smoke flavouring primary products. J. Sep. Sci.
2005;28:871–882. [PubMed]
43. Sung WC, Huang CH, Sun FM. Volatiles components detected in liquid
smoke flavoring preparation from the smoking ingredients of smoked large
yellow croaker. Chia-nan Annu. Bull. 2007;33:21–30.
44. Svobodova D, Krenek P, Fraenkl M, Gaspari J. The colour reaction of phenols
with the Gibbs reagent. The properties of the coloured product and the
optimum reaction conditions. Microchim. Acta. 1978;70:197–211.
45. Toledo RT. International smoked seafood conference proceedings Wood
smoke components and functional properties. In: Kramer DE, Brown L,
editors. Anchorage, AK: Alaska Sea Grant College Program; 2007. pp. 55–
61.
46. Toth L, Potthast K. Chemical aspects of the smoking of meat and meat
products Advances in food research. In: Chichester CO, Mark EM, Schweigert
BS, editors. Orlando, FL: Academic Press; 1984. pp. 87–158.
47. Tucker IW. Estimation of phenols in meat and fat. J. Assoc. Off Anal. Chem.
1942;25:779–782.
48. Underwood G. 1992. High browning liquid smoke composition and method of
making a high browning liquid smoke composition. U.S. Patent 5135770.
49. U.S. Food and Drug Administration. 2003. GRN 138-GRAS Notice 000138:
Fish oil
http://www.accessdata.fda.gov/scripts/fcn/gras_notices/grn000138.pdf
(acessed 27 November 2012)
50. Varlet V, Prost C, Serot T. Volatile aldehydes in smoked fish: analysis
methods, occurence and mechanisms of formation. Food Chem.
2007;105:1536–1556.
51. Varlet V, Serot T, Prost C. Smoke flavoring technology in seafood Handbook
of seafood and seafood products analysis. In: Nollet LML, Toldra F, editors.
Boca Raton, FL: CRC Press; 2010. pp. 233–254.
52. Vitt SM, Himelbloom BH, ACrapo C. Inhibition of Listeria innocuaL.
monocytogenes in a laboratory medium and cold-smoked salmon containing
liquid smoke. J. Food Safety. 2001;21:111–125.
53. Walstra P, Wouters JTM, Geurts J. Boca Raton, FL: CRC Press; 2006. Dairy
science and technology; pp. 160–162.
54. Wasserman AE. Organoleptic evaluation of three phenols present in wood
smoke. J. Food Sci. 1966;31:1005–1010.
55. Wei Q, Ma XH, Dong JE. Preparation, chemical constituents and antimicrobial
activity of pyroligneous acids from walnut tree branches. J. Anal. Appl. Pyrol.
2010;87:24–28.
56. Yi-min W, Zeng-li Z, Hai-bin L, Fang H. Low temperature pyrolysis
characteristics of major components of biomass. J. Fuel Chem. Technol.
2009;37:427–432.
XVI. Exploring the Use of Sensorial LTP/LTD-Like
Stimulation to Modulate Human Performance for
Complex Visual Stimuli
PLoS One. 2016; 11(6): e0158312.
Felipe Pegado,1,2,* Hendrik Vankrunkelsven,1 Jean Steyaert,2,3 Bart Boets,2,3 and
Hans Op de Beeck1,*
Michael H. Herzog, Editor

Abstract

Is it possible to passively induce visual learning/unlearning in humans for complex

stimuli such as faces? We addressed this question in a series of behavioral studies
using passive visual stimulation (flickering of faces at specific temporal frequencies)
inspired by well-known synaptic mechanisms of learning: long-term potentiation
(LTP) vs long-term depression (LTD). We administered a face identity change
detection task before and after a passive stimulation protocol to test for potential
changes in visual performance. First, with bilateral stimulation, subjects undergoing
high-frequency LTP-like stimulation outperformed those submitted to low-frequency
LTD-like stimulation despite equivalent baseline performance (exp. 1). Second,
unilateral stimulation replicated the differential modulation of performance, but in a
hemifield-specific way (exp. 2). Third, for both stimulation groups, a sudden
temporary drop in performance on the stimulated side immediately after the
stimulation, followed by progressive recovering, can suggest either ‘visual fatigue’ or
‘face adaptation’ effects due to the stimulation. Fourth, we tested the life-time of
these modulatory effects, revealing they vanish after one hour delay (exp. 3). Fifth,
a control study (exp. 4) using low-level visual stimuli also failed to show longer-term
effects of sensory stimulation, despite reports of strong effects in the literature.
Future studies should determine the necessary and sufficient conditions enabling
robust long-term modulation of visual performance using this technique. This step is
required to consider further use in fundamental research (e.g., to study neural
circuits involved in selective visual processing) and potential educational or clinical
applications (e.g., inhibiting socially-irrelevant aspects of face processing in autism).

Introduction

The human brain can learn and remember complex visual patterns. These capacities
are enabled by efficient neural circuitries for processing images of objects [1,2],
places [3], printed words [4–6] and faces [7,8]. Neural circuits involved in such
processing are further fine-tuned by learning (for review, see [9]).

Neural mechanisms of learning and memory have been studied extensively for over
40 years in rodent brain slices at the synaptic level [10]. High-frequency electrical
stimulation of neural tissue increases synaptic strength (i.e., Long-Term Potentiation
[LTP]), whereas low-frequency stimulation decreases it (Long-Term Depression
[LTD]). Rodent in vivo studies [11] and also in vitro investigations in human neural
tissue [12] have further validated the LTP/LTD mechanisms as biological models
related to learning in humans by observing increased synaptic strength after
learning. More recently, evidence of LTP-like effects in humans was provided by
using Transcranial Magnetic Stimulation (TMS), a non-invasive method [13,14].
Importantly, interaction between real-life learning and the induced plasticity with
neural stimulation has also been demonstrated [15–17]. Further, evidence suggests
that it is possible to improve sensorial or motor functioning in a bottom-up way [18].
Indeed this translation of LTP/LTD principles towards non-invasive applications in
humans has provoked a growing excitement in the field [19].

Interestingly, instead of modulating a whole sensorial or motor system like in TMS

studies, one can even target specific neural processing through simple sensorial
stimulation. While some studies demonstrated short term (minutes) modulatory
effects of sensorial stimulation [20] others reported long-lasting effects (up to 10
days) for elementary visual processing (black and white bars stimuli) [21,22].
Moreover, longer stimulations (hours) can induce important cortical reorganization,
as shown in cats with peripheral nerve stimulation [23]. Given the common properties
of classical LTP/LTD in vitro and the results obtained with simple sensorial
stimulation (i.e., frequency-dependence, long-lasting effects and glutamatergic
dependency; see [24,22,25,26], both approaches seem to rely on the same synaptic
mechanisms. In accordance, learning induced by natural exposure (perceptual
learning) or by stimulation (LTP) present similar characteristics, such as stimulus-
specificity and a minimal amount of trials/stimulation required to induce learning
[25,27].

But can we passively manipulate visual learning/unlearning in humans for complex

high-level stimuli such as pictures of objects or faces? Here we explored the use of
sensorial stimulation at specific frequencies either to improve or to weaken visual
processing of faces. Our protocol consisted of a challenging face identity change
detection task which was administered three times before and three times after the
visual stimulation (see Fig 1). Visual stimulation consisted of bilateral (exp.1, 3 and
4) or unilateral (exp.2) changes of task-relevant stimuli (face identity in exp. 1, 2 and
3; simple bars in exp.4) at a high (~10Hz) or low (~1Hz) frequency rate (LTP and
LTD-like stimulation, respectively; see Methods). Subjects’ performance were re-
assessed immediately after (exp.1 and 2) or one hour after (exp.3 and 4) the
stimulation.

Experimental design.
A) Visual performance on face processing was assessed before (pre-stimulation)
and after (post-stimulation) passive visual stimulation with two different protocols
(LTP-like or LTD-like). B) Face identity change detection task used in Experiments
1–3. In a sequence of screens, relevant (face identity) and/or irrelevant (face
orientation) changes were randomly presented in each hemifield. Each condition
(“Relevant Change”, “Irrelevant Change”, “Competitive Change” and “Relevant with
Irrelevant”; see Methods) was presented an equal number of trials. One ninth of the
trials presented no changes (not shown). Note the challenging nature of the
competitive change trials where task-relevant identity changes should be detected
despite the presence of a distractor (irrelevant head-orientation change) in the
opposite hemifield. The correct response for these example trials is indicated below
each condition.

Fig 1. Experimental design.

Results

Experiment 1: Bilateral stimulation

In the first experiment, in order to maximize potential effects of stimulation on visual

performance, we administered bilateral stimulation and assessed the effect on face
processing performance immediately after stimulation. As in previous studies
[21,22], the analyses focused on the “competitive trials”, in which a relevant change
at one side was accompanied by an irrelevant change at the other side. As shown
in Fig 2A, “competitive trials” were the most difficult and were clearly far away from
ceiling performance. Error rates (ER) in competitive trials were used as the
dependent measure in a factorial Analysis of variance (ANOVA), declaring
stimulation-type (LTP-like versus LTD-like) as between-subject factor and time-point
(pre versus post-stimulation) as within-subject factor. In all experiments the first of
the pre-stimulation blocks (baseline) was used to accustom subjects with the task
and was not included in the analyses. The ANOVA revealed a significant Group x
Time interaction (F(1,21) = 12.90; p < 0.002), indicating that the LTP-like stimulation
group significantly outperformed the LTD-like group after but not before the
stimulation (see Fig 2B). Post-hoc testing revealed that the groups did not differ on
any of the baseline blocks (respectively p = 0.7, p = 0.7 and p = 0.39) but they did
on all post-stimulation blocks, with LTP outperforming the LTD-like group
(respectively p = 0.02; p < 0.01, and p = 0.02). Despite the limited number of
participants in this experiment, individual results suggest that these differential
modulation effects on visual performance were quite well reproducible at the single-
subject level (see Fig 2C).
Experiment 1 (bilateral stimulation).

A) Overall task performance: Error Rates for the ‘No change’, ‘Irrelevant Change’,
‘Relevant and Irrelevant Change’, ‘Relevant Change’, and ‘Competitive Trials’
conditions. Note that competitive trials (‘compet’) are the most difficult ones. Errors
in this condition were used to monitor the effects of LTP- vs LTD-like stimulation. B)
Performance on ‘competitive trials’ for the three pre-stimulation and three post-
stimulation blocks of the LTP-like and LTD-like groups. C) Single-subject results on
‘competitive trials’ for each stimulation group. D) Error rates on ‘competitive trials’ for
each hemifield. A main effect of hemifield together with a lack of significant
interaction with stimulation suggest a general left hemifield advantage in this face
processing task (see Results and S2 Fig for equivalent results in exp. 2 and 3). Pre-
stim = pre-stimulation (i.e., blocks 2 and 3); post-stim = post-stimulation (i.e., blocks
4, 5 and 6) (see Results). * p< 0.05; ** p< 0.01. Error bars = +/- 1 SEM across
subjects.
Fig 2. Experiment 1 (bilateral stimulation).

Note that immediately after the stimulation (block 4) there was an increase in
average error rates relative to block 3, with a significant Group x Time interaction
(F(1,21) = 6.31; p = 0.02), demonstrating that the LTD-like group presented a higher
increase in errors (15.7%) relative to the LTP-like group (2.6%). Further, when
restricting this analysis to the LTD-like group we observed a significant increase in
errors (F(1,11) = 13.9; p = 0.003) while for the LTP-like group this impoverishment
did not reach statistical significance (F(1,10) < 1). This effect can be due to visual
fatigue or visual adaptation effects (‘face adaptation’) and will be further
characterized in exp. 2 (unilateral stimulation).

Next, to verify if our task was targeting high-level visual processing we tested the
presence of a left hemifield advantage for face processing, as reported in the
literature [28]. This advantage may be due to the lateralization of the Fusiform Face
Area [FFA] predominantly on the right hemisphere [7] with contralaterally biased
receptive fields [29]. We thus added hemifield (Left versus Right) as a within-subject
factor to the previously described ANOVA. A main effect of hemifield (F(1,21) = 8.99;
p < 0.007) with significantly better performance in the left hemifield was found.
Importantly, no interaction with group (F < 1) nor triple interaction (F < 1) were
observed. Thus, these results indicate an equivalent effect of the stimulation on both
contralateral hemispheres. Indeed, when restricting the analysis to each hemifield
separately, we found a significant Group X Time interaction for the left (F(1,21) =
8.08; p = 0.0097) as well as for the right hemifield (F(1,21) = 8.48; p = 0.0083) (see
Fig 2D).

A similar left hemifield advantage was found in the two other experiments involving
face processing (exp.2 and 3) but not in the experiment with oriented bars (exp.4)
(see S2 Fig). These results concordantly suggest that our task paradigm was
targeting high-level visual processing of faces.

Experiment 2: Unilateral stimulation

We conducted a second experiment to test hemifield-specific effects of sensory

stimulation by administrating unilateral stimulation. All other aspects were kept
constant relative to exp.1, including reassessing face processing performance
immediately after the stimulation. First, to verify a potential general effect of the
stimulation on visual performance, independently of stimulation frequency, we
compared performance on the stimulated versus the non-stimulated hemifield (Fig
3A). Even more clearly than in exp. 1, we observed here a temporary drop in
performance immediately after the stimulation (block 4) relative to block 3, on the
stimulated side (F(1,88) = 38.6; p < 0.0001) which was not statistically significant on
the non-stimulated side (F(1,88) = 3.11; p = 0.08). Note that this temporary disruption
was present for both groups (see Fig 3B). As in exp.1, the LTD-like was more
affected than LTP-like (Group x Time interaction: (F(1,88) = 6.7; p = 0.01). However
in contrast to exp.1, here, not only the LTD-like but even the LTP-like group showed
a significant increase in error rates after the stimulation (respectively F(1,42) = 35; p
< 0.0001 and F(1,46) = 7.9; p = 0.007) on the stimulated side, while no effect was
found on the non-stimulated side (F < 1 for both groups). Further, by comparing
stimulated versus non-stimulated sides we found an interaction with Time for both
LTD-like (F(1,42) = 34.7; p < 0.0001) and LTP-like groups (F(1,46) = 4.5; p = 0.04).
This difference in results across experiments for the LTP-like group can possibly be
linked to the competitive nature of the task (‘competitive trials’), i.e., unilateral
reduction in discriminability (exp.2) inducing a more clear bias than bilateral
impoverishments (exp.1).

Experiment 2 (unilateral stimulation).

A) Error Rates on ‘competitive trials’ for the stimulated vs non-stimulated hemifield

(LTP-like and LTD-like stimulation collapsed). B) Error rates on ‘competitive trials’
for each stimulation group and hemifield: LTP stim = errors in the stimulated
hemifield for the LTP-like group; LTP non-stim = errors in the non-stimulated
hemifield for the LTP-like group; LTD stim = errors in the stimulated hemifield for the
LTD-like group; LTD non-stim = errors in the non-stimulated hemifield for the LTD-
like group. C) Stimulation effects outside the “performance-disrupted window” (i.e.,
block 6) compared to pre-stimulation performance (i.e. block 2 and 3). Error bars =
+/- 1 SEM across subjects.

Fig 3. Experiment 2 (unilateral stimulation).

To investigate stimulation effects outside this performance-disrupted window where

visual fatigue or face adaptation could have a strong impact we analyzed error rates
of the last block post-stimulation against the baseline. For the stimulated hemifield
(‘stim’), we observed again the differential modulation of visual performance (Group
x Time: F(1,88) = 5.9, p < 0.017), with the LTP outperforming the LTD-like group on
the post-stimulation but not on the pre-stimulation blocks. This interaction was
absent in the non-stimulated hemifield (‘non-stim’) (F < 1) (Fig 3C). In summary, after
the disruption window we could notice again passive modulation of performance
depending on the stimulation-frequency.

Experiment 3: One hour delay post-stimulation

In the third experiment, we tested the life-time of these stimulation effects by

introducing a one hour delay between the end of the stimulation and the post-
stimulation test blocks. This delay was used for three main reasons: 1) this is the
typical length of delay used in previous experiments [21,22]; 2) it may avoid the
temporary disruption interference noted in exp. 2; 3) it is long enough to allow
theoretical analogy with long-term potentiation/depression observed at the neural
level. In other words, if the LTP/LTD-like sensory stimulation would effectively induce
LTP/LTD at the neural level, we should notice modulatory effects on behavioral
performance after one hour delay.

Furthermore, in the present experiment we also included a no-stimulation condition

and an LTP-like stimulation of the task irrelevant change (orientation of the faces).

Overall, we did not find a significant differential modulation of face processing

performance by LTP vs LTD like stimulation. In particular there was no significant
Group x Time interaction (F(3,92) = 1.85, p = 0.14), suggesting that frequency-
dependent stimulation effects have vanished completely after the one hour delay
period (see Fig 4). The results for the LTP-like relevant-change stimulation were
strikingly similar to the no-stimulation condition, suggesting either no LTP-like effect
at all or a short lasting LTP-like effect (between ~15 minutes and one hour). For the
LTD-like group the pattern of results was less decisive. Despite the absence of any
overall statistical effect when taking all time points into consideration, there was a
suggestive drop in performance in the first testing block after LTD-like stimulation
(see Fig 4). When restricting the analysis to this time point (i.e. block 4), there was
a trend (F(1,93) = 2.99, p = 0.087) towards degraded performance of the LTD-like
stimulation group as compared to the three other groups. This trend already
disappeared from testing block 5 onwards. Thus even if the increased error rates
reflected a long-lasting effect of LTD-like stimulation (present one hour later), it was
transient and quickly undone after one block of performing the face processing task
(see results for blocks 5 and 6). Overall, Experiment 3 does not provide convincing
evidence that the LTP/LTD-like stimulation resulted in long-lasting frequency-
dependent changes in behavioral performance.
Fig 4. Experiment 3 (one hour delay post-stimulation).

Error rates on ‘competitive trials’ for each stimulation condition: LTP relev = high-
frequency relevant stimulation (face identity changes); LTP irrelev = high-frequency
irrelevant stimulation (head orientation changes); No Stim = control group (without
stimulation); LTD = low-frequency relevant stimulation (face identity change). See
Results. Error bars = +/- 1 SEM across subjects.

Experiment 4: Low-level stimuli and one hour delay

Finally, we performed a fourth experiment which aimed at understanding the reason

for a lack of longer-term effects (> one hour), as these have previously been reported
with very similar protocols but using low-level visual stimuli [21]. Essentially, we
wanted to disentangle two hypotheses: 1) Are stimulation effects smaller for higher
level visual processing? or 2) Is our current stimulation and test paradigm insensitive
to changes in visual performance? Consequently, we tried to replicate the original
findings of Beste and colleagues [21] while using our own protocol parameters. Thus,
we replaced the face stimuli by black and white bars, and also added a 20Hz
stimulation condition (since it was the frequency they used in the LTP-like condition),
in order to verify if the stimulation frequency rate could have played a critical role.
(See Methods for an explanation why we used 10 Hz for the face processing
experiments).
As illustrated in Fig 5, we could not replicate any long-term (1 hour delay) frequency-
dependent stimulation effects on low-level visual processing (Group x Time: F(2,59)
= 1.82, p = 0.17). It should be noted that the protocol of exp.4 still slightly differed
from the one used by Beste and colleagues in the following aspects: 1) saliency level
of the distractor: only one level here, while they used two levels (they found
differential strength and duration of stimulation effects depending on the saliency
level, and it is difficult to exactly say where saliency in our experiment falls within
their tested range); 2) number of trials before and after stimulation (432 here vs 512
in Beste et al.); 3) context of conditions: we included an additional “no-change”
condition in 1/9th of the trials); 4) distance of stimuli to fixation cross: 2.1 degrees
here vs 1.1 degree there). However, we tested 10 additional subjects with a 1.1
degree distance between stimuli and fixation cross and the performance did not
change at baseline nor at post-stimulation blocks.
Error Rates on ‘competitive trials’ for each stimulation condition: LTP 10 Hz = task-
relevant stimulation (luminance change) at 10Hz; LTP 20 Hz = task-relevant
stimulation (luminance change) at 20Hz; NoStim = no stimulation. Arrow indicates
when the stimulation was applied and followed by 1 hour delay. Error bars = +/- 1
SEM across subjects.

Fig 5. Experiment 4 (bars instead of faces; one hour delay).

Discussion

In the present work we explored the potential use of LTP/LTD-like non-invasive brain
stimulation via sensorial input to passively manipulate visual sensitivity of humans
to detect changes in highly complex stimuli (faces). We found short term (minutes)
stimulation effects in the predicted direction, with the LTP-like group outperforming
the LTD-like group (exp. 1). This performance modulation was hemifield-specific
(exp. 2), concurring with previous reports on behavioral [21] and neural data [30].
We also noticed a temporary disruption in performance just after the stimulation, only
on the stimulated side but for both types of stimulation (see Fig 3A and 3B of exp.
2). Results of both experiments suggest that these modulatory effects where present
for at least 15 minutes (see block 6), a duration compatible with previously observed
synaptic modulation using spike-timing dependent plasticity techniques on the visual
cortex of the cat [20], and also compatible with human perceptual modulation [20]
including perception of faces [31]. In complement, a third experiment (exp. 3)
demonstrated the vanishing of these effects after a 1 hour delay, suggesting that the
duration of effects found here ranged from 15 minutes to less than one hour. The
lack of longer-term effects contrasts with previous reports using similar protocols but
for low-level visual stimuli [21,30]. We thus conducted a control experiment (exp. 4)
to try to disentangle between two explanatory hypotheses for our findings: A)
reduced influence of sensory stimulation on higher level visual processing (e.g.,
effects may be restricted to low-level areas and may not propagate to higher level
areas) versus B) a lack of sensitivity of our paradigm. The results of Experiment 4
shown a lack of replication of long-term stimulation effects (see Fig 5), possibly
suggesting that the parameters of our test and/or stimulation protocols could have
been suboptimal to demonstrate long-term effects. While we can exclude the
involvement of a number of paradigm factors (see results of exp.4), the importance
of the saliency of the distractors on task sensitivity, for instance, cannot be ruled out
(cf. fig. 2 in [21]).

Overall, we can conclude that it is not straightforward to reveal long-term effects of

LTP/LTD-like sensory stimulation, neither with high-level face stimuli nor with low-
level visual stimuli. At a shorter time scale however, we could demonstrate the
predicted modulation on visual performance for face discrimination as a function of
stimulation protocols. We could also observe a temporary decrease in behavioral
performance on the stimulated side (see Fig 3A). This can be linked to “visual
fatigue” caused by the stimulation. Alternatively it can also be due to ‘visual
adaptation’ to the face stimuli used during stimulation, leading to a temporary
decrease in discriminability. Indeed, disrupted perception of faces can be observed
after face adaptation, typically presenting an exponential decay of the effect and a
duration of a few minutes [32–34], a pattern also observed in the present work. In
any case, this temporary “performance disruption” observed here can possibly shed
light on previous unexpected findings in the literature. In this regard, an fMRI study
investigating LTP-like effects on the visual system recorded reduced instead of
increased visual cortex responses relative to baseline immediately after 9Hz
stimulation [35]. The “temporary disruption” effect may have masked the LTP-like
effects in this time period, and the expected enhanced responses could have been
obtained a few minutes later, outside the “disrupted window”. Indeed, this is exactly
what was found in an EEG study using also 9Hz stimulation when subjects were
retested several minutes after stimulation (up to 52 minutes) [30]. Furthermore, a
recent EEG study using a parametric variation of delays after high-frequency
stimulation (~9Hz) found differential modulatory effects in early components (< 250
ms) of visual evoked potentials: for 2–4 and 4–6 minute delays the evoked potentials
were decreased whereas longer delays (20–22 and 120–122 minutes) revealed
increased visual responses [36]. Besides the temporary disrupted effect immediately
after the stimulation, we should also consider visual adaptation as an alternative
explanation for the differential modulation effects found in the present experiments
(exp. 1 & 2). Indeed, visual adaptation to face stimuli could have played a major role,
given that the total exposure time to faces was higher for the LTD-like compared to
the LTP-like group (see Methods). As a consequence, the former could have
suffered more face adaptation than the later, leading to higher error rates of face
discrimination after the stimulation. Note that even if total presentation time would
be the same, the differential dynamics in the two conditions could give rise to
differential adaptation. One stimulus presented for 5 seconds could elicit different
amounts of adaptation compared to 5 stimuli of 1 second long. Visual adaptation is
a complex phenomenon. Previous studies have shown that the duration of visual
adaptation effects is variable (seconds to days), with mechanisms for short and long-
term changes being dissociated [37] and occurring for both low [38] and high-level
visual stimuli, including faces [39,40]. The underling neural plasticity has been
related to changes at different levels, from cellular membrane changes [41] to
synaptic [42,43] and network modulations [44]. Thus, if visual adaptation is playing
a role here it is probably interfering at different levels of neural hierarchy.

Another point to be considered is the fact that interfering during the initial learning
phase can have complex consequences, with either suppressive or facilitatory
effects, as reported in the perceptual learning literature [45–47]. As can be seen in
the first blocks of the task (baseline), participants were still in the middle of the steep
learning phase when the stimulation was introduced. Thus, it is conceivable that the
effects of the stimulation could have been confounded with this complex interaction
during ongoing learning processes.

Conclusion and Future Perspective

Brain stimulation is getting momentum in the field of cognitive neuroscience as an

alternative tool for research and educational/clinical applications. Here we provide
evidence of short-term effects (minutes) of passive sensorial stimulation on
behavioral performance for high-level stimulus processing. Instead of interfering with
face processing via focal brain stimulation [48] we could manipulate performance of
face processing in a predictable and passive way. Future studies should determine
the critical factors to induce longer-term effects (hours, days) in order to enable a
robust use of this bottom-up method to selectively study neural circuits involved in a
particular visual processing. Given its simplicity (indeed even recreational videos or
video games could integrate the principle), this bottom-up “neuronal education”
holds the potential for a wide range of applications including therapeutic
interventions, as it has been proposed in the tactile sensory domain [18,49]. It may
be of special interest for individuals suffering from neurodevelopmental disorders,
such as autism spectrum disorder; e.g., to inhibit selective irrelevant aspects of face
processing, especially given new evidence suggesting they would be more sensitive
to LTP/LTD-like manipulations [50]. However, long-term effects of this sensory
modulation proved difficult in our study, indicating that we still have to understand
the necessary and sufficient conditions to induce such long-term effects before
considering a larger use of this technique.

Methods

Participants

284 typically developing young adults, essentially undergraduate students from the
bachelor/master programs of psychology at KU Leuven (mean age = 19.4 years, SD
= 2.2; 235 females) participated in one of the four experiments, distributed in the
following way: 26 in exp.1; 98 in exp.2; 98 in exp.3; 62 in exp.4. They were recruited
via the university online experiment system for first year students, social media and
publicity on the campus. Exclusion criteria were neurological disease, age > 27 years
old and (uncorrected) vision problems. Subjects received student credits or 8
euros/hour in vouchers. 14 subjects were excluded because at least one of the
following reasons: incomplete data, performance < 50% during stimulation or
technical problem (excluded: 3 in exp.1; 8 in exp.2; 2 in exp.3; 0 in exp.4). All
participants had normal or corrected-to- normal vision. The study was approved by
the local ethical committee of KU Leuven University (S55601). All subjects provided
written consent.

Face identity task

To monitor face processing performance we used a face identity change detection

task in exp.1, 2 and 3 (see Fig 1B). This attention competition task was modeled
after the one used by Beste et al.(2011), but using faces as stimuli instead of low-
level stimuli (bars). Two human faces were simultaneously presented on a computer
screen, right and left from the central fixation cross. Participants had to detect
changes in face identity and report the side at which they occurred by pressing the
corresponding button (left or right). Irrelevant head orientation changes were also
displayed to distract attention, requiring selective face processing (identity change
detection) to correctly perform the task. Participants were explicitly instructed to
ignore irrelevant face orientation changes. The two faces were presented for 200ms,
followed by a short interval of 50ms, followed by a new screen with two faces for
200ms (see Fig 1B). In the “Relevant Change” condition, one face identity changed
on one side of the screen. In the “Irrelevant Change” condition, only the orientation
of one face changed. In the “Competitive trials” condition, a relevant identity change
was present on one side (that should be reported) while an irrelevant orientation
change was present on the other side. This latter condition was especially
challenging due to the simultaneous presence of a target and a distractor change.
In the “Relevant with Irrelevant” condition both kinds of change occured at the same
side. In the “No Change” condition (not depicted in Fig 1), neither of the two faces
changed. All conditions were equivalently present on the left and right hemifields and
with an equal amount of trials, with exception of the “No Change” condition that only
represented 1/9th of the trials. Three blocks of 144 trials each (~ 5 minutes duration
each) were administered before stimulation (pre-stimulation) and three blocks after
stimulation (post-stimulation). The first trial of each block was not analyzed to avoid
surprise effects of the first stimulus. Conditions were explicitly shown to the subjects
prior to the task. The same procedures were applied in exp. 4 but faces were
replaced by black and white bars which changed in terms of luminance (relevant
change) and orientation (irrelevant change) (cf. Beste et al., 2011). Correctness of
responses was favored instead of speediness. In all four experiments, Response
Time (RT) measures were not sensitive to the stimulation manipulation.

Stimulation protocols

After obtaining baseline (pre-stimulation) measures on the face identity task (three
blocks), participants underwent one of the two visual stimulation protocols (LTP-like
versus LTD-like) for 40 minutes (10 mini-blocks of 4 minutes). During stimulation, for
all four experiments, participants performed an orthogonal task, by detecting subtle
and fast shifts of the central fixation cross to the left or right side of the screen. This
task was used to guarantee that participants were fixating correctly and thus
receiving the stimulation at the same location on their retina but also to insure they
were not paying attention to the peripheral flickering of stimuli. Note that it was an
engaging task since we used subtle fixation shifts. In experiment 1, 3 and 4, subjects
received bilateral stimulation. In experiment 2, they received unilateral stimulation
on only one hemifield (left or right) during the entire stimulation protocol while the
non-stimulated side presented no-stimulus or a static one (counterbalanced across
participants).

The stimulation consisted of face switches between the same pair of stimuli used in
the face identity task (only two different faces were used both in stimulation and task;
see Fig 1B) at different rates. The LTP-like group received stimulus changes at ~10
cycles/second (10.6 with 85Hz monitor for exp.1 & 2; 10.0 with 100Hz monitor for
exp.3 & 4) during 5 seconds-on and 5 seconds off. The LTD-like group received
stimulation at approximately 1.0 cycle/second, presented continuously. These
protocol parameters were chosen given the previously reported effective induction
of LTP and LTD-like effects on visual performance [21]. Note however that for high-
frequency stimulation we preferred to use 10Hz instead of 20Hz for two main raisons:
1) high-level visual areas process visual information in lower frequency ranges than
low-level visual areas [51,52] and 2) to avoid perceptual fusion between the two face
identities when increasing the flickering rate above 10Hz. For each 10 seconds
period, the head position (frontal or lateral view) was fixed and only faces identity
changed. We used both head positions during the stimulation, interleaved across 10
seconds periods, for all stimulation groups.

Two additional groups were included in exp.3: “LTP-irrelevant” where the stimulation
targeted the irrelevant distraction change (head position), i.e., flickering of the same
face in the two head positions; and “No Stim” group, without stimulation but
performing the same orthogonal task on the fixation-cross.

In exp.4, the stimulation consisted of bilateral luminance changes (black and white)
of the bars. As in previous experiments, we used both orientations of the bars,
interleaved across 10 seconds periods.

Apparatus

The visual stimuli were presented with 85Hz (exp.1 and 2) or 100 Hz (exp.3 and 4)
CRT monitors, 15,6”large, placed at approximately 85cm from the participant in a
dark room. Chin fixation was applied in order to keep the distance between the eyes
and the screen constant and to guarantee that stimuli arrived at the same retinal
location for both the task and stimulation. Each face stimulus subtended 7.3 x 4.6
degrees of visual angle (vertically and horizontally respectively). Stimuli were
presented bilaterally and were equidistant from the central fixation cross of
approximately 2.1 degrees. Responses were recorded with a standard keyboard.
Psychtoolbox [53,54] a free toolbox of Matlab (The MathWorks Inc., Natick, MA,
2000) was used to deliver stimuli for the task and stimulation. Statistics were
performed in R software (www.r-project.org). Face stimuli were created with
FaceGen software (http://facegen.com).

Supporting Information

S1 Fig
Error rates (ER) and Response Times (RT) for experiments 1–4.

See Fig 2A for ER of Experiment 1. Conditions: NoChang = ‘No change’; Irrel =

‘Irrelevant Change’; RelIrrel = ‘Relevant and Irrelevant Change’; Relev = ‘Relevant
Change’; Compet = ‘Competitive Trials’. Error bars = +/- 1 SEM across subjects.

(TIFF)

Click here for additional data file.(820K, tiff)

S2 Fig
Main effects of Hemifield.
In all experiments using faces (Exp. 1, 2 & 3) a significant left hemifield advantage
was noticed (p < 0.01 for all), a reported finding in face processing studies, probably
linked to the location of Face Fusiform Area on the contralateral hemisphere. In
contrast, for Exp.4 using low-level stimuli (black and white bars), no hemifield
advantage was noticed. Error bars = +/- 1 SEM across subjects.

(TIFF)

Click here for additional data file.(219K, tiff)

Acknowledgments

We would like to thank Silke Janssens, Julia Mettner and Ilse Delmaere for their help
in data collection, Nicky Daniels and Mathijs Franssen for support at the
organizational level and Detlef Balschun, Marta Bovet-Carmona and Tariq Ahmed
for useful discussions.

Funding Statement

This paper was supported by the following grant(s):

Fonds Wetenschappelijk Onderzoek 12Q4615N to Felipe Pegado.

KU Leuven ZKC5610 / F+/12/029 to Felipe Pegado.
KU Leuven 3H120620 to Felipe Pegado.
European Research Council ERC-2011-Stg-284101 to Hans Op de Beeck.
Interuniversity Attraction Poles IUAP P7/11 to Hans Op de Beeck.
Interdisciplinary Research Fund IDO/10/003 to Hans Op de Beeck.
Interdisciplinary Research Fund IDO/10/003 to Bart Boets.
Interdisciplinary Research Fund IDO/10/003 to Jean Steyaert.

F.P. was funded by KU Leuven University (ZKC5610, F+/12/029, 3H120620) and

FWO (Fonds Wetenschappelijk Onderzoek) postdoc fellowship (12Q4615N). H.O.B.
was supported by ERC-2011-Stg-284101 and IUAP P7/11. B.B., J.S. and H.O.B.
were supported by Interdisciplinary Research Fund (IDO/10/003). The funders had
no role in study design, data collection and analysis, decision to publish, or
preparation of the manuscript.

Data Availability

All relevant data are within the paper and its Supporting Information files. Data is
available in the following link: https://osf.io/ja24u/.
References

1. Op de Beeck HP, DiCarlo JJ, Goense J, Grill-Spector K, Papanastassiou A,

Tanifuji M, et al. Fine-scale spatial organization of face and object selectivity in the
temporal lobe: do functional magnetic resonance imaging, optical imaging, and
electrophysiology agree? J Neurosci. 2008;28: 11796 doi:
10.1523/JNEUROSCI.3799-08.2008 [PubMed]
2. Op de Beeck HP, Torfs K, Wagemans J. Perceived shape similarity among
unfamiliar objects and the organization of the human object vision pathway. J
Neurosci Off J Soc Neurosci. 2008;28: 10111–10123. doi:
10.1523/JNEUROSCI.2511-08.2008 [PubMed]
3. Dilks DD, Julian JB, Kubilius J, Spelke ES, Kanwisher N. Mirror-Image Sensitivity
and Invariance in Object and Scene Processing Pathways. J Neurosci. 2011;31:
11305–11312. doi: 10.1523/JNEUROSCI.1935-11.2011 [PMC free article]
[PubMed]
4. Dehaene S, Pegado F, Braga LW, Ventura P, Nunes Filho G, Jobert A, et al. How
learning to read changes the cortical networks for vision and language. Science.
2010;330: 1359–1364. doi: 10.1126/science.1194140 [PubMed]
5. Pegado F, Nakamura K, Cohen L, Dehaene S. Breaking the symmetry: Mirror
discrimination for single letters but not for pictures in the Visual Word Form Area.
NeuroImage. 2011;55: 742–749. doi: 10.1016/j.neuroimage.2010.11.043 [PubMed]
6. Pegado F, Comerlato E, Ventura F, Jobert A, Nakamura K, Buiatti M, et al. Timing
the impact of literacy on visual processing. Proc Natl Acad Sci U S A. 2014;111:
E5233–5242. doi: 10.1073/pnas.1417347111 [PMC free article] [PubMed]
7. Kanwisher N, McDermott J, Chun MM. The fusiform face area: a module in human
extrastriate cortex specialized for face perception. J Neurosci. 1997;17: 4302
[PubMed]
8. Goesaert E, Op de Beeck HP. Representations of facial identity information in the
ventral visual stream investigated with multivoxel pattern analyses. J Neurosci Off J
Soc Neurosci. 2013;33: 8549–8558. doi: 10.1523/JNEUROSCI.1829-12.2013
[PubMed]
9. Op de Beeck HP, Baker CI. The neural basis of visual object learning. Trends
Cogn Sci. 2010;14: 22–30. doi: 10.1016/j.tics.2009.11.002 [PMC free article]
[PubMed]
10. Bliss TVP, Lømo T. Long-lasting potentiation of synaptic transmission in the
dentate area of the anaesthetized rabbit following stimulation of the perforant path.
J Physiol. 1973;232: 331–356. [PMC free article] [PubMed]
11. Morris RG, Anderson E, Lynch GS, Baudry M. Selective impairment of learning
and blockade of long-term potentiation by an N-methyl-D-aspartate receptor
antagonist, AP5. Nature. 1986;319: 774–776. doi: 10.1038/319774a0 [PubMed]
12. Chen WR, Lee S, Kato K, Spencer DD, Shepherd GM, Williamson A. Long-term
modifications of synaptic efficacy in the human inferior and middle temporal cortex.
Proc Natl Acad Sci U S A. 1996;93: 8011–8015. [PMC free article] [PubMed]
13. Stefan K, Kunesch E, Cohen LG, Benecke R, Classen J. Induction of plasticity
in the human motor cortex by paired associative stimulation. Brain J Neurol.
2000;123 Pt 3: 572–584. [PubMed]
14. Huang Y-Z, Edwards MJ, Rounis E, Bhatia KP, Rothwell JC. Theta burst
stimulation of the human motor cortex. Neuron. 2005;45: 201–206. doi:
10.1016/j.neuron.2004.12.033 [PubMed]
15. Ziemann U. Learning Modifies Subsequent Induction of Long-Term Potentiation-
Like and Long-Term Depression-Like Plasticity in Human Motor Cortex. J Neurosci.
2004;24: 1666–1672. doi: 10.1523/JNEUROSCI.5016-03.2004 [PubMed]
16. Cantarero G, Tang B, O’Malley R, Salas R, Celnik P. Motor learning interference
is proportional to occlusion of LTP-like plasticity. J Neurosci Off J Soc Neurosci.
2013;33: 4634–4641. doi: 10.1523/JNEUROSCI.4706-12.2013 [PMC free article]
[PubMed]
17. Nabavi S, Fox R, Proulx CD, Lin JY, Tsien RY, Malinow R. Engineering a
memory with LTD and LTP. Nature. 2014; doi: 10.1038/nature13294 [PMC free
article] [PubMed]
18. Smith PS, Dinse HR, Kalisch T, Johnson M, Walker-Batson D. Effects of
repetitive electrical stimulation to treat sensory loss in persons poststroke. Arch Phys
Med Rehabil. 2009;90: 2108–2111. doi: 10.1016/j.apmr.2009.07.017 [PubMed]
19. Cooke SF, Bliss TVP. Plasticity in the human central nervous system. Brain.
2006;129: 1659–1673. doi: 10.1093/brain/awl082 [PubMed]
20. Fu Y-X, Djupsund K, Gao H, Hayden B, Shen K, Dan Y. Temporal specificity in
the cortical plasticity of visual space representation. Science. 2002;296: 1999–2003.
[PubMed]
21. Beste C, Wascher E, Güntürkün O, Dinse HR. Improvement and Impairment of
Visually Guided Behavior through LTP-and LTD-like Exposure-Based Visual
Learning. Curr Biol. 2011; [PubMed]
22. Beste C, Wascher E, Dinse HR, Saft C. Faster Perceptual Learning through
Excitotoxic Neurodegeneration. Curr Biol. 2012;22: 1914–1917. doi:
10.1016/j.cub.2012.08.012 [PubMed]
23. Recanzone GH, Allard TT, Jenkins WM, Merzenich MM. Receptive-field changes
induced by peripheral nerve stimulation in SI of adult cats. J Neurophysiol. 1990;63:
1213–1225. [PubMed]
24. Dinse HR, Ragert P, Pleger B, Schwenkreis P, Tegenthoff M. Pharmacological
modulation of perceptual learning and associated cortical reorganization. Science.
2003;301: 91–94. doi: 10.1126/science.1085423 [PubMed]
25. Cooke SF, Bear MF. Visual Experience Induces Long-Term Potentiation in the
Primary Visual Cortex. J Neurosci. 2010;30: 16304–16313. doi:
10.1523/JNEUROSCI.4333-10.2010 [PMC free article] [PubMed]
26. Clapp WC, Eckert MJ, Teyler TJ, Abraham WC. Rapid visual stimulation induces
N-methyl-D-aspartate receptor-dependent sensory long-term potentiation in the rat
cortex. Neuroreport. 2006;17: 511–515. doi: 10.1097/01.wnr.0000209004.63352.10
[PubMed]
27. Aberg KC, Herzog MH. About similar characteristics of visual perceptual learning
and LTP. Vision Res. 2012;61: 100–106. doi: 10.1016/j.visres.2011.12.013
[PubMed]
28. Anderson E, Parkin AJ. On the nature of the left visual field advantage for faces.
Cortex J Devoted Study Nerv Syst Behav. 1985;21: 453–459. [PubMed]
29. Hemond CC, Kanwisher NG, Op de Beeck HP. A Preference for Contralateral
Stimuli in Human Object- and Face-Selective Cortex. PLoS ONE. 2007;2 doi:
10.1371/journal.pone.0000574 [PMC free article] [PubMed]
30. Teyler TJ, Hamm JP, Clapp WC, Johnson BW, Corballis MC, Kirk IJ. Long-term
potentiation of human visual evoked responses. Eur J Neurosci. 2005;21: 2045–
2050. doi: 10.1111/j.1460-9568.2005.04007.x [PMC free article] [PubMed]
31. McMahon DBT, Leopold DA. Stimulus Timing-Dependent Plasticity in High-Level
Vision. Curr Biol. 2012;22: 332–337. doi: 10.1016/j.cub.2012.01.003 [PMC free
article] [PubMed]
32. Leopold DA, Rhodes G, Muller K-M, Jeffery L. The dynamics of visual adaptation
to faces. Proc R Soc B Biol Sci. 2005;272: 897–904. doi: 10.1098/rspb.2004.3022
[PMC free article] [PubMed]
33. Rhodes G, Jeffery L, Clifford CWG, Leopold DA. The timecourse of higher-level
face aftereffects. Vision Res. 2007;47: 2291–2296. doi:
10.1016/j.visres.2007.05.012 [PubMed]
34. Carbon C-C, Leder H. Face adaptation: Changing stable representations of
familiar faces within minutes. Adv Cogn Psychol. 2005;1: 1–7.
35. Clapp WC, Hamm JP, Kirk IJ, Teyler TJ. Translating Long-Term Potentiation
from Animals to Humans: A Novel Method for Noninvasive Assessment of Cortical
Plasticity. Biol Psychiatry. 2012;71: 496–502. doi: 10.1016/j.biopsych.2011.08.021
[PMC free article] [PubMed]
36. Forsyth JK, Bachman P, Mathalon DH, Roach BJ, Asarnow RF. Augmenting
NMDA receptor signaling boosts experience-dependent neuroplasticity in the adult
human brain. Proc Natl Acad Sci U S A. 2015;112: 15331–15336. doi:
10.1073/pnas.1509262112 [PMC free article] [PubMed]
37. Bao M, Engel SA. Distinct mechanism for long-term contrast adaptation. Proc
Natl Acad Sci. 2012;109: 5898–5903. doi: 10.1073/pnas.1113503109 [PMC free
article] [PubMed]
38. Vul E, Krizay E, MacLeod DIA. The McCollough effect reflects permanent and
transient adaptation in early visual cortex. J Vis. 2008;8: 4–4. doi: 10.1167/8.12.4
[PubMed]
39. Carbon C-C, Ditye T. Sustained effects of adaptation on the perception of familiar
faces. J Exp Psychol Hum Percept Perform. 2011;37: 615–625. doi:
10.1037/a0019949 [PubMed]
40. Oruc I, Barton JJS. Adaptation improves discrimination of face identity. Proc R
Soc B Biol Sci. 2011;278: 2591–2597. doi: 10.1098/rspb.2010.2480 [PMC free
article] [PubMed]
41. Sanchez-Vives MV, Nowak LG, McCormick DA. Membrane Mechanisms
Underlying Contrast Adaptation in Cat Area 17In Vivo. J Neurosci. 2000;20: 4267–
4285. [PubMed]
42. Chance FS, Nelson SB, Abbott LF. Synaptic depression and the temporal
response characteristics of V1 cells. J Neurosci Off J Soc Neurosci. 1998;18: 4785–
4799. [PubMed]
43. Carandini M, Heeger DJ, Senn W. A Synaptic Explanation of Suppression in
Visual Cortex. J Neurosci. 2002;22: 10053–10065. [PubMed]
44. Ahmed B, Allison JD, Douglas RJ, Martin KA. An intracellular study of the
contrast-dependence of neuronal activity in cat visual cortex. Cereb Cortex. 1997;7:
559–570. [PubMed]
45. Aberg KC, Tartaglia EM, Herzog MH. Perceptual learning with Chevrons requires
a minimal number of trials, transfers to untrained directions, but does not require
sleep. Vision Res. 2009;49: 2087–2094. doi: 10.1016/j.visres.2009.05.020 [PubMed]
46. Wright BA, Sabin AT. Perceptual learning: how much daily training is enough?
Exp Brain Res. 2007;180: 727–736. doi: 10.1007/s00221-007-0898-z [PubMed]
47. Ashley S, Pearson J. When more equals less: overtraining inhibits perceptual
learning owing to lack of wakeful consolidation. 2012. doi: 10.1098/rspb.2012.1423
[PMC free article] [PubMed]
48. Romanska A, Rezlescu C, Susilo T, Duchaine B, Banissy MJ. High-Frequency
Transcranial Random Noise Stimulation Enhances Perception of Facial Identity.
Cereb Cortex N Y N 1991. 2015;25: 4334–4340. doi: 10.1093/cercor/bhv016 [PMC
free article] [PubMed]
49. Kattenstroth J-C, Kalisch T, Kowalewski R, Tegenthoff M, Dinse HR. Quantitative
assessment of joint position sense recovery in subacute stroke patients: a pilot
study. J Rehabil Med. 2013;45: 1004–1009. doi: 10.2340/16501977-1225 [PubMed]
50. Oberman L, Eldaief M, Fecteau S, Ifert-Miller F, Tormos JM, Pascual-Leone A.
Abnormal modulation of corticospinal excitability in adults with Asperger’s syndrome:
Modulation of excitability in Asperger’s. Eur J Neurosci. 2012;36: 2782–2788. doi:
10.1111/j.1460-9568.2012.08172.x [PMC free article] [PubMed]
51. Liu J, Wandell BA. Specializations for Chromatic and Temporal Signals in Human
Visual Cortex. J Neurosci. 2005;25: 3459–3468. doi: 10.1523/JNEUROSCI.4206-
04.2005 [PubMed]
52. Stigliani A, Weiner KS, Grill-Spector K. Temporal Processing Capacity in High-
Level Visual Cortex Is Domain Specific. J Neurosci Off J Soc Neurosci. 2015;35:
12412–12424. doi: 10.1523/JNEUROSCI.4822-14.2015 [PMC free article]
[PubMed]
53. Brainard DH. The Psychophysics Toolbox. Spat Vis. 1997;10: 433–436.
[PubMed]
54. Pelli DG. The VideoToolbox software for visual psychophysics: transforming
numbers into movies. Spat Vis. 1997;10: 437–442. [PubMed]
XVII. Effect of Radio Frequency Heating on Yoghurt, I:
Technological Applicability, Shelf-Life and Sensorial
Quality
Foods. 2014 Jun; 3(2): 318–335.
Caroline Siefarth,1,2 Thi Bich Thao Tran,2 Peter Mittermaier,2 Thomas Pfeiffer,2 and
Andrea Buettner1,2,*

Abstract

This first part of a two-part study focuses on the technical feasibility of applying radio
frequency (RF) heating at different temperatures (58, 65 and 72 °C) to a stirred
yoghurt gel after culturing. For comparison, a convectional (CV) heating process was
also applied. The aim was to increase the yoghurt shelf-life, by preventing post-
acidification and the growth of yeasts and molds. At the same time, the viability of
lactic acid bacteria (LAB) was investigated in view of existing legal regulations for
yoghurts. Additionally, the yoghurt color, aroma and taste profiles were evaluated. It
was found that the application of RF heating was effective for the rapid attainment
of homogenous temperatures of 58 and 65 °C, respectively. For RF heating at 72
°C, it was not possible to establish a stable heating regime, since in some cases,
there was significant overheating followed by strong contraction of the yoghurt curd
and whey separation. Hence, it was decided not to continue with the RF heating
series at 72 °C. In the case of CV heating, heat transfer limitations were observed,
and prolonged heating was required. Nevertheless, we showed that yeasts and
molds survived neither the RF nor CV heat treatment. LAB were found not to survive
the CV treatment, but these beneficial microorganisms were still present in reduced
numbers after RF heating to 58 and 65 °C. This important observation is most likely
related to the mildness of RF treatment. While post-acidification was not observed
on yoghurt storage, slight color changes occurred after heat treatment. The flavor
and taste profiles were shown to be similar to the reference product. Furthermore, a
trained sensory panel was not able to distinguish between, for example, the
reference yoghurt and the RF 65 °C sample by triangular testing (α = 5%), showing
the potential of novel strategies for further improvements of heat-treated yoghurt.

Keywords: heating, pH, radio frequency, sensory, shelf-life, storage, yoghurt

1. Introduction

Thermal treatment is a common and important strategy in the dairy industry for
inactivating microorganisms and enzymes and, thus, guaranteeing safe products
throughout the predicted shelf-life. However, traditional thermal treatments rely on
heat transfer by conduction and convection, resulting in relatively long heating-up
times, depending on the respective food matrix. These limitations can lead to strong
physicochemical changes within the product, resulting in sensorial and textural
modifications, as well as potentially decreasing the nutritive value. Hence, the dairy
industry is always searching for new technologies. Over the last few decades, new
technologies have been described in scientific publications, but many have not been
broadly transferred to manufacturing processes in the food industry. One of these
techniques is the radio frequency (RF) heating of foods with common frequencies of
13.56 MHz and 27.12 MHz. RF heating was first described in the middle of the last
century in the context of thawing and curing meat [1,2,3]. The advantage of
electromagnetic heating is its ability to generate heat inside the food material by
orientation polarization of dipoles, such as water, or the forced movement of ions [4].
In this way, the limitations of conventional heat transfer and heat diffusion are
overcome and very rapid heating becomes possible, at least in principle. However,
RF heating installations in most cases use very high voltages and are prone to
electric flashovers. More recently, RF heating was applied to bottled or packaged
food in a water bath equipped with electrodes, as described by Bach [5] and Felke
et al. [6]. By using water, with its high relative permittivity or so-called relative
dielectric constant (εr), instead of air as the dielectric field transfer medium between
electrodes and food packages, the electrode voltage can be much reduced without
reducing the heating rate. This fact minimizes the risk of electric flashovers.
Moreover, the εr of water is closer to that of food materials than the εr of air. Exposing
food to the electric field in a dielectric environment with similar εr, field concentrations
at the edges and corners of packages, which cause local overheating, can be
avoided and heating is more uniform. In addition, the preheated water acts as a
thermal buffer and provides additional temperature uniformity to the surface of the
food package. In order to avoid the absorption of RF energy by the water, de-ionized
water is used.

Microwave heating has also been used to heat milk products, and this has resulted
in improved sensorial characteristics compared to conventional heating on a
laboratory scale [7]. However, microwave fields with the commonly used frequency
of 2.45 GHz have a limited ability to penetrate larger food volumes. The limited
penetration is described by the penetration depth parameter. This is the distance
from the food surface into the food volume at which 63% of the electromagnetic
power has already been absorbed. The penetration depth of microwave fields of 2.45
GHz into milk or yoghurt is in the order of 1 cm, while, according to Felke et al. [6],
in the case of 27.12 MHz RF fields, the penetration depths into food materials is
about 20 cm and, thus, leads to the more uniform heating of products of a larger
diameter. A further disadvantage of microwave fields compared to 27.12 MHz RF
fields is the presence of patterns of constructive and destructive wave interference
inside the food, which leads to patterns of hot and cold spots. In the case of RF
fields, the waves are so long, that interference is not relevant on the scale of a food
item or a food package.

Due to the demand of consumers for minimally processed, but safe foods, fuelling
the need for constant technological progress, heating using radio frequencies was
chosen as a promising technological approach for the present study. The aim was
to perform a feasibility study on the technological applicability of RF heating on
stirred yoghurt gels after culturing.

Generally, yoghurt is consumed not only because of its high nutritive value and
appealing organoleptic properties, but also due to its health-promoting effects.
Especially, the living microflora of lactic acid bacteria has a positive influence on
human digestion (probiotic effect). Moreover, bioactive peptides that are released,
e.g., during fermentation as a result of the enzymatic cleavage of milk proteins, are
known to have various effects on human health (biogenic effect) [8]. According to
the Code of Federal Regulations [9], yoghurt may be heat-treated to destroy viable
microorganisms to extend shelf-life. Nevertheless, if dairy ingredients are heat-
treated after culturing, then the name of the food must be followed by the
parenthetical phrase “heat treated after culturing”. To increase the shelf-life of
yoghurt products beyond three to four weeks at refrigeration temperatures, Kessler
[10] describes several temperature ranges that are sufficient for the pasteurization
of the yoghurt curds: 65 to 75 °C and holding times of 30 to 60 s. Based on this, a
comparable approach was chosen in the present study for post-heating stirred
yoghurt filled in glass jars. Three temperature regimes were tested: 58, 65 and 72
°C. In detail, we aimed to prolong the shelf-life by mild heat treatment using an
adapted heating technology, reducing the microbial numbers, while, at the same
time, maintaining the yoghurt’s sensorial and textural profiles as close as possible to
those of the stirred yoghurt reference. RF heating was expected to result in shorter
heating-up times compared to convectional (CV) heating and, thus, to impart less
damage or deterioration with respect to favorable aspects, such as the sensory
properties of the products. A further goal of our study was to determine whether RF
heating can achieve the same required homogeneous temperature distribution in the
yoghurt curd, as expected from CV heating. Accordingly, for comparison, the
yoghurts were also heated via CV treatment using steam as the heat carrier medium.
The properties of the treated products (changes in pH, color and sensory properties)
were then compared to the stirred yoghurt reference without any additional heat
treatment. Additionally, in view of the microbial aspects and the related effects on
yoghurt shelf-life, potential changes in microbial numbers (lactic acid bacteria (LAB),
yeasts and molds) were investigated.

A second, separate part of this feasibility study will focus on the changes in
microstructure and texture caused by a post-fermentative heat treatment [11].

2. Experimental Section

2.1. Yoghurt

Commercially available plain stirred yoghurt (4.4% protein, 5.4% carbohydrates,

3.8% fat) was purchased from a local supermarket. The yoghurt was filled in 500 mL
glass jars, which were closed with a metal screw cap (twist-off, 70 mm) and were
filled up to about 100 mm. The jars had a diameter of 87 mm. All yoghurts were
purchased at the same time and had the same best-before date, which was four
weeks after purchase. According to the manufacturer disclosure, the yoghurts had a
shelf-life of four weeks and, thus, were delivered immediately after manufacturing.

For further investigations, a set-style yoghurt reference (3.8% fat, 3.4% protein, 4.4%
carbohydrates), matured in a cup (150 mL), was purchased from a local supermarket
and stored with the other yoghurt samples.

2.2. Heat Treatment

Post-fermentative heat-treatments were performed within three days after the

samples were purchased. Prior to heating, the yoghurt samples were pre-tempered
to a controlled starting temperature of 40 ± 1 °C. The following target temperatures
were applied: 58, 65 and 72 °C. Immediately after heating, the glass jars were placed
in an ice water bath for rapid cooling. Before and after heat treatment (Week 0), the
samples were placed into a cooling chamber at refrigeration temperatures of 8 ± 1
°C up to a storage period of five weeks. Thus, the storage period exceeded the
yoghurt best-before date by one week. In weekly cycles, various quality parameters
were investigated, as explained in the respective sections (Section 2.3, Section 2.4,
Section 2.5 and Section 2.6). Samples without any additional heating step were held
as references for comparative investigations.
2.2.1. Radio Frequency (RF) Treatment

RF heating of yoghurt was performed in an RF water bath on pilot scale, as displayed

in Figure 1. RF power was provided by a tube generator with 27.12 MHz operating
frequency and 16 kW rated power (Type 16000 K, Kiefel AG, Freilassing, Germany)
together with an impedance matching network by the same manufacturer. The power
flow to the products was controlled by automatically adjusting the electrode voltage,
which was measured directly at the electrodes. Applied electrode voltages during
the yoghurt heating trials were between 2.2 and 2.4 kV; with resulting field intensities
in the water bath between 17.8 kV·m−1 and 19.2 kV·m−1. By the additional control of
exposure time, the average temperature in the product at the end of RF heating
could be determined with a tolerance of ±2 °C.

Figure 1. Schematic set-up of the RF water bath, modified from Felke et al. [6].

Heating procedure: Two glass jars with tightly closed screw caps placed on a support
were immersed in the water bath under atmospheric pressure (Figure 2). The water
bath was preheated to a temperature slightly above the process target temperature.
RF exposure started immediately after immersion and continued for a preset time.
The electrode voltages and the exposure times had been determined in previous
heating experiments according to the respective target temperatures. Samples for
further microbiological, structural, textural and sensory evaluation were exposed to
an additional temperature holding time and were not subjected to inline-temperature
measurements. However, to guarantee temperature control, three to four spare
samples were opened for supervisory purposes during a sample heating series. All
details on the specific heating parameters used to produce the product samples are
compiled in Table 1.

Figure 2. Experimental set-up: Yoghurt jars on a support immersed between the

electrodes of the RF water bath (here: inline-temperature measurement with fiber-
optic temperature sensors).

Table 1. Heating procedure of RF heating in an RF water bath.

Process target Electrode RF exposure Holding time 1 Water bath

temperature voltage time (RF switched off) temperature

58 °C 2.3 ± 0.1 kV 60 s 60 s 63 ± 1 °C
Process target Electrode RF exposure Holding time 1 Water bath

temperature voltage time (RF switched off) temperature

65 °C 2.3 ± 0.1 kV 90 s 60 s 70 ± 1 °C

72 °C 2.3 ± 0.1 kV 120 s 60 s 77 ± 2 °C

1 At the process target temperature.

Temperature measurements: In previous heating experiments, inline-measurements

with fiber-optic temperature sensors (Fotemp 4 fiber-optical thermometer with TS5
0.5-mm diameter sensors, Optocon AG, Dresden, Germany) were performed to
adjust the electrode voltage and the exposure times according to the respective
target temperatures. The fibers were introduced through septa glued onto holes,
which were drilled into the metal screw caps of the jars. Temperatures were
measured in the center of the jars, as well as near the walls, at mid-height of the
filled yoghurt mass.

During the main series of the heating experiments, manual temperature

measurements were performed directly after heating with a fast response
thermocouple (0.5 mm diameter, type K, Thermocoax GmbH, Stapelfeld, Germany)
connected to an electronic thermometer (Ebro TFN 520-SMP, Ebro GmbH,
Ingolstadt, Germany). Temperatures were measured at different height levels within
the yoghurt mass in the jars. In addition, the average temperature of the yoghurt
filling was measured after mixing with a plastic spoon, because of its low thermal
capacity.

The electric conductivity (mS·cm−1) of the stirred reference yoghurt was measured
with a laboratory conductometer (Innolab Cond level 2, WTW GmbH, Weilheim,
Germany) within a temperature range of 25 to 55 °C.

2.2.2. Convectional (CV) Treatment

CV heating of yoghurts was performed in a convection oven (Rational white

efficiency SCC WE61, Rational AG, Landsberg am Lech, Germany) in steam mode
with 100% saturated steam.

Heating procedure: Five glass jars with tightly closed screw caps were placed on a
fence that was positioned in the center of the convection oven. The oven
temperature was set to slightly above the process target temperatures. An inline-
thermocouple was used to control the products temperature, with its tip placed in the
core of one additional yoghurt jar. Therefore, the lid of the jar was perforated to yield
a center hole; this temperature control sample was discarded afterwards. The
samples were exposed to CV heating until the process target temperature was
reached and an additional temperature holding time was maintained. All details on
the respective CV heating parameters used to produce the product samples are
compiled in Table 2.

Table 2. Heating procedure of convectional (CV) heating in a convection oven

(100% saturated steam).

Process target CV exposure Oven/ steam

Holding time 1
temperature time Temperature

58 °C 60.0 ± 0.5 min 60 s 63 °C

65 °C 61.5 ± 4.5 min 60 s 70 °C

72 °C 58.0 ± 0.5 min 60 s 77 °C

1 At process target temperature.

Temperature measurements: In addition to the inline-thermocouple, the entire

temperature profile was recorded once by the use of a fast response thermocouple
(0.5 mm diameter, type K, Thermocoax GmbH, Stapelfeld, Germany), placed in the
core of the product and connected to a data logging instrument (ALMEMO 2890-9,
Ahlborn Mess- und Regelungstechnik GmbH, Holzkirchen, Germany).

2.3. Microbiological Investigations

The microbial numbers of yeasts and molds, as well as lactic acid bacteria (LAB)
were investigated directly after (post-)manufacturing of the different yoghurt samples
(Week 0) and after a storage period of five weeks. No further inoculation of the
samples with LAB or other microorganisms was performed, and the jars were
opened just immediately before the samples were taken for further microbial
investigations. For the preparation of sample dilutions, 10 g of yoghurt were
dissolved in 90 mL of Ringer’s solution (25%, OXOID LTD., Hampshire, U.K.).
Ringer’s solution was also used for any further dilution steps. In detail, for each
treatment and temperature regime, three independent yoghurt jars were randomly
selected and three dilution steps were prepared in threefold repetition to analyze
microbial numbers.

Yeasts and molds: yeast extract glucose chloramphenicol (YGC) agar (Merck KGaA)
was selected as the medium for the viability investigation. The incubation was
performed at 25 °C aerobically for three to five days. LAB: de Man, Rogosa and
Sharpe (MRS) agar (Merck KGaA) was used for enumeration, and aerobic
incubation was performed at 37 °C for 72 h. Microbial numbers were expressed as
colony forming units (CFU) g−1.

2.4. pH and Color Measurement

The determination of the pH of the yoghurt samples was performed over the entire
storage period (Weeks 0, 2, 4 and 5) using a pH meter (pH 538, WTW GmbH,
Weilheim, Germany) together with a pH electrode (BlueLine 11pH, SI Analytics
GmbH, Mainz, Germany) and temperature electrode (TFK 325, WTW GmbH).
Instrumental color analysis was carried out directly after manufacturing (Week 0) by
the use of a Chroma-meter CR-300 (Konika Minolta Inc., Marunouchi, Japan) with a
DP-301 data processor. Calibration was performed on a white standard (CR-A43,
Konika Minolta Inc.). The stirred samples were filled in Petri dishes, and the surface
was sleeked. Each sample was analyzed at ten different points above the surface,
and the color was expressed in L*a*b* mode, in which L* represents the lightness
value and a* and b* values the chromaticity coordinates. For color and pH
measurements, the samples were analyzed in triplicate.

2.5. Aroma and Taste Profile Analysis

Stirred yoghurt samples (20 mL) were filled into sensory glass beakers (140 mL, J.
Weck GmbH u. Co. KG, Wehr, Germany) and closed with a lid. Sensory analyses
were performed in a sensory panel room at 21 ± 1°C. Trained panelists (n = 12,
male/female, age 24 to 45) with normal olfactory and gustatory function participated
in the sensory sessions and exhibited no known illness at the time of examination.
Prior to this study, the assessors were recruited in weekly training sessions in the
recognition of about 100 selected odor-active compounds according to their odor
qualities by means of an in-house developed flavor language. The order of the
presentation of the different yoghurts was randomized, and no information on the
purpose of the experiment or the composition of the samples was given to the
panelists. The results were averaged for each attribute. Sensory analyses were
performed up to Week 4 of storage (best-before date) in intervals of two weeks
(Weeks 0, 2 and 4).

Aroma profile analysis (APA): in the first session, the panelists were asked to
evaluate the yoghurt gels (retronasal), and the named odor attributes of the different
products were collected. Attributes that were detected by more than 50% of the
panelists were selected for subsequent evaluations. In subsequent sessions, the
panelists were asked to score the perceived retronasal intensities of the selected
attributes on a seven-point-scale from 0 (no perception) to 3 (strong perception) in
increments of 0.5.

Taste Profile Analysis: in addition to the APA, the panelists were requested to
evaluate the following taste attributes: sweet, sour, salty and bitter. The panelists
had to score the attributes’ intensities on a visual analogue scale from 0 (not
perceivable) to 10 (strongly perceivable).

2.6. Triangle Test

A triangle test was performed according to DIN EN ISO 4120:2007 [12]. Two
triangles of the following yoghurt gels were tested: reference vs. RF 65 °C and RF
58 °C vs. CV 58 °C. During the tests, the panelists had their eyes bandaged to avoid
any influence of the possible differences in the yoghurts’ appearance. Twelve
panelists evaluated each triad in duplicate, leading to 24 evaluations in total. The
test was performed after a storage period of four weeks (best-before date).

2.7. Statistical Analysis

Statistical analyses were performed by the use of the software OriginPro 9G

(OriginLab Co., Northampton, MA, USA) and Statistica 10 (StatSoft Europe GmbH,
Hamburg, Germany), respectively. For all groups of data, one-way analysis of
variances (ANOVA) and Fisher LSD post-hoc testing were carried out to elaborate
differences between the differently treated yoghurts and during storage (repeated
measures one-way ANOVA). The level of statistical significance was set at 5%.

3. Results and Discussion

This section presents the results of the RF and CV heating of yoghurt products after
culturing and describes the technological applicability, shelf-life, pH, color and
sensorial changes. The details of the textural and microstructural changes are given
elsewhere in the second part of this two-part study [11].

3.1. Technological Applicability of the RF Heating of Yoghurt Gels

To get initial information about the electrical properties of the plain stirred reference
yoghurt, its electric conductivity was measured. As can be seen in Figure 3, the
electric conductivity of the yoghurt was directly proportional to the temperature.
Generally, the measured electric conductivity was slightly higher than the literature
values for milk, most likely due to the contribution of lactic acid to conductivity [11].

Figure 3. Electric conductivity of the plain stirred reference yoghurt.

RF heating was carried out as described in Section 2.2.1, and three different
temperature regimes were applied, which were chosen based on the death line of
vegetative cells [10]. Starting at a temperature of 40 °C, 60 s were necessary to
reach 58 °C, 90 s to obtain 65 °C and 120 s to achieve 72 °C. Thus, the heating rate
for all temperature regimes was 0.28 ± 0.02 K·s−1 (cf. Figure 4.)

Figure 4. RF heat treatment: the temperature-time profile for heating a stirred yoghurt
(in glass jars) to 65 °C in an RF water bath (T = 70 °C).

Besides a fast heating rate, temperatures of 58 and 65 °C could also be applied very
homogeneously to stirred yoghurt in an RF water bath. After heating yoghurt jars to
72 °C, in some cases, there was significant overheating followed by the strong
contraction of the yoghurt curd and whey separation. It was not possible to establish
a stable heating regime for a 72 °C process. The reason for the heating instability is
not clear. A possible explanation could be a sudden change in the dielectric
properties above 70 °C, which provoked faster heating. It was therefore decided not
to undertake an RF heating series at 72 °C, and this is the reason why no
temperature-time profile at 72 °C is shown in Figure 4. Further details concerning
syneresis and the texture analysis of the few yogurt samples that were heated to 72
°C are described in the second part of this two-part study [11].

In the frame of this feasibility study, experiments were performed on a small pilot
scale. However, RF technology with power ratings of up to several 100 kW are
currently implemented on the production scale for manufacturing processes in
general industries, including the food industry. Furthermore, large pilot scale
equipment of the RF water bath heating process used in this study exists, and its
economic feasibility has been estimated [13]. Although investment costs are higher
for the RF heater compared to conventional technologies, the energy costs are
comparable, even when taking into account the electricity usage compared to the
fossil fuel consumption of conventional heaters. Notably, energy costs with emerging
RF power generators constructed with semi-conductor technology are expected to
fall, due to their higher efficiency, and they are more robust than traditional tube
generators.

3.2. Technological Applicability of CV Heating of Yoghurt Gels

Temperatures of 58, 65 and 72 °C were also applied to stirred yoghurt in a

convection oven. Unlike with RF heating, heat transfer limitations were encountered
in the convection oven. This was even so when the heat energy was transferred
using steam as the heat carrier medium rather than air. The result was that the
heating rate was comparatively low, with the heat curve showing a slowly ascending
sigmoidal behavior. Unlike with RF heating, different heating rates for each
temperature regime were observed: 0.30, 0.41 and 0.55 K·min−1 (58, 65 and 72 °C),
respectively (Figure 5).

Figure 5. CV heat treatment: temperature-time profile for heating a stirred yoghurt

(in glass jars) to 65 °C in a convection oven in steam mode (100%, T = 70 °C).

While difficulties were reported for the dielectric heating of yoghurt gels to 72 °C, CV
heating was successfully applied to all temperature regimes. However, the aim of
the present study was to improve yoghurt shelf-life, while maintaining its high quality
with regard to living LAB and the expected texture and sensorial properties. As
significant textural changes occurred in the CV 72 °C samples, as reported in the
second part of this study [11], these products were produced in the same small batch
numbers as the RF 72 °C samples.

3.3. Microbiological Investigations

Bach [5] was the first and, to the best of our knowledge, the only scientist who has
applied mild temperatures to yoghurts filled in plastic containers via electromagnetic
fields. Nevertheless, no detailed information was provided about microbial numbers.

Thus, LAB, as well as yeasts and molds were characterized in the heat-treated
yoghurt products of the present study. Microbiological investigations were performed
directly after manufacture and also after a storage period of five weeks. The results
are shown in Figure 6.

Figure 6. Colony forming units (CFU) g−1 at Week 0 (grey) and Week 5 (black): (A)
yeasts and molds, (B) lactic acid bacteria (LAB). CFU g−1 are expressed as mean
values ± standard deviation (SD), with “>” denoting “more than” or “<” denoting “less
than” the proposed CFU g−1.

In Figure 6, the reference yoghurt without any additional heat treatment (Ref)
demonstrates the initial population (Week 0) of yeasts and molds (A) and LAB (B).
Generally, a reduction in microbial numbers was observed after any additional heat
treatment of the yoghurt (Figure 6). In detail, yeasts and molds did not survive post-
fermentative RF and CV heat treatments, even at 58 °C. On the other hand, LAB
were found to partially survive RF heating to 58 and 65 °C, while they were
inactivated by CV heating at the same temperatures. The decrease in LAB numbers
was in good agreement with the temperature and time of heat treatment (thermal
stress).

Regarding RF and microwave processing, there is an ongoing discussion about the

possible non-thermal effects on microorganisms due to exposure to an electrical
alternating field. It was shown by Heddleson and Doores [14] that the inactivation of
microorganisms can be attributed to the heat energy developed within the product,
which was also described by Datta and Davidson [4] on comparing the inactivation
curves of microorganisms from RF, microwave and conventional heating. These
observations are further supported by our data on LAB numbers, which are in good
agreement with the degree of thermal stress applied to the yoghurts. Accordingly,
our work shows that RF is a mild heating method.

Generally, these results are in agreement with the literature findings on the thermal
lability of these microorganisms (cf. Section 1). Kessler [10] described that mild heat
treatment at 50 to 55 °C for several minutes might be sufficient for fermented milk
products in view of the targeted destruction of yeasts, molds and rods that produce
much acid and whose proteases are able to break down proteins, yielding unwanted
bitter peptides. It might be that even milder temperatures suffice to prolong yoghurt
shelf-life. However, this should be investigated in more detail in a further study.

Within the framework of this feasibility study, microbial numbers were also
investigated after five weeks of storage. In the case of the stirred reference yoghurt,
an increase in the number of yeasts and molds was found in storage (Figure 6A),
while LAB numbers were generally found to decrease in storage (Figure 6B); an
observation that is in accordance with previous reports on LAB numbers [15,16].
Summarizing our results, one can conclude that the RF heating of yoghurt showed
promise: an increased shelf-life can be predicted, since a reduction in microbial
numbers was achieved. Hence, the potential to prolong yoghurt shelf-life is related
not only to the inactivation of yeasts and molds, but also to the prevention of post-
acidification due to reduced LAB numbers. This assumption follows from the work of
Chandan and O’Rell [17], who described that heating to 60–65 °C stabilizes the
product, so that the yoghurt shelf-life is prolonged to eight to 12 weeks (after
manufacture) at a storage temperature of 12 °C or lower. In a follow-up study, the
experimental design should therefore include a storage time up to at least three
months.

3.4. pH and Color Measurement

The pH value was measured over the entire storage period in order to reveal
possible post-acidification due to the activity of living LAB. A storage period of up to
five weeks did not reveal any changes in pH. The pH stayed very constant at a value
of 4.3 ± 0.1 (10.8 ± 0.6 °C) for all samples. Thus, any ongoing post-acidification can
be excluded, also for the reference yoghurt, without any further heat treatment and,
thus, high LAB numbers.
Additionally, color changes in the yoghurt gels due to post-fermentative heat
treatment were investigated directly after manufacture (Table 3). It was found that
the color between the different yoghurt matrices differed slightly. The L* value
(whiteness), negative a* value (green fluorescence) and positive b* value (yellow
fluorescence) significantly increased after an additional heat treatment at 58 and
65°C. A direct correlation to the final temperature applied was shown for the positive
b* value and, thereby, a change towards a yellow color. Typically, yoghurt shows a
blue fluorescence. This fluorescence occurs due to the conversion of riboflavin
(vitamin B2), which has a characteristic yellow to green fluorescence, to colorless
leucoriboflavin in the presence of LAB [18]. This explains the lower b* value of the
reference yoghurt. An increase in the yellow character of the yoghurts with the
increasing temperature of the heat treatment might be due to a slight caramelization
processes; Maillard reactions are less probable, due to the low pH of yoghurt and
the mild temperatures applied. Another reason might be the conversion of riboflavin
to lumiflavin, which also shows a yellow-to-green fluorescence, but again, such
reactions are favored in alkaline solutions [18].

Table 3. Color changes in yoghurt due to post-fermentative heat treatment and

the results of ANOVA (one-way).

CV-treated ANOVA
RF-treated yoghurts
Color yoghurts results
Reference
analysis
yoghurt
(L*a*b*) F p
58 °C 65 °C 72 °C 58 °C 65 °C 72 °C
value value

90.71 90.63 89.51 90.67 90.59 90.45

90.36 ± 3.33 ×
L* ± 0.05 ± 0.02 ± 0.10 ± 0.04 ± 0.09 ± 0.04 119.27 −11
0.08 b 10
d c,d a c,d c b

−3.47 −3.45 −3.73 −3.49 −3.48 −3.42

−3.34 ± 7.05 ×
a* ± 0.04 ± 0.04 ± 0.02 ± 0.01 ± 0.01 ± 0.03 37.99
0.06 a 10−8
b,c b,c d c c b

10.31 10.27 11.12 10.41 10.33 10.66

9.75 ± 1.72 ×
b* ± 0.05 ± 0.06 ± 0.12 ± 0.10 ± 0.09 ± 0.10 47.11
0.17a 10−8
b b d b b c

Intensity values with different letters indicate significant differences between

products (p < 0.05, Fisher LSD post-hoc).

To the best of our knowledge, no other literature describes the color changes in
yoghurt due to an additional heating step after culturing.

3.5. Aroma and Taste Profile Analysis

The aroma profile of yoghurt is a mixture of various odor-active compounds
[19,20,21]. Further characteristics of yoghurt flavor are, for example, the sour
impressions and the specific textural properties that are jointly evaluated when
tasting yoghurt [22,23]. In the present study, the heat-treated yoghurt samples and
the stirred reference yoghurt were evaluated retronasally by means of an APA and
taste profile analysis in order to reveal possible profile changes due to an additional
heating step (Table 4).

Table 4. Retronasal intensity rating of specific flavor attributes by aroma

profile analysis (APA) of yoghurt gels with/without additional heat treatment
after culturing and the results of ANOVA (one-way): Weeks 0, 2 and 4.

RF- CV-
Aroma profile ANOVA
treated treated
results
analysis (APA) Reference yoghurts yoghurts Set
(scale: 0–3; mean yoghurt reference
values) 58 65 65 F p
58 °C
°C °C °C value value

2.0 2.2 1.86 ×

yoghurt-like 2.6 a 1.6 c 2.1 b 2.3 a,b 4.27
b,c a,b 10−3

3.10 ×
fatty 1.3 1.6 1.2 1.1 1.1 0.9 1.22
10−1 *

5.70 ×
buttermilk-like 0.9 1.2 1.3 0.9 1.2 1.4 0.78
Week 10−1 *
0 0.7 1.2 0.7 6.53 ×
cream-like 0.6 c 1.3 a 0.3 c 3.53
b,c a,b b,c 10−3

sour cream- 2.80 ×

0.6 1.1 1.1 0.8 1.1 1.4 1.28
like 10−1 *

cream 1.0 3.24 ×

0.6 b 1.4 a 0.6 b 0.8 b 0.8 b 2.60
cheese-like a,b 10−2

4.12 ×
yoghurt-like 2.3 1.8 2.1 2.0 1.7 - 1.01
10−1 *

Week 5.71 ×
fatty 1.2 0.9 1.1 0.7 0.9 - 0.74
2 10−1 *

7.26 ×
buttermilk-like 0.9 0.8 1.0 1.0 1.3 - 0.51
10−1 *
 RF- CV-
Aroma profile ANOVA
treated treated
results
analysis (APA) Reference yoghurts yoghurts Set
(scale: 0–3; mean yoghurt reference
values) 58 65 65 F p
58 °C
°C °C °C value value

8.82 ×
cream-like 0.8 0.9 1.1 0.9 1.0 - 0.29
10−1 *

sour cream- 8.11 ×

1.3 1.0 1.0 1.1 1.1 - 0.40
like 10−1 *

cream 3.73 ×
0.7 1.2 1.0 1.3 1.3 - 1.08
cheese-like 10−1 *

8.98 ×
yoghurt-like 2.2 2.0 2.2 2.3 2.3 - 0.27
10−1 *

9.70 ×
fatty 1.0 0.8 0.7 1.0 0.8 - 0.13
10−1 *

4.90 ×
buttermilk-like 1.1 0.8 1.5 1.1 1.1 - 0.87
Week 10−1 *
4 5.68 ×
cream-like 0.5 0.6 0.5 0.4 0.3 - 0.74
10−1 *

sour cream- 8.34 ×

0.9 1.0 1.0 0.8 1.2 - 0.36
like 10−1 *

cream 6.72 ×
0.7 1.0 0.9 1.2 1.1 - 0.59
cheese-like 10−1 *

* No post-hoc test necessary (p ≥ 0.10, n.s.). Intensity values with different letters
indicate significant differences between products (p < 0.05, Fisher LSD post-hoc).

Regarding the APA, the following descriptors were detected in the yoghurts by more
than 50% of the panelists and were hence selected for subsequent evaluations:
yoghurt-like, fatty, buttermilk-like, cream-like, sour cream-like and cream cheese-
like. At Week 0, the odor attribute “yoghurt-like” was described for all samples as
being clearly to strongly perceivable. All the other odor attributes were perceived
rather weakly in the yoghurts, with just a few exceptions: the attributes, fatty, cream-
like and cream cheese-like, were described for the RF 58 °C yoghurt as being slightly
more strongly perceivable than in the other yoghurts. The attributes, buttermilk-like
and sour cream-like, were evaluated highest in the case of the set-style reference
yoghurt. A closer look at Table 4 reveals that only three attributes were evaluated as
being significantly different between the products (Week 0): yoghurt-like, cream-like
and cream cheese-like. In detail, those variances arose either between the
references and the heat-treated samples (yoghurt-like, cream-like) or between the
yoghurts heated to 58 °C and the other products (cream cheese-like). After two and
four weeks of storage, these differences decreased and were no longer significant.
Hence, during storage, the aroma profiles of the reference and heated yoghurts were
found to be very similar to each other.

A taste profile analysis was additionally performed with the results being shown in
Table 5. At Week 0, all yoghurt products were described as being sour with medium
intensity (4.5–6.3); thereby, the set-style reference was evaluated as being the
sourest (6.3). The attributes sweet, salty and bitter were just slightly perceivable and,
in general, no significant differences between the samples could be detected. Thus,
no formation of bitter peptides in the course of the additional heating step was
observed. Additionally, no significant differences between the products became
evident. In storage, the taste profiles of the samples remained quite close to each
other, and no significant differences arose. Accordingly, it can be concluded that the
trained sensory panel not only evaluated the aroma, but also the taste profiles of all
heat-treated samples as being very close to the stirred reference product.

Table 5. The intensity rating of specific taste attributes by taste profile analysis
of yoghurt gels with/without additional heat treatment after culturing and the
results of ANOVA (one-way): Weeks 0, 2 and 4.

Taste profile RF-treated CV-treated ANOVA

(visual Reference yoghurts yoghurts Set results
analysis
analogue scale, yoghurt 65 65 reference F p
mean values) 58 °C 58 °C
°C °C value value

7.37 ×
sweet 1.8 2.0 2.2 2.3 2.1 1.3 0.55
10−1 *

3.85 ×
sour 5.3 5.6 4.9 4.5 5.2 6.3 1.07
10−1 *
Week 0
9.93 ×
salty 1.2 0.8 1.0 1.0 0.9 1.1 0.09
10−1 *

8.84 ×
bitter 0.7 0.9 1.0 0.5 0.5 0.9 0.34
10−1 *

5.55 ×
Week 2 sweet 1.5 0.9 1.8 1.9 1.2 - 0.76
10−1 *
Taste profile RF-treated CV-treated ANOVA
(visual Reference yoghurts yoghurts Set results
analysis
analogue scale, yoghurt 65 65 reference F p
mean values) 58 °C 58 °C
°C °C value value

7.56 ×
sour 5.9 5.8 5.8 5.3 6.5 - 0.47
10−1 *

9.97 ×
salty 1.2 1.4 1.3 1.4 1.5 - 0.04
10−1 *

9.05 ×
bitter 0.4 0.2 0.2 0.2 0.2 - 0.26
10−1 *

3.45 ×
sweet 0.9 1.7 2.0 1.3 1.2 - 1.15
10−1 *

9.76 ×
sour 5.1 5.6 5.5 5.2 5.4 - 0.12
10−1 *
Week 4
7.29 ×
salty 0.9 1.1 0.9 1.5 0.7 - 0.51
10−1 *

9.74 ×
bitter 0.5 0.5 0.6 0.5 0.7 - 0.12
10−1 *

* No post-hoc test necessary (p ≥ 0.10, n.s.). Intensity values with different letters
indicate significant differences between products (p < 0.05, Fisher LSD post-hoc).

Moreover, with the storage of the individual yoghurts, no significant changes in the
flavor or taste profiles were detected. The only exceptions were the 58 °C samples
(RF, CV) in the case of the attribute cream-like: its intensity was evaluated as
significantly higher in the samples at Week 0, but a subsequent decrease occurred
thereafter. Generally, the observed lack of changes in flavor (intensity) in storage is
in good agreement with previous finding [16,24]. Martin et al. [24] reported very low
differences in the typical retronasal and orthonasal odor profiles of yoghurt (e.g.,
yoghurt-like, cream-like, butter-like) during 21 days of storage. Only the attribute,
cream-like, was rated significantly lower in storage, and the same was observed for
the 58°C heated products in the present study. Overall, Martin et al. [24] reported
only minor olfactory differences in the aroma profiles of such plain yoghurt products
fermented using different starter cultures. In that work, only the odor attribute,
yoghurt-like, was ranked with medium intensity, and all other attributes were found
to be only weakly perceivable, showing, again, the overall agreement with the
findings of the present study.
In the present study, triangular testing was additionally performed to determine
whether general differences in the sensory profiles of the differently treated yoghurts
were statistically significant at the end of their predicted shelf-life (Week 4). To avoid
any influence of the yoghurts appearance on the results of the triangular evaluation
(e.g., due to slight color differences, as reported in Section 3.4), the testing was
performed blindfolded by the panelists. In detail, two triads were tested: reference
vs. RF 65 °C and RF 58 °C vs. CV 58 °C. The first triad was chosen to compare the
reference yoghurt with an RF heated yoghurt, since this treatment was generally
found to be very mild in the case of 58 and 65 °C (Section 3.3). The second triad
was chosen to reveal the differences between RF and CV heated products for the
mildest temperature of 58 °C, because huge varieties in heating rates were found
(Section 3.1 and Section 3.2). In accordance with the results of the APA and the
taste profile analysis, the tests confirmed that all of the samples were very close to
each other with respect to their overall sensory properties: for none of the triads was
the divergent sample identified with statistical significance. Out of 24 evaluations, 13
correct answers would have been necessary to indicate a significant difference
between the samples at a level of significance of 5% (alpha and beta error,
respectively). In the case of the triad, reference vs. RF 65 °C, 12 out of 24
evaluations led to the correct identification of the divergent sample; for the triad, RF
58 °C vs. CV 58 °C, only seven evaluations identified the divergent sample correctly.
Accordingly, the samples were not distinguishable based on their overall sensory
properties.

4. Conclusions

The present study evaluated the effect of additional heat treatment (CV and RF) after
the fermentation process in yoghurt production. Such additional heat treatment has
been little studied previously. Both heating methods showed promise concerning the
homogeneity of the temperature distribution. However, RF appeared to be superior,
due to a very fast heating rate of 0.3 K·s−1. This study also focused on the yoghurt
shelf-life, pH, color and sensorial quality. Microbial characterization demonstrated
the mildness of the RF treatment, because LAB partially survived. Nevertheless, a
reduction in LAB highlighted the potential of this method for prolonging the yoghurt
shelf-life, not only by inactivating the yeasts and molds, but also by preventing strong
post-acidification. Whereas generally, no significant changes in pH and sensorial
quality were observed, slight color changes were found to occur after heat treatment,
possibly due to the onset of the caramelization processes. Based on these
observations, we believe RF heating is a promising technological tool for future
applications in the dairy industry (e.g., milk pasteurization, sterilization or post-
heating of fermented products). However, further improvements with regard to batch-
wise manufacturing are necessary in order to open up this technique to industry,
such as the application of radio frequencies in a continuous process.

Acknowledgements

The present study was awarded and financially supported by the Danone Innovation
Prize 2012. The authors would like to thank Danone GmbH (Haar, Germany) for this
support. Furthermore, the authors would like to thank Angela López Pinar from the
Friedrich-Alexander University of Erlangen-Nürnberg, Department of Chemistry and
Pharmacy, as well as Monika Steinbuechl and Sigrid Regnet from the Fraunhofer
IVV, Department Retention of Food Quality, for their technical support. We
acknowledge support by Deutsche Forschungsgemeinschaft and Friedrich-
Alexander-Universität Erlangen-Nürnberg (FAU) within the funding programme
Open Access Publishing.

Abbreviations

ANOVA analysis of variance

APA aroma profile analysis

CFU colony forming units

CV convectional heating

εr relative permittivity/relative dielectric constant

LAB lactic acid bacteria

n.s. not significant

RF radio frequency heating

SD standard deviation

sig. significant

Conflicts of Interest

The authors declare no conflict of interest.

References

1. Sanders H.R. Dielectric thawing of meat and meat products. Int. J. Food Sci.
Technol. 1966;1:183–192. doi: 10.1111/j.1365-2621.1966.tb01805.x. [Cross Ref]
2. Cathcart W.H., Parker J.J. Defrosting frozen foods by high-frequency heat. J.
Food Sci. 1946;11:341–344. [PubMed]
3. Pircon L.J., Loquercio P., Doty D.M. High-frequency cooking. High-frequency
heating as a unit operation in meat processing. J. Agric. Food Chem. 1953;1:844–
847.
4. Datta A.K., Davidson P.M. Microwave and radio frequency processing. J. Food
Sci. 2000;65:32–41. doi: 10.1111/j.1750-3841.2000.tb00616.x. [Cross Ref]
5. Bach J. Improvements in or Relating to Methods for Effecting Uniform Heating of
Organic Products. 1 507 069. British Patent. 1978
6. Felke K., Pfeiffer T., Eisner P. Neues Verfahren zur schnellen und schonenden
Erhitzung von verpackten Lebensmitteln: Hochfrequenzerhitzung im Wasserbad.
Chem. Ing. Tech. 2009;81:1815–1821.
7. Villamiel M., Schutyser M.A.I., de Jong P. Novel methods of milk processing. In:
Tamime A.Y., editor. Milk Processing and Quality Management. 1st ed. Blackwell
Publishing; Oxford, UK: 2009. pp. 205–236.
8. Shiby V.K., Mishra H.N. Fermented milks and milk products as functional foods—
A review. Crit. Rev. Food Sci. Nutr. 2013;53:482–496. doi:
10.1080/10408398.2010.547398. [PubMed] [Cross Ref]
9. Food and Drug Administration (FDA) CFR—Code of Federal Regulations, Title
21—Foods and Drugs, Subchapter B—Food for Human Consuption, Part 131—Milk
and Cream, Subpart B—Requirements for Specific Standardized Milk and Cream,
Section 131.200—Yogurt. [(accessed on 27 February 2014)]. Available online:
http://www.gpo.gov/fdsys/pkg/CFR-1999-title21-vol2/pdf/CFR-1999-title21-vol2-
sec131-200.pdf.
10. Kessler H.G. Food and Bio Process Engineering: Dairy Technology. 5th ed.
Verlag A. Kessler; München, Germany: 2002.
11. Siefarth C., Tran T.B.T., Mittermaier P., Pfeiffer T., Buettner A. Effect of radio
frequency heating on yoghurt—II. Microstructure and texture. Foods. 2014 submitted
for publication.
12. Deutsches Institut für Normung (DIN) DIN-EN-ISO-4120:2007-10. Sensory
Analysis—Methodology—Triangle Test (German version) DIN; Berlin, Germany:
2007.
13. Cost Comparison Retort versus RF Heater, Internal Report: 24.1.2014.
Fraunhofer Institute for Process Engineering and Packaging (IVV); Freising,
Germany: 2014.
14. Heddleson R.A., Doores S. Factors affecting microwave heating of foods and
microwave induced destruction of foodborne pathogens—A review. J. Food Prot.
1994;57:1025–1037.
15. Damin M.R., Minowa E., Alcântara M.R., Oliveira M.N. Effect of cold storage on
culture viability and some rheological properties of fermented milk prepared with
yogurt and probiotic bacteria. J. Texture Stud. 2008;39:40–55.
16. Salvador A., Fiszman S.M. Textural and sensory characteristics of whole and
skimmed flavored set-type yogurt during long storage. J. Dairy Sci. 2004;87:4033–
4041. doi: 10.3168/jds.S0022-0302(04)73544-4. [PubMed] [Cross Ref]
17. Chandan R.C., O’Rell K.R. Manufacture of various types of yogurt. In: Chandan
R.C., White C.H., Kilara A., Hui Y.H., editors. Manufacturing Yogurt and Fermented
Milks. 1st ed. Blackwell Publishing; Oxford, UK: 2006. pp. 211–236.
18. Töpel A. Chemie und Physik der Milch: Naturstoff, Rohstoff, Lebensmittel. 1st
ed. Behr’s Verlag; Hamburg, Germany: 2004. pp. 350–351.
19. Ott A., Fay L.B., Chaintreau A. Determination and origin of the aroma impact
compounds of yogurt flavor. J. Agric. Food Chem. 1997;45:850–858. doi:
10.1021/jf960508e. [Cross Ref]
20. Ott A., Germond J.-E., Baumgartner M., Chaintreau A. Aroma comparisons of
traditional and mild yogurts:  Headspace gas chromatography quantification of
volatiles and origin of α-diketones. J. Agric. Food Chem. 1999;47:2379–2385.
[PubMed]
21. Pollien P., Ott A., Montigon F., Baumgartner M., Muñoz-Box R., Chaintreau A.
Hyphenated headspace-gas chromatography-sniffing technique:  Screening of
impact odorants and quantitative aromagram comparisons. J. Agric. Food Chem.
1997;45:2630–2637. doi: 10.1021/jf960885r. [Cross Ref]
22. Belitz H.-D., Grosch W., Schieberle P. Lehrbuch der Lebensmittelchemie. 6th
ed. Springer; Berlin Heidelberg, Germany: 2008.
23. Buettner A. Aroma Release and Perception during Consumption of Food Taking
into Account Physiological Aspects. Verlag Dr. Hut; Munich, Germany: 2007.
24. Martin N.C., Skokanova J., Latrille E., Beal C., Corrieu G. Influence of
fermentation and storage conditions on the sensory properties of plain low fat stirred
yogurts. J. Sens. Stud. 1999;14:139–160
XVIII. First Approach to the Analytical Characterization
of Barrel-Aged Grape Marc Distillates Using Phenolic
Compounds and Colour Parameters
Food Technol Biotechnol. 2014 Dec; 52(4): 391–402.
Raquel Rodríguez-Solana,1,2 José Manuel Salgado,3 José Manuel Domínguez,1,2
and Sandra Cortés-Diéguez 1,2

Summary

Phenolic compounds (benzoic and cinnamic acid derivatives) were determined by

high-performance liquid chromatography with multiple wavelength detector (HPLC-
-MWD) in grape marc distillates aged in Quercus petraea, Quercus robur and
Quercus alba wooden barrels. In addition to colour indices and evaluable
polyphenols, all samples were described by sensorial analysis. There were
significant differences in the mean concentrations of the majority of phenolic
compounds among the samples. Gallic and benzoic acids were the most abundant
and samples aged in Q. robur from Galicia (NW of Spain) had the highest
concentration of most of the determined phenols. Grape marc distillates aged in Q.
robur obtained the highest values of all sensorial attributes, whereas samples aged
in Q. petraea and Q. alba obtained similar scores. Principal component analysis
accounted for 88.32% of total variance, showing a good separation of aged distillates
in terms of phenolic compounds and colour characteristics, according to the species
and origin of the oak wood used in the ageing process.

Key words: ageing, grape marc distillate, HPLC-MWD, phenols, sensory analysis,
wooden barrel

Introduction

Different high-alcohol drinks can be obtained after distilling some previously

fermented raw material. The most popular alcoholic beverages obtained from grape
marc are orujo (Spain), bagaçeira (Portugal), grappa (Italy) and tsipouro (Greece).
In the majority of these cases, freshly distilled beverages are defined with sensory
attributes like sharp, alcoholic, rude and bitter, therefore ageing in oak barrels is
essential to give them the sensory characteristics that consumers like (1, 2).

During the maturation process, several physicochemical reactions occur in

distillates, such as extraction of wood components, loss of low-boiling-point
compounds from the immatured distillate by evaporation, and interactions among
components from wood and beverage. As a result of different reactions such as
polymerization, esterification, acetalisation, hydrolysis and oxidation (3, 4), the initial
product modifies its chemical composition and the sensory characteristics, visual
aspect (colour and limpidity), taste and flavour (5). Low molecular mass phenolic
compounds can be pointed out among the released components from wood during
ageing, since they are not present in fresh distillates (6). The identification and
quantification of phenols are crucial due to their influence on the chemical
composition, sensory characteristics of the resulting aged distillate and also on their
antioxidant activity (7, 8).

The extension of these physicochemical changes depends on several factors, such

as the botanical species and geographical origin of the wood (region, type of forest,
climate, soil type, etc.), the technology of the barrel-making process (mainly the
toasting intensity), the size of the barrel, their previous uses, ageing time, ageing
conditions (cellar temperature and humidity) and the pH, alcohol content and total
acidity of the initial wine or distillate (9–15).

Oak (genus Quercus) is the most suitable botanical species of wood for alcoholic
drink maturation (16). Quercus alba (American oak), which grows in different areas
in the United States, Quercus petraea (sessile oak) and Quercus robur (pedunculate
oak) from Europe, are the most common oak species used in barrel-making (17).
French oak is the most popular wood for the maturation of beverages around the
world (18), whereas Spanish oaks, mainly grown in the North of Spain, are used at
a lower scale for ageing of wines and a few studies about their behaviour have been
published (19–21).

Some distilleries in Galicia (NW of Spain) also use oak from Q. robur grown in this
area to age the grape marc distillate (called orujo). However, only a small part of the
production of orujo is subjected to an ageing process, due to significant expenses:
wooden barrels are expensive, and this cost increases in proportion to the length of
the ageing period, mainly due to alcohol evaporation.

The physicochemical parameters of these beverages are fixed by the Regulatory

Council in the control program (22). These parameters are the same as in young
orujo and only change the minimum and maximum values fixed for them, although
the ageing process changes the chemical composition of young distillate and as
consequence its sensory profile (23). The influence of the species of oak wood and
time of ageing on the mineral composition of distillate was also observed (24), and
the results showed a great influence of the species of oak wood on the final
composition of the distillate.

The current study has been undertaken to obtain the descriptive phenolic and
sensory attributes of aged orujo distillates, in order to characterize these alcoholic
beverages for the first time. The results can also be useful to valuate the suitability
of employing oak barrels grown in Spain as an alternative to the barrels bought from
other countries and thus contributing to the improvement of the quality of these
distillates and the attempt to reduce the high initial cost.

Materials and Methods

Samples
Ten commercial grape marc distillates from Galicia (orujo) aged in 225-litre wooden
barrels from the species Quercus robur (pedunculate oak; origin: Limousin, France
and Galicia, Spain), Quercus alba (American oak) and Quercus petraea (origin:
Allier, France) were analysed. These samples correspond to all the types of barrels
currently present on the market. They were collected and bottled in glass bottles
after an ageing period between 1 and 6 years. General information about aged orujo
(species of oak wood, time of ageing, etc.) was supplied by the producers. In Table
1 all the information concerning the analysed samples is summarized.

Table 1. Main characteristics of the analysed orujo distillates aged in different oak
wood barrels

Distillatio
Oak Geographic ϕ(alcoho t(ageing
Sample Grape Produce n
specie al l) )
code variety r code* equipme
s location % month
nt

Limousin
QRL13 Mencía A steam 38.7 13
(France)

Albariń Limousin
QRL16 B steam 40.5 16
Quercu o (France)
s
robur Albariń Galicia
QRG60 C alembic 49.0 60
o (Spain)

Albariń Galicia
QRG72 C alembic 44.8 72
o (Spain)

Albariń Allier
QPA60 D steam 47.6 60
o (France)

mix of
red
Allier
QPA72 grape A alembic 50.0 72
Quercu (France)
varietie
s
s
petraea
QPA72 Albariń Allier
F steam 40.3 72
B o (France)

QPA14 Godell Allier

G steam 43.4 144
4 o (France)
 Distillatio
Oak Geographic ϕ(alcoho t(ageing
Sample Grape Produce n
specie al l) )
code variety r code* equipme
s location % month
nt

Albariń
QA72A F steam USA 44.8 72
o
Quercu
s alba
Albariń
QA72B F steam USA 46.1 72
o

*Different letters represent different companies

Reagents

The chemical standards used: gallic acid (3,4,5-trihydroxybenzoic acid)

monohydrate, sinapic acid, syringaldehyde (3,5-dimethoxy-4-
hydroxybenzaldehyde), syringic acid (3,5-dimethoxy-4-hydroxybenzoic acid) and
vanillin were purchased from Fluka (Madrid, Spain). Benzoic acid, coniferaldehyde
(4’-hydroxy-3’-methoxycinnamaldehyde), ferulic acid (trans-4-hydroxy-3-
methoxycinnamic acid), isoferulic acid (3-hydroxy-4-methoxycinnamic acid), p-
coumaric acid (4-hydroxycinnamic acid), sinapaldehyde (trans-3,5-dimethoxy-4-
hydroxycinnamaldehyde), and 4-hydroxybenzaldehyde were supplied by Sigma-
Aldrich Chemie GmbH (Steinheim, Germany). Protocatechualdehyde (3,4-
dihydroxybenzaldehyde), vanillyl alcohol (4-hydroxy-3-methoxybenzyl alcohol), and
vanillic acid (4-hydroxy-3-methoxybenzoic acid) were purchased from SAFC Sigma-
Aldrich. Absolute ethanol was purchased from Merck (Darmstadt, Germany).
Methanol (HPLC-gradient grade) and formic acid were from Panreac (Barcelona,
Spain), and Milli-Q water from Millipore system (Bedford, MA, USA).

Total phenolic content

According to Spigno et al. (25), total phenols were determined applying two methods:
(i) Singleton and Rossi procedure (26) using Folin-Ciocalteu reagent, the best
method to determine total phenols, including tanins, and (ii) direct measure of the
absorbance of each sample at 280 nm (A280 nm) employing a one-centimetre quartz
cuvette (27). All samples were previously diluted 20 or 100 times. Total phenolic
index (TPI) was determined by the following equation:

The quantification of both parameters was carried out using a calibration curve with
known concentrations of gallic acid and expressed as gallic acid equivalents (GAE).
Colour intensity and hue

Colour intensity and hue, determined by absorbance measurements of diluted or

undiluted samples at the wavelengths of 420 (yellow), 520 (red) and 620 nm (violet),
were also evaluated employing one-centimetre polystyrene cuvettes (28, 29). The
equations for the calculation of colour intensity (CI) and hue (h) were as follows:

and

Spectrophotometric analyses were performed using a UV-VIS Cintra 6

spectrophotometer (GBC Scientific Equipment, Madrid, Spain).

HPLC-MWD analysis

All samples of aged grape marc spirits were filtered through 0.45-µm pore
membranes of cellulose acetate (Sartorius, Goettingen, Germany) before the
analysis using high-performance liquid chromatography (HPLC). An Agilent
Technologies 1200 series system consisted of a quaternary pump (G1311A), an
injector, a degasser (G1322A), a multiple wavelength detector (MWD, UV/VIS;
Agilent, Palo Alto, CA, USA) and a Zorbax SB-Aq reversed-phase column, 5 µm,
150 mm×4.6 mm i.d. (Agilent) with a guard column. Samples of 20 µL of aged spirit
or calibration standards were injected into the column and eluted with the following
gradient: solvent A (methanol) and solvent B (2.5% formic acid in Milli-Q water, by
volume) at a flow rate of 1 mL/min. Zero time conditions were 100% B, after 35 min
the pumps were adjusted to 52% B and 48% A, at 56 min to 100% A until the end of
the analysis at 65 min. Detection was carried out at (276±4) nm. The identification
of each compound was done by comparing the retention times with those of pure
standards. All determinations were made in duplicate.

Sensory analysis

The panel for sensory analysis consisted of eight assessors, five males and three
females aged from 35 to 55, all of them members of the official panel of Geographic
Denomination of Spirits and Liqueurs from Galicia, expert tasters in sensory analysis
of this kind of alcoholic beverages. The selected judges worked directly for wineries
and distilleries either as winemakers or in marketing. The sensory analysis was
performed in a laboratory, containing 20 independent tasting booths, and designed
according to the International Organization for Standardization, standard ISO 8589
(30).
Before evaluation, during three training sessions (12 h), a collection of six
representative samples was tested by the panellists in order to generate relevant
appearance and taste attributes. All samples were commercially available. The aim
of these sessions was to develop a common vocabulary for the description of the
sensory attributes of aged orujo samples. In the first phase of this training, the judges
identified thirty-three descriptors (13 for appearance and 20 for taste). After a round-
table discussion and by consensus, the panel selected and refined the attributes that
best describe their perceptions. Synonymous, hedonic and irrelevant descriptors
were also eliminated by using statistical methods described in ISO 11035 standard
(31). Finally, the generated attributes were reduced to eight (three for appearance
and five for taste).

In the formal session, the intensity of each attribute was rated on a five-point scale,
where 0 indicated that the descriptor was not perceived, 1=low, 2=low-medium,
3=medium, 4=medium-high and 5=high. Each sample was previously coded and
presented to judges in random order. The sensory evaluation was carried out at
room temperature, using the official glasses of the corresponding Regulation
Commission. Tasting was carried out in the morning during five sessions on different
days to avoid fatigue of the tasters due to the high degree of alcohol in the aged
grape marc distillates (37.5–50%, by volume).

Statistical analysis

The obtained results were analysed using XLSTAT-Pro (Addinsoft, Paris, France).
One-way analysis of variance (ANOVA) was applied to establish whether significant
differences (p<0.05) existed between the values obtained for the mean
concentration of each compound in the analysed aged grape marc distillates. The
multiple range test (LSD) was applied to confirm the obtained results. Pearson’s
correlations among all identified phenolic compounds, between sensorial descriptors
and phenolic compounds, and between sensorial attributes and colour parameters
were also calculated. In order to determine the influence of the oak species on the
composition of grape marc distillates, a multivariate principal component analysis
was carried out.

Results and Discussion

Spectrophotometric parameters of aged orujo distillates

Table 2 lists all phenolic indices and chromatic characteristics of the analysed
samples. Taking into account only the species of oak, no significant differences were
observed in the Folin-Ciocalteu index among the samples; however, the values of
other spectrophotometric parameters were significantly different. Higher contents of
total phenols were observed in the group of distillates aged in Q. robur (Galicia),
showing significantly different value with respect to the other analysed distillates,
including those aged in the same oak species of other geographical origin. Total
phenolic index (TPI) determined in each group of distillates showed higher values
than the corresponding total phenols evaluated with Folin-Ciocalteu method.
Table 2. Concentration of total phenols and chromatic characteristics of orujo
distillates after ageing in barrels

Oak species Quercus robur Quercus Quercus

petraea alba

Geographical Limousin Galicia Allier USA

location

FCI/(mg of GAE per

626±12 451±33 482±63 402±303
L)

TPI/(mg of GAE per

(1232±1052)a (5590±352)b (1728±1051)a (1211±44)a
L)

CI (0.9±0.5)ac (2.3±0.3)b (1.5±0.2)a (0.7±0.2)c

h (4.8±0.3)a (5.9±0.3)b (5.3±0.5)ab (5.4±0.1)ab

*Different letters in superscript within the same row indicate statistically significant
differences (p≤0.05) according to LSD test FCI=Folin-Ciocalteu index, TPI=total
phenolic index, GAE=gallic acid equivalent, CI=colour intensity, h=hue

Orujo aged in Q. robur (Galicia) showed the highest values of colour intensity,
whereas the samples aged in Q. alba showed the lowest. Hue value was lower in
the samples aged in Q. robur (Limousin); however, other samples did not show
significant differences. The results showed a great influence of the wood origin on
the chromatic characteristics of the samples, since significant differences were found
among distillates aged in Quercus robur from different areas.

Validation of the HPLC method

The validation of the method was done based on linearity and analytical limits (limits
of detection and quantification).

Linearity

The study of the linearity was performed using the HPLC analysis of seven standard
solutions containing increasing concentrations of their respective standards covering
the range of linearity. These solutions were prepared in triplicate in ultrapure water
with 40% (by volume) of absolute ethanol and filtered through 0.45-µm pore
membranes of cellulose acetate. Data in Table 2 show that linearity is satisfactory in
almost all cases, with the correlation coefficient (R2) ranging from 0.9712 (4-
hydroxybenzaldehyde) to 0.9999 (coniferaldehyde).
Analytical limits of detection and quantification

The limit of detection (LOD) and limit of quantification (LOQ) were determined from
the parameters of the analytical curves. Both limits were calculated according to the
following mathematical relationships:

and

where S.D. is the estimation of the standard deviation of the regression line, and m
is the slope of the calibration curve (32). The LOD and LOQ were low for all
determined phenolic compounds (Table 3).

Table 3. Family, compound, retention time, linearity parameters and analytical limits
of the HPLC method for the quantification of phenolic compounds in aged grape
marc distillates

tR/mi S.D. LO LO
Family Compound a S.D.a b b R2
n D Q

gallic acid 32.5 17.0 0.998

5.1 60.30 0.89 2.93 9.78
2 8 7

vanillic acid 0.997

17.59 39.04 3.74 4.83 0.85 0.48 1.61
2
Benzoic
acids
benzoic acid 0.999
20.02 0.94 2.72 1.83 0.10 0.39 1.31
4

syringic acid 0.999

21.75 58.76 1.7 1.11 0.36 0.17 0.57
7

LO LO
Family Compound tR/min a S.D.a b S.D.b R2
D Q

protocatechualdehy 417.6 – 30.3 0.984

10.91 6.26 0.05 0.17
de 4 2.63 3 6
 tR/mi S.D. LO LO
Family Compound a S.D.a b b R2
n D Q

4-
513.0 10.3 39.0 0.971
hydroxybenzaldehyd 15.03 2.44 0.07 0.22
6 6 1 2
e
Benzoic
aldehyde vanillin 0.999
20.62 72.98 4.07 4.54 0.59 0.27 0.92
s 6

syringaldehyde 0.999
24.36 28.83 4.02 4.86 0.32 0.59 1.97
4

p-coumaric acid –
22.82 74.09 3.59 2.77 0.993 0.51 1.7
7.82

ferulic acid –
227.3 11.4 0.999
25.9 99.4 6.15 0.19 0.64
4 3 8
Cinnamic 2
acids
isoferulic acid – 0.998
27 60.01 1.1 1.1 0.07 0.25
1.02 3

sinapic acid – 0.995

28.26 20.88 6.19 0.64 1.22 4.06
4.87 4

coniferaldehyde 0.999
27.49 13.52 0.65 1.24 0.04 0.21 0.68
Cinnamic 9
aldehyde
s sinapaldehyde – 0.999
29.71 24.06 1.23 0.19 0.16 0.55
0.40 8

Benzoic vanillyl alcohol 0.993

9.98 17.77 0.94 2.60 0.38 0.23 0.76
alcohol 5

tR=retention time (min), a=slope, S.D.a=standard deviation of the slope, b=intercept,

S.D.b=standard deviation of the intercept, R2=correlation coefficient of linear range,
LOD=limit of detection, LOQ=limit of quantification

Concentration of phenolic compounds in grape marc distillates aged in

different wooden barrels

Fig. 1 shows the molecular structures of the compounds extracted from the wood.
The phenolic aldehydes are produced by thermodegradation of the terminal
monomer units of lignin: the cinnamic aldehydes convert to benzoic aldehydes, and
then they are oxidized to phenolic acids (19). Table 4 reports the concentrations of
the quantified phenolic compounds in the analysed aged distillates. The
concentration of gallic acid in aged beverages depends on the toast level of barrels,
since gallic acid is degraded at high temperatures (33). Consequently, this
compound is more abundant in distillates aged in barrels with light or medium toast
levels (7). Besides gallic acid, the concentration of ferulic and vanillic acids also
decreased with higher temperature during the toasting process (34). The content of
gallic acid was significantly higher in distillates aged in Quercus robur from Galicia
(>50 mg/L). This compound was also present, but at low concentration, in distillates
aged in Quercus petraea from Allier. Vanillic acid can be directly extracted from oak
wood or be formed by oxidation of vanillin during the ageing process, whereas
syringic acid is formed during toasting by the oxidation of the corresponding
aldehyde. Both compounds were present at higher concentrations in distillates aged
in Quercus robur from Galicia. In contrast, distillates in the same species from
Limousin showed lower values of these compounds.

Fig. 1 Molecular structures of the phenolic compounds extracted to the grape marc
distillates after ageing in oak barrels. The structures circled with dashed line in bold are
guaiacyl compounds, while the structures circled with double dashed line are syringyl- -type
compounds
Table 4. Concentration of phenolic compounds in ten grape marc distillates aged in
three different oak species during different periods of time (ANOVA results are also
shown)

γ/(mg/mL)

Quercus Quercus
Oak species Quercus robur
petraea alba

Geographical location Limousin Galicia Allier USA

Samples 2 2 4 2

(15.6±1.3)
gallic acid <LOQ (58.2±9.2)a b <LOQ

vanillic acid <LOQ (2.8±1.5)a (2.4±0.7)a (2.1±0.0)a

Benzoic
acids
(12.2±13.0) (30.1±10.9) (14.3±4.5) (15.7±4.1)
benzoic acid a a a a

syringic acid (1.2±0.1)a (5.8±0.7)b (3.6±1.2)c (3.1±0.9)c

protocatechualdehyd
<LOQ <LOQ <LOQ <LOQ
e

Benzoic 4-
<LOQ (0.3±0.0)a (0.4±0.2)a <LOQ
aldehyde hydroxybenzaldehyde
s
vanillin (1.3±0.4)a (5.7±1.6)b (3.8±1.2)b (3.8±1.1)ab

syringaldehyde (3.2±1.2)a (12.2±3.7)b (8.2±2.2)b (8.2±2.4)b

p-coumaric acid <LOQ <LOQ <LOQ <LOD

ferulic acid <LOQ 0.9±0.0 <LOQ <LOQ

Cinnamic
acids
isoferulic acid (0.3±0.0)a (0.8±0.0)b (0.4±0.1)a (0.3±0.0)a

sinapic acid <LOQ (12.6±0.3)a (5.4±0.4)b <LOQ

Cinnamic coniferaldehyde (1.4±0.9)a (13.5±0.7)b (9.1±7.5)b (2.0±0.4)a

aldehyde
s sinapaldehyde (3.6±1.4)a (7.7±0.7)b (6.3±2.6)ab (4.2±1.2)ab
 γ/(mg/mL)

Benzoic
vanillyl alcohol <LOQ (2.5±0.4)a (0.9±0.1)b <LOD
alcohol

Ratio 1 gallic acid/vanillin – 10.2 4.1 –

syringaldehyde/vanilli
Ratio 2 2.5 2.1 2.1 2.1
n

*Different letters in superscripts within the same row indicate statistically significant
differences (p≤0.05) according to LSD test LOQ=limit of quantification, LOD=limit of
detection

Vanillin and syringaldehyde are phenolic compounds related to lignin. Vanillin was
detected in all analysed samples, with significantly lower concentration in the
samples aged in Quercus robur from Limousin and higher in distillates aged in the
same species from Galicia. Vanillin is the phenolic aldehyde which greatly influences
the aroma of distillates because of its low threshold value (320 µg/L) and adds
positive vanilla notes (35). In the analysed samples, syringaldehyde was the most
abundant phenolic aldehyde.

Thermal degradation of lignin leads to the formation of some phenolic alcohols (19).
The concentration of vanillyl alcohol was higher in the spirit aged in Quercus robur
(Galicia), whereas this compound was not present in distillates aged in Quercus
robur (Limousin) or Quercus alba.

Ferulic acid was used in this study as a discriminating compound of the oak wood
species because it can only be quantified in Quercus robur (Galicia). Similar results
were obtained by Canas et al. (7), showing that the brandies aged in Portuguese
oak (Quercus pyrenaica Willd.) contain higher values of this compound than other
brandies aged in other oak species. The geographical proximity and similar climatic
conditions of both areas (Galicia and northern Portugal) may be the main reasons to
explain these similarities, despite the fact that the spirits aged in two different species
of Quercus.

The presence and concentration of benzoic and cinnamic aldehydes, which arise
from lignin degradation, depend on the temperature applied during toasting (16, 36,
37). Results in this study showed that coniferaldehyde and sinapaldehyde were
present at low concentrations in samples from Quercus robur (Limousin) and
Quercus alba. In all cases, their individual values were higher than the corresponding
benzoic aldehydes, 4-hydroxybenzaldehyde and vanillin.

Fig. 2 shows the differences between the four groups of grape marc distillates
according to the total concentration of phenols from each family. Most of the
identified phenolic compounds were at higher concentrations in the orujo samples
aged in Quercus robur (Galicia) than in the samples aged in the same species of
Quercus from Limousin. Between the other two species, orujo distillates aged in
Quercus alba had lower concentrations of most of the determined phenolic
compounds than the distillates aged in Quercus petraea.

Fig. 2. Mean and standard deviation of phenolic compound families in the different
species of Quercus. QRL=Quercus robur (Limousin), QRG=Quercus robur (Galicia),
QPA=Quercus petraea (Allier), QA=Quercus alba (USA)

A significant effect of oak wood species was observed on 11 of the 15 studied

compounds (Table 4). Benzoic acid, gallic acid and syringaldehyde were the main
low molecular mass compounds in the analysed samples. The results are in
agreement with those previously obtained by dos Anjos et al. (32) in a study about
cachaça. The concentration of minor compounds such as vanillin, syringic and ferulic
acids was significantly higher in distillates aged in Quercus robur from Galicia.
Ferulic acid was only quantified in these samples. Distillates aged in Quercus robur
from Limousin had the lowest concentration of all determined phenolic compounds.
In the majority of cases, their concentrations were lower than their corresponding
detection and quantification limits.

Quality and authenticity of aged grape marc distillates determined by phenol

ratios

Van Jaarsveld et al. (18) and Gimenez-Martínez et al. (38) showed that the gallic
acid/vanillin ratio is influenced by the type of wood and it has been used to define
the quality of spirits. A higher gallic acid/vanillin ratio indicates medium to high quality
brandy (39). According to van Jaarsveld et al. (40), the gallic acid/vanillin ratio
increases with the level of toasting. In this study (Table 4) this relationship was
significantly higher in distillates aged in Quercus robur from Galicia (10.2) than in
distillates aged in Quercus petraea from Allier (4.1).

Other authors (39, 41) also used the syringaldehyde/vanillin ratio to evaluate the
quality of aged beverages. Usually the syringaldehyde content exceeds more than
twice the vanillin content in oak wood species (42). In this study, the
syringaldehyde/vanillin ratio values were in all cases near 2. A relationship between
syringaldehyde and vanillin in the range from 1.4 to 2.5 shows a balanced lignin
composition (38). No differences were found in this ratio among the species of oak
wood. The syringaldehyde/vanillin ratio may also be used to evaluate the possible
addition of commercial vanillin as flavouring to increase the aroma of aged distillate.
In addition, the relationship between benzoic aldehydes (vanillin and
syringaldehyde) and cinnamic aldehydes (coniferaldehyde and sinapaldehyde) can
be used to evaluate the authenticity of the aged distillates. Canas et al. (37) showed
that the relationship between both aldehyde families allows differentiating the type
of wood (chestnut or oak) used in the ageing process. In this study, the relationship
between both groups of aldehydes is shown in Fig. 3. Orujo aged in Quercus alba
had the highest ratio of benzoic/cinnamic aldehydes (1.95), whereas samples aged
in Quercus petraea had lower ratio than the unity (0.79). Similar values were
observed in the samples of orujo aged in Quercus robur from Galicia and from
Limousin, with a ratio of 1.07 and 0.84, respectively.

Fig. 3. Proportion of benzoic and cinnamic aldehydes as a function of the botanical

species. QRL=Quercus robur (Limousine), QRG=Quercus robur (Galicia),
QPA=Quercus petraea (Allier), QA=Quercus alba (USA)

Sensory analysis of aged grape marc distillates

Eight common descriptors (three for appearance and five for taste) were defined by
the judges to describe the samples. The mean intensity of each attribute was used
to define the sensory profile of the evaluated orujo samples (Fig. 4). The obtained
results showed that the oujo samples aged in Quercus robur from Galicia reached
the highest values of all descriptive parameters, which were close to the values for
samples aged in Quercus robur from Limousin. Samples aged in Quercus alba had
lower colour intensity and lower scores of most of the evaluated attributes. Orujo
samples aged in Quercus petraea and Quercus alba showed similar values in
positive taste parameters (sweet, dense and oily), but the ageing process in Quercus
petraea increased some negative notes, giving these distillates more spicy/pungent
and alcoholic taste. Samples aged in Quercus petraea had the lowest visual score,
having poor brightness and transparency.

Fig. 4. Mean sensory profile of the appearance and taste of aged grape marc
distillates. QRL=Quercus robur (Limousin), QRG= Quercus robur (Galicia),
QPA=Quercus petraea (Allier), QA= Quercus alba (USA)

Pearson correlation coefficients

Pearson correlation coefficients among all identified phenolic compounds are shown
in Table 5. The majority of the compounds were positively correlated, with values of
R above 0.6. Vanillin and syringaldehyde are compounds related to lignin, but no
correlation was established between them. However, vanillic and syringic acids, both
products of lignin degradation, showed a high positive correlation (0.714). Both
benzoic acids were also positively correlated with guaiacyl-type compounds
(syringaldehyde and sinapaldehyde). Gallic acid showed high positive correlation
with the majority of phenolic and cinnamic acids and with the corresponding
aldehydes. Cinnamic acids (ferulic, isoferulic, p-coumaric and sinapic acids) were
also highly correlated.
Table 5. Pearson correlation matrix (r) among phenolic compounds

Va 4- Co
Va Syr Be p- Fe Isof Sin Proto Syri Sin
nill Hydro nife
nill ing nz Cou rul erul api catec Va ng- ap-
yl xy- r-
ic ic oic mar ic ic c hu- nill ald ald
alc benza ald
aci aci aci ic aci aci aci aldeh in ehy ehy
oh ldehy ehy
d d d acid d d d yde de de
ol de de

– –
Gallic 0.5 0.7 0.9 0.58 0.9 0.92 0.9 0.6 0.5 0.5
0.1 0.584 0.0 0.536
acid 23 14 29 7 69 3 77 39 96 14
39 59

Vanillic 1.0 0.7 0.1 0.6 0.45 0.6 0.55 0.6 0.3 0.9 0.9 0.6
0.753 0.464
acid 00 14 17 67 8 34 5 82 14 64 14 81

–
Syringic 1.0 0.1 0.7 0.64 0.6 0.78 0.7 0.7 0.7 0.4
0.841 0.0 0.288
acid 00 32 64 4 97 7 79 66 04 71
12

– – –
Vanillyl 1.0 0.23 0.0 0.06 0.2 0.3 – 0.0
0.1 0.1 0.324 0.1
alcohol 00 4 03 1 43 01 0.140 49
57 20 29

Benzoic 1.0 0.44 0.9 0.92 0.9 0.2 0.7 0.7 0.6
0.721 0.694
acid 00 7 18 0 65 01 18 29 85

p- –
1.00 0.5 0.65 0.5 0.5 0.4 – 0.2
Coumaric 0.344 0.3
0 88 9 78 91 30 0.087 31
acid 06

–
Ferulic 1.0 0.91 0.9 0.7 0.7 0.5
0.671 0.0 0.536
acid 00 3 76 48 01 37
61

–
Isoferulic 1.00 0.9 0.6 0.6 0.5
0.704 0.0 0.465
acid 0 25 68 56 58
54

Sinapic 1.0 0.0 0.7 0.7 0.5

0.695 0.572
acid 00 25 61 14 92
 Va 4- Co
Va Syr Be p- Fe Isof Sin Proto Syri Sin
nill Hydro nife
nill ing nz Cou rul erul api catec Va ng- ap-
yl xy- r-
ic ic oic mar ic ic c hu- nill ald ald
alc benza ald
aci aci aci ic aci aci aci aldeh in ehy ehy
oh ldehy ehy
d d d acid d d d yde de de
ol de de

Protocate
0.0 0.7 0.7 0.4
chualdeh 1.000 0.352
13 78 71 42
yde

1.0 0.2 0.4 0.7

Vanillin 0.754
00 16 23 51

Syringald 1.0 0.9 0.6

0.470
ehyde 00 51 85

Sinapald 1.0 0.8

0.655
ehyde 00 38

4-
Hydroxyb 0.8
1.000
enzaldeh 72
yde

Coniferal 1.0
dehyde 00

View it in a separate window

Correlations higher than ±0.6 are shown in bold

Pearson correlations between phenolic compounds and taste attributes of aged

grape marc distillates were also evaluated (Table 6). Strong positive correlations
were found between, benzoic (gallic, syringic and benzoic) and cinnamic (ferulic,
isoferulic and sinapic) acids with negative descriptors of mouthfeel (astringent and
alcoholic notes). On the other hand, positive taste attributes, sweet, dense and oily,
showed strong positive correlations with the corresponding benzoic
(protocatechualdehyde and syringaldehyde) and cinnamic (sinapaldehyde)
aldehydes. The 4-hydroxybenzaldehyde and p-coumaric acid were negatively
correlated with the positive attributes, whereas vanillin, as alcohol, aldehyde and
acid showed strong positive correlation with the alcoholic note.

Table 6. Pearson correlation matrix (r) among phenolic compounds and sensory
attributes of taste
 Sweet Dense/oily Spicy/pungent Astringent Alcoholic

Gallic acid 0.098 0.256 0.363 0.726 0.841

Vanillic acid 0.183 0.487 –0.077 0.317 0.587

Syringic acid 0.100 0.358 0.190 0.591 0.869

Benzoic acid – –0.112 0.259 0.781 0.652

0.150

Protocatechualdehyde 0.438 0.714 –0.162 0.479 0.316

4- – –0.731 –0.090 –0.334 0.300

Hydroxybenzaldehyde 0.724

Vanillin 0.143 0.451 –0.038 0.549 0.726

Syringaldehyde 0.622 0.820 0.346 0.243 0.428

p-Coumaric acid – –0.826 –0.638 0.365 –0.023

0.799

Ferulic acid 0.119 0.303 0.274 0.782 0.819

Isoferulic acid 0.270 0.447 0.241 0.798 0.673

Sinapic acid 0.153 0.358 0.302 0.700 0.846

Coniferaldehyde – –0.577 0.354 0.307 0.583

0.500

Sinapaldehyde 0.779 0.736 0.882 –0.227 0.147

Vanillyl alcohol 0.366 0.555 0.422 0.543 0.735

Correlations higher than ±0.6 are shown in bold

Colour parameters and the three visual attributes were also correlated (Table 7). The
results showed that hue is the colour parameter that has the most influence on the
positive valorisation of the aged distillates.

Table 7. Pearson correlation matrix (r) among colour parameters and sensory
attributes of appearance
 Transparency Brightness Colour

Total phenolic index 0.191 0.161 0.454

Total phenols 0.724 0.590 –0.017

CI 0.536 0.496 0.584

h 0.851 0.876 0.660

Correlations higher than ±0.6 are shown in bold; CI=colour intensity, h=hue

Principal component analysis

Fig. 5 shows the score plot of the first two PCs, obtained with the individual phenols
and chromatic characteristics as variables, which explain 88.32% of the variability
among the samples. Fig. 5a shows that the first principal component, PC1 (77.91%),
was positively correlated with all studied variables, whereas the second principal
component, PC2 (10.40%), was mainly positively correlated with the benzoic and
cinnamic aldehydes and negatively with gallic and isoferulic acids and total phenols.
Four groups of samples plotted on the plane defined by the two first principal
components can be observed in Fig. 5b. Samples from group 1 (QRG60 and
QRG72) were better characterized by all variables associated with the positive side
of PC1, mainly by gallic, syringic and sinapic acids and by total phenols and colour
intensity. In contrast, samples included in group 2 (QRL13 and QRL16) were
scarcely characterized by these variables, showing differences among samples
aged in the same species of oak wood of different origin. Group 3 (QA72A and
QA72B) was characterized by volatiles on the positive side of PC2, mainly
syringaldehyde and sinapaldehyde. Group 4 was composed of distillates aged in Q.
petraea (QPA30, QPA72A, QPA72B and QPA144) in the centre of the plot. The PCA
analysis clearly showed a good separation of the aged orujo samples according to
the species and origin of the oak wood employed in the ageing process, independent
of the time of ageing.
Fig. 5. Principal component analysis score plot of: a) aged grape marc distillates, and b)
phenolic compound variables. QRL= Quercus robur (Limousin), QRG=Quercus robur
(Galicia), QPA= Quercus petraea (Allier), QA=Quercus alba (USA). The number reflects the
ageing time. A=sinapaldehyde, B=syringaldehyde, C=vanillin, D=syringic acid, E=hue,
F=colour intensity, G=vanillyl alcohol, H=sinapic acid, I=gallic acid, J=total phenols, K=
isoferulic acid

Conclusions
The results obtained in this study provide the first data on the phenolic composition
of the aged grape marc distillate (orujo) and contribute to the knowledge about this
alcoholic beverage. Benzoic acid, gallic acid and syringaldehyde were the main low
molecular mass compounds. Distillates aged in Quercus robur from Galicia showed
the highest concentration of the majority of the determined phenolic compounds,
whereas samples aged in Quercus robur from Limousin had the lowest
corresponding values. These results showed the influence of the growth origin on
the oak composition. Ferulic acid was only detected in orujo samples aged in
Quercus robur from Galicia. Consequently, this compound can be used as a
discriminant among the three oak wood species in this study. Most of the determined
phenolic compounds and colour parameters were positively correlated with each
other and with the sensory attributes defined by the tasters. No significant
differences of the Folin-Ciocalteu indices were shown among the analysed aged
orujo beverages; however, total phenols, colour intensity and hue were significantly
higher in orujo aged in Quercus robur from Galicia. Principal component analysis
allowed the classification of the aged distillate samples according to the origin and
species of oak wood. However, the results shown in this study must be completed
with those obtained from an experimental design using a unique distillate aged in
different oak species during the same time. Other variables such as toast level and
contact time (oak-distillate) must also be taken into account in deciding which type
of oak is the most suitable for ageing grape marc spirits. In this research analytical
characterization of orujo alcoholic beverages is presented for the first time.

Acknowledgements

We are grateful for the financial support of this work to the Spanish Ministry of
Science and Innovation (project CTQ2011-28967), and partial financial support from
the FEDER funds of the European Union and to the Regulatory Board of
Geographical Names of the Traditional Galician Spirits and Liqueurs for providing
the samples. José Manuel Salgado is grateful for postdoctoral fellowship (EX-2010-
0402) of Education Ministry of Spanish Government.

References

1. De Rosso M, Panighel A, Dalla Vedova A, Stella L, Flamini R. Changes in

chemical composition of a red wine aged in acacia, cherry, chestnut, mulberry and
oak wood barrels. J Agric Food Chem. 2009;57:1915–20. 10.1021/jf803161r
[PubMed] [Cross Ref]
2. Lee KYM, Paterson A, Piggot JR, Richardson GD. Origins of flavour in whiskies
and a revised flavour wheel: A review. J Inst Brew. 2001;107:287–313.
10.1002/j.2050-0416.2001.tb00099.x [Cross Ref]
3. Litchev V. Influence of oxidation processes on the development of the taste and
flavour of wine distillates. Am J Enol Vitic. 1989;40:31–5.
4. Onishi MJ, Guymon JF, Crowlell EA. Changes in some volatile constituents of
brandy during aging. Am J Enol Vitic. 1977;28:152–8.
5. Rodríguez Dodero MC, Guillén-Sánchez DA, Schwarz-Rodríguez M, García-
Barroso C. Phenolic compounds and furanic derivates in the characterization and
quality control of Brandy de Jerez. J Agric Food Chem. 2010;58:990–7.
10.1021/jf902965p [PubMed] [Cross Ref]
6. Ebeler SE, Terrien MB, Butzke CE. Analysis of brandy aroma by solid-phase
microextraction and liquid–liquid extraction. J Sci Food Agric. 2000;80:625–30.
10.1002/(SICI)1097-0010(200004)80:5<625::AID-JSFA584>3.0.CO;2-5 [Cross Ref]
7. Canas S, Casanova V, Belchior AP. Antioxidant activity and phenolic content of
Portuguese wine age brandies. J Food Compos Anal. 2008;21:626–33.
10.1016/j.jfca.2008.07.001 [Cross Ref]
8. Goldberg DM, Hoffman B, Yang J, Soleas GJ. Phenolic constituents, furans, and
total antioxidant status of distilled spirits. J Agric Food Chem. 1999;47:3978–85.
10.1021/jf9811626 [PubMed] [Cross Ref]
9. Garde-Cerdán T, Lorenzo C, Carot JM, Esteve MD, Climent MD, Salinas MR.
Effect of composition, storage time, geographic origin and oak type on the
accumulation of some volatile oak compounds and ethylphenols in wines. Food
Chem. 2010;122:1076–82. 10.1016/j.foodchem.2010.03.077 [Cross Ref]
10. Matricardi L, Waterhouse AL. Influence of toasting technique on color and
ellagitannins of oak wood in barrel making. Am J Enol Vitic. 1999;50:519–26.
11. Miller DP, Howell GS, Michaelis CS, Dickmann DI. The content of phenolic acid
and aldehyde flavor components of white oak as affected by site and species. Am J
Enol Vitic. 1992;43:333–8.
12. Prida A, Puech JL. Influence of geographical origin and botanical species on the
content of extractives in American, French, and East European oak woods. J Agric
Food Chem. 2006;54:8115–26. 10.1021/jf0616098 [PubMed] [Cross Ref]
13. Rodríguez-Rodríguez P, Gómez-Plaza E. Differences in the extraction of volatile
compounds from oak chips in wine and model solutions. Am J Enol Vitic.
2011;62:127–32. 10.5344/ajev.2010.10045 [Cross Ref]
14. Rodríguez-Rodríguez P, Gómez-Plaza E. Dependence of oak-related volatile
compounds on the physicochemical characteristics of barrel-aged wines. Food
Technol Biotechnol. 2012;50:59–65.
15. Spillman PJ, Sefton MA, Gawel R. The effect of oak wood source, location of
seasoning and coopering on the composition of volatile compounds in oak-matured
wines. Aust J Grape Wine Res. 2004;10:216–26. 10.1111/j.1755-
0238.2004.tb00025.x [Cross Ref]
16. Canas S, Belchior AP, Spranger MI, Bruno de Sousa R. HPLC method for the
quantification of phenolic acids, phenolic aldehydes, coumarins and furanic
derivatives in different kinds of toasted wood used for the ageing of brandies. Anal
Methods. 2011;3:186–91. 10.1039/C0AY00269K [Cross Ref]
17. Chatonnet P, Dubourdieu D. Comparative study of the characteristics of
American white oak (Quercus alba) and European oak (Quercus petraea and
Quercus robur) for production of barrels used in barrel ageing wines. Am J Enol Vitic.
1998;49:79–85.
18. van Jaarsveld FP, Hattingh S, Minnaar P. Rapid induction of ageing character in
brandy products. Part II. Influence of type of oak. S. Afr. J. Enol. Vitic. 2009;30:16–
23.
19. Cadahía E, Muńoz L, Fernández de Simón B, García-Vallejo MC. Changes in
low molecular weight phenolic compounds in Spanish, French and American oak
woods during natural seasoning and toasting. J Agric Food Chem. 2001;49:1790–8.
10.1021/jf0006168 [PubMed] [Cross Ref]
20. de Simón B, Cadahía E, Jalocha J. Volatile compounds in Spanish red wine aged
in barrels made of Spanish, French and American oak wood. J Agric Food Chem.
2003;51:7671–8. 10.1021/jf030287u [PubMed] [Cross Ref]
21. Guchu E, Díaz-Maroto MC, Díaz-Maroto IJ, Vila-Lameiro P, Pérez-Coello MS.
Influence of the species and geographical location on volatile composition of spanish
oak wood (Quercus petraea Liebl. and Quercus robur L.). J Agric Food Chem.
2006;54:3062–6. 10.1021/jf053055z [PubMed] [Cross Ref]
22. Regulation of the Geographic Indications of Traditional Spirits and Liqueurs from
Galicia, DOG No. 10 (2012) 2516-2545.
23. Rodríguez-Solana R, Rodríguez N, Dominguez JM, Cortés S. Characterization
by chemical and sensory analysis of commercial grape marc distillate (orujo) aged
in oak wood. J Inst Brew. 2012;118:205–12. 10.1002/jib.25 [Cross Ref]
24. Rodríguez-Solana R, Salgado JM, Domínguez JM, Cortés S. Assessment of
minerals in aged grape marc distillates by FAAS/FAES and ICP-MS.
Characterization and safety evaluation. Food Contr. 2014;35:49–55.
10.1016/j.foodcont.2013.06.031 [Cross Ref]
25. Spigno G, Tramelli L, De Faveri DM. Effects of extraction time, temperature and
solvent on concentration and antioxidant activity of grape marc phenolics. J Food
Eng. 2007;81:200–8. 10.1016/j.jfoodeng.2006.10.021 [Cross Ref]
26. Singleton VL, Rossi JA., Jr Colourimetry of total phenols with phosphomolybdic-
phosphotungstic acid reagents. Am J Enol Vitic. 1965;16:144–58.
27. Ribéreau-Gayon P. The amount of total phenolic compounds in red wines. Chim.
Anal. 1970;52:627–31. [in French]
28. Pérez-Magarińo S, González-San José ML. Evolution of flavanols, anthocyanins,
and their derivatives during the aging of red wines elaborated from grapes harvested
at different stages of ripening. J Agric Food Chem. 2004;52:1181–9.
10.1021/jf035099i [PubMed] [Cross Ref]
29. Del Alamo Sanza M, Fernández Escudero JA, De Castro Torío R. Changes in
phenolic compounds and colour parameters of red wine aged with oak chips and in
oak barrels. Food Sci Technol Int. 2004;10:233–41. 10.1177/1082013204046095
[Cross Ref]
30. Analysis S. General Guidance for the Design of Test Rooms, ISO 8589:2007,
International Organization for Standarization (ISO), Geneva, Switzerland.
31. Analysis S. Identification and Selection of Descriptors for Establishing a Sensory
Profile by a Multidimensional Approach ISO 11035:1994, International Organization
for Standarization (ISO), Geneva, Switzerland.
32. Anjos JP, das Graças Cardoso MD, Saczk AA, Dórea HS, Santiago WD,
Machado AMR, et al. Evolution of the concentration of phenolic compounds in
cachaça during aging in an oak (Quercus sp.) barrel. J Braz Chem Soc.
2011;22:1307–14. 10.1590/S0103-50532011000700016 [Cross Ref]
33. Cabrita MJ, Barrocas Dias C, Costa Freitas AM. Phenolic acids, phenolic
aldehydes and furanic derivatives in oak chips: American vs. French oaks. S. Afr. J.
Enol. Vitic. 2011;32:204–10.
34. Canas S, Belchior AP, Falcăo A, Gonçalves JA, Spranger MI, Bruno de Sousa
R. Effect of heat treatment on the termal and chemical modifications of oak and
chestnut wood used in brandy ageing, Cięncia Téc. Vitiv. 2007;22:5–14.
35. Boidron JN, Chatonnet P, Pons M. Effects of wood on aroma compounds of
wine. Conn. Vigne Vin. 1988;22:275–94.
36. Canas S, Leandro MC, Spranger MI, Belchior AP. Low molecular weight organic
compounds of chestnut wood (Castanea sativa L.) and corresponding aged
brandies. J Agric Food Chem. 1999;47:5023–30. 10.1021/jf9900480 [PubMed]
[Cross Ref]
37. Canas S, Quaresma H, Belchior AP, Spranger MI, Bruno de Sousa R. Evaluation
of wine brandies authenticity by the relationships between benzoic and cinnamic
aldehydes and between furanic aldehydes, Cięncia Téc. Vitiv. 2004;19:13–27.
38. Giménez-Martínez R, López-García de la Serrana H, Villalón-Mir M, Navaro-
Alarcón M, Olalla-Herrera M, Cabrera-Vique C, et al. Study of vanillin,
syringaldehyde and gallic acid content in oak wood and wine spirits mixtures:
Influence of heat treatment and chip size. J Wine Res. 2001;12:175–82.
10.1080/09571260120106811 [Cross Ref]
39. Gómez-Cordovés C, Bartolomé B, Jimeno ML. Identification of 2,3-dihydroxy-1-
guaiacylpropan-1-one in brandies. J Agric Food Chem. 1997;45:873–6.
10.1021/jf9605613 [Cross Ref]
40. van Jaarsveld FP, Hattingh S, Minnaar P. Rapid induction of ageing character in
brandy products. Part III. Influence of toasting. S. Afr. J. Enol. Vitic. 2009;30:24–37.
41. Puech JL, Moutounet M. Phenolic compounds in an ethanol-water extract of oak
wood and in a brandy. Lebenson Wiss Technol. 1992;25:350–2.
42. Vlassov VN, Maruzhenkov DS. Application of GC/MS method for the
identification of brandies and cognacs. Analusis. 1999;27:663–7.
10.1051/analusis:1999136 [Cross Ref]

XIX. Shelf Life of Anchovy Products (Engraulis

Encrasicolus): Evaluation of Sensory, Microbiological
and Chemical Properties
Ital J Food Saf. 2014 Jan 21; 3(1): 1678.
Andrea Ariano, 1 Luigi Scarano,1 Amalia Mormile,1 Maria Barile,1 Giuseppe Palma,2
and Nicoletta Murru1

Abstract

Fishery products have always been an important food in Italy. In the past, increased
consumption was mainly due to the good quality of the products, easiness of use
and their beneficial effects on health. Recently, owing to the national financial crisis,
there has been a decline in the consumption of fish. In fact, in 2013, according to
data from ISMEA, the consumption of fresh fish suffered a sharp contraction (-5%).
This decline also concerns anchovy (Engraulis encrasicolus). This species, partly
because of its low price, is a mainstay of traditional Italian food. The aim of this study
was to evaluate sensorial, chemical and microbiological properties of anchovy-
based (Engraulis encrasicolus) products during storage at 4 and -20°C. Fresh
anchovies, obtained from the wholesale fish market of Pozzuoli (Southern Italy) were
cut into small pieces and hand-prepared using bread, eggs, cheese and lemon juice.
Samples were analysed after 0, 2, 4, 6 and 8 days of storage at 4°C. An aliquot was
quickly frozen and analysed after 34 days at -20°C. Sensory assessment,
microbiological (specific spoilage organisms, Listeria spp. and Salmonella spp.) and
chemical (histamine, total volatile basic nitrogen, trimethylamine, thiobarbituric acid,
pH and aw) analyses were performed. Results showed that the shelf life of anchovy
products was less than 5 days for the samples stored at 4°C. At -20°C, all anchovies
preparations showed good sensory, microbiological and chemical properties for 34
days.

Key words: Engraulis encrasicolus, Fish preparations, Shelf life, SSOs, Histamine

Introduction

Fishery products have always been an important food In Italy. In the past, increased
consumption was mainly due to the good quality of the products, ease of use and
their beneficial effects on health.

Currently, there is a strongly decline in consumption of fish mainly related to the

financial crisis so much so that in 2013 the consumption of fresh fish by Italian people
suffered a sharp decline (-5%) (ISMEA, 2013). This negative trend also regards the
anchovy (Engraulis encrasicolus) market. This fish species has always been a part
of traditional, poor cuisine because, despite its status as the most prized of small
fish, it actually costs little. Although the anchovy market is historically linked to the
sale of fresh produce and produce preserved in salt and oil, it could be important to
diversify this by developing new kind of products. The aim of this trial was to assess
the shelf-life of anchovy (Engraulis encrasilocus) meatballs stored at 4°C for 8 days
and at -20°C for 1 month by evaluating their microbiological, chemical and sensorial
features.

Materials and Methods

For the study, a batch of fishballs was made at the workshop at Pozzuoli fish market
using anchovies (Engraulis encrasilocus) fished no more than 5 h before in the
nearby Gulf. The fishballs, which were bigger than a walnut, were obtained by
manually mixing anchovies (cut with scissors), bread, eggs, cheese and lemon juice
and coating the same with bread crumbs at the end. Then, 8 meatballs at a time
were packed in aluminum semi airtight containers covered with foil and stocked at
4°C. One container was frozen and stored at -20°C for more than a month. All the
refrigerated samples and raw materials were submitted to sensorial, chemical and
microbiological analyses after 0, 2, 6, 8 storage days. The frozen sample, on the
other hand, was analysed after 34 days of storage. All the analyses were performed
at a food inspection laboratory.

Sensorial analyses

All samples were subjected to sensorial analyses by a test panel of 5 qualified people
who evaluated the colour, flavour, texture and the presence of liquid in the
containers. Afterwards, the meatballs were cooked in hot oil as preparation for a
taste test.

Microbiological analyses
For the microbiological analysis, 25 gr of sample were homogenised in 225 mL of
(w/v) sterilised 0.2% peptone water at 45°C in a Stomacher Lab-Blender 400 (PBI
International, Milan, Italy) and consecutive ten-fold serial dilutions were made in
sterilised 0.2% peptone water. Aerobic mesophilic counts (AMC) were enumerated
on plate count agar (PCA; Oxoid, Milan, Italy) after 48 h and 15 days of aerobic
incubation at 32 and 4°C, respectively, enterobacteriaceae on violet red bile glucose
agar (VRBG; OXOID), using the pour-plate and overlay technique, after 48 h of
incubation at 37°C, Pseudomonas spp. on pseudomonas agar base+Pseudomonas
C-N-C Supplement (OXOID) at 22°Cx48 h; Aeromonas spp. on Aeromonas Medium
Base+Ampicilin Supplement (OXOID) at 30°Cx48 h, Brochothrix thermosphacta on
SPAA Agar Base+STAA selective Supplement (OXOID) at 22°C x 48 h, Listeria spp.
and Salmonella spp. according to ISO 6579:2008 (ISO, 2008) and ISO 11290-
1:1996 (ISO, 1996), respectively. All analyses were performed in duplicate and the
mean value of each count was expressed in Log cfu/g.

Chemical analyses

In all samples, pH (pH-meter FE20; Mettler, Toledo, Spain) were determined: a w

(activity water, con HigroLab rotonic PBI), total volatile basic nitrogen (TVB-N)
(Antonacopoulos and Vincke, 1989), trimethylamine nitrogen (TMA-N) (AOAC,
1984), Thiobarbituric acid (TBA) (Pearson, 1973), Histamine (Emborg and Dalgaard,
2006).

Results

Sensorial analyses

After 4 days of storage, anchovy fishballs stored at 4°C showed a lively and bright
colour, a firm texture, had a pleasing smell and there was no liquid in the package.
From the 6th day of storage, they were grayish in colour, had a pungent odour, less
firm texture and there was a turbid liquid in the package. On the 8th storage day, a
complete loss of freshness was observed. Regarding the taste test, the anchovy
fishballs were acceptable up to the 4th day of storage. After thawing, the frozen
samples showed sensorial features similar to those of the refrigerated fishballs on
the first day of storage.

Microbiological analyses

In the refrigerated samples, TVC at 32 and 4°C, and enterobacteriaceae after 8 days
of storage reached values of 8.63, 6.23 and 4.89 log CFU/g, respectively (Figure 1).
Aeromonas spp. and Pseudomonas spp. increased gradually and constantly during
storage (Figure 2).
Figure 1.. Average values of total viable count at 32 and 4°C, and Enterobacteria in
anchovy fishballs.

Figure 2. Average values of Aeromonas spp. and Pseudomonas spp. in anchovy

fishballs.

In the frozen sample, TVC at 32°C and enterobacteriaceae showed lower values like
those registered at the beginning of storage. On the contrary, TVC at 4°C increased
by 4.59 log CFU/g compared to the values of raw material (Figure 1). Aeromonas
spp. and Pseudomonas spp. were absent in the frozen sample (Figure 2).
Brochothrix thermosphacta, Listeria spp. and Salmonella spp.

All bacteria results are shown in Figures 1 and and22.

Chemical analyses

TVB-N and TMA reached 12.14 and 34.42 mg/100gr on the 6 th day, 22.14 and
35.42/100 gr after 8 days of storage at 4°C (Figure 3). In the chilled product, on 34th
day values of 20.45 and 4.8 mg/100 gr for TVB-N and TMA respectively were
revealed. From an initial value of 5 mg/100 gr, histamine increased during
refrigerated storage up to 20 mg/100 g on 8th day (Figure 3). As regards TBA,
concentrations of malonaldehyde of 0.98 mg/Kg in the refrigerated samples on 8th
day and of 0.16 mg/Kg in the chilled sample were detected. In all samples, pH was
6.17 on first day and then decreased and showed the mean value of 5.64 until the
end of storage. Aw showed the mean value of 0.95 during entire storage in all
samples.

Figure 3. Average values of total volatile basic nitrogen, trimethylamine and

hystamine in anchovy fishballs.

Chemical results are shown in Figure 3.

Discussion and Conclusions

Pseudomonas spp. and Aeromonas spp. are widely recognised as specific spoilage
organisms (SSOs) of fish products (Gram and Huss, 1996; Gram et al., 2002;
Dalgaard et al., 1995) deteriorating them when exceeding the value of 5x105 UFC/g
(ICMSF, 1986) In this trial, these microorganisms showed a low charge in the raw
material. From the 5th day, the gradual increase of TVC at 32 and 4°C, Aeromonas
spp. and Pseudomonas spp. was accompanied by a gradual loss in freshness
(Alasalvar et al., 2001).

TVB-N and TMA was higher than the limits of 35 and 20 mg/100g in the refrigerated
samples on the 8th day of storage. Histamine showed values lower than the limits of
Reg. 2073/2005 (European Commission, 2005) until the end of storage. Regarding
TBA, the concentration of malonaldehyde was always lower than the limit of 5 mg/Kg
for products widely oxidised.

The frozen product had good sensorial, microbiological and chemical features after
34 days of storage at -20°C thanks to the rapid reduction of the temperature and to
early processing times after the capture.

Our results allow us to define a shelf life of 4 days for refrigerated anchovy meatballs
and of 34 days for the chilled product.

The anchovy is a fish species which is still under appreciated despite its high
nutritional value. The development of new anchovy preparations could promote the
consumption of this fish species especially by particular categories of consumers
such as the elderly and children. Of course, in order to deliver a flawless product, it
is necessary to start with fresh raw material and to respect the cold chain.

References

1. Alasalvar C, Taylor KDA, Öksüz A, Garthwaite T, Alexis MN, Grigorakis K,

2001. Freshness assessment of cultured sea bream (Sparus aurata) by
chemical, physical and sensory methods. Food Chemistry 72:33-40.
2. Antonacopoulos N, Vyncke W, 1989. Determination of volatile basic in fish. Z
Lebensm Unters For 189:309-16.
3. AOAC, 1984. Official methods of analysis of the Association of Official
Analytical Chemist. 14th edition Association of Official Analytical Chemist ed.,
Washington.
4. Dalgaard P, 1995. Qualitative and quantitative characterization of spoilage
bacteria from packed fish. Int J Food Microbiol 26:319-33. [PubMed]
5. Emborg J, Dalgaard P, 2006. Significant histamine formation in tuna
(Thunnus albacares) at 2°C, effect of vacuum and modified atmosphere
packing on psychrotolerant bacteria. J Food Protect 694:897-906. [PubMed]
6. European Commission, 2005. Regulation of the European Parliament and of
the Council of 15 November 2005 on microbiological criteria for foodstuffs,
2073/2005/EC. In: Official Journal, L 338, 22/12/2002.
7. Gram L, Huss HH, 1996. Microbiological spoilage of fish and fish products.
Int J Food Microbiol 33:121-37. [PubMed]
8. Gram L, Ravn L, Rasch M, Bruhn JB, Christensen AB, Givskov M, 2002. Food
spoilage: interactions between food spoilage bacteria. Int J Food Microbiol
78:79-97. [PubMed]
9. ICMSF, 1986. Microorganisms in foods 2. Sampling for microbical analysis:
profiles and specific applications. 2nd ed. University of Toronto Press, Buffalo,
NY.
10. ISMEA, 2013. Trend of consumption of fish in Italy. Istituto di servizi per il
mercato agricolo alimentare ed., Rome.
11. ISO, 1996. Microbiology of food and animal feeding stuffs. Horizontal method
for the detection and enumeration of Listeria monocytogenes. Part 1:
 detection method. ISO norm 11290-1:1996. International Organization for
Standardization Publ., Geneva.
12. ISO, 2008. Microbiology of food and animal feeding stuffs. Horizontal method
for the detection of Salmonella spp. ISO norm 6579:2008. International
Organization for Standardization Publ., Geneva.
13. Pearson D, 1973. Laboratory techniques in food analysis. Butterworths,
London.

XX. Sensorial evolution of cassava flour (Manihot

esculenta crantz) added to protein concentrate
cassava leaves
Food Sci Nutr. 2013 Sep; 1(5): 357–362.
Elaine C S Lima,1 Márcia B S Feijo,2 Maria C J Freitas,2 Edna R dos Santos,3
Armando U O Sabaa-Srur,1 and Luciana S M Moura1
Abstract

Cassava is regarded as the nutritional base of populations in developing countries,

and flour, product made of cassava, is the most consumed in the world. The cassava
leaves are very rich in vegetable proteins, but a big amount is lost in processing the
crop. The objective of this study was to do a sensory evaluation of cassava flour to
which a protein concentrate obtained from cassava leaves (CPML) was added. The
CPML was obtained from cassava leaves by isoelectric precipitation and added to
cassava paste for preparation of flour in three parts 2.5, 5, and 10%. The acceptance
test was done by 93 consumers of flour, using hedonic scale of 7 points to evaluate
characteristics like color, scent, flavor, bitterness, texture, and overall score. By the
method of quantitative descriptive analysis (QDA), eight trained tasters evaluated
the following characteristics: whitish color, greenish color, cassava flavor, bitter
flavor, characteristic flavor, lumpiness, raw texture, leaf scent, and cassava scent.
The acceptability test indicated that flour cassava with 2.5 was preferred. Whitish
color, greenish color, cassava flavor, bitter flavor, salty flavor, characteristic flavor,
lumpiness texture, raw texture, and the smell of the leaves and cassava flour were
the main descriptors defined for flour cassava with CPML has better characteristics.

Keywords: Acceptance test, protein concentrate, sensorial

Introduction

The cassava is considered an alimentary base for people in Africa, Asia, and Latin
America assuming socioeconomical position in the world, due to its high capacity to
adapt to climatic conditions. Its roots can be easily cultivated and it is the principal
source of carbohydrates for needy people with protein deficiency, which is one of
primary factors of human malnutrition that affects a big part of population. This
deficiency can be the result of lack of protein containing foods, of both animal and
vegetative origin, and its high price.

This vegetable is characterized by its high concentration of carbohydrates that is

why it is considered a caloric aliment. The cassava roots also contain vitamin C,
carotenoids, thiamine, riboflavin, and nicotinic acid. They also represent
considerable quantities of calcium and phosphorus (NEPA/UNICAMP 2006).

The cassava leaves have been examined in Brazil and different countries due to its
nutritional characteristics and big waste in the field (Modesti 2006). High levels of
protein in the cassava leaves were mentioned in different works, with its variety from
20.77 to 35.9 g/100 g (Madruga and Câmara 2000; Ortega-Flores et al. 2003; Wobeto
2003).

The utilization of biomass became a world concern, especially in Brazil, the country,
which is rich in variety and quantity of foods. However, thousands of Brazilians are
still starving, while a big part of biomass is wasted as in the case of subsistence
plants, and the cassava (Manihot esculents Crantz) is one of them. During the
harvest, it has its aerial part left in the field (Ferri 2006). The leaves contain important
nutrients which could be used in a diet as a protein portion (Corrêa et al. 2004).

Sensorial evaluation is a very important tool to evaluate new dietary products

because it helps to obtain aliments that are pleased a consumer. Hence, the
objective of this work was to evaluate the addition of protein concentrate cassava
leaves (CPML) to cassava flour by acceptance test to consumer level and by
quantitative descriptive analysis (QDA) with trained tasters.

Material and Methods

Obtaining of cassava flour with addition of CPML

The CPML was obtained from cassava leaves by isolectric precipitation (Lima et al.
2011). Processing of cassava leaves with CPML consisted of the following steps:

selection of cassava leaves, washing to remove the dirt, manual peeling, washing in
the running water, sanitation, triturating, pressing, adding of CPML to mass in three
concentrations (2.5, 5.0, and 10%), drying in ventilated stove at 60°C, milling,
screening, roasting with homogenization with manual mixer, cooling, and filling.
Therefore, four formulations of cassava flour were produced: 0.0% CPML – F0; 2.5%
CPML – F1; 5.0% CPML – F2, and 10% CPML – F3.

This sensorial analysis study was previously approved by The Committee of Ethics
of Federal University of Rio de Janeiro (UFRJ), No. 069/2010.

Acceptability test

The acceptance test was done with one sample of cassava flour without CPML
(control) and three samples of cassava flour with 2.5, 5, and 10% CPML,
respectively. There were 93 nontrained tasters in completely randomized groups, 10
g of encoded samples with three digits, accompanied with one glass of water at room
temperature to clean the mouth and tongue before each evaluation (Dutcosky 2007).

To study the acceptance of the cassava flour, a hedonic scale of 7 points was used
(7 = liked a lot and 1 = disliked a lot) to evaluate appearance, scent, flavor, texture,
and overall score. The hedonic results were expressed in graphics of frequency
distribution and variety analysis (analysis of variance [ANOVA]) and comparative
test of averages (Turkey), according to Stone (1994) and Stone and Sidel (1992) using
statistic program XLSTAT, estimation error with of 5%.

Quantitative descriptive analysis

A QDA test was used to characterize and quantity the sensory attributes of control
flour and with 2.5, 5, and 10% CPML, respectively, was determined using the
technique of QDA, according to Stone (1994). The tests were performed at the
laboratory of Aliments Analysis and Processing, in the Nutrition Institute of Josue de
Castro of UFRJ, which constructions include individual cabins and illumination
control.

Recruiting of tasters

There were recruited 16 individuals among students and employees of the Nutrition
Institute of Josue Castro of UFRJ. The criteria to form a team were to be a consumer
of cassava flour and to be selected in teams by selective tests as its sensorial
perceptiveness is normal (recognition of basic flavors and smells).

Development of descriptive terminology

The following products were chosen to establish descriptive terms: distillated water,
wheat flour, toast, lemon gelatin powder, cooked cassava, 0.06% solution of caffeine
in water, cookies cream cracker, corn flour, and cassava leaf.

The products were served in disposable materials, encoded with numbers of three
digits, containing 10–20 g of each product. Panelists were asked to describe the
characteristics of the samples related to the color, flavor, texture, and smell. Next, a
list of all used terms and number of times that they were reported was prepared. This
led to discussions in groups with the objective of creating descriptive terms and
reference samples of scale extremes.

The result of four evaluation sessions of reference samples and discussions in group
was the evaluation list of control flour and flour with 2.5, 5, and 10% CPML.

Panelist selection and descriptive analysis of samples

The samples were analyzed initially by 16 panelists, in triplicate, in four sessions in

subsequent days. Cards for recording descriptions were used by all team members.
Each descriptive term was evaluated in an unstructured line scale, anchored at
extremes with the following terms of intensity: strong and weak. After the panelists
had completed their test, the distance from the left end of the line to the point marked
by the panelist was measured.

The flour was served in paper cups, encoded with numbers of three digits,
accompanied by a glass of water at room temperature. The tests were done as open
trials. ANOVA was done on the results of each taster for every evaluated
characteristic, considering variation point samples and repetition. The final
descriptive panel was chosen, those panelist that represented discriminative ability
Psample < 0.05, good reproducibility at trials (Prepetition > 0.05), and accordance with
other members. So, the team was formed of eight tasters.

The obtained data were submitted to variance analysis (ANOVA) and comparative
test of averages (Turkey), estimation error with of 5%, according to Stone (1994) and
Stone and Sidel (1992) using statistic program XLSTAT.
Results and Discussion

Acceptability test

In Table Table1,1, the average of global impression of elaborated cassava flour for
the following sensorial characteristics: color, smell, flavor, bitterness, texture, and
overall score are presented. These average rates were assigned to the hedonic
scale of 7 points. So in Figure Figure1,1, the distribution of testers according to the
analyzed characteristics of elaborated cassava flour is represented. F0 (0% CPML),
F1 (2.5% CPML), F2 (5% CPML), and F3 (10% CPML).

Table 1. Average grades of testers for sensorial characteristics of elaborated

cassava flour

Characteristics F0 (0% CPML) F1 (2.5% CPML) F2 (5% CPML) F3 (10% CPML)

Color 5.18ab 5.28a 5.08ab 4.83b

Smell 4.80a 4.68a 4.95a 4.30a

Flavor 4.25ab 4.62a 4.00b 3.14c

Bitterness 4.10a 4.41a 4.08a 3.02b

Texture 4.57ab 4.73a 4.47b 4.41ab

Overall score 4.89a 5.15a 4.67a 3.75b

The same letters in the same line mean that there was not a significant difference (P
> 0.05) Turkey test; 7, liked very much; 6, liked a lot; 5, liked; 4, indifferent; 3, did not
like; 2, did not like a lot; 1, did not like very much.
Figure 1. Distribution of providers by the evaluated characteristics: color, smell,
flavor, bitterness, texture, and overall score, using hedonic scale.

The evaluations of color ranged from liked to liked a lot for flour F0, F1, and F2, but
for the flour F3 they varied from indifferent and liked. It was noticed that there was
no difference in the smell of the flour, the data of tasters varied from indifferent and
liked. The flavor was pleasant for flour with 2.5% CPML, followed by flour with 0%
CPML, 5% CPML, and 10% CPML. While regarding texture, this characteristic
represented the impressions varying between indifferent and liked.

Quantitative descriptive analysis

Descriptive terminology

Nine verbal terms were developed by the team of tasters to describe the similarities
and differences among evaluated cassava flour samples. Each one of descriptors
used was defined by the team of tasters and qualitative references were associated
to each term (Table (Table22).

Table 2. List of descriptive terms, definitions, and references for each

characteristic

Descriptive
Definition Reference
term

Color

Characteristic color of raw Strong – wheat flour Weak –

Whitish color
manioc flour toasted manioc flour

Strong – lemon gelatin powder

Greenish color Characteristic color of CPML
No one – wheat flour

Flavor

Strong – cooked manioc No one

Manioc flavor Characteristic manioc flavor
– distillated water

Characteristic bitter flavor Strong – 0.06% caffeine solution

Bitter flavor presented in the caffeine in water No one – distillated
solution water

Strong – cookie cream cracker

Salty flavor Flavor of saline solution
No one – distillated water

Strong – manioc flavor

Characteristic Characteristic flavor of manioc
commercial No one – distillated
flavor flour
water

Texture

Thick granules were noticed Weak – wheat flour Strong – corn

Lumpiness
during the act of swallowing flour

Characteristic texture of wheat Strong – row wheat flour Weak –

Raw
flour toasted manioc flour

Smell
Descriptive
Definition Reference
term

Characteristic smell of fresh Strong – 1 cut leave Weak –

Leave
leave distillated water

Characteristic smell of cooked Strong – cooked manioc Weak –

Manioc
manioc distillated water

Sensorial profile of samples

The results show that the most whitish cassava flour was F 0 (without addition of
CPML), while the most greenish cassava flour was F3 with 10% CPML. Related to
flavor, it was noticed that the less intensive flavor was in flour F2 and F3. On the other
hand, the bitter flavor was more intensive in the samples with higher concentrations
of CPML (F3, F2, F1, and F0). The characteristic flavor of cassava flour was more
intensive in flour F1, F0, F2, and F3, respectively. For the characteristic of texture,
lumpiness was more intensive in the sample F0, while the texture was less intensive
in flour F2. The smell of the leaves was stronger in samples with higher
concentrations of CPML. It was more noticeable in flour containing 5% CPML, while
the smell of cassava was noticed better in samples F1, F0, F3, and F2, respectively
(Fig. (Fig.22).

Figure 2. Configuration of quantitative descriptive analysis (QDA) for the

characteristics of color, flavor, texture, and scent of elaborated cassava flour. F0 (0%
CPML), F1 (2.5% CPML), F2 (5% CPML), and F3 (10% CPML).

The results show that the most whitish cassava flour was F 0 (without addition of
CPML), while the most greenish cassava flour was F3 with 10% CPML. Related to
flavor, it was noticed that the less intense flavor was in flours F2 and F3. On the other
hand, the bitter flavor was stronger in the samples with higher concentrations of
CPML (F3, F2, F1, and F0). The characteristic flavor of cassava flour was more
intense in flour F1, F0, F2, and F3, respectively. For the characteristic of texture, there
was more lumpiness in the sample F0, while the row texture was less in flour F2. The
scent of a leave was stronger in samples with higher concentrations of CPML. It was
detected better in flour with 5% CPML, while the scent of cassava was noticeably
better in samples F1, F0, F3, and F2, respectively (Table (Table33).

Table 3. Average levels of sensorial descriptors of flour obtained by

descriptive analysis

F0 (0% F1 (2.5% F2 (5% F3 (10%

Characteristics
CPML) CPML) CPML) CPML)

Whitish color 0.58a 0.55a 0.51a 0.34a

Greenish color 0.30c 2.59b 2.46b 8.06a

Manioc flavor 3.66a 5.49b 1.31c 1.43c

Bitter flavor 1.31b 5.09a 6.93c 8.28c

Characteristic flavor 7.64c 8.43b 3.79ab 1.70a

Lumpiness 7.34a 6.66a 4.81b 7.00c

Raw texture 1.09a 4.35a 0.44b 0.91a

Leaf scent 0.33b 0.44a 9.03b 8.78b

Manioc scent 1.76b 4.05b 0.70a 0.85a

The same letters in the same line mean that there was not a significant difference (P
> 0.05) Turkey test.

The acceptance test has demonstrated that the cassava flour with 2.5% CPML was
more appreciated by tasters without significant difference for flour with 0% and 5%
CPML. This result is similar to data obtained by QDA. When more CPML is added
to cassava flour, its characteristics become more distinct from that of the control
flour, thus less preference by tasters.

Conclusion
An addition of 2.5% CPML was preferred by tasters. However, there was no
significant difference between control flour (without addition of CPML) and flour with
2.5% CPML and 5% CPML.

QDA has shown that higher concentrations of CPML in cassava flour negatively
affected the greenish color, scent of the leaves, cassava flavor, and characteristics
of cassava flour. However, the characteristics of flour with 2.5% and 5% CPML were
close those of the control flour (without CPML addition), which means it is possible
that flour prepared with CPML had better sensorial characteristics according to
sensorial terms.

Conflict of Interest

None declared.

References

1. Corrêa AD, Santos SR, Abreu CMP, Jokl L, Santos CD. Remoção de
polifenóis da farinha de folhas de mandioca. Sci. Food Technol.
2004;24:159–164.
2. Dutcosky SD. Análise sensorial de alimentos. Curitiba: Champagnat; 2007.
p. 123.
3. Ferri P. Cascavel, Brazil: Centro de Ciências Exatas e Tecnológicas,
Universidade Estadual do Oeste do Paraná; 2006. Extração de proteínas de
folhas de mandioca para obtenção do concentrado proteico. M.Sc. thesis.
4. Lima ECS, Bastos MLA, Nanscimento CCSF, Feijó MBS, Sabaa-Srur AUO.
Caracterização do concentrado proteico de folhas de mandioca (Manihot
esculenta cv. saracura) obtido por precipitação isoelétrica. Higiene Alimentar.
2011;25:194–195.
5. Madruga MS, Câmara FS. The chemical composition of multimistura as a
food supplement. Food Chem. (Oxford) 2000;68:41–44.
6. Modesti CF. Brazil: Universidade Federal de Lavras; 2006. p. 73. Obtenção
e caracterização de concentrado proteico de folhas de mandioca submetido
a diferentes tratamentos. M.Sc. thesis.
7. NEPA – Núcleo De Estudos E Pesquisas Em Alimentação. 2nd ed.
Campinas, Brazil: UNICAMP; 2006. p. 113. Tabela brasileira de composição
de alimentos – TACO. Versão II.
8. Ortega-Flores CI, Costa MAI, Cereda MP, Penteado MVC. Avaliação da
qualidade proteica da folha desidratada de mandioca (Manihot esculenta
Crantz) J. Brazilian Soc. Food Nutr. 2003;25:47–59.
9. Stone H, et al. Sensory evaluation by quantitative descriptive analysis. Food
Technol. 1994;28:24–34.
10. Stone H, Sidel JL. Sensory evaluation practices. 2nd ed. San Diego, CA:
Academic Press; 1992. p. 308.
11. Wobeto C. Lavras, Brazil: Universidade Federal de Lavras; 2003. p. 82.
Nutrientes e antinutrientes da farinha de folhas de mandioca (Manihot
esculenta Crantz) em três idades da planta. M.Sc. thesis.
XXI. Visual Sensorial Impairments in Neurodevelopmental
Disorders: Evidence for a Retinal Phenotype in Fragile
X Syndrome
PLoS One. 2014; 9(8): e105996.
Rafaëlle Rossignol,#1,2,¶ Isabelle Ranchon-Cole,#4,¶ Arnaud Pâris,#1,2,¶ Ameziane
Herzine,1,2 Astrid Perche,3 David Laurenceau,3 Pauline Bertrand,4 Christine Cercy,4
Jacques Pichon,1,2 Stéphane Mortaud,1,2 Sylvain Briault,1,2,3 Arnaud Menuet,1,2 and
Olivier Perche1,2,3,*
Barbara Bardoni, Editor

Abstract

Visual sensory impairments are common in Mental Deficiency (MD) and Autism
Spectrum Disorder (ASD). These defects are linked to cerebral dysfunction in the
visual cortical area characterized by the deregulation of axon growth/guidance
and dendrite spine immaturity of neurons. However, visual perception had not
been addressed, although the retina is part of the central nervous system with a
common embryonic origin. Therefore, we investigated retinal perception, the first
event of vision, in a murine model of MD with autistic features. We document that
retinal function is altered in Fmr1 KO mice, a model of human Fragile X
Syndrome. Indeed, In Fmr1 KO mice had a lower retinal function characterized
by a decreased photoreceptors neuron response, due to a 40% decrease in
Rhodopsin content and to Rod Outer Segment destabilization. In addition, we
observed an alteration of the visual signal transmission between photoreceptors
and the inner retina which could be attributed to deregulations of pre- and post-
synaptic proteins resulting in retinal neurons synaptic destabilization and to
retinal neurons immaturity. Thus, for the first time, we demonstrated that retinal
perception is altered in a murine model of MD with autistic features and that there
are strong similarities between cerebral and retinal cellular and molecular
defects. Our results suggest that both visual perception and integration must be
taken into account in assessing visual sensory impairments in MD and ASD.

Introduction

Mental Deficiency (MD) and Autism Spectrum Disorders (ASD) are frequent
pathologies (more than 2% of worldwide population [1], [2]) appearing during
childhood. They are both characterized by impairments in cognitive functions, social
integration, and/or communication. Apart from this “core” deficit, hallmark for
diagnosis of MD and ASD, patients frequently demonstrated a range of sensory
abnormalities [1], [3]–[5]. Indeed, atypical responses to sensory stimulation have
been reported in approximately 70–90% of individuals with MD or ASD [1], [6]–[9].
Sensory impairment is defined as the inability to interpret outside stimuli such as
visual, auditory, verbal, sense of touch, taste, smell or feeling pain, and may manifest
as both hyper and hyposensitivity to stimulation. Among hypothesis underlying the
neurophysiological basis of such impairments, the mis-wiring of neuronal
connections in the developing brain and synaptic destabilization had been reported.
Indeed, MD or ASD have been linked to cerebral dysregulation of axon
growth/guidance and dendrite spine maturation [10]–[12] leading to synaptic defects
[13]–[15]. Currently, sensory impairments are attributed to a cerebral phenotype.

Fragile X Syndrome (FXS) is the most common form of MD with ASD features [16],
[17]. This X-linked disorder is caused by the absence of FMRP (Fragile X Mental
Retardation Protein) due to the transcriptional silencing of FMR1 gene (Fragile X
Mental Retardation gene 1). This FMRP defect leads to a large number of synaptic
proteins alterations [18],[19] and consequently to neuronal dendrite spine immaturity
and synaptic impairment at brain level [19]–[21]. It has been shown that vision
integration is particularly affected in FXS patients, with alteration of spatiotemporal
visual processing, reduction of contrast sensitivity for visual stimuli presented at high
temporal frequencies, and visual sensitivity for both static (texture difference) and
moving images [22], [23]. These defects in visual sensory were associated with
cerebral neuron immaturity [12], [15] especially in primary visual cortex [11].

However, before being integrated at the cerebral level, the visual signal has to be
detected by- and transmitted through- the retina. At this level, the physical signal,
light, is transformed into an electrophysiological signal by highly specialized cells,
the photoreceptors. The electrophysiological signal is then transmitted and
modulated through the retina before reaching the optic nerve to the brain. So far, no
data have been collected on light perception by the retina in MD, ASD or FXS.
Despites its peripheral location, the retina is part of the central nervous system and
shares a common neurodevelopmental origin with the diencephalon [24], [25].
Therefore, the retina presents a strong similarity with brain neurons in terms of
neurotransmitters, composition in highly differentiated neurons and functional
processes [26]. Based on these similarities, we hypothesized that the retinal
perception itself might also be altered.

In order to address this question, we investigated retinal features in a murine model

of MD with autistic features, the Fmr1 KO mice, model of Fragile X Syndrome. This
mouse is considered as a relevant model of FXS since it presents behavioral, cellular
and molecular phenotypes similar to human FXS [12], [27]–[29]. Therefore, in the
Fmr1 KO mice, we studied whether the absence of Fmrp expression could affect
retinal structure, function and molecular markers of neurotransmission.

Methods

Animals
Male Fmr1 Knock-Out (KO) and their wild-type (WT) littermates (adult: 6-months-
old) were generated by breeding heterozygous Fmr1+/− females with C57BL/6J
background WT males [28]. Mice were weaned at 21 days of age and group-housed
with their same-sex littermates. On the same day, tail samples were collected for
DNA extraction and for subsequent PCR assessment of genotypes as previously
described [28], [30]. Food and water were provided ad libitum. Animals were
maintained under temperature (22°C) and humidity (55%) controlled conditions with
a 12∶12 hr dim light–dark cycle (25 lux, lights on at 7 a.m.). For tissue collection,
mice were euthanized by CO2 gas inhalation or anesthetized by a single
intraperitoneal injection of 44 mg/kg ketamine and 8 mg/kg xylazine in saline
followed by cervical dislocation. The present experimental protocol received full
review and approval by the regional animal care and use committee (Comité
Régional d’Ethique à l’Expérimentation Animale - CREEA) prior to conducting the
experiments.

Electroretinography

After overnight dark adaptation, mice were anesthetized with ketamine (50 mg/kg)
and xylazine (2 mg/kg). Eye drops were used to dilate the pupil (Atropine sulfate 1%,
ALCON). Mice were placed on a temperature-regulated heating pad throughout the
recording session. Strobe flash ElectroRetinoGrams (ERGs) (10 µs) were recorded
using an Ag/AgCl electrode in contact with the corneal surface. An Ag/AgCl electrode
was placed on the tong and a copper reference screen under the animal. Dark-
adapted responses were presented within an integrating sphere (Labsphere,
France) that mimics a ganzfeld and allows to illuminate uniformly the all retina. ERGs
are recorded using flash intensities ranging from −3.47 to +0.46 log cd s/m 2. Stimuli
were presented in order of increasing intensity. Conversely, the duration of the
interstimulus interval is 30 s since this interval has been shown to be sufficient for a
flash not to alter the next flash response. Responses were differentially amplified
(0.3–10,000 Hz), averaged, and stored. Intensity–response functions were obtained
in a single session.

ERG analysis

The leading edge of the a-waves obtained in response to the highest intensity stimuli
was analyzed with a modified form of the Lamb–Pugh model of rod
phototransduction [31]–[33] equation: P3 = {1−exp[−iSA(t−td)2]}Amax (1) where P3
represents the massed response of the rod photoreceptors and is analogous to the
PIII component of Granit [31]. The amplitude of P3 is expressed as a function of flash
energy (i) and time (t) after flash onset. SA is the gain of phototransduction, Amax is
the maximum response, and td is a brief delay. The amplitude of the b-wave is
calculated from the minimum of the a-wave to the maximum of the b-wave. Intensity–
response function of the b-wave amplitude was fitted with the Naka–Rushton
equation: B/Bmax = In/(In+Kn) (2) where I is the stimulus luminance of the flash (2,88
cd.s.m−2), B is the b-wave amplitude of ERG at I luminance, Bmax is the asymptotic
b-wave amplitude, K is the half-saturation constant corresponding to retinal
sensitivity and n is a dimensionless constant controlling the slope of the function.
The latency is the time interval between the stimulation and the peak of the b-wave
or the a-wave. Oscillatory Potentials (OPs) were recorded by using a band-pass
between 30 and 300 Hz. For each OPs, the amplitude from the baseline to the peak
and the latency were calculated.

Rod response recovery

To evaluate rod response recovery after bleaching, a single test flash of 2.88
(cd.s.m−2) was presented on dark-adapted retina, then mice were exposed to a
steady light for 2 min to bleach the retina. Immediately after bleaching and then every
10 min for 90 min, a single test flash of 2.88 (cd.s.m−2) was presented. The a-wave
response at the indicated time after bleaching was normalized to the initial dark-
adapted response for each mouse.

Quantitative RT-PCR and Western Blotting

Quantitative RT-PCR, performed using Taqman technologies (Applied

technologies), and Western blotting were realized as described previously [34], [35].
Further experimental details were described in Supporting Information S1.

Retinal Histology and Golgi staining

Retinal histology was done as described previously [36]. Retinal layers thicknesses
were measured every 0.78 mm from the optic nerve to the inferior and to the superior
ora serrata.

A modified Golgi staining method based on FD Rapid GolgiStain Kit

(FDNeuroTechnologies, Ellicott, USA) was used. WT and Fmr1 KO eyes cup were
dissected in 4% PFA pH 7.4, and cornea and lens were removed. Eye cup was
incubated in the dark for 2 days (room temperature) in impregnation solution of the
kit, and then placed 1 day in post-fixative solution (in the dark, room temperature).
Free-floating 150 µm vibratome retinal sections obtained from eye cup were cut
along the meridian through the optic nerve. Sections were incubated 30 minutes at
room temperature in revelation solution of the kit. Retinal sections were dehydrated
in graded ethanol, cleared in xylene, and mounted in paramount (Fisher Scientific).
Every 100 µm of retina from the optic nerve to the inferior and to the superior ora
serrata, number of neurons with or without exuberant and disorganised dendrites
was counted under microscope (x400, Leica, Paris, France). Mean values in each
group were presented as percentage of WT.

Rhodopsin Quantification

Rhodopsin quantification was realized as described before [37]. We measured total

Rhodopsin levels by spectral analysis (300–700 nm) using the differential
absorbance of Rhodopsin at 498 nm before and after bleaching (10 min). Whole
eyes were homogenized in 240 µL ROS buffer (1 mM HEPES, pH 7.4; 3 mM NaCl;
6 mM KCl; 0.2 mM MgCl2; 0.1 mM DTT) supplemented with 50 mM hydroxylamine,
1.5% maltoside and proteinase inhibitor cocktail (Pierce, Paris, France). The
samples were spun down at 3000 g for 5 min at 4°C and the supernatant was
assayed. Rhodopsin concentration was calculated by difference absorbance at 498
nm using the molar extinction coefficient of 42,700 M−1.cm−1.

PSD95, Syt1a, mGluR5 and Rhodopsin immunohistochemistry

PSD95, Syt1a and mGluR5 immunochemistry were done on eye cryosections (14
µm), whereas Rhodopsin was tested on paraffin sections (4 µm). Briefly, sections
were heated 30 minutes in sodium citrate buffer pH 6.0 and then blocked 2 hours in
blocking buffer (10% FBS, 2% BSA, 0.3% Triton X-100 and 0.1% NaN3) at room
temperature. Sections were incubated with a 1∶500 dilution of PSD95, Syt1a,
mGluR5 or Rhodopsin antibodies (Abcam, Paris, France) overnight at 4°C in
blocking buffer. After washing them three times in TBS pH 7.6, sections were stained
with 1∶100 dilution goat anti-mouse secondary antibody (Life Technologies,
Carlsbad, USA) (FITC for PSD95, Syt1a and mGluR5 and TRITC for Rhodopsin).
DAPI at 10 µg/mL (Roche Applied Science, Indianapolis, USA) was then applied on
sections for 10 minutes at room temperature. Sections were washed three times in
TBS pH 7.6, mounted with Fluoromount (Vector Laboratories, Burlingame, USA) and
stored at 4°C. Imaging was realised on a Carl Zeiss apotome (40X, 40/1.3) and
analysed on AxioVision software thank to the Cytometry and Imagery platform of the
CBM (Orléans, France).

Electron microscopy

Electron microscopy was realized in Centre d’Imagerie Cellulaire Santé (Clermont-

Ferrand, France). Experimental procedure is described in Supporting Information
S1.

Statistical analysis

All results are expressed as mean ± SEM. Data analysis was performed using
GraphPad Prism 6.00. Statistical comparisons among groups were conducted using
Student’s unpaired t test. Statistical significance was defined as p<0.05. Significant
differences between groups are noted by *. One symbol for p<0.05; two symbols for
p<0.01; three symbols for p<0.001; four symbols for p<0.0001.

Results

Fmrp is expressed in WT mice retinas

Quantitative PCR and Western-blot showed that Fmr1 (Fig. 1A) and its coded protein
Fmrp (Fig. 1B) are expressed in the WT retina. In Fmr1 KO mice, no significant
mRNA or protein were observed, as expected (Fig. 1A, 1B).
Scotopic a-wave characteristic in WT and Fmr1 KO mice.

Retinal function was evaluated using to ElectroRetinoGram (ERG) (n = 20 for WT and
n = 17 for Fmr1 KO). For each (A) typical ERG obtained at light intensity −2.88
log(cd.s.m−2), the decreasing part of the a-wave is fitted to calculate the extrapolated
(B) maximal a-wave amplitude (Amax) and (C) SA parameter reflecting the
photoreceptor sensitivity. In Fmr1 KO mice compared to WT mice, Amax is
significantly decreased by 26% (p<0.05), and the SA parameter is not significantly
different (p = 0.08). (D) a-wave latency was not different between groups. Data are
presented as Mean ± SEM. Student’s t test, *p<0.05.

Figure 1. Fmr1/Fmrp expression in WT and Fmr1 KO retinas.

Retinal function is altered in Fmr1 KO mice

In order to evaluate the effect of Fmrp defect on retinal function, we recorded the
electrophysiological response of the retina to a light stimulation, called
ElectroRetinoGram (ERG). Typically, an ERG is characterized by a negative
deflection termed the a-wave, which is initiated by the activity of light-sensitive cells,
the photoreceptors (Fig. 2A, Fig. 1S in Supporting Information S1). The following
positive deflection, termed the b-wave, reflects signal transmission to the inner retina
(Fig. 2A, Fig. 1S in Supporting Information S1), mainly due to bipolar and Müller
Cells activities [38]. The small ripples on ascending part of the b-wave, called
oscillatory potentials, involved multiple components, presumably including outer and
inner retinal circuitry [38].

Scotopic a-wave characteristic in WT and Fmr1 KO mice.

Retinal function was evaluated using to ElectroRetinoGram (ERG) (n = 20 for WT and
n = 17 for Fmr1 KO). For each (A) typical ERG obtained at light intensity −2.88
log(cd.s.m−2), the decreasing part of the a-wave is fitted to calculate the extrapolated
(B) maximal a-wave amplitude (Amax) and (C) SA parameter reflecting the
photoreceptor sensitivity. In Fmr1 KO mice compared to WT mice, Amax is
significantly decreased by 26% (p<0.05), and the SA parameter is not significantly
different (p = 0.08). (D) a-wave latency was not different between groups. Data are
presented as Mean ± SEM. Student’s t test, *p<0.05.

Figure 2. Scotopic a-wave characteristic in WT and Fmr1 KO mice.

For each ERG, at the highest light stimulus, the decreasing part of the a-wave was
fitted to calculate the maximal a-wave amplitude (Amax) and the parameter SA
reflecting photoreceptor sensitivity. Amax vary from −568±55 µV in WT mice to
−354±43 µV in Fmr1 KO mice (Fig. 2B) indicating that its amplitude was significantly
(p = 0.027) decreased by 26%, whereas SA was not significantly affected (Fig. 2C).
The a-wave latency was not different between WT and Fmr1 KO mice (Fig. 2D).
These observations demonstrated that Fmr1 KO retinas have a lower photoreceptor
response without modification of photoreceptor sensitivity to light.

For each ERG (Fig. 3A), the b-wave sensitivity curve was fitted to calculate the
maximal b-wave amplitude (Bmax) reflecting the maximal retinal response, the half
saturation luminance (K) reflecting the light intensity generating half Bmax, and the
slope of the curve in its linear part (n) reflecting the contrast sensitivity of retina. The
mean b-wave sensitivity curve of Fmr1 KO mice was slightly lower than the one from
WT mice (Fig. 3A) leading to a significant decrease (p = 0.0085) of Bmax from 853±33
µV in WT mice to 709±39 µV in Fmr1 KO mice (Fig. 3B). No significant alteration of
K was observed (Fig. 3C), whereas there was a significant (p = 0.02) increase of n
from 0.73±0.04 in WT mice to 0.94±0.08 in Fmr1 KO mice (Fig. 3D). Neither b-wave
latency (Fig. 3E) nor the oscillatory potentials (Fig. 3F) were affected in Fmr1 KO
mice. Therefore, Fmr1 KO retinas have a lower maximal b-wave amplitude and an
increase in the slope of the curve in its linear part.

Scotopic b-wave characteristic in WT and Fmr1 KO mice.

Retinal function was evaluated using ElectroRetinoGram (ERG). (A) Serial
responses to increasing flash stimuli (−3.47 log(cd.s.m −2) to 0.6 log(cd.s.m−2)) were
obtained for WT and Fmr1 KO mice under dark-adapted conditions (n = 20 for WT
and n = 17 for Fmr1 KO). ERG response (µV) was plotted against light intensities to
obtain the b-wave sensitivity curve. In Fmr1 KO mice, the b-wave sensitivity curve is
slightly collapsed compared to WT mice. The b-wave sensitivity curve was then fitted
to calculate (B) the saturated b-wave amplitude (Bmax), (C) the K parameter (intensity
providing half saturation) and (D) the n parameter (representing the b-wave
sensitivity curves slope). In Fmr1 KO mice, Bmax is significantly decreased by 20%,
the n parameter is significantly increased by 29%, whereas K parameter remained
unchanged compared to WT mice. (E) b-wave latency and (F) oscillatory potential
(OP) amplitude were not different between groups. Data are presented as Mean ±
SEM. Student’s t test, *p<0.05.

Figure 3. Scotopic b-wave characteristic in WT and Fmr1 KO mice.

Interestingly, the ratio Bmax/Amax was similar between WT (−1.63±0.13 AU) and Fmr1
KO (−1.64±0.19 AU) mice, suggesting that the decrease of the maximal b-wave
amplitude is mostly due to the decrease of the maximal a-wave amplitude. However,
the increase in the slope of b-wave sensitivity curve (Fig. 3D) indicates that a given
increase of light-stimulation induced a higher increase of the b-wave in Fmr1 KO
than in WT mice.

All these data demonstrated that Fmr1 KO mice have an alteration of retinal function
characterized by a reduction of the maximal photoreceptor response and an
alteration of signal transmission between photoreceptors and the inner retina leading
to an increased sensitivity to contrast.

Retinal synaptic structure is altered in Fmr1 KO mice

We firstly investigated global retinal histology by measuring retinal layer thicknesses.

These thicknesses were not different between WT and Fmr1 KO mice whatever was
the considered layer: ROS (Rod Outer Segment), ONL (Outer Nuclear Layer), OPL
(Outer Plexiform Layer), INL (Inner Nuclear Layer) or Total Retina (Fig. 4A, Fig. 2S
in Supporting Information S1).
 (A) Retinal layer structure was investigated thank to histology. No significant
variation of layer thicknesses was observed. Scale bar, 15 µm. (B) Retinal synaptic
structure was investigated through pre- and post-synaptic markers in WT and Fmr1
KO retinas. PSD95, Syt1a and mGluR5 expressions were assessed by Western-blot
analysis (n = 8 per group). Proteins variations were normalized to β-actin and data
are expressed as percentage of WT. PSD95 and Syt1a proteins were decreased by
37% and 20% (*p<0.05) respectively in Fmr1 KO mice compared to WT retinas
whereas mGluR5 was upexpressed by 30% (*p<0.05). Three independent
experiments were performed with similar results. (C) PSD95, Syt1a and mGluR5
immunolocalisations were realized in Fmr1 KO and WT retinas (n = 4 per group).
PSD95 showed a labelling in Outer Plexiform Layer (OPL), Inner Nuclear Layer (INL)
and Ganglion Cells Layer (GCL) for WT retinas. In Fmr1 KO retinas, all staining was
weaker. Syt1a showed a labeling in the OPL for WT retinas as for Fmr1 KO retinas.
mGluR5 showed a labeling in OPL, INL, the Inner Plexiform Layer (IPL) and GCL for
WT retinas and for Fmr1 KO retinas. Scale bar, 20 µm. (D) Retinal synaptic
immaturity was investigated through Golgi staining. Number of (i) mature neurons
and (ii) immature neurons presenting exuberant and disorganized dendrites were
counted on 100 µm of WT and Fmr1 KO retinas and presented as percentage of
WT. Number of immature neurons presenting exuberant and disorganized dendrites
was significantly (p = 0.046) increased in Fmr1 KO (176±27%) compared to WT
(100±27%). Data are presented as Mean ± SEM. Student’s t test, *p<0.05.

Figure 4. Characterization of neuronal synaptic structure and neuronal

immaturity.

Secondly, to investigate synaptic structure, we focused on Synaptotagmin1a (Syt1a

- pre-synaptic) [18], metabotropic glutamate receptor 5 (mGluR5 - post-synaptic)
[39], and Post-Synaptic Density protein 95 (PSD95 - post-synaptic) [41]–[43]
expressions since their deregulation had been associated with synaptic structure
alteration. No variation was observed between WT and Fmr1 KO retinas for Syt1a,
Psd95 and mGluR5 mRNA (Fig. 3S in Supporting Information S1). In Fmr1 KO,
PSD95 and Syt1a proteins were significantly (p<0.02) down-regulated by 37±10%
and 20±4% respectively, compared to WT retinas (Fig. 4B). PSD95 was regularly
localised in Fmr1 KO retina but its staining appeared weaker in the ONL, INL and
GCL than for WT retinas (Fig. 4C). Similarly, Syt1a was expressed in the same
retinal layers in WT and Fmr1 KO mice, but the staining seemed more smoothed in
Fmr1 KO mice (Fig. 4C). mGluR5 was expressed in the same layers (OPL, INL, IPL
and GCL) in WT as in Fmr1 KO (Fig. 4C). In the INL, mGluR5 labeling formed vertical
streaks that were clearly more intense in Fmr1 KO retinas compared to WT. This
was associated with a significant (p = 0.01) increased from 100±11% in WT to
130±10% in Fmr1 KO retina of the mGluR5 expression (Fig. 4B).

Thirdly, we investigated synaptic immaturity through Golgi staining. In WT as in Fmr1

KO retinas, Golgi-stained cells were amacrine cells mainly localized in the INL.
However, number of immature neurons presenting exuberant and disorganized
dendrites was significantly (p = 0.046) increased in Fmr1 KO (176±27%) compared
to WT (100±27%) (Fig. 4D).

Therefore, the absence of Fmrp had no impact on gross retinal structure suggesting
a similar number of retinal neurons in Fmr1 KO and WT retinas. However, synaptic
connections are disrupted. Indeed, deregulation of key pre- and post-synaptic
markers, as well as the increased proportion of retinal neurons with exuberant
dendrites, clearly evidence synaptic structure alteration and immaturity of retinal
neurons in Fmr1 KO mice.

Rhodopsin content is decreased in Fmr1 KO mice

In order to better understand the origin of the lower photoreceptor response, we

measured Rhodopsin (photopigment) content. Indeed, Rhodopsin is a light-sensing
G protein-coupled receptor localised in the membrane of rod photoreceptors that is
responsible for light absorption. No difference was observed between WT and Fmr1
KO retinas for Rhodopsin mRNA expression (Fig. 3S in Supporting Information S1).
In Fmr1 KO retinas, Rhodopsin protein content was significantly (p<0.04) decreased
by 37±9% as shown by Western-blot analysis (Fig. 5A) and by 47% as shown by
spectrophotometric dosage (1320±28 pmoles/retina in WT mice vs 696±15
pmoles/retina in Fmr1 KO mice) (Fig. 5B). The remaining Rhodopsin protein showed
an usual recovery after bleaching (Fig. 5C).

Rhodopsin characterization in adult WT and Fmr1 KO mice retinas.

(A) Rhodopsin protein amount was assessed by Western-blot analysis (n = 12 per
group). All protein variations were normalized to β-actin. Rhodopsin protein was
decreased by 37% (*p<0.05) in Fmr1 KO mice compared to WT retinas. Three
independent experiments were performed with similar results. (B) Rhodopsin protein
decrease is confirmed by spectrophotometric quantification (n = 6 per group). Indeed,
Rhodopsin amount was decreased by 47% (**p<0.01) in Fmr1 KO mice compared
to WT retinas. (C) Rhodopsin isomerisation was investigated by studing a-wave
recovering following 2 minutes bleach (n = 5 per group). Forty minutes after bleaching
the relative a-wave amplitude was returned to dark-adapted value. No difference
was observed between WT and Fmr1 KO retinas. (D) Localization of Rhodopsin was
done by immunostaining. Immunohistochemistry analysis demonstrated that
Rhodopsin is distributed in ROS and between nucleus in the ONL in WT mice,
whereas in Fmr1 KO mice only the ROS labelling could be observed (n = 4 per
group). Scale bar, 40 µm. (E) ROS ultrastructure was investigate by Transmission
Electron Microscopy in adult WT and Fmr1 KO mice retinas. The images were set
at 16000X magnification, with scale bars indicating 50 nm. These representatives
pictures showed that density and linear organization of disks in Fmr1 KO mice ROS
are altered compared to WT mice ROS. Data are presented as Mean ± SEM.
Student’s t test, *p<0.05, **p<0.01.

Figure 5. Rhodopsin characterization in adult WT and Fmr1 KO mice retinas.

Immunohistochemistry analysis demonstrated that Rhodopsin is distributed in ROS

and between nucleus in the ONL in WT mice, whereas in Fmr1 KO mice only the
ROS labelling could be observed (Fig. 5D). Although immunohistochemistry is not a
quantitative method, we noticed that Rhodopsin staining in Fmr1 KO was weaker
than in WT. Interestingly, although the length of ROS was fairly similar between WT
and Fmr1 KO (data not shown), density and linear organisation of disks in Fmr1 KO-
ROS were clearly altered (Fig. 5E).

Thus, we could show that in Fmr1 KO retinas, the Rhodopsin content is decreased
and ROS structure is altered but Rhodopsin recovery after bleaching is not affected.

Discussion

The starting point of vision is the detection of light by the retina and more specifically
the absorption of light by photoreceptors cells photopigment. In these cells, light is
transformed into an electrophysiological signal by a process named
phototransduction. This electrophysiological signal, after going through the retina
and the optic nerve, reaches the brain and is integrated. Surprisingly, although visual
sensory impairments were described in Mental Deficiency (MD) and Autism
Spectrum Disorders (ASD), including Fragile X Syndrome (FXS) [22], [23], no data
had been collected on light perception at the retinal level even if the retina is a neural
tissue with the same embryonic origin as diencephalon [44]. It is even more
interesting since our experiments showed that Fmrp is expressed in the WT retina.
Therefore, we hypothesized that a lack of Fmrp could induced similar cellular and
functional defects in the retina as it does in cerebral neurons.

Retinal function, recorded by ElectroRetinoGram (ERG), is defined by all the

electrophysiological manifestations between Rhodopsin activation by light and the
electrophysiological message sent through the optic nerve to the brain [38], [45],
[46]. As expected, Fmr1 KO mice showed altered ERG recordings characterized by
a decrease in the a and b waves, and an increase in the slope of the b-wave
sensitivity curve. These data indicate retinal impairments in Fmr1 KO mice. Because
the Bmax/Amax ratio was similar between Fmr1 KO and WT mice, we can assume that
the b-wave decrease and so the amplitude of the signal transmitted from the
photoreceptors to the inner retina is mainly due to the decrease of the a-wave. In
addition, a-wave reduction was not due to a loss of photoreceptors, since the ONL
thickness was similar between Fmr1 KO and WT mice, but linked to decreased in
Rhodopsin content as shown by Western-blot and spectrophotometric analysis.
Indeed, Rhodopsin is the specific rod-photoreceptor protein responsible for the first
events in the perception of light [47], and its concentration is directly correlated to a-
wave amplitude [38], [48]. In Fmr1 KO retinas, Rhodopsin was normally localized in
the ROS but displayed a ∼40% concentration reduction which results in a 26%
reduction in the a-wave. Similar results had already been described in heterozygote
Rhodopsin knockout mice characterized by a ∼20% Amax decreased associated with
a ∼40% reduced Rhodopsin content [37], [49]. Although we did not notice any
reduction of the ROS thickness in Fmr1 KO retinas, in opposition to Liang et al. study
[49], the electronic microscopy showed disorganized and reduced disk density in the
photoreceptors outer segment. The lack of Rhodopsin might participate to the
destabilization of outer segment [37], [49]. These results raised the question: is the
Rhodopsin mRNA a target of Fmrp? Further experiments should help to identify
other Fmrp targets involved in organ specific impairments.

Electrophysiological data also revealed that Fmr1 KO mice present an increased

slope (n) of the b-wave sensitivity curve. This increase indicates that a given raise in
light-stimulation induced a higher raise in retinal response which can be interpretable
as an alteration in contrast sensitivity/perception. There was no significant difference
in the slope for the a-wave amplitude (data not shown) indicating that the alteration
in contrast sensitivity/perception could be linked to the transmission of the signal
between the photoreceptors and the inner retina rather than an alteration of
phototransduction. To explore further this possibility, we looked at pre- (Syt1a) and
post- (PSD95 and mGluR5) synaptic markers expression. Indeed, Fmrp is a RNA-
binding protein specifically regulating dendrite mRNA translation [50]–[52]. In its
absence, Fmr1 KO neurons in vitro [19] or Fmr1 KO hippocampi [18], [40] and in
somatosensory cortex [29] in vivo, show a deregulation in several pre- and post-
synaptic proteins with, as consequences, a destabilization of synapse structure,
immaturity of dendrite spines and an alteration of neurons plasticity [15], [53]. In
Fmr1 KO mice retinas, both pre- and post-synaptic proteins were deregulated, Syt1a
and PSD95 being down-regulated whereas mGluR5 was up-regulated. The PSD95
down-regulation is striking since it had been previously linked to plasticity defect and
loss of neuron - neuron communication [41], [54]. As a consequence of all these
retinal protein deregulations, the immature state of retinal neuron development in
Fmr1 KO mice is characterized by an increase proportion of exuberant and
disorganized dendrites. These retinal molecular and cellular phenotypes are clearly
similar to those previously described in Fmr1 KO mice brain, such as synaptic
alterations [29], [55] or Fmrp-induced synaptic proteins translation deregulation [19],
[56]–[58].
Interestingly, PSD95 had been shown to be unchanged in Fmr1 KO cerebellum and
cortex structures [59] whereas it is up-regulated in cortical neurons culture [60] or
whole brain (data not shown), and down-regulated in hippocampus [59] or retina.
Since Psd95 mRNA is one of the Fmrp target [61], we can suggest that the absence
of Fmrp is leading to a tissue-specific protein translation deregulation as observed
by the different profile in retina/hippocampus and cerebellum/cortex. The specific
translation defect is even more reinforced since Psd95 mRNA level was not altered
between WT and Fmr1 KO retinas. Further molecular experiments on
transcriptional/translation modulation linked to Fmrp should help to better
understand this tissue specific Fmrp translation regulation.

In agreement with FXS phenotypes, MD or ASD, adult Fmr1 KO mice exhibit many
behavioral impairments on both social and cognitive components, characterized by
lower levels of social affiliative behaviors or preferences [62]–[64] and by spatial
memory impairments [65], [66]. In this study, we demonstrate for the first time that
Fmr1 deficiency affects retinal function. Therefore, this begs the question of how far
vision defects are involved in the recorded behavioral impairments of the Fmr1 KO
mice. Indeed, animal behaviors clearly rely on sensorial processing allowing mice to
integrate environmental stimuli and to adapt their action. Based on our study, we can
suggest that vision impairments on both perception and integration sides should be
taken into account in the behavioural analyses of Fmr1 KO mice.

As a conclusion, we highlighted for the first time in a mouse model of MD and ASD
that sensory impairments involved both peripheral perception and central integration
defects. Our findings clearly state that the mis-wiring of neuronal connections and
synaptic destabilization in the brain as in the retina lead to similar cellular and
functional phenotypes. Based on our data, MD and/or ASD patients, and especially
FXS patients, should be investigated on their visual perception.

Supporting Information

Supporting Information S1

Figure 1S: Representative ERGs obtained from WT or Fmr1 KO mice. Figure 2S:
Retinal layers thicknesses in WT and Fmr1 KO retinas. Figure 3S: Synaptic markers
mRNA expression in WT and Fmr1 KO retinas.

(DOC)

Click here for additional data file.(613K, doc)

Acknowledgments

We would like to thank Claire Szczepaniak and Christelle Blavignac of the Centre
Imagerie Cellulaire Santé of the Auvergne University for their excellent technical
support and expertise in electronic microscopy, Alexandre Herpin and Jerôme
Larrigaldie for animal breeding, and Valérie Quesniaux, François Erard and Jennifer
Palomo for the English editing.

Funding Statement

Research was supported by CNRS, Regional Hospital of Orléans, University of

Orléans, FEDER 35106, and FRAXA Research Foundation. The funders had no role
in study design, data collection and analysis, decision to publish, or preparation of
the manuscript.

Data Availability

The authors confirm that all data underlying the findings are fully available
without restriction. All data are included within the paper.

 References
 1. Battaglia A (2011) Sensory impairment in mental retardation: a potential
role for NGF. Arch Ital Biol 149, 193–203. [PubMed]
 2. Leonard H, Wen X (2002) The epidemiology of mental retardation:
challenges and opportunities in the new millennium. Ment Retard Dev Disabil
Res Rev 8, 117–34. [PubMed]
 3. Biersdorff KK (1994) Incidence of significantly altered pain experience
among individuals with developmental disabilities. Am J Ment Retard 98, 619–
31. [PubMed]
 4. Nader R, Oberlander TF, Chambers CT, Craig KD (2004) Expression of
pain in children with autism. Clin J Pain 20, 88–97. [PubMed]
 5. Lane AE, Young RL, Baker AE, Angley MT (2010) Sensory processing
subtypes in autism: association with adaptive behavior. J Autism Dev Disord
40, 112–22. [PubMed]
 6. Lane NE, Schnitzer TJ, Birbara CA, Mokhtarani M, Shelton DL, et al. ..
(2010) Tanezumab for the treatment of pain from osteoarthritis of the knee. N
Engl J Med 363, 1521–31. [PubMed]
 7. Baranek GT, Chin YH, Hess LM, Yankee JG, Hatton DD, et al. .. (2002)
Sensory processing correlates of occupational performance in children with
fragile X syndrome: preliminary findings. Am J Occup Ther 56, 538–46.
[PubMed]
 8. Baranek GT, David FJ, Poe MD, Stone WL, Watson LR (2006) Sensory
Experiences Questionnaire: discriminating sensory features in young children
with autism, developmental delays, and typical development. J Child Psychol
Psychiatry 47, 591–601. [PubMed]
 9. Baron-Cohen S, Ashwin E, Ashwin C, Tavassoli T, Chakrabarti B (2009)
Talent in autism: hyper-systemizing, hyper-attention to detail and sensory
hypersensitivity. Philos Trans R Soc Lond B Biol Sci 364, 1377–83. [PMC free
article] [PubMed]
 10. Volpe JJ (2001) Perinatal brain injury: from pathogenesis to
neuroprotection. Ment Retard Dev Disabil Res Rev 7, 56–64. [PubMed]
 11. Berman RF, Murray KD, Arque G, Hunsaker MR, Wenzel HJ (2012)
Abnormal dendrite and spine morphology in primary visual cortex in the CGG
knock-in mouse model of the fragile X premutation. Epilepsia 53 Suppl 1,
150–60. [PMC free article] [PubMed]
 12. Irwin SA, Idupulapati M, Gilbert ME, Harris JB, Chakravarti AB, et al. ..
(2002) Dendritic spine and dendritic field characteristics of layer V pyramidal
neurons in the visual cortex of fragile-X knockout mice. Am J Med Genet 111,
140–6. [PubMed]
 13. Davidovic L, Navratil V, Bonaccorso CM, Catania MV, Bardoni B, et al. ..
(2011) A metabolomic and systems biology perspective on the brain of the
fragile X syndrome mouse model. Genome Res 21, 2190–202. [PMC free
article] [PubMed]
 14. Jamain S, Radyushkin K, Hammerschmidt K, Granon S, Boretius S, et al.
.. (2008) Reduced social interaction and ultrasonic communication in a mouse
model of monogenic heritable autism. Proc Natl Acad Sci U S A 105, 1710–
5. [PMC free article] [PubMed]
 15. Bilousova TV, Dansie L, Ngo M, Aye J, Charles JR, et al. .. (2009)
Minocycline promotes dendritic spine maturation and improves behavioural
performance in the fragile X mouse model. J Med Genet 46, 94–102.
[PubMed]
 16. Abrahams BS, Geschwind DH (2008) Advances in autism genetics: on
the threshold of a new neurobiology. Nat Rev Genet 9, 341–55. [PMC free
article] [PubMed]
 17. Penagarikano O, Mulle JG, Warren ST (2007) The pathophysiology of
fragile x syndrome. Annu Rev Genomics Hum Genet 8, 109–29. [PubMed]
 18. Klemmer P, Meredith RM, Holmgren CD, Klychnikov OI, Stahl-Zeng J, et
al. .. (2011) Proteomics, ultrastructure, and physiology of hippocampal
synapses in a fragile X syndrome mouse model reveal presynaptic
phenotype. J Biol Chem 286, 25495–504. [PMC free article] [PubMed]
 19. Liao L, Park SK, Xu T, Vanderklish P, Yates JR 3rd (2008) Quantitative
proteomic analysis of primary neurons reveals diverse changes in synaptic
protein content in fmr1 knockout mice. Proc Natl Acad Sci U S A 105, 15281–
6. [PMC free article] [PubMed]
 20. Bear MF, Huber KM, Warren ST (2004) The mGluR theory of fragile X
mental retardation. Trends Neurosci 27, 370–7. [PubMed]
 21. Vanderklish PW, Edelman GM (2005) Differential translation and fragile
X syndrome. Genes Brain Behav 4, 360–84. [PubMed]
 22. Farzin F, Rivera SM, Whitney D (2011) Resolution of spatial and temporal
visual attention in infants with fragile X syndrome. Brain 134, 3355–68. [PMC
free article] [PubMed]
 23. Kogan CS, Boutet I, Cornish K, Zangenehpour S, Mullen KT, et al. ..
(2004) Differential impact of the FMR1 gene on visual processing in fragile X
syndrome. Brain 127, 591–601. [PubMed]
 24. Pei YF, Rhodin JA (1970) The prenatal development of the mouse eye.
Anat Rec 168, 105–25. [PubMed]
 25. Kaneko A (1979) Physiology of the retina. Annu Rev Neurosci 2, 169–91.
[PubMed]
 26. Purves D (2001) Neuroscience. 2nd edition. Sunderland (MA): Sinauer
Associates; 2001.
 27. Bernardet M, Crusio WE (2006) Fmr1 KO mice as a possible model of
autistic features. ScientificWorldJournal 6, 1164–76. [PubMed]
 28. Consortium TD-BFX (1994) Fmr1 knockout mice: a model to study fragile
X mental retardation. Cell 78, 23–33. [PubMed]
 29. Nimchinsky EA, Oberlander AM, Svoboda K (2001) Abnormal
development of dendritic spines in FMR1 knock-out mice. J Neurosci 21,
5139–46. [PubMed]
 30. Durtch-Belgium Fragile X Consortium (1994) Fmr1 knockout mice: a
model to study fragile X mental retardation. The Dutch-Belgian Fragile X
Consortium. Cell 78, 23–33. [PubMed]
 31. Granit R (1933) The components of the retinal action potential in
mammals and their relation to the discharge in the optic nerve. J Physiol 77,
207–39. [PMC free article] [PubMed]
 32. Lamb TD, Pugh EN Jr. (1992) G-protein cascades: gain and kinetics.
Trends Neurosci 15, 291–8. [PubMed]
 33. Lamb TD, Pugh EN Jr. (1992) A quantitative account of the activation
steps involved in phototransduction in amphibian photoreceptors. J Physiol
449, 719–58. [PMC free article] [PubMed]
 34. Perche O, Menuet A, Marcos M, Liu L, Paris A, et al. .. (2013) Combined
deletion of two Condensin II system genes (NCAPG2 and MCPH1) in a case
of severe microcephaly and mental deficiency. Eur J Med Genet 56, 635–41.
[PubMed]
 35. Perche O, Doly M, Ranchon-Cole I (2009) Calpains are activated by light
but their inhibition has no neuroprotective effect against light-damage. Exp
Eye Res 89, 989–94. [PubMed]
 36. Perche O, Doly M, Ranchon-Cole I (2007) Caspase-dependent apoptosis
in light-induced retinal degeneration. Invest Ophthalmol Vis Sci 48, 2753–9.
[PubMed]
 37. Price BA, Sandoval IM, Chan F, Nichols R, Roman-Sanchez R, et al. ..
(2012) Rhodopsin gene expression determines rod outer segment size and
rod cell resistance to a dominant-negative neurodegeneration mutant. PLoS
One 7, e49889. [PMC free article] [PubMed]
 38. Frishman LJ (2006) Origins of the electroretinogram. Principles and
pratice of clinical electrophysiology of vision. Ed 2 (Heckenlivery JR, Arden
GB, eds) Cambridge, MA:MIT., 139–183.
 39. Dolen G, Bear MF (2009) Fragile x syndrome and autism: from disease
model to therapeutic targets. J Neurodev Disord 1, 133–40. [PMC free article]
[PubMed]
 40. Levenga J, de Vrij FM, Buijsen RA, Li T, Nieuwenhuizen IM, et al. .. (2011)
Subregion-specific dendritic spine abnormalities in the hippocampus of Fmr1
KO mice. Neurobiol Learn Mem 95, 467–72. [PubMed]
 41. Tsai NP, Wilkerson JR, Guo W, Maksimova MA, DeMartino GN, et al. ..
(2012) Multiple autism-linked genes mediate synapse elimination via
proteasomal degradation of a synaptic scaffold PSD-95. Cell 151, 1581–94.
[PMC free article] [PubMed]
 42. Cline H (2005) Synaptogenesis: a balancing act between excitation and
inhibition. Curr Biol 15, R203–5. [PubMed]
 43. Westmark CJ (2013) FMRP: a triple threat to PSD-95. Front Cell Neurosci
7, 57. [PMC free article] [PubMed]
 44. Bonetti C, Surace EM (2010) Mouse embryonic retina delivers information
controlling cortical neurogenesis. PLoS One 5, e15211. [PMC free article]
[PubMed]
 45. Fulton AB, Dodge J, Hansen RM, Williams TP (1999) The rhodopsin
content of human eyes. Invest Ophthalmol Vis Sci 40, 1878–83. [PubMed]
 46. Fox DA, Rubinstein SD (1989) Age-related changes in retinal sensitivity,
rhodopsin content and rod outer segment length in hooded rats following low-
level lead exposure during development. Exp Eye Res 48, 237–49. [PubMed]
 47. Hofmann KP, Reichert J (1985) Chemical probing of the light-induced
interaction between rhodopsin and G-protein. Near-infrared light-scattering
and sulfhydryl modifications. J Biol Chem 260, 7990–5. [PubMed]
 48. Lee ES, Flannery JG (2007) Transport of truncated rhodopsin and its
effects on rod function and degeneration. Invest Ophthalmol Vis Sci 48, 2868–
76. [PMC free article] [PubMed]
 49. Liang Y, Fotiadis D, Maeda T, Maeda A, Modzelewska A, et al. .. (2004)
Rhodopsin signaling and organization in heterozygote rhodopsin knockout
mice. J Biol Chem 279, 48189–96. [PMC free article] [PubMed]
 50. Bassell GJ, Warren ST (2008) Fragile X syndrome: loss of local mRNA
regulation alters synaptic development and function. Neuron 60, 201–14.
[PMC free article] [PubMed]
 51. Braun K, Segal M (2000) FMRP involvement in formation of synapses
among cultured hippocampal neurons. Cereb Cortex 10, 1045–52. [PubMed]
 52. Zukin RS, Richter JD, Bagni C (2009) Signals, synapses, and synthesis:
how new proteins control plasticity. Front Neural Circuits 3, 14. [PMC free
article] [PubMed]
 53. Chen PC, Bhattacharyya BJ, Hanna J, Minkel H, Wilson JA, et al. .. (2011)
Ubiquitin homeostasis is critical for synaptic development and function. J
Neurosci 31, 17505–13. [PMC free article] [PubMed]
 54. Shi D, Xu S, Waddell J, Scafidi S, Roys S, et al. .. (2012) Longitudinal in
vivo developmental changes of metabolites in the hippocampus of Fmr1
knockout mice. J Neurochem 123, 971–81. [PMC free article] [PubMed]
 55. Comery TA, Harris JB, Willems PJ, Oostra BA, Irwin SA, et al. .. (1997)
Abnormal dendritic spines in fragile X knockout mice: maturation and pruning
deficits. Proc Natl Acad Sci U S A 94, 5401–4. [PMC free article] [PubMed]
 56. Krueger DD, Osterweil EK, Chen SP, Tye LD, Bear MF (2011) Cognitive
dysfunction and prefrontal synaptic abnormalities in a mouse model of fragile
X syndrome. Proc Natl Acad Sci U S A 108, 2587–92. [PMC free article]
[PubMed]
 57. Muddashetty RS, Kelic S, Gross C, Xu M, Bassell GJ (2007) Dysregulated
metabotropic glutamate receptor-dependent translation of AMPA receptor
and postsynaptic density-95 mRNAs at synapses in a mouse model of fragile
X syndrome. J Neurosci 27, 5338–48. [PubMed]
 58. Todd PK, Mack KJ, Malter JS (2003) The fragile X mental retardation
protein is required for type-I metabotropic glutamate receptor-dependent
translation of PSD-95. Proc Natl Acad Sci U S A 100, 14374–8. [PMC free
article] [PubMed]
 59. Zhu ZW, Xu Q, Zhao ZY, Gu WZ, Wu DW (2011) Spatiotemporal
expression of PSD-95 in Fmr1 knockout mice brain. Neuropathology 31, 223–
9. [PubMed]
 60. Pfeiffer BE, Zang T, Wilkerson JR, Taniguchi M, Maksimova MA, et al. ..
(2010) Fragile X mental retardation protein is required for synapse elimination
by the activity-dependent transcription factor MEF2. Neuron 66, 191–7. [PMC
free article] [PubMed]
 61. Zalfa F, Eleuteri B, Dickson KS, Mercaldo V, De Rubeis S, et al. .. (2007)
A new function for the fragile X mental retardation protein in regulation of
PSD-95 mRNA stability. Nat Neurosci 10, 578–87. [PMC free article]
[PubMed]
 62. McNaughton CH (2008) Evidence for social anxiety and impaired social
cognition in a mouse model of fragile X syndrome. Behav. Neurosci. 122,
293–300. [PubMed]
 63. Mines MA, Yuskaitis CJ, King MK, Beurel E, Jope RS (2010) GSK3
influences social preference and anxiety-related behaviors during social
interaction in a mouse model of fragile X syndrome and autism. PLoS ONE
5, e9706. [PMC free article] [PubMed]
 64. Moy SS, Nadler JJ, Young NB, Nonneman RJ, Grossman AW, et al. ..
(2009) Social approach in genetically engineered mouse lines relevant to
autism. Genes Brain Behav 8, 129–42. [PMC free article] [PubMed]
  65. Mineur YS, Huynh LX, Crusio WE (2006) Social behavior deficits in the
Fmr1 mutant mouse. Behav. Brain Res. 168, 172–175. [PubMed]
 66. Yan QJ, Rammal M, Tranfaglia M, Bauchwitz RP (2005) Suppression of
two major fragile X syndrome mouse model phenotypes by the mGluR5
antagonist MPEP. Neuropharmacology 49, 1053–1066. [PubMed]


XXII. Effect of selected spices on chemical and sensory

markers in fortified rye‐buckwheat cakes
Food Sci Nutr. 2016 Jul; 4(4): 651–660.
Małgorzata Przygodzka, 1 Henryk Zieliński, 1 Zuzana Ciesarová, 2 Kristina Kukurová,
2 and Grzegorz Lamparski 1

Abstract

The aim of this study was to find out the effect of selected spices on chemical and
sensorial markers in cakes formulated on rye and light buckwheat flour fortified with
spices. Among collection of spices, rye‐buckwheat cakes fortified individually with
cloves, nutmeg, allspice, cinnamon, vanilla, and spice mix revealed the highest
sensory characteristics and overall quality. Cakes fortified with cloves, allspice, and
spice mix showed the highest antioxidant capacity, total phenolics, rutin, and almost
threefold higher available lysine contents. The reduced furosine content as well as
free and total fluorescent intermediatory compounds were observed as compared to
nonfortified cakes. The FAST index was significantly lowered in all cakes enriched
with spices, especially with cloves, allspice, and mix. In contrast, browning index
increased in compare to cakes without spices. It can be suggested that clove,
allspice, vanilla, and spice mix should be used for production of safety and good
quality cakes.

Keywords: Antioxidant activity, buckwheat flour, cakes, Maillard reaction, sensory

evaluation, spices

Introduction

Spices have been used by human since ancient times. According to the U.S. Food
and Drug Administration (US FDA) spice is an “aromatic vegetable substance in the
whole, broken, or ground form, the significant function of which in food is seasoning
rather than nutrition” and from which “no portion of any volatile oil or other flavoring
principle has been removed” (Sung et al. 2012). The compilation of current trends in
bakery technology to enhance antioxidant activity of bakery products was widely
described by Dziki et al. (2014). At the top of the list, spices have been suggested as
a well‐recognized source of compounds with antioxidant potential (Hinneburg et al.
2006; Wojdyło et al. 2007; Charles 2013). Recently, the inquisitive studies on antioxidant

capacities of spices employing updated analytical methods were reported by

Przygodzka et al. (2014). According to data collected in Food Frequency
Questionnaire, the average of spices/herbs intake was estimated as 1.1 grams per
day for one person what revealed that spices are important contributor of
antioxidants to our diet (Carlsen et al. 2011). Spices are mainly employed as flavoring
and color agents, whereas potential use to preservation food and disease prevention
has been already studied (Kaefer and Milner 2008; Cazzola and Cestaro 2014;
Embuscado 2015). Spice application was demonstrated by Illupapalayam et al. (2014).
Their probiotic‐yogurt with cardamom, cinnamon, and nutmeg has increased
sensorial acceptability among consumers, besides spice addition increased overall
antioxidant activity of this functional product.

Presently, consumers seeking for new food products are focused on joining two
aspects: a taste and functional properties (Wójtowicz et al. 2013). The functional
properties of innovative products in prevention or therapy support in selected
diseases are desirable. Anticancer, antiallergic, antiviral, cholesterol‐reducing, blood
pressure‐reducing, and arteriosclerosis‐reducing were ascribed as buckwheat's
healing effects (Krkošková and Mrázová 2005). In this trend, buckwheat‐based
product with spices addition can be a good alternative to inclusion in varied and
balanced diet. Moreover, several studies approved consumer acceptability of
buckwheat‐based products (Wronkowska et al. 2008; Filipčev et al. 2011; Sedej et al.
2011; Chlopicka et al. 2012). The high sensorial acceptability of 30% buckwheat flour

incorporation in baked products was reached.

In this study, the recipe of rye‐buckwheat cakes (RBC) was enriched with one spice
form the list including: anise, allspice, cardamom, cinnamon, cloves, coriander,
fennel, ginger, nutmeg, star anise, vanilla, white pepper, and commercial spice mix
for ginger cakes. The sensory evaluation of cakes was used as a tool for selection
cakes accepted by sensory panel. It seems to be rationale to use Maillard reaction
(MR) products as markers for description quality of RBC fortified with spices. It is
well‐known that MR products are responsible for the development of color, taste,
and aroma as well as the nutrients loss of thermally treated food (Markowicz Bastos
and Gugliucci 2015). Virág et al. (2013) stated that remaining lysine after baking process
is a good indicator of the progress of MR and important to monitor its content as
essential amino acids. Several unfavorable food contaminants are simultaneously
formed in thermal processing. During early step of MR, the nutritionally valued
available lysine can be converted into furosine, a heat‐treatment marker (Gökmen
et al. 2008; Giannetti et al. 2014). The advanced stage of MR is characterized by the
formation of fluorescence compounds with regard to advanced glycation end‐
products formation and monitoring protein degradation by FAST index (Delgado‐
Andrade et al. 2007; Liogier de Sereys et al. 2014). Positively, melanoidins formed in
the final stage of MR are responsible for the color formation and possess the ability
to scavenge free radicals (Langner and Rzeski 2014). It was concerned that MR
products formation in a model systems and food products can be reduced/increased
by an application of substances having a high antioxidant potential (Marková et al.
2012; Oral et al. 2014; Cheng et al. 2015).

The aim of this study was to find out an impact of selected spices on Maillard reaction
progress and sensory quality of RBC fortified with spices. Therefore, analysis of
selected chemical and sensorial markers such as quercetin 3‐rhamnosylglucoside
(rutin)—the main buckwheat flavonoid, available lysine, total phenolics contents
(TPC), antioxidant capacity (AC) of cakes using extracts scavenging activity against
ABTS•+ radical cation and against superoxide anion radicals (O2 •−) measured by the
photochemiluminescence method (PCL), and furosine, fluorescent compounds, and
melanoidins, were addressed in this study. To determine the impact of thermal
treatment on protein damage, FAST index was calculated.

Materials and Methods

Chemicals and reagents

2,2′‐Azinobis(3‐ethylbenzothiazoline‐6‐sulphonic acid) diammonium salt (ABTS), 6‐

hydroxy‐2,5,7,8‐tetramethylchroman‐2‐carboxylic acid (Trolox), rutin (quercetin‐3‐
rutinoside), lysine (Nα‐acetyl‐L‐lysine), and pronase E (Streptomyces griseus lyoph.)
were purchased from Sigma (Sigma Chemical Co., St. Louis, MO). PCL ACW
(Antioxidant Capacity of Water‐soluble substances) kit for PCL assay was from
Analytik Jena AG (Jena, Germany). o‐phtaldialdehyde for fluorescence (OPA) and
sodium dodecylsulfonate (SDS) were supplied by Fluka (Buchs, Switzerland).
Furosine (2‐furoylmethyl‐lysine) was purchased from PolyPeptide (Strasbourg,
France). Acetonitrile and methanol (HPLC purity) were provided by POCh (Gliwice,
Poland). Water was purified with Mili‐Q‐system (Millipore, Bedford, MA).

Formulation of rye‐buckwheat ginger cakes enhanced with spices

The cakes were baked using rye flour blended with light buckwheat flour in ratio
70:30 (w/w). The making process involved dough preparation by mixing flours,
honey, and sugar. Each one of selected spices (2% on flour mixture basis; w/w) from
the list: anise, allspice, cardamom, cinnamon, cloves, coriander, fennel, ginger,
nutmeg, star anise, vanilla, white pepper, and commercial spice mix for ginger cake,
was used in RBC recipe. According to the producer's declaration, commercial spice
mix contained cinnamon, pepper, clove, anise, coriander, fennel, and nutmeg. The
amounts of ingredients added to make each type of cake are presented in Table 1.
The dough was cut into 0.5‐cm‐thick disks of 5.5 cm diameter and baked at 180°C
for 18 min in a DC‐32E electric oven (Sveba‐Dahlen, Fristad, Sweden). Finally, the
cakes were freeze‐dried and grounded into powder. The powdered samples were
sieved through a 60‐mesh screen and then stored at −20°C until analyzed.

Table 1. Formula of rye‐buckwheat cakes fortified with selected spices: anise,

allspice, cardamom, cinnamon, cloves, coriander, fennel, ginger, nutmeg, star anise,
vanilla, white pepper, and commercial mix of spices for ginger cake
Ingredients Control cake Rye‐buckwheat cake with spice addition
Rye flour (T‐720) (g) 70 70
Light buckwheat flour (g) 30 30
Buckwheat honey (g) 50 50
Sugar (g) 20 20
Baking powder (g) 3 3
Butter (g) 25 25
Selected spice (g) 0 2

Sensory evaluation

Twenty‐four attributes related to the appearance, odor, taste, and texture of rye‐
buckwheat ginger cakes with spices were selected and thoroughly used to during
profiling procedure. Sensory characteristics and overall quality of ginger cakes were
evaluated according to international unified standards (ISO/DIS 1998). A six‐member
trained panel judged ginger cakes in a 10‐point scale (0—for weak, 10—for very
good) using quantitative descriptive analysis to determine differences between each
type of ginger cakes (Stone et al. 2012). The description of sample preparation and
standardized procedure of sensory evaluation were in details presented by Zieliński
et al. (2012).

Overall acceptability of each sample was evaluated in relation to the sensory

preferences on the basis of overall appearance, aroma, taste, and texture, in a 10‐
point hedonic scale, where: not accept, and 10 fully accept. The profiling analysis of
all samples was run in duplicate (two series) proceeded by introduction session.
Ginger cakes were considered as acceptable if their mean scores for overall
acceptability were above 6 (Kowalska et al. 2012).

Preparation of extracts from RBC

Rye‐buckwheat cake powders (100 mg) were extracted with 1 mL of 65% (v/v)
ethanol. After ultrasonic vibration for 30 sec, the solution was mixed and centrifuged
for 5 min at 5000× g at 4°C. That step was repeated five times and the supernatants
were collected into 5‐mL flask. Final extracts concentration was 20 mg/mL. Ethanol
extracts were prepared in triplicate. Next extracts were stored at −20°C until analysis
of rutin content, total phenolic compounds (TPC), and AC by ABTS and PCL ACW
assays.

Determination of total phenolic content (TPC) and rutin

The TPC was determined with Folic‐Ciocalteu reagent as it was described in details
by Przygodzka et al. (2014). TPC was standardized against gallic acid and expressed
in terms of mg gallic acid equivalents (GAE)/g dry matter. The content of rutin in
ginger cakes was determined with HPLC (Shimadzu, Japan) with UV detector (SPD‐
10A) set up 330 nm as it was recently described by Zielińska et al. (2010). For
quantitative analysis, rutin standard was prepared in triplicate at five concentrations
within the range 1.0–40 μM. All solutions were filtered through a 0.45 μm nylon
membrane before use. The results were expressed in μg per g of dry matter.

Antioxidant capacity determination

The AC of RBC enhanced with spices was determined by ABTS and

photochemiluminescence (PCL ACW) assays as it was described in details by
Przygodzka et al. (2014). The results provided by ABTS and PCL ACW methods were
expressed as μmol of Trolox equivalents (TE)/g DM.

Available lysine content determination

The OPA assay as described by Michalska et al. (2008) was employed to determine
available lysine content using the microplate reader (Infinite ® M1000 PRO, Tecan,
Switzerland). Exactly 50 μL of sample, 100 μL of OPA reagent, and 100 μL of water
were added to well and incubated for 3 min (96‐well microplate; Porvair Sciences,
Norfolk, UK). Then the fluorescence reading was measured at extinction wavelength
340 nm and emission wavelength 455 nm. Quantitative analysis was performed by
the external standard method, employing a calibration curve of N α‐acetyl‐L‐lysine
ranged from 10 to 250 μM. Each result is a mean of three independent extractions.

Maillard reaction products determination

Furosine content determination

According to Delgado‐Andrade et al. (2007), 30 mg of cake sample was hydrolyzed

with 4 mL of 4.9 M HCl at 110°C for 23 h in a Pyrex screw‐capped vial with PTFE‐
faced septa. Hydrolysis tubes must be sealed under nitrogen. After that the
hydrolysates was centrifuged for 10 min. A 0.5 mL portion of the supernatant was
applied to a Sep‐pak C18 cartrigde (Millipore) conditioned with 5 mL of methanol
and 10 mL of distilled water, then eluted with 3 mL of 3M HCL and evaporated under
vacuum. The dried sample was dissolved in 1 mL of a mixture of water, acetonitrile,
and formic acid (95:5:0.2) before HPLC analysis.

The furosine was quantified by HPLC system (Shimadzu, Japan) comprised of a

controller (SCL‐10AVP), a PDA detector (SPD‐M10AVP). A Cadenza CD‐C18
column (250 × 2 mm, 3 μm, Imtakt, Kyoto, Japan) at 35°C. The mobile phase
consisted of a solution of 5 mM sodium heptanases sulfonate containing 20% of
acetonitrile and 0.2% of formic acid. The elution was isocratic and the flow rate was
0.2 mL/min. The UV detector was set at 280 nm. Calibration curve was made by the
external standard of furosine 0.2–9 μg/mL.

Measurement of MR fluorescence intermediatory compounds and FAST index

calculation
The fluorescence of free, linked‐to‐protein, and total intermediary compounds (FIC)
was determined after sample extraction and further enzymatic hydrolysis using
pronase E according to Delgado‐Andrade et al. (2006). Readings were recorded in a
luminescent spectrofluorimeter (LS 50B; Perkin Elmer, Waltham, USA) setting at λ
ext. = 347 nm and λ em. = 415 nm. Tryptophan fluorescence TrpFL was measured at λ
ext. = 290 nm and λ em. = 340 nm. Results are expressed in fluorescence intensity (FI)
per mg of sample DM. The FAST index was calculated as recently reported by
Zieliński et al. (2012) with a one novelty modification based on the use of fluorescent
compounds linked‐to‐proteins for index calculation. The samples were analyzed in
triplicate and FAST index data were expressed as a percentage (%).

Brown pigments assay

Formation of brown pigments was estimated as reported in details by Zieliński et al.

(2012). All measurements were performed in triplicate. Results were expressed as
arbitrary absorbance units.

Statistical analysis

The results of the chemical analyses are given as the means and the standard
deviation of three independent measurements. Statistical one‐way analysis of
variance (ANOVA) using Fischer test was performed. The significance level was set
at P < 0.05. The correlation test between rutin content, antioxidant ability, and MRPs
formation was performed and the Pearson correlation coefficients were calculated.
Statistical analyses were performed using software package (StatSoft Inc., v. 7.1,
Tulsa, OK).

Results and Discussion

Consumer acceptance

The overall quality of RBC made of light buckwheat flour encorporated with selected
spices is presented on Figure 1. The overall acceptability for control cake was 6.4,
in comparison its sensorial score was higher than for cakes made of buckwheat and
wheat flour proposed by Kaur et al. (2014). The RBC fortified with spices showed
following rank of acceptability: cakes with vanilla (7.9), with spice mix (7.5), with
cinnamon (6.9), and with nutmeg (6.6). Also high acceptability showed cakes fortified
with allspice (6.2) and cloves (6.1). Taking into account the overall acceptability
rating, it was decided to use RBC fortified with allspice, cinnamon, cloves, spice mix,
nutmeg, and vanilla for further chemical analysis.
Figure 1 The overall quality of rye‐buckwheat cakes fortified with spices addition.

Sensory evaluation

In order to observe the above differences in the analyzed samples more clearly, the
sensory profiles of RBC as control and cakes enriched with clove, cinnamon,
allspice, nutmeg, vanilla, and spice mix were displayed as spider diagrams in
Figure 2A,B. The mean sensory ratings for the samples and the analysis of variance
are presented in Table 2. The buckwheat honey, sugar as well as cinnamon and
vanilla usage can contribute on high level of sweet taste and odor in presented
cakes. The sweetness can be mitigated bitter taste and aftertaste of buckwheat flour.
According to Pauly et al. (2013), the high values of hardness can be linked to type of
buckwheat flours used in recipe. ANOVA showed that there were significant
differences in the intensity of attributes such as brown color, odor descriptors:
“sweet,” “biscuit,” “cinnamic,” “cloves,” “vanilla,” and “spicy”; taste descriptors: bitter,
“cinnamic,” “cloves,” “vanilla,” “spicy,” pungent, and aftertaste. The high score of
acceptability for cake with vanilla was involved with significantly low contribution of
negative attributes as spicy, pungent taste, and aftertaste, masked by intensive
biscuit and vanilla odor, and vanilla taste. The highest scores of cinnamic odor and
cinnamic taste, vanilla odor and vanilla taste as well as cloves odor and cloves taste
were characteristic for cakes with cinnamon, vanilla, and cloves, respectively. It can
be concluded that 2% of spice addition was sufficiently for differentiation between
samples. Therefore, the sensory evaluation proved that addition of spices to RBC
formulation increased the sensory quality of products.
Figure 2 Sensory profiles of rye‐buckwheat cake without spice (control cake) and
rye‐buckwheat cakes fortified with selected spices: nutmeg, cinnamon, spice mix,
cloves, allspice, and vanilla. O, attributes of odor; t, attributes of taste.

Table 2 Descriptive analysis of results based on the analysis of variance (ANOVA)

performed on rye‐buckwheat cakes (RBC) fortified with selected spices
Attribute Rye‐buckwheat cakes
Control Cloves Nutmeg Allspice Cinnamon Vanilla Spice
mix
1 Brown color 5.67cd 4.81de 6.58bc 5.01cde 8.51a 3.98e 7.28ab
2 Sweet o. 3.38c 3.93bc 4.21bc 5.08ab 6.26a 6.08a 5.41ab
Dried
3 1.82a 0.25a 1.99a 0.54a 0.89a 0.36a 0.60a
mushroom o.
4 Biscuit o. 3.23bc 0.97d 2.42cd 3.95abc 4.67ab 5.32a 4.31ab
5 Anise o. 0.01a 0.01a 0.20a 0.14a 0.03a 0.01a 0.02a
6 Cinnamic o. 0.09b 0.01b 0.00b 0.51b 3.10a 0.02b 2.66a
7 Cloves o. 0.00c 4.81a 0.02c 0.00c 0.13c 0.01c 0.88b
White pepper
8 0.07a 0.01a 0.01a 0.02a 0.02a 0.02a 0.02a
o.
9 Vanilla o. 0.09c 0.01c 0.08c 0.02c 0.12c 4.66a 0.34b
10 Spicy o. 3.61a 4.92a 4.62a 3.76a 4.27a 1.45b 3.48a
11 Sweet t. 6.53a 5.53a 5.23a 6.61a 6.35a 7.25a 6.04a
12 Bitter t. 2.06ab 0.60b 2.51a 0.70b 1.95ab 0.51b 3.41a
13 Anise t. 0.51a 0.02a 0.46a 0.02a 0.02a 0.02a 0.02a
14 Cinnamic t. 0.02b 0.02b 0.02b 0.85b 3.92a 0.02b 1.40b
15 Cloves t. 0.02c 5.79a 0.02c 0.02c 0.00c 0.01c 0.62b
16 White pepper t. 0.01a 0.02a 0.44a 0.38a 0.02a 0.02a 0.02a
17 Vanilla t. 0.34b 0.03b 0.01b 0.02b 0.02b 5.01a 0.06b
18 Ginger t. 0.01a 1.68a 0.83a 0.02a 0.01a 0.03a 0.84a
19 Spicy t. 4.68ab 5.17ab 5.54a 3.18c 4.06bc 1.13d 4.08bc
20 Pungent t. 0.13b 1.99a 0.72b 0.26b 0.30b 0.02b 0.61b
21 Aftertaste 2.15bc 4.26a 3.85ab 2.17bc 3.13abc 1.57c 2.64abc
22 Hardness 7.98a 8.19a 8.19a 8.41a 7.66a 8.03a 7.96a
23 Crispness 6.42a 6.08a 6.59a 6.24a 6.86a 7.88a 7.51a

The cakes were marked in a 10‐point scale (0—for weak, 10—for very good). Means
in each row with the same letters do not have significant differences (Fisher test,
P < 0.05). o, attributes of odor; t, attributes of taste.

The total phenolic content (TPC) and rutin determination

The content of rutin and total phenolic compounds in RBC fortified with selected
spices is compiled in Table 3. The addition of spices to RBC formula resulted in
increase in TPC in comparison to control cake. The significantly highest TPC values
in cakes with spice mix, cinnamon and cloves were observed, 4.84‐fold, 2.03‐fold,
and 1.97‐fold, respectively (P < 0.05). Our results are in accordance to Przygodzka
et al. (2014) who noted high TPC values for spice mix, cloves, and cinnamon. The
 level of rutin was significantly higher in cakes after cloves and allspice application in
comparison to control cake (3.07 and 1.66 times, respectively). The novel RBC
possess rutin content 5.5 times and 6 times higher than in gluten‐free rice‐light
buckwheat and rice‐wholegrain buckwheat (70:30, w/w) breads, respectively (Sakač
et al. 2011), which might be related to differentiations of rutin content in buckwheat
flours and application of buckwheat honey in the cakes recipe. Additionally, our TPC
results for control cake are in agreement with the results of these gluten‐free breads
(Sakač et al. 2011).

Table 3 The content of rutin, total phenolic compounds, and antioxidant capacity of
rye‐buckwheat cakes enhanced with selected spices

Type of cake Rutin (μg/g TPC (mg Antioxidant capacity

DM) GAE/g DM) (μmol TE/g DM)
PCL ACW ABTS
Rye‐buckwheat control cake 104.36 ± 3.75 1.12 ± 0.03 6.15 ± 0.51 21.13 ± 0.88f
cd g d

Rye‐buckwheat cake with

100.63 ± 2.33d 1.32 ± 0.08f 5.17 ± 0.83d 21.87 ± 1.21f
vanilla
Rye‐buckwheat cake with
101.62 ± 6.17d 2.28 ± 0.05b 8.69 ± 0.87c 49.38 ± 0.19c
cinnamon
Rye‐buckwheat cake with
319.80 ± 3.51a 2.11 ± 0.04c 23.30 ± 1.00b 55.52 ± 2.73b
cloves
Rye‐buckwheat cake with
173.19 ± 7.52b 1.84 ± 0.17d 9.25 ± 0.22c 40.86 ± 2.28d
allspice
Rye‐buckwheat cake with
100.84 ± 3.53d 1.56 ± 0.10e 5.78 ± 0.13d 30.49 ± 0.84e
nutmeg
Rye‐buckwheat cake with spice
111.16 ± 3.50c 2.70 ± 0.09a 31.56 ± 0.05a 63.24 ± 1.31a
mix

TPC (total phenolic content) is expressed in mg of gallic acid equivalents/g of dry

matter (mg GAE/g DM). Antioxidant capacity measured by ABTS and PCL ACW
methods is expressed in μmol of Trolox equivalents (TE)/g DM. Values are means ±
standard deviation (n = 3). Values in each column with different small superscript
letters are significantly different (Fisher test, P < 0.05). DM, dry matter.

A weak correlation between rutin and TPC was noted (r = 0.23). It can be suggested
that other flavonoid compounds extracted from spices have higher contribution on
the antioxidant properties of RBC. Moreover, the negative correlation was found
between TPC and bitter taste (r = −0.61). It can be said that phenolic compounds
increased bitterness, our findings are in accordance to information collected by
Shahidi and Naczk (1995).

Antioxidant properties
The 2% spices substitution in the formulation of cakes made of rye and light
buckwheat flour resulted in significant differences (P < 0.05) in the AC determined
against scavenging ability of ABTS•+ and O2 −• (PCL ACW method) radicals. The
results for AC determination are presented in Table 3. Sorted by AC measured by
ABTS method RBC supplemented with spice mix has the highest antioxidant value
followed by cloves, cinnamon, then allspice, nutmeg and finally vanilla cakes.
Significantly highest results were obtained for RBC with spice mix, cloves, cinnamon
and allspice, 2.99, 2.63, 1.93 –times higher than in control cake. The antioxidant
potential evaluation by PCL ACW method for RBC was listed as follows: spice mix>
cloves> allspice ≈ cinnamon> nutmeg≈ vanilla. The addition of spice mix and cloves
was more effective in enhancing antioxidant activity, as evaluated by means of PCL
ACW, which increased 6.10‐fold and 4.50‐fold, respectively. These results are in
agreement with findings of Hossain et al. (2008), which indicated that cloves and
cinnamon have the highest AC among other spices.

Moreover, the TPC and rutin contribution on AC overall was expressed by correlation
coefficient. The strong correlation between TPC/ABTS and TPC/PCL ACW data
(r = 0.97 and r = 0.81, respectively) were observed. According to studies of Bi et al.
(2015), the strong correlation between TPC and AC measured by ABTS was observed
for cloves extracts. It may suggest that active compounds from cloves have high
contribution to antioxidant overall capacity of RBC. However, the weaker correlations
for rutin versus PCL (r = 0.43) and rutin versus ABTS (r = 0.44) were noted.

Available lysine determination

The results for available lysine amount after thermal processing are shown in
Table 4. Available lysine values of 0.52 mg/g DM was found in control RBC without
condiments supplementation. According to the obtained results for available lysine
in rye‐buckwheat cake with condiments addition a protective effect on lysine
blockage was found. The statistically significantly most high lysine blockage content
in cloves, allspice, and spice mix was noted (2.75, 2.64, 2.20 times higher). The
observation of protective effect of spices on lysine blockage in cakes was confirmed
by positive correlation between OPA values and rutin and TPC contents (r = 0.74
and 0.63, respectively), as well as AC measured by ABTS (r = 0.74) and PCL ACW
(r = 0.62). On the basis of these results, it can be concluded that spices positively
influenced the baking process and increased the nutritional value of the product.

Table 4 Data on Maillard reaction products in rye‐buckwheat cakes fortified with

spices
Type Availab Furosin Free FIC Total FIC Linked‐ Tryptop FAST Browni
of le e (μg/g) (FI/mg (FI/mg to‐ han (%) ng (AU)
cake lysine DM) DM) protein (FI/mg
fortifi (mg/g (FI/mg DM)
ed DM) DM)
with
spice
s
Contro 0.59 ± 0 510.8 ± 77.44 ± 1 166.57 ± 89.13 ± 1 19.30 ± 1 462 ± 0.36 ± 0
l .04d 12.0a .44d 2.20c .76bc .00cd 8b .02e
0.67 ± 0 488.4 ± 116.14 ± 201.68 ± 85.54 ± 5 17.68 ± 0 484 ± 0.41 ± 0
Vanilla
.09d 6.5b 0.97a 6.12a .15c .73d 2a .01d
Cinna 0.95 ± 0 450.2 ± 69.06 ± 3 182.60 ± 113.54 ± 30.87 ± 0 368 ± 0.49 ± 0
mon .03c 9.7c .35b 4.18b 7.53a .81b 4e .01c
1.62 ± 0 111.7 ± 69.21 ± 1 142.36 ± 73.15 ± 7 21.92 ± 1 230 ± 0.68 ± 0
Cloves
.06a 7.2f .04b 6.23e .27d .21c 9g .01a
Allspic 1.56 ± 0 292.9 ± 65.63 ± 1 153.26 ± 87.63 ± 5 19.44 ± 0 451 ± 0.52 ± 0
e .07a 5.6d .65c 3.76d .46bc .22cd 3c .01b
Nutme 0.99 ± 0 458.6 ± 114.91 ± 200.92 ± 86.01 ± 7 31.87 ± 3 392 ± 0.48 ± 0
g .02c 10.8c 1.11a 8.34a .23c .61b 4d .02c
Spice 1.30 ± 0 220.5 ± 54.02 ± 2 151.47 ± 97.45 ± 3 38.00 ± 2 256 ± 0.52 ± 0
mix .16b 10.1e .05e 1.23d .28b .26a 2f .01b

FIC is expressed in fluorescence intensity (FI) per mg of sample DM. Browning is

expressed as absorbance units (AU). Values are means ± standard deviation
(n = 3). Values in each column with different small superscript letters are significantly
different (Fisher test, P ≤ 0.05). DM, dry matter.

Maillard reaction products evaluation

As shown in Table 4, furosine content decreased after spices addition from 4 up to

78% in comparison to cake without spices addition. The furosine contents in rye‐
buckwheat ginger cakes after spices addition were significantly lower than the
maximum allowable tolerance of furosine in milk proposed by Martysiak‐Żurowska
and Stołyhwo (2007). Moreover the furosine content is even twice lower than
determined in commercial breakfast cereals (Rada‐Mendoza et al. 2004) and five
times lower in cookies made of wheat flour (Gökmen et al. 2008). The observation of
inhibition effect of spices on furosine formation in cakes with spices was confirmed
by high correlation between furosine and rutin (r = −0.80) as well as AC measured
by ABTS (r = −0.81) and PCL ACW (r = −0.84). Whereas the weaker correlation
(r = −0.68) between furosine and TPC contents was calculated. The strong
relationship between furosine and OPA values and melanoidins formation were
observed. The negative correlation coefficients were calculated (r = −0.91, r = −0.90,
respectively), that can suggest that available lysine is a dominant precursor of MR
progress in early stage and furosine formation is competitive to melanoidins.
Whereas the positive correlation between furosine and FAST index data was noted
(r = 0.81). It can be said that, in a great percent of furosine is converted into
fluorescent compounds.

Collected in Table 4, the total, free, and linked‐to‐protein FIC found in all cakes were
within the range of 142.4–201.7, 54.0–116.1, and 73.1–113.5 FI/mg sample DM,
respectively. The total FIC values significantly decreased after cloves, spice mix,
and allspice supplementation (15%, 10%, and 8%, respectively), the same effect
was observed for free FIC data. It can be said that some spices promote
fluorescence compounds formation, however between total FIC and rutin negative
correlation was noted (r = −0.66). Between total FIC and OPA data according to
calculated correlation coefficient (r = −0.72).

To describe protein nutritional loss, FAST index was calculated as a ratio between
linked‐to‐protein fluorescence/tryptophan fluorescence and expressed in
percentage. The strong correlation between tryptophan and TPC data and
antioxidant ability measured by PCL ACW and ABTS assays (r = 0.92, 0.87, and
0.93, respectively) were noted. According to tryptophan results, it can be said that
compounds from RBC supplemented with spices with strong antioxidant ability, have
a potential to increase nutritional value. Table 4 shows the FAST index values. The
values ranged from 230% to 484%. FAST index cloves, spice mix, cinnamon,
nutmeg, and allspice being 2.0, 1.92, 1.8, 1.2, and 1.0 times lower. The positive
influence of spices on FAST index decreasing was proved by correlation coefficient
calculated between FAST/TPC, FAST/PCL, and FAST/ABTS (r = −0.80, r = −0.89,
r = −0.81, respectively). According to results presented by Zieliński et al. (2012), rye
ginger cakes showed a higher FAST index values in comparison to rye‐buckwheat
ginger cakes. However the FAST index for rye ginger cakes before storage is twice
lower than indexes for rye‐buckwheat ginger cakes. This indicates that measuring
the FAST index in rye‐buckwheat ginger cakes before storage and then in a
determined time intervals can be important to monitor ongoing changes.

Brown high molecular polymers of MRP pigments formation were determined and
presented in Table 4. Addition of cloves, spice mix, allspice, cinnamon, nutmeg, and
vanilla to the recipe significant increased (P < 0.05) browning index by 88%, 44%
(both spice mix and allspice), 36%, 33%, and 14%, respectively. It has been proven
that melanoidin formation is positively correlated with AC measured by ABTS test
(r = 0.76) and PCL ACW assay (r = 0.65) and with TPC and rutin contents in the
cakes (r = 0.63 and 0.86). These findings are in agreement with Zieliński et al. (2010),
who also evaluated the positive correlation between the AC and melanoidin content
in wheat‐rye ginger cakes. Moreover, browning index and OPA data were positively
correlated (r = 0.89). The correlation value between FAST and browning indexes
suggests that there is no relationship between loss of nutritional and melanoidin
formation (r = ‐0.81).

Conclusions
Among collection of spices, the RBC fortified individually with cloves, nutmeg,
allspice, cinnamon, vanilla, and spice mix addition revealed the highest sensory
characteristics and overall quality. Cakes fortified with cloves, allspice, and spice mix
showed the highest AC, total phenolics, rutin, and almost threefold higher available
lysine contents. The reduction in furosine content as well as free and total fluorescent
intermediatory compounds was observed as compared to control nonfortified cakes.
In contrast, browning index was increased as compared to cakes without spices. In
this study, the chemical and sensorial markers were fully applicable for description
of the quality of RBC fortified with spices. It can be suggested that cloves, allspice,
vanilla, and spice mix should be used for production of safety and good quality
cakes.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of
this paper.

Acknowledgment

This research was funded by grant No. 2012/07/N/NZ9/02250 from National Science
Centre. The article is a part of the Ph.D. thesis of Małgorzata Przygodzka.

Notes

This paper was supported by the following grant(s):

National Science Centre 2012/07/N/NZ9/02250.

References

1. Bi X., Soong Y. Y., Lim S. W., and Henry C. J.. 2015. Evaluation of antioxidant
capacity of Chinese five‐spice ingredients. Int. J. Food Sci. Nutr. 66:289–292.
[PubMed]
2. Carlsen M.H., Blomhoff R., and Andersen L.F.. 2011. Intakes of culinary herbs
and spices from a food frequency questionnaire evaluated against 28‐days
estimated records. Nutr. J. doi: 10.1186/1475‐2891‐10‐50. [PMC free article]
[PubMed]
3. Cazzola R., and Cestaro B.. (2014). Antioxidant spices and herbs used in
diabetes Pp. 110–119 in Preedy V., editor. , ed. Aging. Diabetes: oxidative
stress and dietary antioxidants. Academic Press, London, UK.
4. Charles D. J. 2013. Natural antioxidants Pp. 39–64 in Charles J.D., editor. ,
ed. Antioxidant properties of spices, herbs and other sources. Springer, New
York, USA.
5. Cheng J., Chen X., Zhao S., and Zhan Y.. 2015. Antioxidant‐capacity‐based
models for the prediction of acrylamide reduction by flavonoids. Food Chem.
168:90–99. [PubMed]
6. Chlopicka J., Pasko P., Gorinstein S., Jedryas A., and Zagrodzki P.. 2012.
Total phenolic and total flavonoid content, antioxidant activity and sensory
evaluation of pseudocereals breads. LWT Food Sci. Technol. 46:548–555.
7. Delgado‐Andrade C., Rufián‐Henares J. A., and Morales F. J.. 2006. Study
on fluorescence of Maillard reaction compounds in breakfast cereals. Mol.
Nutr. Food Res. 50:799–804. [PubMed]
8. Delgado‐Andrade C., Seiquer I., Navarro M. P., and Morales F. J.. 2007.
Maillard reaction indicators in diets usually consumed by adolescent
population. Mol. Nutr. Food Res. 51:341–351. [PubMed]
9. Dziki D., Różyło R., Gawlik‐Dziki U., and Świeca M.. 2014. Current trends in
the enhancement of antioxidant activity of wheat bread by the addition of plant
materials rich in phenolic compounds. Trends Food Sci. Technol. 40:48–61.
10. Embuscado M.E. (2015). Herbs and spices as antioxidants for food
preservation Pp. 251–283. in Shahidi F., editor. , ed. Handbook of
antioxidants for food preservation. Woodhead Publishing Series in Food
Science, Technology and Nutrition, Cambridge, UK.
11. Filipčev B., Šimurina O., Sakač M., Sedej I., Jovanov P., Pestorić M., et al.
2011. Feasibility of use of buckwheat flour as an ingredient in ginger nut
biscuit formulation. Food Chem. 125:164–170.
12. Giannetti V., Boccacci Mariani M., Mannino P., and Testani E.. 2014. Furosine
and flavor compounds in durum wheat pasta produced under different
manufacturing conditions: multivariate chemometric characterization. LWT
Food Sci. Technol. 56:15–20.
13. Gökmen V., Serpen A., Cetinkaya Acar O¨., and Morales F. J.. 2008.
Significance of furosine as heat‐induced marker in cookies. J. Cereal Sci.
48:843–847.
14. Hinneburg I., Damien Dorman H. J., and Hiltunen R.. 2006. Antioxidant
activities of extracts from selected culinary herbs and spices. Food Chem.
97:122–129.
15. Hossain M. B., Brunton N. P., Barry‐Ryan C., Martin‐Diana A. B., and
Wilkinson M.. 2008. Antioxidant activity of spice extracts and phenolics in
comparison to synthetic antioxidants. Rasayan J. Chem. 4:751–756.
16. Illupapalayam V. V., Smith S. C., and Gamlath S.. 2014. Consumer
acceptability and antioxidant potential of probiotic‐yogurt with spices. LWT
Food Sci. Technol. 55:255–262.
17. ISO/DIS 13299. (1998). Sensory analysis – methodology – general guidance
for establishing a sensory profile. International Organization for
Standardization, Geneva, Switzerland.
18. Kaefer C. M., and Milner J. A.. 2008. The role of herbs and spices in cancer
prevention. J. Nutr. Biochem. 19:347–361. [PubMed]
19. Kaur M., Singh Sandhu K., Arora A., and Sharma A. 2015. Gluten free biscuits
prepared from buckwheat flour by incorporation of various gums:
physicochemical and sensory properties. LWT Food Sci. Technol. 62:628–
632.
20. Kowalska H., Marzec A., and Mucha M.. 2012. Sensory evaluation of some
types of bread and functional consumer preferences among the bread.
 Zeszyty Problemowe Postępu Nauk Rolniczych 571:67–78 (abstract in
english).
21. Krkošková B., and Mrázová Z.. 2005. Prophylactic components of buckwheat.
Food Res. Int. 38:561–568.
22. Langner E., and Rzeski W.. 2014. Biological properties of melanoidins: a
review. Int. J. Food Prop. 17:344–353.
23. Liogier de Sereys A., Muller S., Desic S. D., Troise A. D., Fogliano V., Acharid
A.. 2014. Potential of the FAST index to characterize infant formula quality
Pp. 457–475 in Preedy V.R., editor; , Watson R.R., editor; , Zibadi S., editor.
, eds. Handbook of dietary and nutritional aspects of bottle feeding, human
health handbooks. Academic Publishers, Wageningen, the Netherlands.
24. Marková L., Ciesarová Z., Kukurová K., et al. 2012. Influence of various
spices on acrylamide content in buckwheat ginger cakes. Chem. Pap.
66:949–954.
25. Markowicz Bastos D. H., and Gugliucci A.. 2015. Contemporary and
controversial aspects of the Maillard reaction products. Curr. Opin. Food Sci.
1:13–20.
26. Martysiak‐Żurowska D., and Stołyhwo A.. 2007. Content of furosine in infant
formulae and follow‐on formulae. Polish J. Food Nutr. Sci. 57:185–190.
27. Michalska A., Amigo‐Benavent M., Zielinski H., and del Castillo M. D.. 2008.
Effect of bread making on formation of Maillard reaction products contributing
to the overall antioxidant activity of rye bread. J. Cereal Sci. 48:123–132.
28. Oral R. A., Dogan M., and Sarioglu K.. 2014. Effect of certain polyphenols
and extracts on furans and acrylamide formation in model system and total
furans during storage. Food Chem. 142:423–429. [PubMed]
29. Pauly A., Pareyt B., Lambrecht M. A., Fierens E., and Delcour J. A.. 2013.
Flours from wheat cultivars of varying hardness produces semi‐sweet biscuits
with varying textural and structural properties. LWT Food Sci. Technol.
53:452–457.
30. Przygodzka M., Zielińska D., Ciesarová Z., Kukurová K., and Zieliński H..
2014. Comparison of methods for evaluation of the antioxidant capacity and
phenolic compounds in common spices. LWT Food Sci. Technol. 58:321–
326.
31. Rada‐Mendoza M., García‐Bañosa J. L., Villamiela M., and Olano A.. 2004.
Study on nonenzymatic browning in cookies, crackers and breakfast cereals
by maltulose and furosine determination. J. Cereal Sci. 39:167–173.
32. Sakač M., Torbica A., Sedej I., and Hadnađev M.. 2011. Influence of
breadmaking on antioxidant capacity of gluten free breads based on rice and
buckwheat flours. Food Res. Int. 44:2806–2813.
33. Sedej I., Sakać M., Mandić A., Mišan A., Pestorić M., Šimurina O., et al. 2011.
Quality assessment of gluten‐free crackers based on buckwheat flour. LWT
Food Sci. Technol. 44:694–699.
34. Shahidi F., and Naczk M.. 1995. Contribution of phenolic compounds to
sensory characteristics of food Pp. 199–226 in Shahidi F., editor. , ed. Food
phenolics. Sources. Chemistry. Effects. Applications. Taylor and Francis,
London, UK.
 35. Stone H., Bleibaum R. N., and Thomas H. A.. 2012. Sensory evaluation
practices. Academic Press, Orlando, FL.
36. Sung B., Prasad S., Yadav V. R., and Aggarwal B. B.. 2012. Cancer cell
signaling pathways targeted by spice‐derived nutraceuticals. Nutr. Cancer
64:173–197. [PubMed]
37. Virág D., Kiss A., Forgó P., Csutorás C., and Szabolcs M.. 2013. Study on
Maillard‐reaction driven transformations and increase of antioxidant activity in
lysine fortified biscuits. Microchem. J. 107:172–177.
38. Wojdyło A., Oszmiański J., and Czemerys R.. 2007. Antioxidant activity and
phenolic compounds in 32 selected herbs. Food Chem. 105:940–949.
39. Wójtowicz A., Kolasa A., and Mościcki L.. 2013. Influence of buckwheat
addition on physical properties, texture and sensory characteristics of
extruded corn snacks. Polish J. Food Nutr. Sci. 63:239–244.
40. Wronkowska M., Troszyńska A., Soral‐Śmietana M., and Wołejszo A.. 2008.
Effects of buckwheat flour (Fagopyrum esculentum Moench) on the quality of
gluten‐free bread. Polish J. Food Nutr. Sci. 58:211–216.
41. Zielińska D., Szawara‐Nowak D., and Zieliński H.. 2010. Determination of the
antioxidant activity of rutin and its contribution to the antioxidant capacity of
diversified buckwheat origin material by updated analytical strategies. Polish
J. Food Nutr. Sci. 60:315–321.
42. Zieliński H., Amigo‐Benavent M., del Castillo M. D., Horszwald A., and
Zielińska D.. 2010. Formulation and baking process affect Maillard reaction
development and antioxidant capacity of ginger cakes. J. Food Nutr. Res.
49:140–148.
43. Zieliński H., del Castillo M. D., Przygodzka M., Ciesarova Z., Kukurova K.,
and Zielińska D.. 2012. Changes in chemical composition and antioxidative
properties of rye ginger cakes during their shelf‐life. Food Chem. 135:2965–
2973. [PubMed]

XXIII. Compositional profiling and sensorial analysis of

multi-wholegrain extruded puffs as affected by fructan
inclusion
J Food Sci Technol. 2015 Sep; 52(9): 5975–5981.
C. Handa and S. Goomer

Abstract

Rice grits, corn grits, pulse, wholegrain - finger millet and sorghum were utilized in
the production of multigrain extruded puffs using a single screw extruder. The effect
of inclusion of fructan - fructoligosaccharide in multi-wholegrain (MWG) extruded
puffs was examined. MWG fructan enriched puffs puffs had 450 % higher dietary
fiber content than the control puff (CP). These puffs can be categorized as ‘Good
Source’ of fiber as it suffices 17.2 % DV of fiber. Puffs were rated 8.1 ± 0.6, 8.3 ± 0.7,
8.1 ± 0.6, 7.5 ± 0.5 and 8.2 ± 0.6 for color, flavor, texture, appearance and overall
acceptability respectively. The scores for all the attributes were found to be not
significantly different (p < 0.05) from CP. The MWG fructan puffs were rated higher
on flavor than the CP having a score of 8.3 ± 0.7 as opposed to 8.2 ± 0.4 for CP. This
indicates that the nutritional quality and acceptability of MWG extruded puffs could
be improved by the inclusion of fructans.

Keywords: Extrusion, Proximate composition, Sensory analysis, Snack, Fiber

Introduction

Extrusion cooking is a modern, high temperature, short-time processing technology

to manufacture snack foods (White 1994). Food extruders provide thermo-
mechanical and mechanical energy (shear) necessary to cause physicochemical
changes of raw materials with an intense mixing for dispersion and homogenization
of ingredients including conveying, mixing, shearing, heating or cooling, shaping,
venting volatiles and moisture, flavor generation, encapsulation, and sterilization
(Linko et al. 1981; Wiedman and Strobel 1987). Extruded foods are composed
mainly of cereals, starches, and/or vegetable proteins. The major role of these
ingredients is to give structure, texture, mouth feel, bulk, and many other
characteristics desired for specific finished products (Launay and Lisch 1983;
Jamora et al. 2002; Tahnoven et al. 1998). These snacks generally lack dietary fiber.

Snack food extrusion includes subjecting selected grains to a variety of complex

physical processes to yield snacks with varied shapes and textures (White 1994).
Different types of snack foods are available of which the new generation snacks fall
into several categories such as low fat, baked but not fried, high fiber products made
from pure dietary fiber and natural fiber from wholegrain etc.

Consumer acceptance of extruded snack foods is mainly due to the convenience,

value, attractive appearance and texture found to be particular for these foods,
especially when it concerns to snack products (Harper 1981). The success or failure
of a new extruded snack food product is directly related to sensory attributes, where
texture plays a major role (Pamies et al. 2000). Together with the sensory attributes,
today’s consumers are also looking at the nutritional aspects of these snack foods.
The demand for healthier snack foods is on the rise.

In the midst of the current debate on functional foods and the proven health effects
of fructans, it is important that food manufacturers and retailers introduce new
products with added health benefits in the market. In the food industry, inulin or
fructo-oligosaccharides from chicory (Cichorium intubus) roots or Jerusalem
artichoke (Helianthus tuberosus) tubers are being utilized to optimize the quality of
diet. Inulin and fructoligosaccharide (FOS) have been classed as functional food
ingredients (Arai 2002). Inulin is a polydisperse β (2–1) fructan. The degree of
polymerization (DP) is in the range of 2 to 60. The low molecular weight fractions of
inulin (DP 2–20) are known collectively as FOS. These have been generally
recognized as safe (GRAS) by FDA. From an analytical and a physiological point of
view both inulin and FOS have been classed as dietary fibers (Coussement 1999).
They meet the analytical definition of dietary fiber used by AOAC (Report 2001).

Due to the non digestibility of inulin and FOS, they are found to be suitable for
consumption by diabetics (Yamashita et al. 1984; Rumessen et al. 1990). Inulin and
FOS have been termed “prebiotics” (Gibson et al. 1995) because they are non
digestible food ingredients that selectively stimulate growth of health-stimulating
intestinal bacteria such as bifidobacteria (Bouhnik et al. 2004; Wiele et al. 2007). The
caloric value of inulin and FOS is estimated to be 1.0 and 1.5 kcal/g respectively
(Roberfroid 1999). They can be used for either their nutritional advantages or
technological properties, but they are often applied to offer a dual benefit: an
improved organoleptic quality and a better-balanced nutritional composition
(Coussement 1999).

Extruded snacks/puffs presently available in the market are mainly refined grain
based such as corn and rice and thus lack dietary fiber. These puffs are mainly fried
in nature and contain very high amount of fat (upto 30 %). The present study aimed
to develop multigrain, fiber enriched, low calorie extruded puffs using wholegrain
together with fructan. Finger millet is rich in calcium and dietary fiber. Sorghum is
rich in iron and fiber. These are nutritious millets known for their health benefits. But
the coarse texture of these grains limits their application. Using soluble fiber - fructan
(fructoligosaccharide) is one way to reduce gritty texture because they are more
palatable. Direct puffing technology which totally eliminates the frying operation in
snack production has been utilized for the production of puffs. The novel feature of
the products is that these are not only low calorie but also contains dietary fiber both
soluble and insoluble.

Thus, in the present research the concept of optimal nutrition, by way of introduction
of functional ingredients in the extruded puffed snacks has been addressed. The
imperative sensorial properties and proximate composition which are decisive to the
acceptability of extruded puffs have been investigated.

Materials and methods

Single screw extrusion of multi-wholegrain puffs

A tailor made single screw extruder with a capacity of up to 25 kg (G.L. Extrusion

System Private Limited, Delhi, India) founded on direct expansion technology was
used for production of puffs. The standard formulation for control puff included rice
grits: corn grits: green gram pulse in the ratio of 80:10:10. Rice (Oryza spp.) grits
and corn (Zea mays) grits (bag size of 50 kg each) were procured from Ahaar
Consumer Products (P) Limited, Lawrence Road, Delhi, India. The specifications of
the product included description of the raw material, sensory characteristics,
physicochemical specification and particle size distribution. Green gram pulse was
procured locally and was standardized as per Prevention of Food Adulteration Act
(PFA), 1955. All raw material including rice grits, corn grits, green gram pulse, finger
millet and sorghum were analysed for proximate composition using standard AOAC
2005 procedures. Extrusion of control puffs (CP) was carried out at a temperature of
110 °C. Extrudates were cylindrical puffs. The diameter of the extrudates varied
between 12.5 and 13.5 mm whereas the length varied between 27.9 and 29.3 mm
Extrudates were spread on perforated aluminum trays and dried in an air oven
maintained at 100 °C for 15 min to moisture content of 2.5–3 %.

Multi-wholegrain (MWG) puffs were developed using wholegrain millets-finger millet

and sorghum by successively replacing rice and corn grits. Finger millet (Eleusine
coracana) and Sorghum (Sorghum vulgare) were purchased locally and were
standardized as per IS 4333 (Part I and II): 1996 specification “Method of Analysis
for food grains”. All puffs were prepared in a batch size of 4 kg. Temperature for
extrusion of multi-wholegrain (MWG) puff was standardized at 110 °C in relation to
optimum expansion. Extrudates were spread on perforated aluminum trays and
dried in an air oven maintained at 100 °C for 15 min to moisture content of 2.5–3 %.

MWG puffs with 20 % finger millet and 30 % sorghum were enriched with fructan
(FOS) at 10 and 15 % levels. FOS (Beneo Raftilose® P95, Orafti, Belgium, U.K) was
procured from DPO Food Specialities Pvt. Ltd, Thane, Maharashtra, India. Beneo
Raftilose P95 is a powder containing mainly oligofructose produced by partial
enzymatic hydrolysis of chicory inulin. It is a food ingredient composed of
oligofructose, fructose, glucose and sucrose. The total number of fructose or glucose
units (=Degree of Polymerization or DP) of oligofructose ranges mainly between 2
and 8. The technical specifications of FOS are provided in Table 1. FOS was
incorporated at the initial blending (preconditioning) stage together with other raw
materials. The temperature for extrusion of FOS enriched multigrain puff was
standardized to be 130 °C. Extrudates were spread on perforated aluminum trays
and dried in an air oven maintained at 100 °C for 15 min to moisture content of 2.5–
3 %. Extrusion could be effectively undertaken with 10 % FOS level whereas with
15 % level some complexity was encountered in running the extruder, as FOS is
hygroscopic in nature and because of intense heat the likelihood of charring
enhances. Therefore, inclusion of FOS in MWG puffs was limited to 10 % level.
Figure 1 illustrates control puffs, multi-whole grain puffs and multi-whole grain puffs
with fructan.

Table 1 Technical specifications of FOS (Beneo Raftilose® P95)

Parameter Analysis
Oligofructose ≥93.2 %
Glucose+fructose+Sucrose <6.8 %
Carbohydrate >99.5 %
Caloric value 1.5 Kcal/g
Degree of polymerization (DP) 2–8
Appearance Fine white Powder
Parameter Analysis
Slightly sweet, without
Taste
after taste
Behaviour Hygroscopic
Dispersability in water Excellent
Solubility in water >750 g/l
Density Approx. 700 ± 70 g/l

Orafti, Belgium, U.K

Fig. 1 Control puffs, multi-whole grain puffs and multi-whole grain puffs with fructan

Packaging of extruded puffs

All puffs were packed in 20 g three layered metallized pouches (procured from Jain
Flexipack Private Limited, Azadpur, Delhi, India) comprising of the top layer made
up of polyester film (12 μm thick) printed on reverse having a grammage of
16.8 g/m2, the middle layer was made up of plain metallized polyester film having a
grammage of 16.8 g/m2 laminated to the top layer with the help of 2 μm adhesive
layer and the third layer made up of low density polyethylene (LDPE) film plain
(30 μm thick) having a grammage of 27.9 g/m2 laminated to middle layer with the
help of 2 μm adhesive layer. Packaged puffs were stored at ambient temperature
until needed for nutritional and sensorial analysis.

Compositional profiling of extruded puffs using standard analytical methods

Complete compositional profiling of the CP and MWG fructan enriched puff was
undertaken as follows:
Fat, protein, moisture, ash and carbohydrate content

Fat content was determined using soxhlet extraction, protein using Kelvac (Kes 12
l) Nitrogen Estimator (Pelican Equipments, India), moisture by air oven method (IS
1011 2002), and ash using IS 1011:2002. Carbohydrate content was calculated by
the difference.

Dietary fiber and fructan estimation

The total dietary fiber was estimated by AOAC 991.43:2005 method Total fructan
content of the products and fructan content of corn and rice grits, sorghum, finger
millet and green gram pulse were determined by AOAC 999.03:2005 enzymatic
spectrophotometric method using Megazyme K-Fruc Kit, Ireland.

Fatty acid profiling including trans fat using Gas Chromatography

Fatty acid profile of the puff fat was determined using Report 2001 method as
protocol based on Gas Chromatography. From this the trans fat content was
estimated.

Minerals - calcium and iron estimation by AAS

Calcium and iron content of puffs was estimated using atomic absorption
spectrophotometer (SPECTRAA - 220 FS from Varian, Inc.).

Caloric value of puffs

Caloric value of puffs was estimated using Atwater factors by multiplying the
proportion of protein, fat and carbohydrate by their respective physiological fuel
values of 4, 9 and 4 kcal/g and taking the sum of all the products. FOS calorific value
was taken as 1.5 kcal/g.

Sensorial analysis of extruded puffs

Sensorial analysis of extruded puffs was performed by a twenty five member

untrained panel consisting of postgraduate students of the Department of Foods and
Nutrition, Lady Irwin College, University of Delhi, New Delhi, India. The panelists,
aged between 18 and 22 years, were regular consumers of extruded puffs. Hedonic
sensory attributes evaluated in this study were colour, texture, flavour, appearance
and overall acceptability (OAA) of extruded puffs. A 9 point hedonic scale anchored
by “dislike extremely” and “like extremely” was employed (Amerine et al. 1965).
Session was conducted at room temperature in a sensory room equipped with white
fluorescent lighting. The panel session was held mid-morning, about 4 h after
breakfast. Puffs were served in bowls to panelists. Samples were provided randomly
to the panelists. Sensory evaluation procedures were explained to the panelists
before testing commenced. Sensory evaluation score cards were prepared and
panelists were asked to read through the instructions and the questions on the
sensory form and the meaning of each attribute was explained to the panelists to
avoid misinterpretation. The panelists were given time to ask for clarification of the
sensory evaluation procedure when uncertain or unclear about the process. Water
at room temperature was provided to rinse the mouth between evaluations. A
panelist score of 7 or higher (like moderately) given to the product was considered
as a sign of acceptability (Chavez-Jauregui et al. 2003). Mean ± SD scores of
panelists for each variant of extruded puffs were estimated.

Statistical analysis

Analysis of variance (ANOVA) with the least significant difference test (LSD-test)
was applied. The level of significance used was 95 % (p  < 0.05). All the statistical
analysis was performed using software SPSS 16.0 for windows version. For the
graphical treatment data was imported into graphics package of MS Excel.

Results and discussion

This paper caters to today’s consumer need for healthy natural ingredients with
higher nutritional-physiological values and the expected taste enjoyment in snacks.
The various sensory qualities attributes of snack foods like appearance, texture,
taste, color and flavour together with compositional profiling were addressed in the
current research.

Compositional profiling of raw material for extruded puffs

The results of the compositional profiling of raw material are presented in Table 2.
The fat content of rice grits was found to be 0.004 % while that of corn grits was
estimated to be 2.1 %. The protein content of rice and corn grits was comparable.
The total dietary fiber content of rice and corn grits was estimate to be 3.3 and 3.8 %
respectively. The fructan content of rice grits (0.35 %) was found to be higher than
that of corn grits (0.15 %). The protein content of green gram pulse was estimated
to be 23.2 %. The total dietary fiber and fructan content were estimated to be 8.4
and 2.0 % respectively. The fat content of finger millet was found to be 1.3 % while
that of sorghum was estimated to be 1.6 %. The protein content of finger millet was
found to be lower than that of sorghum. The values obtained were 7.4 and 10.5 %
respectively. The total dietary fiber content of finger millet and sorghum was
estimated to be 11.2 and 9.9 % respectively. Thus, finger millet was found to contain
higher total dietary fiber than sorghum. The fructan content of sorghum (0.6 %) was
found to be higher than that of finger millet (0.3 %) (Table 2).

Table 2 Compositional profile of the raw material for puffs (%)#

Proximate Rice Corn Green gram Finger

Sorghum
composition grits grits pulse millet
Protein 10.1 9.8 23.2 7.4 10.5
Proximate Rice Corn Green gram Finger
Sorghum
composition grits grits pulse millet
Carbohydrate 73.5 70.9 49.7 64.1 63.0
Fat 0.004 2.1 1.1 1.3 1.6
Total dietary fiber 3.3 3.8 8.4 11.2 9.9
Fructan 0.35 0.15 2.0 0.3 0.6
Ash 0.3 0.4 3.4 2.8 1.8
Moisture 12.5 12.9 12.2 12.9 12.6

#Results are expressed as mean of three determinations

Compositional profiling of extruded puffs

The compositional profiling of the CP and the final MWG fructan enriched puff is
presented in Table 3. The USP of MWG fructan enriched extruded puff is as follows:

 Total dietary fiber including oligofructose is 450 % higher than the control puff.
 Carbohydrate content is 14.1 % lesser than the control puff.
 Calorie content is 9.3 % lesser than the control puff.
 Iron content of multigrain puff is 100 % higher than the control puff.
 Calcium content of multigrain puff is 238 % higher than the control puff.

Table 3 Compositional profile of extruded puffs#

MWG fructan MWG fructan MWG fructan

CP (per
Nutrition facts enriched puff enriched puff (per enriched puff
100 g)
(per 100 g) serving =30 g) (% DV) a
Energy, Kcal 388 352 106
Protein, g 11.4 11 3.3 6.6 %
Carbohydrate, g 76.4 65.6 19.7 5%
Total Fat, g 4 4 1.2 0.8 %
Saturated Fat, g 0.6 0.6 0.2 0.04 %
Trans Fat, g 0 0 0
Monounsaturated
2.3 2.3 0.7
Fat, g
Polyunsaturated
1.1 1.1 0.3
Fat, g
Ca, mg 21 71 21 2.1 %
Fe, mg 1.3 2.6 0.8 4.4 %
Total Dietary Fiber,
1.8 8.2 2.5
g
Insoluble Fiber, g 1.2 6.6 2.0
 MWG fructan MWG fructan MWG fructan
CP (per
Nutrition facts enriched puff enriched puff (per enriched puff
100 g)
(per 100 g) serving =30 g) (% DV) a
Soluble Fiber, g 0.6 1.6 0.5
FOS (Fructan), g 0.8 6.1 1.8
Total Fiber, g 2.6 14.3 4.3 17.2 %

 #Results are expressed as mean ± SD of three determinations

 aPercent DV are based on a 2000 cal diet

As per FDA a food to be labeled as ‘Good Source’ of particular nutrient should

provide 2.5 to 4.9 g/serving i.e. 10 to 19 % of DV and as ‘High Fiber’ or an ‘Excellent
Source’ should provide 5 g and above/serving i.e. 20 % or more of DV. Based on
these criteria MWG fructan enriched puff can be categorized as ‘Good source’ as it
suffice 17.2 % DV of fiber.

Also products to be labeled as ‘Low’ in a particular nutrient should provide 5 % or

less of the DV. The carbohydrate content of MWG fructan enriched puff can be
categorized as ‘Low’ as they will suffice for only 5 % of the DV. However, FDA
doesn’t authorize a ‘Low’ nutrient content claim for carbohydrates. Secondly these
puffs can be categorized as ‘Low’ in fat as these would provide only 1.2 g total fat
per RACC and only 0.2 g saturated fat per RACC. As per FDA foods to be labeled
as ‘Low’ in fat and ‘Low’ in saturated fat should provide 3 g or less of total fat and
1 g or less of saturated fat per RACC respectively.

Sensorial analysis of extruded puffs

Table 4 shows the sensorial score ratings on a nine point hedonic scale of all the
formulations of extruded puffs. The results of analysis revealed that CP (Sample 1)
was rated higher on attributes like color, texture, appearance and overall
acceptability by the panelists. The color and OAA components were rated highest
amongst all with scores of 8.4 ± 0.7 and 8.4 ± 0.5 respectively on a nine point scale.
The ANOVA results indicated that all the components of sensory evaluation were
not significantly affected till 10 % finger millet and 30 % sorghum incorporation level.
Above this level color, flavor, texture and OAA for all MWG puffs were found to be
significantly different (p <  0.05) from the control. The OAA mean score rating for
20 % finger millet and 30 % sorghum inclusion level (Sample 6) was 7.7 ± 0.5 on a
nine point scale. The sample with 30 % finger millet (Sample 7) was rated slightly
below the ‘like moderately’ cutoff of 7, having an OAA score of 6.9 ± 0.7 on a nine
point scale.

Table 4 Sensory evaluation scores (n = 25) of extruded puffs#

Sample† Color Flavor Texture Appearance OAA

1(Control) 8.4 ± 0.7 8.2 ± 0.4 8.3 ± 0.5 8.0 ± 0.7 8.4 ± 0.5
Sample† Color Flavor Texture Appearance OAA
2 8.0 ± 0.7 8.1 ± 0.6 8.0 ± 0.7 7.8 ± 0.6 8.1 ± 0.7
3 8.0 ± 0.7 7.9 ± 0.6 8.0 ± 0.7 7.8 ± 0.6 8.0 ± 0.7
4 7.9 ± 0.8 7.8 ± 0.8 7.8 ± 0.8 7.7 ± 0.7 7.8 ± 0.4
5 7.4 ± 0.7* 7.6 ± 0.5* 7.7 ± 0.7* 7.5 ± 0.7 7.7 ± 0.8*
6 7.4 ± 0.7* 7.6 ± 0.7* 7.7 ± 0.5* 7.5 ± 0.7 7.7 ± 0.5*
7 7.0 ± 0.7* 6.8 ± 0.6* 7.5 ± 0.7* 6.6 ± 0.7* 6.9 ± 0.7*
8 8.1 ± 0.6 8.3 ± 0.7 8.1 ± 0.6 7.5 ± 0.5 8.2 ± 0.6

#Results are expressed as mean ± SD scores on 9 point hedonic scale

* Significantly different from control (p < 0.05)

† Sample 1 (control) contains no finger millet, sorghum or fructan, Samples 2–7

contains finger millet and sorghum in varied proportion and no fructan, Sample 8
contains finger millet, sorghum together with fructan

MWG fructan enriched puff (Sample 8) was rated 8.1 ± 0.6, 8.3 ± 0.7, 8.1 ± 0.6,
7.5 ± 0.5 and 8.2 ± 0.6 for color, flavor, texture, appearance and OAA respectively
(Table 4). The scores for all the attributes were found to be not significantly different
(p  < 0.05) from CP. The OAA score of 8.2 ± 0.6 for MWG fructan enriched puff
(Sample 8) was close to the OAA score of 8.4 ± 0.5 for the CP. The MWG fructan
enriched puffs (Sample 8) were rated higher for flavor than the CP having a score of
8.3 ± 0.7 as opposed to 8.2 ± 0.4 for control puff. This indicates that the acceptability
of multigrain puffs could be improved by the incorporation of fructan (FOS). This
better acceptability of MWG fructan enriched puffs could be attributed to their
improved texture and flavor as FOS has ability to blend with other fibers to create a
bland flavor profile (http://www.orafti.com). Oligofructose contributes towards
improved mouthfeel (Crittenden and Playne 1996), which could have lead to
improved scores for texture of puffs. Also, the use of inulin or oligofructose as a fiber
ingredient has been stated to lead to an improved taste and texture (Franck and
Coussement 1997).

MWG fructan enriched puffs (Sample 8) was rated 8.1 ± 0.6, 8.3 ± 0.7, 8.1 ± 0.6,
7.5 ± 0.5 and 8.2 ± 0.6 for color, flavor, texture, appearance and OAA respectively.
The corresponding MWG puff (Sample 6) had mean scores of 7.4 ± 0.7, 7.6 ± 0.7,
7.7 ± 0.5, 7.5 ± 0.7 and 7.7 ± 0.5 for color, flavor, texture, appearance and OAA
respectively (Table 4). Thus, MWG fructan enriched puff was rated higher for all the
attributes as compared to corresponding MWG puff without FOS. Therefore,
incorporation of FOS led to better acceptability of MWG puffs. Spider plots
visualizing the relative sensory scores of CP, MWG puffs and MWG fructan enriched
puffs are shown in Fig. 2.
Fig. 2 Sensory evaluation scores (n = 25) of control, MWG and MWG fructan
enriched puffs on 9 point hedonic scale

Conclusions

Inclusion of fructan provided improved desirable sensory/organoleptic

characteristics to the limited palatability of wholegrain snacks. The acceptability of
these products was increased further by improving the nutritional contour principally
the dietary fiber content (both soluble and insoluble) since low fiber consumption has
been implicated as a risk factor in many diseases. Thus, such extruded snack foods
can be used as a convenient vehicle to the overall success of a functional food
because of their popularity amongst subjects. This research provides the health
conscious individuals with a healthier snack option over the regular refined grain
snack - A multi-wholegrain, novel functional fiber enriched extruded snack.

Acknowledgments

The cooperation of the human subjects who participated in the study is gratefully
acknowledged. This study was supported by research grant from University Grants
Commission (UGC), New Delhi, India.

References

1. Amerine MA, Pangborn RM, Roseller EB. Principles of sensory evaluation of

foods. Ist. New York: Academic Press; 1965.
2. AOAC (2005) Official Methods of Analysis of Association of Analytical
Communities Published by Association of Analytical Communities, Virginia,
USA, Eighteenth Edition
3. Arai S. Global view on functional foods: Asian perspectives. Brit J Nutr.
2002;88(Suppl 2):S139–S143. doi: 10.1079/BJN2002678. [PubMed] [Cross
Ref]
4. Bouhnik Y, Raskine L, Simoneau G, Vicaut E, Neut C, Flourie B. The capacity
of nondigestible carbohydrates to stimulate fecal bifidobacteria in healthy
humans: a double-blind, randomized, placebo-controlled, parallel-group,
dose-response relation study. Am J Clin Nutr. 2004;80:1658–1664. [PubMed]
5. Chavez-Jauregui R, Cardoso-Santiago RA, Pinto E, Silva MEP, Areas JAG.
Acceptability of snacks produced by the extrusion of amaranth and blends of
chickpea and bovine lung. Int J Food Sci Technol. 2003;38:795–798. doi:
10.1046/j.1365-2621.2003.00734.x. [Cross Ref]
6. Coussement PAA. Inulin and oligofructose: safe intakes and legal status. J
Nutr. 1999;129:S1412–S1417. [PubMed]
7. Crittenden RG, Playne MJ. Production, properties and applications of food-
grade oligosaccharides. Trends Food Sci Technol. 1996;7:353–361. doi:
10.1016/S0924-2244(96)10038-8. [Cross Ref]
8. Franck A, Coussement P. Multi-functional inulin. Food Ingredients Analysis
Inter. 1997;10:8–10.
9. Gibson GR, Beatty ER, Wang CJH. Selective stimulation of bifidobacteria in
the human colon by oligofructose and inulin. Gastroenterology.
1995;108:975–982. doi: 10.1016/0016-5085(95)90192-2. [PubMed] [Cross
Ref]
10. Harper JM. Extrusion of foods, vols. 1 and 2. Boca Raton: CRC Press; 1981.
11. IS 1011 . Indian standard specification for biscuits. New Delhi: Bureau of
Indian Standards; 2002.
12. IS 4333 (Part I and II) (1996) Indian standard specification for method of
analysis for food grains. New Delhi, India: Bureau of Indian Standards
13. Jamora JJ, Rhee KS, Rhee KC. Chemical and sensory properties of
expanded extrudates from pork meat-defatted soy flour blends with onion,
carrot and oat. J Food Sci Nutr. 2002;6:158–162.
14. Launay B, Lisch JM. Twin-screw extrusion cooking of starches: flow
behaviour of starch pastes, expansion and mechanical properties of
extrudates. J Food Eng. 1983;2:259–280. doi: 10.1016/0260-8774(83)90015-
8. [Cross Ref]
15. Linko P, Colonna P, Mercier C. High temperature short time extrusion-
cooking. Adv Cereal Sci Technol. 1981;4:145–235.
16. Pamies BV, Roudaut G, Dacremont C, Meste M, Mitchell JR. Understanding
the texture of low moisture cereal products: mechanical and sensory
measurements of crispness. J Sci Food Agric. 2000;80:1679–1685. doi:
10.1002/1097-0010(20000901)80:11<1679::AID-JSFA697>3.0.CO;2-P.
[Cross Ref]
17. PFA (1955) Prevention of food adulteration act with prevention of food
adulteration rules, 1955 and commodity index with exhaustive short notes.
Delhi, India: International Law Book Company, 2008
18. AACC Report (2001) The definition of dietary fiber. Cereal Food World 46:
112–26
19. Roberfroid MB. Caloric value of inulin and oligofructose. J Nutr.
1999;129:S1436–S1437. [PubMed]
20. Rumessen JJ, Bode S, Hamberg O, Gudmand HE. Frutans of Jerusalem
artichokes: intestinal transport, absorption, fermentation and influence on
 blood glucose, insulin and C-peptide responses in healthy subjects. Am J Clin
Nutr. 1990;52:675–681. [PubMed]
21. Tahnoven R, Hietanen A, Sankelo T, Kortaniemi VM, Laakso P, Kallio H.
Snack foods. Lebensm Unters Forsch. 1998;206:360–363. doi:
10.1007/s002170050273. [Cross Ref]
22. White G. Defining the true meaning of snacks. Food Tech Int Eur.
1994;2:115–117.
23. Wiedman W, Strobel E. Processing and economic advantages of extrusion
cooking in comparison with conventional processing in the food industry. In:
O’Connor C, editor. Extrusion technology for the food industry. New York:
Elsevier Applied science; 1987. pp. 132–169.
24. Wiele T, Boon N, Possemiers S, Jacobs H, Verstraete W. Inulin-type fructans
of longer degree of polymerization exert more pronounced in vitro prebiotic
effect. J Appl Microbiol. 2007;102:452–460. [PubMed]
25. Yamashita K, Kawai K, Itakura M. Effect of fructoligosaccharides on blood
glucose and serum lipids in diabetic subjects. Nutr Res. 1984;4:961–966. doi:
10.1016/S0271-5317(84)80075-5. [Cross Ref]

XXIV. Physicochemical Characterization and Sensory

Analysis of Yeast-leavened and Sourdough Soy
Breads
J Food Sci. 2013 Oct; 78(10): 10.1111/1750-3841.12246.
Gabrielle Yezbick, Jennifer Ahn-Jarvis, Steven J. Schwartz, and Yael Vodovotz

Abstract

Sourdough fermentation has been shown to have numerous beneficial effects on

bread quality, and nutritionally enhance soy-supplemented bread by altering
isoflavone chemical forms. Given this, the objective of this study was to compare the
loaf quality and shelf life of sourdough and yeast-leavened soy breads by various
physical, thermal, and sensorial methods, and to assess the effects of fermentation
by various microorganisms on isoflavone profile in dough and breads using high-
performance liquid chromatography analysis. Sourdough fermentation yielded a less
extensible dough compared to yeast-leavened soy dough (P < 0.001), and resulted
in a harder bread crumb (P < 0.05) and lighter crust color (P < 0.001), compared to
yeast-leavened soy bread (Y-B). Sensory analysis revealed a significantly higher
overall liking of Y-B compared to sourdough soy bread (SD-B) (P < 0.001).
Segmentation analysis of the cohort suggests that overall liking and bread
consumption frequency may be determinants of Y-B or SD-B preference. SD-B and
Y-B exhibited similar shelf-life properties. Despite significantly different enthalpies
associated with the melting of amylose-lipid complexes, thermal analysis of the 2
soy breads stored for 10 d (ambient conditions) demonstrated no significant
difference in water distribution and starch retrogradation (P < 0.05). Lastly, SD-B
was determined to have 32% of total isoflavones occurring in the aglycone form
compared to 17% in Y-B. These findings warrant further investigation of sourdough
fermentation as a processing technique for quality and nutritional enhancement of
soy-based baked goods.

Keywords: bread, fermentation, isoflavone, sourdough, soy

Introduction

Sourdough fermentation is an ancient practice, in which a symbiotic culture of

bacteria and yeast (approximate ratio of 100:1) is used to ferment grain (typically
wheat or rye), resulting in the production of acids and other flavor molecules, as well
as CO2 for leavening (Corsetti and Settanni 2007). Sourdough fermentation,
however, requires more time for CO2 production compared to baker’s yeast
(Saccharomyces cerevisiae), which led to the industrialization of yeast-leavened
bread at the turn of the 20th century and reduced sourdough bread to a niche, artisan
market (Carnevali and others 2007). Experts in the field, however, believe that
sourdough fermentation can impart enhancing effects on loaf quality, such as the
generation of favorable flavor and aroma compounds (Gobbetti and others 2005)
and improvement of gas retention and loaf volume (Gänzle and others 2007).
Additionally, sourdough fermentation may result in shelf-life extension by the
generation of antimicrobial compounds, retarding of starch retrogradation, and the
formation of acids during the fermentation process (Gobbetti and others 2005).

Differences in loaf quality can be quantified by numerous methods. For example,

dough extensibility is frequently utilized to quantify large deformations that correlate
to baking performance, such as gas retention and loaf volume, as well as gluten
composition of doughs and the effect of mixing time on dough extensibility
(Anderssen and others 2004; Abang Zaidel and others 2008). Additionally, texture
and thermal analyses of bread are frequently utilized to study key events of the bread
staling process, such as increasing bread crumb hardness, moisture migration from
bread crumb to crust, and amylopectin recrystallization occurring with storage time
(Davidou and others 1996; Baik and Chinachoti 2000; Vittadini and Vodovotz 2003;
Lodi and Vodovotz 2008).

Sourdough fermentation of alternative flours, such as soy, may offer nutritional

benefits in addition to physical benefits on loaf quality. Soy phytochemicals such as
isoflavones (genistein, daidzein, and glycitein) are hypothesized to mimic the action
of 17-β-estradiol, and their consumption has been correlated with reducing the risk
of various diseases such as hormone-related cancers (Messina and others 2006),
cardiovascular disease, and osteoporosis (Scheiber and others 2001). Soy-
supplemented bread has been shown to be an effective carrier of soy isoflavones
compared to other soy-based foods, delivering as much as 31.6 ± 2.1 mg
isoflavones/slice (Ahn-Jarvis and others 2013).

Some research suggests that conjugated isoflavones must be deglucosylated by the

action of β-glucosidase to their aglycone forms for absorption via passive diffusion
across human enterocytes and, therefore, derivation of their purported health
benefits (Xu and others 1995; Scalbert and Williamson 2000). The fermenting culture
of sourdough typically contains autochthonous lactic acid bacteria (LAB), and
several strains of LAB have been shown to exhibit β-glucosidase activity (Di Cagno
and others 2010). Additionally, fermented soy foods have been shown to contain
significantly greater aglycone forms compared to nonfermented soy foods, which is
generally attributed to the β-glucosidase activity of the fermenting culture (Coward
and others 1993, Pyo and others 2005).

The objectives of this study were to compare the bread quality and shelf life of
sourdough and yeast-leavened soy breads by various physical, thermal, and
sensorial methods, and to assess the isoflavone profile of the fermenting doughs
and final baked products by HPLC analysis.

Materials and Methods

Materials

Ingredients (% wet basis) for yeast-leavened soy bread (Y-B) and sourdough soy
bread (SD-B) formulation are detailed in Table 1. HPLC-grade acetonitrile (ACN),
acetic acid, methanol (MeOH), and water were purchased from Fisher Scientific (Fair
Lawn, N.J., U.S.A.). Dimethyl sulfoxide (ACS spectrophotometric grade) and
isoflavone standards daidzein, genistein, glycitein, daidzin, and glycitin were
purchased from LC laboratories division (PKC Pharmaceuticals, Inc., Woburn,
Mass., U.S.A.) and malonyldaidzin, acetyldaidzin, malonylgenistin, acetylgenistin,
malonylglycitin, acetylglycitin from Wako Chemicals USA, Inc. (Richmond, Va.,
U.S.A.), and glycitein from Indofine (Hillsborough, N.J., U.S.A.).

Table 1 Ingredients and formulations (% wet basis) for Y-B and SD-B.

Formulation (% wet
basis)

Ingredient Manufacturer, location Y-B SD-B

Water – 36.6 26.2
Bread flour (13% Bay State Milling Co., Quincy, Mass., 40.7 30.2
protein) U.S.A.
Soy flour (53% ADM, Decatur, Ill., U.S.A. 12.2 12.3
protein)
Soymilk powder (49% Devansoy, Carroll, Iowa, U.S.A. 4.1 4.1
protein)
Sugar Gordon Food Service, Grand 2.0 2.0
Rapids, Mich., U.S.A.
Shortening The J.M. Smucker Co., Orrville, 2.0 2.0
Ohio, U.S.A.
 Formulation (% wet
basis)

Ingredient Manufacturer, location Y-B SD-B

Wheat gluten Bob’s Red Mill, Milwaukie, Oreg., 1.0 1.0
U.S.A.
Salt U.S. Foodservice, Columbia, Md., 0.8 0.8
U.S.A.
Dough conditioner Caravan Ingredients, Lenexa, Kans., 0.1 0.1
U.S.A.
Yeast Lallemand, Montréal, QC, Canada 0.4 0.0
Preferment – 0.0 21.2

Y-B, yeast-leavened soy bread; SD-B, sourdough soy bread.

Bread production

Y-B was prepared by a sponge-dough method (Vodovotz and Ballard 2009). SD-B
was generated using a modified formulation of Y-B, in which the preferment (dough
yield = 200) inoculated with Lactobacillus brevis, Lactobacillus plantarum, and S.
cerevisiae at 0.1% of the preferment flour weight (Lallemand, Montréal, QC, Canada)
was incorporated at 21.2% of the total dough weight (wet basis). Final proofing of Y-
B was at 40 °C and 95% relative humidity for 1 h (CM2000 combination module,
InterMetro Industries Corp., Wilkes-Barre, Pa., U.S.A.), and 30 °C for 9 h for SD-B
(Isotemp® Oven 200 Series, Model 215F, Fisher Scientific, Fair Lawn, N.J., U.S.A.),
and dough analyses immediately followed. Proofed dough was baked for 50 min at
152 °C (Jet air oven, JA14, Doyon, Linière, QC, Canada), cooled at room
temperature for 3 h, sliced (Doyon SM302 bread slicer, Linière, QC, Canada), and
sealed in polyethylene bags prior to analyses of fresh bread samples. Y-B and SD-
B were stored unsliced in polyethylene bags for 10 d under ambient conditions
(approximately 20 °C) prior to slicing and analyses.

Physical properties

Dough extensibility

Dough extensibility methods were based on work conducted by Frazier and others
(1985). The Instron Universal Testing Machine 5542 (Instron Corp., Norwood, Mass.,
U.S.A.) equipped with a 100N load cell was used to test uniaxial dough extensibility.
Proofed dough was flattened to 5-mm sheets using a Carver press (Fred S. Carver,
Inc., Summit, N.J., U.S.A.), which applied approximately 4000 lbs/in 2 over a 2-h
period. Sample strips (3.0 × 0.5 cm) were placed on the Instron stage, secured by
clamps, and pulled at a crosshead speed of 150 mm/min by a dough hook
attachment. Bluehill® 2 software version 2.17 (Instron Corp., Norwood, Mass.,
U.S.A.) recorded maximum force (N) and extension at maximum load (mm), and the
test was stopped upon dough rupture.

Bread crumb hardness

Bread crumb hardness was determined using the Instron Universal Testing Machine
5542, Bluehill 2 software version 2.17, and a modified AACC method 74-09 (AACC
2000), utilizing a 57-mm dia plunger, as conducted by Crockett and others (2011). A
uniaxial compression with crosshead speed of 100mm/min was applied to 25 × 25 ×
25 mm samples to mimic mastication, with crumb hardness corresponding to the
force (N) required for 40% compression.

Specific loaf volume

The specific loaf volume (cm3/g) was determined using the AACC method 10-05
(rapeseed displacement) and the loaf mass (g) (AACC 2000). Measurements of loaf
mass were taken immediately following a 3-h cooling period.

Crust color

The crust color of SD-B and Y-B was determined using a Chroma Meter CR-300
(Konica Minolta Sensing Americas, Inc., Ramsey, N.J., U.S.A.), which provided
measures of lightness (L), ranging from 0 (black) to 100 (white), and the chromatic
components, a* and b*, which range from green to red and blue to yellow,
respectively, on a scale of −120 to +120.

Sensory analysis

Ohio State Univ. Institutional Review Board exemption was approved (Protocol Nr:
2012E0320) and consumers who had allergies to wheat and/or soy were excluded
from participation. A total of 60 participants were enrolled and 55 adults (≥18 y)
completed all aspects of the sensory evaluation. Although untrained, panelists were
familiar with product sensory evaluations and most had participated in previous
related projects (Wszelaki and others 2005; Kamerund and Delwiche 2007; Ahn-
Jarvis and others 2013). Panelist evaluated overall liking and preference of a single
pair of Y-B and SD-B using a 9-point hedonic scale and paired difference test,
respectively. Y-B and SD-B samples were baked and prepared the day prior to
testing. Bread samples were cut into cubes (1.0 × 1.0 × 0.5 in), included bread crumb
and crust, stored in tightly covered plastic soufflé cups labeled with 3-digit codes,
and counterbalanced and randomized in their presentation. Sensory tests were
conducted in a facility where temperature, airflow, and noise were controlled.
Ambient fluorescent light and incandescent light illuminated each sensory booth.
Overall liking and demographic information was collected using Compusense ® 5.2
software (Compusense, Inc., Guelph, ON, Canada), and preference testing was
conducted using paper ballots.
Thermal analyses

Thermogravimetric analysis (TGA)

Thermogravimetric Analyzer Q5000 (TA Instruments, New Castle, Del., U.S.A.) was
used to evaluate moisture content of Y-B and SD-B bread crumbs. Samples (15 to
20 mg) were placed on platinum pans (PerkinElmer Life and Analytical Sciences,
Inc., Boston, Mass., U.S.A.) and a linear heat ramp of 5 °C/min from approximately
20 °C to 180 °C was applied. Thermograms were analyzed using Universal
Analysis™ 2000 software (TA Instruments) and moisture content of samples was
determined as the difference between initial and final weights of the sample (Fessas
and Schiraldi 2001).

Differential scanning calorimetry (DSC)

Y-B and SD-B bread crumb samples (approximately 10 mg) and references (empty)
were hermetically sealed in stainless steel pans (PerkinElmer Life and Analytical
Sciences, Inc.). DSC Q100 (TA Instruments) was calibrated with indium and purged
with nitrogen prior to loading the cell with food samples. Samples were first cooled
to −50 °C and held isothermally for 3 min before being heated to 150 °C at a linear
rate of 5 °C/min.

Thermograms were generated by plotting heat flow (J/g) against temperature and
were analyzed using Universal Analysis 2000 software (TA Instruments) to
determine the enthalpies (J/g) associated with thermal transitions. Endothermic
peaks occurring below 0 °C were attributed to ice melting (Reid and others 1993;
Vittadini and Vodovotz 2003), allowing for the determination of the percent
“freezable” water (%FW) (Eq. 1). Percent “unfreezable” water (%UFW) was defined
as the difference between the sample moisture content (obtained from TGA) and
%FW. Endothermic peaks occurring between 40 and 60 °C were attributed to
amylopectin melting, and endothermic peaks observed between 100 and 130 °C
were associated with melting of amylose-lipid complexes (Lodi and Vodovotz 2008).

%FW=Enthalpyat0°C(J/g)Latentheatoffusionofwater(333.2J/g)×100
(1)

Isoflavone analyses

Dough formulations for isoflavone analysis

To investigate the effects of sourdough compared with straight-dough techniques on

isoflavone profiles, 4 dough formulations were examined. SD-B variants were soy
sourdough (SD-D), containing S. cerevisiae, Lb. brevis, and Lb. plantarum, and LAB
soy sourdough (LAB-D), containing Lb. brevis and Lb. plantarum. The other 2
straight-dough (Y-B) variants were yeast-leavened soy dough (Y-D), which
contained S. cerevisiae, and control soy dough (C-D), which contained no fermenting
microorganisms.

Isoflavone extraction from bread and dough

Bread crumb (0.5 g) was homogenized in 5 mL of ACN (60% aq, v/v) using a PT
3100 Polytron homogenizer (Kinematica, Inc., Bohemia, N.Y., U.S.A.), whereas in
dough, to adjust for the higher content of water, 1.0 g sample was homogenized with
5 mL of ACN (100%). Homogenized samples were vortexed, sonicated for 15 min
(Mechanical Ultrasonic Cleaner FS30H, 100 watts at 42 kHz output, Fisher Scientific,
Fair Lawn, N.J., U.S.A.), and centrifuged at 3000 rpm for 30 min (IEC HN-SII
Centrifuge, Damon/IEC Div., Needhamhts, Mass., U.S.A.). The extraction was
repeated twice and supernatant pooled for each sample. Extracts (2.0 mL aliquot)
were dried under nitrogen and stored at −25 °C until HPLC analysis. Each sample
was redissolved in 80% MeOH, sonicated, and filtered with 0.2 μm (13 mm dia)
Grace syringe filter (Grace Davison Discovery Sciences, Deerfield, Ill., U.S.A.) prior
to HPLC analysis. Triplicate batches of dough and bread were analyzed in triplicate.

Isoflavone separation and quantification

Bread and dough extracts were analyzed utilizing Waters model 2690 HPLC
equipped with Waters 2996 photodiode array detector, autosampler (10 °C), and
column heater at 30 °C (Waters Assoc., Milford, Mass., U.S.A.). Reversed phase
separation was performed using a 3.0 × 100 mm, 3.5 μm particle, Symmetry C18
column (Waters Assoc.). The secondary mobile phase (1% aqueous acetic acid:
ACN) gradient began at 90:10 progressing linearly to 65:35 in 25 min, 25:75 by 26
min, and 90:10 by 27 min for a total run time of 27 min. Injection volume was 10 μL.

Stock solutions of isoflavone standards were prepared as described by Ahn-Jarvis

and others (2012). In brief, retention times of standards in HPLC chromatograms
and previously published UV-visible spectral signatures were used for identification
of unknowns (Murphy and others 2002). A working mixture was used to generate
calibration curves with correlation coefficients (R2 = 0.998 ± 0.002). HPLC peak area
of each isoflavone was analyzed and quantified using Empower Pro (version 5.0
Waters Assoc.) software.

Statistical analyses

Statistical analyses were performed using Minitab 15 Statistical Software (Minitab

Inc., State College, Pa., U.S.A.) on data collected from sensory evaluation,
physicochemical characterization, and isoflavone profiles from dough fermentation
with various microorganisms and in finished soy breads. Mean ± standard deviation
was reported and reflect replicate samples (3 replicates/batch, 3 batches).
Significant differences (P-value ≤ 0.05) in the physical attributes of the Y-B and SD-
B were discriminated using an independent t-test, whereas a paired t-test was used
to discern significant differences of sensory evaluation tests. A repeated measures
analysis of variance was used to model the data for thermal analysis measures with
respect to storage duration and bread type as well isoflavone chemical forms in
various fermentation strategies in dough and in finished bread. If significant effects
were found, then multiple comparisons were performed using Tukey’s for pairwise
comparisons. Segmentation analysis of soy preference was performed using an
Euclidean, single linkage, hierarchical clustering procedure. Analysis of variance
(ANOVA) was used to evaluate any significant differences among the various
consumer segments.

Results and Discussion

Physical analysis

Dough extensibility was measured to assess differences between sourdough

fermentation (SD-D) and yeast-leavening (Y-D) on the gluten performance of soy-
supplemented wheat dough. Y-D displayed a significantly greater maximum load
and extension at maximum load compared to SD-D (P < 0.001) (Table 2), suggesting
a greater polymer network and, therefore, greater dough strength in Y-D compared
to SD-D (Abang Zaidel and others 2008). These observations are in accordance with
similar studies in which the effects of sourdough fermentation or acidification on
wheat dough extensibility was investigated, with these observations generally being
attributed to microbial proteolysis and the reduction of intra- and intermolecular
disulfide bonds of the gluten macropolymer (Thiele and others 2002; Clarke and
others 2004).

Table 2 Physical properties for Y-D/Y-B and SD-D/SD-B.

Y-D/Y-B SD-D/SD-B

Dough maximum load (N) 0.61 ± 0.11§ 0.41 ± 0.13§

Dough extension at maximum load (mm) 29.71 ± 4.76§ 17.13 ± 2.45§

Dough pH 6.07 ± 0.05§ 4.22 ± 0.14§

Specific loaf volume (cm3/g) 2.3 ± 0.1 2.2 ± 0.1

Bread crumb hardness (N) 13.54 ± 1.60§ 17.04 ± 1.81§

Crust lightness (L) 33.46 ± 1.03§ 36.69 ± 1.65§

Crust chromatic component a* +14.95 ± 0.95 +15.37 ± 1.18

Crust chromatic component b* +16.71 ± 1.79§ +21.88 ± 1.05§

Y-D/Y-B, yeast-leavened soy dough and bread; SD-D/SD-B, sourdough soy dough
and bread.
Mean ± SD reported of triplicate analyses from 3 batches of dough or bread.
§Statistical significance (P ≤ 0.05) determined by independent t-test.

The dough pH of SD-D was significantly less than that of Y-D (P < 0.001) (Table 2),
therefore, the net positive charge of the sourdough system may be partially
responsible for the observed decreased dough extensibility. Furthermore, the denser
crumb of SD-B, as implied by the decreased dough extensibility of SD-D, may have
been responsible for a significantly harder bread crumb of SD-B compared to Y-B (P
< 0.05) (Anderssen and others 2004). The lower pH of SD-D compared to Y-D may
have, additionally, been the causative agent for the lighter crust color (L) observed
for SD-B compared to Y-B (P < 0.001) (Table 2). Maillard browning, the reaction
responsible for the formation of brown pigments of bread crust during baking, is
largely influenced by pH, with an optimum Maillard browning rate occurring at pH 10
and decreasing with increasing acidity (Wolfrom and others 1974).

Sensory analysis

The overall liking of Y-B and SD-B was determined using a 9-point hedonic scale (9
= like extremely; 5 = neither like nor dislike; 1 = dislike extremely), and a paired
difference test was used to determine preference between Y-B and SD-B. Hedonic
scores indicated that participants moderately liked Y-B (6.9 ± 1.3) and neither liked
nor disliked SD-B (4.9 ± 2.0) (P < 0.001). Using a paired preference test, 46 out of
55 subjects with significant probability of less than 0.1% selected preference for Y-
B (Roessler and others 1978). Sensory evaluation of Y-B and SD-B was designed
to assess the overall liking and preference of the 2 soy breads and not sufficiently
powered for segmentation analysis. However, segmentation analysis was performed
and discernible differences between those with Y-B preference compared to those
with SD-B preference were found in our cohort of 55 adults. Those with SD-B
preference consumed bread less than once/wk and their bread variety of preference
was sourdough breads. This behavior was significantly distinct (P < 0.01) from those
who preferred Y-B. Moreover, of those who preferred SD-B, hedonic scores for both
SD-B and Y-B were very distinct (P = 0.005). Hedonic scores for either SD-B or Y-B
was greater than 6 in 89% (8/9) of those who preferred SD-B, whereas of those who
preferred Y-B, hedonic scores greater than 6 was observed in 15% (7/46). Although
our segmentation analysis may be limited in representing analyzable behavior of the
general population, findings warrant the need for a larger sample size of sourdough
bread consumers and suggest that a participant base that equally prefers sourdough
and yeast-leavened bread varieties may provide a more level comparison of these
2 different bread types.

Thermal analysis

TGA and DSC were utilized to determine the moisture content (%) and water
distribution of fresh and stored Y-B and SD-B. Additionally, DSC thermograms of
fresh and stored bread samples were analyzed for transitions associated with bread
crumb staling, including the melting of amylopectin crystals and amylose-lipid
complexes. The moisture properties and thermal transitions of interest were fairly
similar between fresh Y-B and SD-B, and exhibited similar trends upon storage
(Table 3). Exceptions to this generalization were the lower peak temperature
associated with ice melting in fresh SD-B compared to fresh Y-B (P < 0.05), and
lower enthalpy associated with melting of amylose-lipid complexes in fresh and
stored SD-B compared to fresh and stored Y-B (P < 0.05).

Table 3 Moisture and thermal properties for fresh and stored Y-B and SD-B.

Y-B SD-B

Fresh Stored Fresh Stored

Moisture content (%) 42.33 a
± 41.66 ± 0.59 41.99 ± 41.95 ±
0.55a 0.98a 0.52a
Ice melt enthalpy (J/g) 82.45 ± 70.61 ± 4.53b 83.62 ± 75.28 ±
3.82 a 2.12a 4.67b
Ice melt peak temperature −2.57 ± −3.62 ± −3.45 ± −4.14 ±
(°C) 0.37 a 0.53 bc 0.41 b 0.32c

FW (%) 24.74 b
± 21.19 ± 1.36 25.10 ± 22.59 ±
1.15a 0.64a 1.40b
UFW (%) 17.58 ± 20.47 ± 1.88b 16.89 ± 19.36 ±
0.99 a 1.11a 1.71b
Amylopectin melt enthalpy 0.38 ± 0.13a 1.51 ± 0.23b 0.40 ± 0.14a 1.54 ± 0.16b
(J/g)
Amylose-lipid melt enthalpy 1.39 ± 0.15a 1.09 ± 0.12b 0.68 ± 0.20c 0.71 ± 0.07c
(J/g)

Y-B, yeast-leavened soy bread; fresh (day 0) and stored (day 10).

SD-B, sourdough soy bread; fresh (day 0) and stored (day 10).

Mean ± SD reported of triplicate analyses from 3 batches of bread.

Significant difference is indicated by different superscripts, ANOVA and Tukey’s

HSD post hoc test (P ≤ 0.05).

Amylose-lipid complex formation is hypothesized to occur during and/or immediately

after baking (Czuchajowska and Pomeranz 1989), and remains stable with storage
time (Davidou and others 1996; Lodi and Vodovotz 2008). Given that the initial
ingredient composition of SD-D was minimally modified from that of Y-D, the
discrepancy in enthalpies associated with the melting of amylose-lipid complexes of
SD-B and Y-B is likely attributable to the sourdough fermentation process. The
decreased concentration of amylose-lipid complexes in SD-B may have occurred as
a result of amylolytic activity of the LAB present in the sourdough starter culture.
While amylolytic activity of LAB is strain specific, several strains of Lb. plantarum
have been shown to exhibit this activity (Giraud and others 1991; Corsetti and others
1998); therefore, Lb. plantarum present in the sourdough starter culture may have
resulted in a smaller amylose pool available for complexation with lipids.

The complexation of lipid with amylose is believed to result in a decrease in starch

retrogradation; however, this has only been shown to be effective upon extended
storage (Davidou and others 1996). This theory is supported by the similar trends
observed for amylopectin recrystallization occurring over short-term (10 d) storage
in 2 breads (Y-B and SD-B) with significantly different concentrations of amylose-
lipid complexes (Figure 1). Furthermore, amylopectin recrystallization observed for
Y-B and SD-B was much less than that observed for wheat breads not containing
soy following storage (Vittadini and Vodovotz 2003). This observation was previously
noted by Vittadini and Vodovotz (2003), who reported an increase in the enthalpy
associated with the melting of amylopectin crystals from 0.6 to 3.8 W/g after only 7
d of ambient temperature storage in a control wheat bread.

Typical DSC thermograms for fresh and stored SD-B and Y-B displaying the
enthalpy associated with the melting of amylopectin, and the melting of amylose-lipid
complexes in fresh bread (inset).

Figure 1 Typical DSC thermograms for fresh and stored SD-B and Y-B displaying
the enthalpy associated with the melting of amylopectin, and the melting of amylose-
lipid complexes in fresh bread (inset).
Isoflavone analysis in soy bread dough

β-glucosidase is the primary enzyme responsible for the chemical conversion of

isoflavone glycosides to their respective aglycones, which are hypothesized to be
more bioavailable by the body for derivation of their purported health benefits. Given
this, changes in the isoflavone profile of soy bread doughs were examined using
sourdough fermentation and straight dough techniques (Figure 2). β-glucosidase
activity has been observed in previous studies in wheat flour, soy ingredients (soy
flour and soy milk powder), yeast (S. cerevisiae), and LAB (Zhang and others 2004;
Riedl and others 2005). Of the dough types analyzed, the major effects on isoflavone
profile were from fermentation type and duration of fermentation. Among the
chemical forms examined, time and fermentation interaction was observed only in
malonylglucoside forms. LAB-D resulted in the greatest proportional increase in
isoflavone aglycones and correspondingly, the greatest proportional decrease in
isoflavone simple β-glucosides during proofing (Figure 3A). This change in
isoflavone profile may be attributed to the β-glucosidase activity of the particular Lb.
brevis and Lb. plantarum strains present in the preferment. Several strains within
each of these species have been identified as possessing β-glucosidase activity
(Pyo and others 2005; Di Cagno and others 2010), with Lb. plantarum strains
frequently displaying the highest activity of LAB species analyzed (Di Cagno and
others 2010).

Overlay of chromatograms displaying a representative isoflavone profile of soy

sourdough (SD-D) at the beginning (0 h) and end (9 h) of fermentation at 30 °C. The
identities of the labeled peaks are as follows: (1) daidzin, (2) glycitin, (3) genistin, (4)
malonyldaidzin, (5) malonylglycitin, (6) acetyldaidzin, (7) acetylglycitin, (8)
malonylgenistin, (9) daidzein, (10) glycitein, (11) acetylgenistin, (12) genistein.

Figure 2 Overlay of chromatograms displaying a representative isoflavone profile of

soy sourdough (SD-D) at the beginning (0 h) and end (9 h) of fermentation at 30 °C.
The identities of the labeled peaks are as follows: (1) daidzin, (2) glycitin, (3) genistin,
...

Isoflavone profile of 4 chemical forms at start (0 h) and end (9 h) of fermentation.

ANOVA and Tukey’s post hoc test was used to determine significant differences.
Letters represent where differences were observed.

Figure 3 Isoflavone profile of 4 chemical forms at start (0 h) and end (9 h) of

fermentation. ANOVA and Tukey’s post hoc test was used to determine significant
differences. Letters represent where differences were observed.

The conversion of isoflavone conjugates to aglycones was less pronounced in the

remaining dough types (C-D, Y-D, and SD-D) over the 9 h fermentation period
(Figure 3). These findings suggest that the fermentative microorganisms are not the
primary catalysts of the observed conversion, since the dough containing no
fermenting culture (C-D) also displayed substantial increases in isoflavone
aglycones during proofing. Furthermore, the addition of S. cerevisiae to the
preferment containing LAB appears to hinder the conversion of conjugated
isoflavones to aglycones, evidenced by a significantly lower increase in isoflavone
aglycones observed in SD-D compared to LAB-D.

Isoflavone profiles in soy bread

The proportion of aglycones in SD-B was almost double of that in Y-B (Figure 4).
Correspondingly, the simple β-glucoside pool of SD-B was significantly smaller than
that of Y-B (P < 0.001). A similar increase in the aglycone pool of soy bread was
observed upon the addition of almond powder (5% w/w), which is a naturally rich
source of β-glucosidase (Vodovotz 2007). The deglucosylation of isoflavones by
sourdough fermentation, however, is advantageous to the baking industry, as this
technique does not require the addition of an allergen.

Isoflavone profiles (aglycones and glucosides expressed as percent of total

isoflavone concentration) and total isoflavone concentration for Y-B and SD-B. Mean
± SD of triplicate analyses of 3 batches of soy breads. ANOVA and Tukey’s post hoc
test was used to discern significantly different (P ≤ 0.05) values and are annotated
by differing letters.

Figure 4 Isoflavone profiles (aglycones and glucosides expressed as percent of total

isoflavone concentration) and total isoflavone concentration for Y-B and SD-B. Mean
± SD of triplicate analyses of 3 batches of soy breads. ANOVA and Tukey’s ...

The proportion of isoflavones reported as aglycones in SD-B was significantly lower

than that of SD-D at T9 of fermentation, which is likely attributable to ongoing
conversion of isoflavones by β-glucosidase during the lengthy (approximately 4 h)
extraction process within the dough matrix. Moreover, the proportions of acetyl- and
malonyl-glucosides varied between bread types (P < 0.001), with the malonyl-
glucosides representing one of the largest pools of isoflavones in both Y-B and SD-
B. Notably, the isoflavone profiles of Y-B and SD-B did not change after 10 d of room
temperature storage (data not shown). This finding suggests that the isoflavone
profile within a dough system is stabilized once baked and sustained over short-term
storage.

Conclusion

The physical, thermal, and chemical properties of sourdough and yeast-leavened

soy breads, as well as overall liking and preference between these products, were
evaluated to determine the effect of sourdough fermentation on the overall loaf
quality of a soy-supplemented wheat bread. Physical analyses displayed several
significant differences between bread types, including a less extensible dough,
harder bread crumb, and lighter crust color of SD-B compared to Y-B, all of which
may be attributable to a significantly lower pH of SD-D compared to Y-D. Sensory
analysis revealed a higher overall liking and preference for Y-B compared to SD-B,
however, this may have been a reflection of overall participant preference for yeast-
leavened bread varieties. Thermal analyses of Y-B and SD-B displayed similar
moisture and starch retrogradation patterns after 10 d of room temperature storage,
despite significantly different enthalpies associated with the melting of amylose-lipid
complexes. HPLC analysis of SD-B, however, displayed almost double the
proportion of isoflavones occurring in the aglycone form compared to Y-B (32% and
17%, respectively). These findings warrant further investigation of sourdough
fermentation as a processing technique for enhancement of isoflavone aglycones in
soy-based baked goods.

Acknowledgments

This research was made possible by funding from the Center for Innovative Food
Technology (CIFT) and use of Compusense® sensory software.

Notes

This paper was supported by the following grant(s):

National Cancer Institute : NCI P30 CA016058 || CA.

References

1. AACC. Approved methods of the American Association of Cereal Chemists.

10. St. Paul, Minn: American Association of Cereal Chemists; 2000. p. 1200.
2. Abang Zaidel DN, Chin NL, Abdul Rahman R, Karim R. Rheological
characterization of gluten from extensibility measurement. J Food Eng.
2008;86:549–56.
3. Ahn-Jarvis JH, Clinton SK, Riedl KM, Vodovotz Y, Schwartz SJ. Impact of
food matrix on isoflavone metabolism and cardiovascular biomarkers in adults
with hypercholesterolemia. Food Funct. 2012;3:1051–8. [PMC free article]
[PubMed]
4. Ahn-Jarvis JH, Riedl KM, Schwartz SJ, Vodovotz Y. Design and selection of
soy breads used for evaluating isoflavone bioavailability in clinical trials. J Agr
Food Chem. 2013;61:3111–20. [PMC free article] [PubMed]
5. Anderssen RS, Bekes F, Gras PW, Nikolov A, Wood JT. Wheat-flour dough
extensibility as a discriminator for wheat varieties. J Cereal Sci. 2004;39:195–
203.
6. Baik M, Chinachoti P. Moisture redistribution and phase transitions during
bread staling. Cereal Chem. 2000;77(4):484–8.
7. Carnevali P, Ciati R, Leporati A, Paese M. Liquid sourdough fermentation:
Industrial application perspectives. Food Microbiol. 2007;24(2):150–4.
[PubMed]
8. Clarke CI, Schober TJ, Dockery P, O’Sullivan K, Arendt EK. Wheat sourdough
fermentation: effects of time and acidification on fundamental rheological
properties. Cereal Chem. 2004;81(3):409–17.
9. Corsetti A, Gobbetti M, Balestrieri F, Paoletti F, Russi L, Rossi J. Sourdough
lactic acid bacteria effects bread firmness and staling. J Food Sci.
1998;63(2):347–51.
10. Corsetti A, Settanni L. Lactobacilli in sourdough fermentation. Food Res Int.
2007;40:539–58.
11. Coward L, Barnes N, Setchell K, Barnes S. Genistein, daidzein, and their β-
glycoside conjugates: antitumor isoflavones in soybean foods from American
and Asian Diets. J Agr Food Chem. 1993;41(11):1961–7.
12. Crockett R, Ie P, Vodovotz Y. Effects of soy protein isolate and egg white
solids on the physicochemical properties of gluten-free bread. Food Chem.
2011;129:84–91.
13. Czuchajowska Z, Pomeranz Y. Differential scanning calorimetry, water
activity, and moisture contents in crumb center and near crust zones of bread
during storage. Cereal Chem. 1989;66:305–9.
14. Davidou S, Meste ML, Debever E, Bekaert D. A contribution to the study of
staling of white bread: effect of water and hydrocolloid. Food Hydrocolloid.
1996;10(4):375–83.
15. Di Cagno R, Mazzacane F, Rizzello CG, Vincentini O, Silano M, Giuliani G,
De Angelis M, Gobbetti M. Synthesis of isoflavone aglycones and equol in
soy milks fermented by food-related lactic acid bacteria and their effect on
human intestinal Caco-2 cells. J Agr Food Chem. 2010;58:10338–46.
[PubMed]
16. Fessas D, Schiraldi A. Water properties in wheat flour dough I: classical
thermogravimetry approach. Food Chem. 2001;72:237–44.
17. Frazier PJ, Fitchett CS, Russell Eggitt PW. Laboratory measurement of dough
development. In: Faridi H, editor. Rheology of wheat products. St. Paul, Minn:
American Association of Cereal Chemists; 1985. pp. 151–175.
18. Gänzle MG, Vermeulen N, Vogel RF. Carbohydrate, peptide and lipid
metabolism of lactic acid bacteria in sourdough. Food Microbiol.
2007;24:128–38. [PubMed]
19. Giraud E, Brauman A, Keleke S, Lelong B, Raimbault M. Isolation and
physiological study of an amylolytic strain of Lactobacillus plantarum. Appl
Microbiol Biot. 1991;36:379–83.
20. Gobbetti M, De Angelis M, Corsetti A, Di Cagno R. Biochemistry and
physiology of sourdough lactic acid bacteria. Trends Food Sci Tech.
2005;16:57–69.
21. Kamerud JK, Delwiche JF. Individual differences in perceived bitterness
predict liking of sweeteners. Chem Senses. 2007;32(9):803–10. [PubMed]
22. Lodi A, Vodovotz Y. Physical properties and water state changes during
storage in soy bread with and without almond. Food Chem. 2008;110:554–
61.
23. Messina M, Kucuk O, Lampe JW. An overview of the health effects of
isoflavones with an emphasis on prostate cancer risk and prostate-specific
antigen levels. J AOAC Int. 2006;89(4):1121–34. [PubMed]
24. Murphy PA, Barua K, Hauck CC. Solvent extraction selection in the
determination of isoflavones in soy foods. J Chromatogr B. 2002;777:129–38.
[PubMed]
25. Pyo Y, Lee T, Lee Y. Enrichment of bioactive isoflavones in soymilk fermented
with β-glucosidase-producing lactic acid bacteria. Food Res Int.
2005;38:551–9.
26. Reid DS, Hsu J, Kerr W. Calorimetry. In: Blanshard JMV, Lillford PJ, editors.
The glassy state in foods. Nottingham, U.K: Nottingham Univ. Press; 1993.
pp. 123–32.
27. Riedl KM, Zhang YC, Schwartz SJ, Vodovotz Y. Optimizing dough proofing
conditions to enhance isoflavone aglycones in soy bread. J Agr Food Chem.
2005;53:8253–8. [PubMed]
28. Roessler EB, Pangborn RM, Sidel JL, Stone H. Expanded statistical tables
for estimating significance in paired-preference, paired-difference, duo-trio
and triangle tests. J Food Sci. 1978;43:940–7.
29. Scalbert A, Williamson G. Dietary intake and bioavailability of polyphenols. J
Nutr. 2000;130:2073S–85S. [PubMed]
30. Scheiber MD, Liu JH, Subbiah MTR, Rebar RW, Setchell KD. Dietary
inclusion of whole soy foods results in significant reductions in clinical risk
factors for osteoporosis and cardiovascular disease in normal
postmenopausal women. Menopause. 2001;8:384–92. [PubMed]
31. Thiele C, Gänzle MG, Vogel RF. Contribution of sourdough lactobacilli, yeast,
and cereal enzymes to the generation of amino acids in dough relevant for
bread flavor. Cereal Chem. 2002;79(1):45–51.
32. Vittadini E, Vodovotz Y. Changes in the physicochemical properties of wheat-
and soy-containing breads during storage as studied by thermal analyses. J
Food Sci. 2003;68:2022–7.
33. Vodovotz Y. Soy enriched bread. In: Hamaker B, editor. Technology of
functional cereal products. Cambridge, U.K: Woodhead Publishing Ltd; 2007.
pp. 388–408.
34. Vodovotz Y, Ballard C, inventors. The Ohio State Univ. Research Foundation,
assignee. Compositions and processes for making high soy protein-
containing bakery products. 7,592,028. US Patent. 2009 Sep 22;:B2.
35. Wolfrom ML, Kashimura N, Horton D. Factors affecting the Maillard browning
reaction between sugars and amino acids. Studies on the nonenzymatic
browning of dehydrated orange juice. J Agr Food Chem. 1974;22(5):796–800.
36. Wszelaki AL, Delwiche JF, Walker SD, Liggett RE, Miller SA, Kleinhenz MD.
Consumer liking and descriptive analysis of six varieties of organically grown
edamame-type soybean. Food Qual Prefer. 2005;16(8):651–8.
37. Xu X, Harris KS, Wang H, Murphy PA, Hendrich S. Bioavailability of soybean
isoflavones depends upon gut microflora in women. J Nutr. 1995;125:2307–
15. [PubMed]
38. Zhang YC, Lee JH, Vodovotz Y, Schwartz SJ. Changes in distribution of
isoflavones and β-glucosidase activity during soy bread proofing and baking.
Cereal Chem. 2004;81(6):741–5.
XXV. World Market Development and Consumer
Acceptance of Irradiation Technology
Foods. 2016 Dec; 5(4): 79.
Behnoush Maherani, Farah Hossain, Paula Criado, Yosra Ben-Fadhel, Stephane
Salmieri, and Monique Lacroix*
J. Scott Smith, Academic Editor

Abstract

Food irradiation is an efficient technology that can be used to ensure food safety by
eliminating insects and pathogens to prolong the shelf life. The process could be
applied to fresh or frozen products without affecting the nutritional value. Presently
more than 60 countries have adopted the technology. However, the technology
adaptation differs from one country to another and, in some cases, consumers’
misunderstanding and lack of acceptance may hinder the technology adaptation
process. This review summarizes the development of irradiation treatment
worldwide and consumer attitudes towards the introduction of this technology. Also,
the wholesomeness, beneficial effects, and regulation of irradiation are assessed.

Keywords: irradiated foods, food quality, nutritional value, bacterial radiosensitivity,

consumer acceptance, consumer response, consumer communication, global
perception

1. Consumers’ Acceptance of Novel Agri-Food Technologies

In recent decades novel technologies have emerged worldwide in food production,

processing, and preservation. These technical innovations are in development as a
result of modern demands for foods that are fresher, have higher nutritional value,
and are more natural with minimum food additives and no toxins or allergens [1,2].
As a result of these emerging technologies, higher-quality foods are produced with
safer attributes since they have an extended shelf life and are sold at a reasonable
cost. According to Rollin et al. [3], the use of novel foods or novel food ingredients in
Europe and their marketing within the European Community was first defined by
Regulation (EC) No. 258/97. In this legislation, novel foods or food ingredients are
defined as those containing or produced from genetically modified organisms
(GMOs), with a new modified primary molecular structure, consisting of or isolated
from microorganisms, fungi or algae, plants, or animals not obtained by traditional
propagation or breeding practices, and having a history of safe use. They also
encompass processes that give rise to significant changes in composition or
structure of a food or its ingredients. Research in this field is expanding, with food
safety being a major driver for the development of new food technologies in order to
reduce, control, and eliminate foodborne pathogens.

These novel technologies have the ability to enhance the nutritional values of food,
lower the carbon footprint in food production, and reduce water consumption in food
production lines. According to Agriculture and Agri-Food Canada [4], new processing
and production technologies are being applied in the form of extraction methods and
ingredient processing such as the development of water-soluble lipids for addition to
foods and beverages, the application of pulsed ultraviolet light to improve the
nutritional content of mushrooms, the conducting of genetic manipulation, stem cell
applications, and cloning in breeding techniques, and the enhancement of delivery
systems for bioactive ingredients through the use of nanoemulsions or
microencapsulated probiotic cultures or organogels to preserve and extend food
shelf life. Some of the novel food technologies highlighted by Frewer et al. [5] include
genetic modification, animal cloning, nutrigenomics, nanotechnology, high-pressure
processing (HPP), and pulsed electric field processing (PEF). These technical
innovations are revolutionizing the food industry and impart a more competitive edge
to agri-business.

The human diet has drastically changed over the years since the Industrial
Revolution due to changes in agricultural production and animal breeding methods
[6]. With the diversification of agricultural food practices and improvement in
transport, farm products have since then moved easily between cities and across
continental borders. Moreover, some means of conservation technologies such as
refrigeration were welcomed with enthusiasm by consumers. However, an
increasing body of literature suggests that consumers tend to oppose newly
developed food technologies such as genetic modification of crops, which has stirred
much controversy among consumers [6,7]. According to Slovic et al. [8], general
consumers often evaluate the risks of novel technologies differently from experts.
Erdem et al. [9] are also of the same view, and opined that the general public often
diverges from experts in the evaluation of food processed by new technologies.
Consumers are mostly of the view that risks stem from farming practices and
processing, while farmers, on the other hand, believe that the greatest food safety
risks occur as a result of consumer and processor actions. The lack of knowledge
on the part of consumers and poor communication from the farmers and food
processing technologists or engineers increases the misunderstanding between
these two groups. To exemplify this aspect, a joint meeting in 1997 involving the
World Health Organization (WHO), the Food and Agricultural Organization (FAO),
and the International Atomic Energy Agency (IAEA) clearly determined that food
irradiated with an appropriate dose to achieve the intended objective was both safe
to consume and nutritionally adequate [10]. In fact, irradiation technology has proved
to be efficient in reducing bacterial contamination to produce sterile food, which is
particularly important for patients with impaired immunity such as those suffering
from AIDS and cancer [3]. However, research has shown that the public tends to be
averse to irradiated food despite the fact that it has been recognized as safe by
authorities [11]. A number of factors may explain this reluctance, one of which is the
lack of proper knowledge about the technology employed to process the food.
According to Frewer et al. [5], public awareness is not high about food irradiation,
and many people do not even know what irradiation is. A recent study by Nayga et
al. [12] showed that awareness about the nature and benefits of food irradiation led
to positive changes in consumers’ perception and influenced their decisions to buy
irradiated food.

Novel agri-food technologies has led to the development of food preferences and
food neophilia (individuals being willing to try new foods) versus neophobia
(individuals being reluctant to try novel foods) [13]. Based on these concepts, Pliner
and Hobden [14] even introduced a scale to measure food neophobia. Cox and
Evans [15] modified this approach to establish a measure of food technology
neophobia. This scale gauges the fear and reluctance of consumers to eat foods
produced by innovative technologies. In a study conducted by Capiola and
Raudenbush [16], it was found that food neophilics and food neophobics tend to
exhibit different sensory evaluations, psychophysical ratings, stimulus sampling,
physiological responses, and genetic predispositions.

2. Factors Driving Consumer Response to Emerging Technologies

For years investigators have conducted research in view of assessing consumer

responses to novel food technologies, and examined a variety of factors that may
influence the perception of consumers towards these emerging technologies. In this
context, the risks associated with these responses have been evaluated based on
different scenarios such as voluntary and involuntary, immediate or delayed,
observable or unseen, fatal or non-fatal, the degree to which the risk is known or not,
and the degree of control that the consumers have over the risk [11,17]. A range of
foods processed by different technologies such as irradiated food, genetically
modified food, food treated by pulsed electric fields and ultraviolet laser, and
microbially contaminated foods have been under scrutiny to assess their perceived
risks and consumers’ concerns [18,19,20]. Based on a study conducted by Cardello
et al. [21], 20 different traditional and novel food processing technologies were
evaluated for their acceptance by consumers. Genetic food manipulation was at the
top of the list and elicited the highest level of concern from consumers, followed by
the addition of bacteriocins, irradiation, and pulsed X-rays. Consumers displayed
less concern about technologies such as UV light, pulsed electric fields, and
oscillating magnetic fields. The perception of food technologies by consumers in turn
plays a crucial role in their choices, purchasing behavior, and acceptance of these
foods. From an economic point of view, it is essential to optimize the sensory quality
of food products so as to foster their consumption by the public. According to Bruhn
[22], good flavor or unique flavor combinations of food products largely determine
their success with consumers. Such intrinsic features also include the nutritional
value of the food products such as fiber, beneficial fatty acids, lycopene, vitamin C,
and probiotic, among others [23]. An increasing number of consumers are opting for
more “natural” food products with no food additives or those that have been
produced in environmentally friendly or sustainable ways. In addition, food
convenience also plays an important role, such that over 80% of consumers, in a
study, indicated that their purchase was heavily dependent on convenience [24].

However, relying only on the sensorial quality of foods does not guarantee their
success in the marketplace since there are other factors that contribute to their
acceptance by consumers. According to Cardello [21], contextual, cognitive, social,
cultural, and attitudinal attributes also drive consumers’ food choices. With regards
to novel food technologies, consumers show concern about the nature of the
resulting processed food or the nature of the processing technology itself, and these
play a crucial role in determining whether consumers will buy the food or not. Studies
have shown that a lack of knowledge among consumers regarding novel food
processing technologies is a major impediment to their acceptance [25]. Hence,
consumer communication is essential for consumers to accept innovative food
technologies. To some extent, the dread/control framework may explain the aversion
of consumers to new food technologies. Many consumers have little knowledge
about modern production agriculture; according to Campbell and Fitzgerald [26], the
new technologies applied in food processing are foreign to contemporary
consumers, and the low literacy of consumers often limits their acceptance.

According to Li-Cohen and Bruhn [27], information about the processing of food
should be presented to consumers through different routes depending on age and
gender. Based on this study, it was found that men and younger consumers tend to
prefer web-based sources than women or middle to older people, who rely more on
television, newspapers, magazines, and supermarket brochures. In addition,
consumers appear to be cautious about accepting novel technologies applied to food
based on the perceived risks and lack of benefits. A study conducted by Cox et al.
[28] showed that males, on average, display less concern about these emerging
technologies but give more importance to the cost and size of the food products. On
the other hand, females show more negative beliefs about innovative food
technologies.

3. Overcoming Aversion of Novel Foods: Communication

In an effort to mitigate the negative perception of the impact of foods derived from
novel technologies and inculcate trust among consumers, food industry researchers
are focusing their attention on public education and dissemination of information
regarding these food products [29,30,31]. Educational programs should be set up to
impart the right information to the public regarding food produced from novel
technologies. Very often, the future benefits of the technology are praised rather than
the immediate, direct consumer benefits. According to Bruhn [22], emphasis should
be laid on immediate consumer benefits, and the information should be imparted to
them using layman-suitable terminology. In addition, to maintain the trust of the
public, it is imperative to be transparent and share both what is known and what is
not known with regard to the risks and benefits. The food industry should also
anticipate and respond in a timely manner to concerns about potential risks
expressed by the public. According to Costa-Font et al. [32], proper labeling can
effectively provide information about the technology employed and its benefits to
raise awareness and enhance transparency. Curtis et al. [33] and Frewer et al. [34]
prefer food products to display clear and detailed labels. Consumers tend to accept
the associated risks if they are aware of them and have control of them. Studies
have shown that if these approaches are implemented, the aversion of consumers
towards novel products decreases and their acceptance increases in general
[35,36]. Educational and other information-based approaches to changing consumer
attitudes appear to work in most cases. However, studies performed by Grunert et
al. [19] and Wilson et al. [37] on consumer responses to genetic modification of food
have shown that addressing the information deficit does not completely reassure
consumers. Nonetheless, it has not been shown that the same situation applies to
other emerging food processing technologies such as food irradiation. In the latter
case, Nayga et al. [12] showed that educating consumers about the nature and
benefits of food irradiation may effectively induce a positive response and improve
its acceptance.

4. Food Irradiation as a Safe Technology

Food irradiation is a processing technique that involves exposing food to ionizing

radiation such as electron beams, X-rays, or gamma radiation to induce the demise
of bacteria that can cause food poisoning, control insect infestation, delay fruit
ripening, or prevent vegetables from sprouting [38,39]. Studies have shown that this
technology can prevent the proliferation of microorganisms that cause food spoilage,
such as bacteria and molds, by changing their molecular structure [40]. Also
commonly known as “cold pasteurization,” it offers a wide range of benefits to the
food industry and the consumer by ensuring the hygienic quality of solid or semi-
solid foods through inactivation of foodborne pathogens [41]. Interest in irradiation
food technologies is increasing because of persistently high food losses from
infestation, contamination, and spoilage by bacteria and fungi, rising concern about
foodborne diseases, and a growing international trade in food products that must
meet strict import standards of quality and quarantine. In all these areas, food
irradiation has demonstrated valuable and practical benefits when integrated within
an established system for the safe handling and distribution of food products [42]. In
addition, with increasingly restrictive regulations or complete prohibition on the use
of a number of chemical fumigants for insect and microbial control in the food
industry, irradiation is becoming a preferred alternative to protecting food against
insect damage and as a quarantine treatment for fresh produce [43,44]. As such,
irradiation can help to ensure a safer and more plentiful food supply by extending
food shelf life through the control of pests and pathogens. Importantly, according to
the World Health Organization (WHO) and Food and Agriculture Organization (FAO)
[10,39], it is a safe technology for the processing of food commodities when the
appropriate radiation dose is respected.
5. Wholesomeness of Irradiated Food

Many studies have been conducted over the last 100 years on the ‘wholesomeness
of irradiated foods’, a terminology developed during those efforts [45]. As mentioned
before, irradiation is a non-thermal process utilized to achieve the preservation of
food. Irradiation (even for radurization at 0.4–10 kGy and radicidation at 40–45 kGy)
does not impart heat to the food and the nutritional quality of the food is generally
unaffected [46]. The irradiation process can reduce microbial contamination in food,
resulting in improved microbial safety as well as extended shelf life of the food. There
is an established framework of international standards for food irradiation covering
human health, plant protection, labeling, dose delivery, quality assurance, and
facility management. Approximately 60 countries permit irradiation of one or more
foods or food classes [47].

The Codex Alimentarius Commission (Codex) is the body responsible for standards
related to human health. Food irradiation must be conducted according to good
management practice and comply with the Codex Alimentarius General Principles
of Food Hygiene [48]. The foundation for food irradiation was set with the adoption
of the Codex World-wide General Standard for Irradiated Foods in 1983 [49] and a
significant revision in 2003 [50]. The General Standard states that the minimum
absorbed dose should be sufficient to achieve the technological purpose and the
maximum absorbed dose should be less than that which would compromise
consumer safety of wholesomeness or would adversely affect the structural and
functional properties or nutritional and sensory attributes [50]. In 1983, the Codex
Alimentarius Commission accepted that foods irradiated up to 10 kGy were safe and
therefore toxicological testing was no longer necessary. In 1997, the United Nations
confirmed that foods could be treated at any dose without any detrimental effect on
the food’s wholesomeness. The study group concluded that high-dose irradiation,
conducted in accordance with good manufacturing and irradiation practices, could
be applied to several types of foods to improve their hygienic quality, make them
shelf stable, and produce special products [51] (Table 1).

Table 1 Foods permitted to be irradiated under FDA regulations (21 CFR

179.26). Data were updated by Komolprasert [46].

Food Purpose Dose

Fresh, non-heated processed 0.3 kGy min. to

Control of Trichinella spiralis
pork 1 kGy max.

Fresh foods Growth and maturation inhibition 1 kGy max.

Foods Arthropod disinfection 1 kGy max.

Dry or dehydrated enzyme

Microbial disinfection 10 kGy max.
preparations
Food Purpose Dose

Dry or dehydrated
Microbial disinfection 30 kGy max.
spices/seasonings

Fresh or frozen, uncooked

Pathogen control 3 kGy max.
poultry products

Frozen packaged meats

Sterilization 44 kGy min.
(solely NASA)

Refrigerated, uncooked meat

Pathogen control 4.5 kGy max.
products

Frozen uncooked meat

Pathogen control 7 kGy max.
products

Fresh shell eggs Control of Salmonella 3.0 kGy max.

Seeds for sprouting Control of microbial pathogens 8.0 kGy max.

Fresh or frozen molluscan Control of Vibrio species and other

5.5 kGy max.
shellfish 1 foodborne pathogens

1 Data provided by FDA [51].

The FDA considers four broad areas to establish the safety of irradiated foods:
radiological safety, toxicological safety, microbiological safety, and nutritional
adequacy. No evidence of toxicity and radioactivity attributable to irradiation of food
was found. Furthermore, under realistic conditions, radiation has been approved to
achieve the intended microbiological effect by eliminating the Clostridium botulinum
and its toxin as the most resistant bacterium and will not increase the microbiological
risk [52,53,54].

5.1. Nutritional Aspects

Food irradiation is a technology that addresses both food quality and safety because
of its ability to: inactivate the parasites, spoilage, and food-borne pathogenic
microorganisms; and, under certain conditions, deactivate viruses, delay the
ripening of fruits, inhibit germination (e.g., onion, garlic), and control the post-harvest
losses caused by insect infestation without significantly affecting the sensory or other
organoleptic attributes of food, thus contributing to improvements in food hygiene
and enhancing public health [40,55,56]. Furthermore, irradiation treatment can
stimulate the biosynthesis of bioactive compounds [38]. In some conditions,
irradiation can also activate the synthesis of phenolic compounds and enhance the
vitamin content in fruits and vegetables [38]. Recent studies showed that radiation
treatment generally increased the levels of certain beneficial phytochemicals and
enhanced the biological properties of some plants with nutritional value [57]. Indeed,
the addition of any energy to food can break down its nutrients. Foods are irradiated
to provide the same benefits (such as destroying pathogenic bacteria) as when they
are processed by other technologies such as heat, refrigeration, freezing, or
chemical treatment. Indeed, non-thermal food treatments have no potentially harmful
residues and, also in these techniques, nutrient losses are relatively small and often
substantially less than the nutrient losses associated with other methods of
preservation, such as canning, drying, and heat pasteurization and sterilization
[56,58]. Furthermore, it should be noted that the main advantage of food irradiation
is that it can be used to treat packaged foods, which will remain safe and protected
from microbial contamination after treatment [59]. From a nutritional point of view,
trace elements and minerals are not affected by irradiation. Macronutrients such as
protein, carbohydrates, and fats are not significantly affected by doses up to 50 kGy
[52,55]. Saturated and mono-saturated fatty acids represent the essential content of
neutral lipids in meat. Different studies on meat irradiation and its effect on lipids
have shown that at low radiation doses, lipids in the presence of their natural
protectors are not particularly sensitive to radiation-induced peroxidation. They also
found no significant difference in total saturated and unsaturated fatty acids between
irradiated (1, 3, or 6 kGy) and un-irradiated frozen chicken muscle [60,61]. Proteins
are built of amino acids, which are the essential nutrients for the body. The effect of
radiation on protein is related to their state, structure, and composition, whether
native or denatured, whether dry or in solution, whether liquid or frozen, and to the
presence or absence of other substances. However, long-term feeding studies also
concluded that irradiation of raw and prepared meat, including precooked shrimp
and chicken, to prolong shelf life, does not lead to a reduction in their protein
nutritional value and no distinct decrease of the biological value of proteins was
observed [62,63,64]. The amount of vitamin loss due to food irradiation is affected
by several factors, including doses, temperature, presence of oxygen, and food type.
Generally, radiation at low temperatures in the absence of oxygen reduces any
vitamin loss in foods, and the storage of irradiated foods in sealed packages at low
temperatures also helps prevent future vitamin loss. However, not all vitamins have
the same sensitivity to irradiation [65] (Table 2).

Table 2 Relative sensitivity of vitamins to irradiation.

High Sensitivity Low Sensitivity

Vitamin C * Carotene

Vitamin B1 (thiamin) * Vitamin D

Vitamin E Vitamin K

Vitamin A Vitamin B6 (pyridoxine) *

High Sensitivity Low Sensitivity

Vitamin B2 (riboflavin) *

Vitamin B12 (cobolamin) *

Vitamin B3 (niacin) *

Vitamin B9 (folate) *

Pantothenic acid *

* Water-soluble vitamins, Fat-soluble vitamin. Updated from Woodside 2015 [55].

Most of the studies confirmed that irradiated foods are generally nutritionally
equivalent or even better than non-irradiated foods that are subjected to normal
processing [57,64,66,67]. Finally, according to a collective agreement between the
Food and Agriculture Organization (FAO) of the United Nations, the International
Atomic Energy Agency (IAEA), and the World Health Organization (WHO), on the
basis of knowledge derived from over 50 years of research, irradiated foods were
considered safe and wholesome at the specified radiation dose [68]. A joint
FAO/IAEA/WHO Study Group on High-Dose Irradiation (JSGHDI) stated that any
food treated at any high dose is acceptable and healthy as long as it is palatable.
This statement acknowledges that any food destroyed by inappropriate irradiation
treatment may have lost its essential properties but is not necessarily hazardous for
consumption [45,69].

5.2. Sensory Aspects

The consumer attitude towards food is very complex as it is influenced by sensory

and non-sensory attributes, as well as by the interactions between them. Recently,
many studies on sensory acceptance of radiated foods and the influence of food
irradiation on consumer behavior have been performed [70]. The findings confirmed
that food irradiation is a technology that addresses both food quality and safety
because of its ability to control spoilage and food-borne pathogenic microorganisms
without significantly affecting the sensory attributes or other organoleptic attributes
of the food [56,71]. Many studies found no significant difference in sensory quality
and protein content of stir fry chicken dices and ground meat after irradiation during
the storage time [71,72].

The irradiation of vegetables, nuts, green tea, grains, and fresh and dried fruits (such
as spinach leaves, carrots, lettuce, broccoli, red kidney beans, raisins, pistachios,
dried figs, apricots, apples, and pears, and fresh strawberries, pineapples,
clementines, and mangoes) was also shown to lead to good sensory and
organoleptic quality acceptance [40,66,67,70,73,74]. Furthermore, in some cases,
radiation processing also leads to an increase in the nutritional values of irradiated
fruits and vegetables, such as vitamin C content and phenolic compounds
[38,75,76].

5.3. New Perspectives

Irradiation combined with other processes can contribute to food safety, improve the
nutritional value of products, and control losses during transportation and
commercialization [77,78,79]. Many studies have demonstrated that depending on
various added compounds such as essential oils (EOs) extracted from plants and
the combined treatment used (e.g., modified atmosphere packaging (MAP), mild
heat treatment), the relative bacterial radiosensitivity (RBR) increased 2 to 4-fold
[79,80]. The findings showed that the combined treatment leads to a decrease in the
needed dose of radiation, obviates the need for high-heat treatment, and finally
protects the nutritional values and sensory quality of natural products, thereby
obtaining higher quality products. Recent studies have approved radiation in
combination with mild heating treatment or the addition of some EOs such as
carvacrol or cinnamon to increase bacterial radiosensitivity (RBR) [79,80].

Many studies have shown that irradiation technology in combination with other
treatments such as mild heat treatment can be used as an innovative and effective
method to reduce or eliminate the growth of bacteria and parasites and subsequently
extend the shelf life of food products with acceptable nutritional values [81].

6. Food Irradiation around the World

According to the Institute of Food Science & Technology (IFST) [82], more than 50
countries have given approval for over than 60 products to be irradiated in the world.
In Asia the use of irradiation for food decontamination and phytosanitary purposes
was estimated to 285,223 tons per year in 2010. In the European Union, the quantity
of irradiated foods was estimated to 9264 tons, especially for spice decontamination.
In the USA the total was estimated at 103 tons [83]. The USA, China, The
Netherlands, Belgium, Brazil, Thailand, and Australia are the major countries that
have adopted the technology commercially [82]. The use of irradiation for
phytosanitary purposes is important around the world. More than 18,446 tons of food
are irradiated worldwide for phytosanitary purposes, representing 5734 tons in
Hawaii, 493 tons in Australia, 100 tons in India, 951 tons in Thailand, 850 tons in
Vietnam, and 10,318 tons in Mexico, mostly for export to the USA [83]. Australia was
the first user of irradiation for phytosanitary purposes in 2004, especially to export to
New Zealand. India started to export to the USA in 2007, followed by Thailand and
Vietnam. Mexico started to ship irradiated foods to the USA in 2008 and the export
increased from 257 tons in 2008 to 3521 tons in 2009, making it now the most
important exporter to the USA.

6.1. Food Irradiation in America

According to Kume et al. [83], the USA has an important commercial irradiation
program and information is distributed via an update newsletter. Around 120,000
tons of food are irradiated annually in the USA for human and animal consumption
[84]. The most important irradiated products in the USA are spices (80,000 tons), pet
treats (20,000 tons), fresh products (14,000 tons), and ground beef (8000 tons).
Approval for food irradiation started in the USA in 1963 for wheat and wheat flour at
a dose of 0.5 kGy, and in 1985 for parasite elimination at 1 kGy. Then, from 1985 to
1992, the irradiation of dry enzymes, fresh products, spices, and poultry at doses of
10, 1, 30, and 3 kGy, respectively, was accepted. From 2000 to 2008 red meat, eggs,
seeds for sprouting, pet food, sweet potatoes, shellfish, lettuce, and spinach were
added in the list of approved irradiated foods at doses of 8, 50, 1, 5.5, 4, and 1 kGy,
respectively. In 2012, the Food and Drug Association of the USA extended its
approval to cover irradiated unrefrigerated meat. However, according to IFST [82],
an outbreak of E. coli O157:H7 in 1993 that resulted in four deaths and hundreds of
hospitalizations, attributed to undercooked hamburgers, was a major stimulus to
adopt the irradiation of foods. Following a 2006 outbreak of spinach contaminated
with E. coli, approval of the irradiation of spinach and lettuce was given in the USA
in 2008. In 2009, approval of irradiation of oysters to eliminate Vibrio vulnificus was
given. Presently, more than 6000 tons of products including papaya, sweet potatoes,
basil, ginger, melons, taro leaves, curry leaves, longan, litchi, mangosteen, and
rambutan are irradiated annually in Hawaii [85]. The irradiated products are exported
to the U.S. mainland, Germany, and Switzerland.

In Canada around 2000 tons of spices are irradiated annually. Only potatoes, onions,
wheat, flour, flour, whole and ground spices, and dehydrated seasoning preparation
are approved for irradiation. The regulations were adopted in 1960 and 1965 to treat
potatoes and onions, respectively, at a dose of 0.15 kGy; in 1969 to treat wheat and
flour at a dose of 0.75 kGy; and in 1984 to treat spices and seasoning at a dose of
10 kGy. A regulatory proposal was submitted in 2002 for ground beef, poultry,
shrimp, prawns, and mangoes but is still under consideration.

In Mexico, around 7500 tons of irradiated guavas, mangoes, and peppers are
irradiated for export to the USA [82]. According to Kume and Todoriki [83], exports
increased to more than 10,318 tons, especially guava (9121 tons), sweet lime (600
tons), mangoes (239 tons), grapefruit (101 tons), and manzano peppers (257 tons).

In Brazil, quantity of irradiated foods was estimated at 20,000 tons for spices, 3000
tons for fruits, and 23,000 tons in total in 2009 [86].

The total quantity of irradiated foods in the USA, Canada, and Brazil in 2009 was
around 101,400 tons of spices, 7000 tons of fruits, and 8000 tons of meat, for a total
of 116,400 tons of food [86]. According to this research, as spices are mainly used
in industrial processed foods, special labeling is not required for them; however,
irradiated fruits and meat should be labeled.

6.2. Food Irradiation in Asia

More than 285,223 tons of foods were irradiated in Asia in 2010 [83]. According to
this research, China is the largest Asian producer of irradiated foods, with more than
100 irradiators to irradiate more than 200,000 tons of garlic, spices, dried vegetables,
cooked meats, fruits, and grain [83]. Vietnam also irradiated more than 66,000 tons
of foods, including frozen seafood and fruit in 2010. Japan irradiated around 6246
tons of potatoes. In Indonesia similar amounts of food were irradiated in 2010,
especially cocoa, frozen seafood, spices, and others. Then, India and Thailand
irradiated around 2000 tons of spices, dried vegetables, fruits (Mango, Mangosteen,
Logan), fermented sausage (Nham), herbs, sweet tamarind, and others in 2010.
Pakistan, Malaysia, and the Philippines irradiated around 1000 tons of food,
including spices, fruits, nutritional drinks, herbs, and dried vegetables [83]. The
authorization can, however, include more items. For example, in Pakistan the
regulations include potatoes, onions, fresh fruits, grains, chicken and meat, fish and
seafood, spices, herbs, and dry food for animals [87]. In Korea, 5394 tons of spices
and dry vegetables were irradiated in 2009 [86]. However, the authorized products
in Korea include potatoes, onions, garlic, chestnuts, fresh and dried mushrooms,
dried meats, powdered fish, shellfish, soybean paste powder, hot pepper powder,
soybean sauce powder, starch, dried spices and vegetables, yeast and enzymes,
powdered aloe, ginseng, and sterile meals for hospital patients. The irradiation of
these products was accepted from 1987 to 1995 [87]. Bangladesh obtained
authorization for food irradiation in 1995. In 1998, around 1300 tons of different foods
including frozen foods were irradiated [87].

6.3. Food Irradiation in Australia

From 2004 to 2010, 256–1205 tons of mangoes, litchi, and papaya were imported to
New Zealand. In 2010, the irradiation of mango and litchi was 460 tons and 33 tons,
respectively [83]. In 2012, the Food Standard Authority in Australia and New Zealand
also approved the use of irradiation for capsicum and tomatoes [88].

6.4. Food Irradiation in Africa and Other Regions

Eighteen thousand tons of spices and honey were irradiated in 2009 in South Africa.
Egypt also uses the technology for the irradiation of spices and dehydrated
vegetables (550 tons/year). In the Ukraine, more than 70,000 tons of grain and fruits
are treated by irradiation annually [86].

6.5. Food Irradiation in the European Union

Around 9264 tons of food were irradiated in the European Union in 2010 [89]. Ten
countries are doing commercial application of food irradiation. The most important
countries are Belgium, France, and The Netherlands. In Belgium, irradiation
treatment is especially performed for frog legs, poultry, herbs and spices, dehydrated
blood, fish and shellfish, and meat (7279 tons/year). In The Netherlands, irradiation
is practiced for dehydrated vegetables, frog parts, spices, herbs, egg white, poultry,
and shrimp (3299 tons/year). In France, frozen frog legs, poultry, Arabic gum, herbs,
spices, and dried vegetables are treated by irradiation (3111 tons/year). Spain,
Poland, Hungary, Germany, the Czech Republic, Romania, and Estonia especially
treat herbs, spices, and vegetable seasoning (about 369, 687, 151, 127, 27, 10, and
17 tons/year, respectively) [83,86].

7. Global Perceptions of Food Irradiation

Food irradiation has been approved since 1989 by the USDA and FDA. Irradiation
treatment is widely applied for blood and spices. However, this technology is still
controversial due to its bad reputation such as modification of food properties,
formation of dangerous substances, and fall out dangerous process or accidents and
its association with the nuclear establishment. Actually, new research has
demonstrated that almost all of these prejudices are misleading statements and
overestimated. However, recent studies have shown that consumers still remain
reluctant to purchase irradiated products. This is intimately related to the lack of
information about irradiation process and the natural human resistance to change.
In fact, the perception of irradiated food by consumers depends on the degree of
awareness of people about irradiation technology. However, due to the growing
number of recalls after food poisoning incidents, it is important to revise the
marketing policy of irradiated food to make consumers more conscious of the
benefits of this technology for human wellbeing. Based on a broad search, Heddle
et al. [90] proposed six recommendations to increase the acceptance of irradiated
food by consumers

 1 Set up a public education campaign to address needs.

 2 Develop a knowledge translation strategy for health care professionals.
 3 Develop risk communication strategies to address risk perceptions.
 4 Identify a strategy and focus of ongoing research and surveillance related
to pathogen reduction.
 5 Explore society’s willingness to pay attention to pathogen-reduction
technology, by considering the economic impacts associated with this
technology, including direct and indirect costs and the potential for offsetting
additional costs by eliminating redundancy.
 6 Consider the issue of choice.

Most studies on consumers’ perception about irradiated food have shown that
education seems to be the key to consumer acceptance. Numerous consumer
studies clearly showed that when given a choice and even a small amount of
accurate information, consumers are not only willing to buy irradiated foods but also
often prefer them over food treated by conventional means. A variety of market
research studies conducted over the past four decades demonstrated that the
majority of consumers will choose irradiated products over non-irradiated ones after
they learn the facts and understand the benefits [91].

7.1. America
7.1.1. South America

Thirty years ago, a study demonstrated that Argentinian people were impressed by
irradiated onions and garlic. Six tons of onions and one ton of garlic were sold
between 1985 and 1986 and global consumer appreciation gave interesting results.
Recently, the work of Finten et al. [74] confirmed that, in Argentina, people have little
awareness about this technology. Approximately 39% of respondents believed
misleading myths about food irradiation and had doubts about it. However, after
being supplied with informative materials, 42% of respondents were willing to
purchase or eat irradiated ready to eat (RTE) spinach leaves, while 35% were
doubtful. This emphasizes the importance of having well-informed and more aware
consumers.

7.1.2. North America

In the USA in the 1980s, irradiated apples sold well even at a higher price than
unirradiated apples ($0.1/lb) [92]. This demonstrated that consumers are looking for
a product of good quality and irradiation treatment is not a real barrier.

The approval of fresh meat and meat products irradiation in February 2000 allowed
the control of meat pathogens. The challenge was that only half of adult resident
were willing to buy irradiated ground beef or chicken and only a quarter were willing
to pay extra for these products [93]. These statistics were discouraging for
companies.

On the other hand, consumers have recently become more concerned about
irradiation risks. According to Crowley et al. [94], acceptance of meat irradiation was
clearly driven by concerns about the risks of irradiation, but not the risks of bacterial
contamination, confirming differences in the perception of natural and technological
risks. Thus, a general lack of concern about foodborne illness and the fear of
perceived, possibly negative “radiation” side-effects impede willingness to endorse
this food processing technology. The respondents are afraid of different aspects of
the technological risks associated with irradiation such as over-irradiated meat,
health risks associated with eating irradiated meat, radiation exposure due to
accidents, and the belief that the negative health consequences of meat irradiation
would be worse than its potential benefits.

Efforts are needed to educate people to improve their perception about irradiated
foods. Thomson et al. [95] demonstrated that educators’ beliefs about the safety of
food irradiation were influenced by their perceived understanding of it. In the USA,
after a relatively short explanation about irradiation and alternative processes,
consumers generally become more accepting of irradiation especially when
compared to treatments that involve exposure of food to chemical additives and
residues. The marketing of irradiated hamburger, Hawaiian papaya, and sweet
potato was a great success for at least 10 years. New irradiated exotic fruits from
Mexico and several Asian countries are now available in markets [69]. In Michigan
and Florida, public education efforts have achieved some success in changing
peoples’ attitudes about purchasing irradiated foods. One store’s efforts have
enabled it to sell a variety of irradiated produce including grapefruits, oranges,
onions, tomatoes, mushrooms, and blackberries [38]. The report of Hunter [96]
showed that consumers’ most important motivations when buying irradiated food are
killing foodborne pathogens (77%), controlling insect infestation (64%), and reducing
the use of insecticides (60%). A study performed by the National Cattlemen’s Beef
Association in 2002 reported that 85% of participants would accept irradiated beef if
some improvements were made: 1—replacement of the word “irradiation”; 2—
explaining the irradiation process; 3—giving consumers a choice between irradiated
and non-irradiated beef; and 4—improvement of the quality of the final product [97].

Also, Canadian consumers are not really informed, as 57% of Canadians had not
previously heard about irradiation. However, Canadian consumers of different
genders and ages also appear to behave differently about accepting novel
technologies, as men and people aged over 55 years old (48%) were more awarded.
After a brief presentation about irradiation techniques, 74% of people aged over 55
years were ready to support food irradiation. In total, 66% of respondents supported
food irradiation, against 34% who opposed this option [98].

7.2. Europe

Despite the effectiveness of irradiation for food decontamination, the limited diffusion
between EU member countries of ionizing radiations for the treatment of agri-food
reduces its popularity.

Italian people, for example, have a historical fear of nuclear technology, and too often
consumers are misinformed about food irradiation technology. However, contrary to
what one might imagine, in 1976 irradiated potatoes were appreciated by the Italian
people because of their better quality and storability.

Irradiated foods have sold well in Poland (irradiated potatoes and onions in 1987–
1988), with 90% overall acceptance for potatoes and 95% for onions; in France,
irradiated strawberries sold well in 1987 even an additional 30% added to the
conventional price, which indicates a good acceptance (only 2% of people rejected
strawberry irradiation) [92]. Irradiated frog’s legs were also successfully retailed in
France and Belgium.

In the United Kingdom, no significant difference was observed between chilled ready
meals irradiated at 2 kGy and non-irradiated meals (carrots, broccoli, beef and gravy,
roast potatoes, and Yorkshire pudding) [99]. Irradiation was found to be the least
acceptable intervention by Scottish people [100].

Turkish people also have a lack of information about the irradiation process (only
29% of consumers are aware of food irradiation) [101]. In an evaluation, between
69% and 80% of respondents were very concerned and uncertain about the safety
of irradiated foods and were cautious about what they purchase [101,102]. Only 11%
expressed that irradiated foods are safe.

According to Parlato et al. [103], the anti-irradiation message can be effectively

counteracted and consumer confidence in the safety of irradiation process can be
restored by detailed science-based information on irradiation. This starts with a huge
effort from the health authorities and other institutions to allow correct understanding
of the potential advantages of irradiation for the necessary investment to occur.

7.3. Africa

Few studies have been carried out on the African continent. Between 1978 and
1988, 90% of South African consumers of irradiated potatoes, mangoes, papaya,
and strawberries judged that they were satisfied with the quality of the irradiated
items [92].

7.4. Asia

In Thailand, irradiated onions were well accepted by consumers in 1987 and

respondents were ready to purchase them even at a slightly higher price than
unirradiated onions [92]. Fermented sausage was also well accepted and over 94%
of consumers indicated a willingness to buy irradiated sausage again.

In Bangladesh, irradiated fish was well accepted by consumers, especially for its
better quality and appearance [92]. Also, when onions were sold, consumers were
in favor of the irradiated onions. In China, spicy chicken feet were also well accepted
by consumers. In the Philippines, for similar prices, irradiated onions and garlic were
preferred to unirradiated ones.

In Korea, studies on irradiation acceptability to women demonstrated that it is more

convincing to hear a lecture by an expert followed by watching a video- and reading
a book with a group. In addition, acceptance of irradiated food has been shown to
lead to support for the nuclear industry [104].

Japan invested in public education about radiation for schoolchildren and their
parents. More than 60% of kids were satisfied with the information given.
Consumers’ perceptions about irradiation thereby seemed to shift to become more
positive [105,106].

7.5. Oceania

In New Zealand, irradiated mango and litchi have been imported and sold since
2005. Presently a significant volume of irradiated mangoes, lychees, tomatoes, and
capsicums are now purchased by consumers in New Zealand. It is clear that
irradiation fulfills not only a technological need but also a consumer need by making
quality produce available at competitive prices. Consequently, a significant
proportion of the New Zealand public will consistently buy irradiated fresh products
when they are made available [107].

8. Communication: An Important Factor in Consumer Acceptance of Irradiated

Food

In the last decade, with the awareness of health problems caused by food spoilage,
the food industry has utilized new technologies to improve the safety and shelf life
of food. Some of the new technologies being applied are biotechnology, ionizing
radiation, pulsed electronic fields, ultraviolet laser treatment, etc. However, those
emerging technologies pose challenges to the industry in terms of consumer choice
and acceptability. One of the major barriers of commercialization is the influence on
sensorial characteristics that influence the purchase of the product, but there are
also some extrinsic factors that might influence consumers, including contextual,
social, cultural, and attitudinal variables [21].

To evaluate the reaction of people, studies have evaluated the perceived risk to
consumers of new food technologies [108,109]. Data showed that consumers have
significant levels of concern about the hazards of new technology, which from a
technical/rational point of view was evaluated as a low risk. In general, consumers
perceived gamma irradiation as a risky technology because of (1) the carcinogenic
nature of irradiated food; (2) the risks of using irradiation technology; (3) the risk of
irradiation escaping; and (4) the risks associated with the transportation of
radioactive material [110]. This negative perception of new technology was studied
by Cardello [21], who related the perceived risks to the consumer’s lack of
awareness about any processes applied to the purchased food—either because
these are out of the consumer’s control or because they are unobservable.
Therefore, consumers do not know if what they are consuming is irradiated or not or
if it might have negative effects on their health.

Several authors [101,111,112] studied people’s opinions and readiness to accept

irradiated products. Junqueira-Gonçalves et al. [113] indicated that the people
surveyed had a lack of information and understanding of food irradiation. Briefly,
45.9% of the responders answered that irradiated food means radioactive food, and
57.1% of people were uncertain if gamma irradiation can cause damage to human
health and/or the environment. Nevertheless, marketing has shown that consumers
are willing to purchase irradiated products if they are informed about the effects and
the process [91].

Several studies have demonstrated that food irradiation can mitigate the
development of food-borne diseases, which continue to grow with approximately 76
million illnesses, 325,000 hospitalizations, and 5000 deaths in the United States
annually [114], and 1.6 million illnesses, 4000 hospitalizations, and 105 deaths in
Canada [115]. Even though scientists recognize food irradiation as a safe and
effective process, there can still be either a negative bias or a lack of information,
which can be a limiting factor when addressing consumer resistance towards food
irradiation.
Scientists have realized that consumer perception of novel technologies relies
heavily on the communication approach employed. It was proven by Furuta et al.
[106] that more than half of interviewed people present at a “Radiation fair” held in
Japan recognized the word “radiation,” which was taught to them in elementary
school and by the mass media. In addition, results showed that image can be
improved if correct information about radiation is relayed to the public. Hence, factors
such as teaching, labeling, and food retail could play an important role in diffusing
new technology and the creation of a positive link with consumers.

8.1. Teaching

Although gamma irradiation has shown many beneficial effects such as insect
disinfection, extension of food shelf life, and reduction of bacteria [69], irradiation has
not been widely adopted as a commercial process. It was suggested by Zimmerman
et al. [116] that in terms of new food technology either cognitive or affective
perceptions can have an influence on consumer attitudes. Similar findings were
mentioned by Edwards [117], who demonstrated that people’s perceptions are
usually based on the global attitude towards new technology when knowledge of the
specific topic is low [118,119].

However, not only the global adoption of the novelty but also educational programs
can have significant impact on consumer attitudes. Recent studies point out that
once consumers are educated, they will buy irradiated food and adopt a positive
attitude towards food irradiation [120,121]. With the aim of accomplishing this goal,
scientists must take into account the non-technical perceptions of people about the
believed “risk” of the gamma irradiation technique [122]. In comparison to other food
procedures, gamma irradiation was ranked alongside food preservatives and sodium
nitrate as a less-feared factor.

Although nowadays consumers are looking for food safety and “freshness,” there is
evidence that consumers regard food safety as a basic requirement. Thus,
consumers do not value the extension of product shelf life that results from gamma
irradiation. In 1988, the United Kingdom was confronted with Salmonella infection in
egg production, while consumers were not aware that this crisis could occur in eggs.
This is the reason why highlighting improvements caused by irradiation in food safety
can serve as a proof of the advantages of this technology.

Some authors such as Rogers [123] have proposed that the topic of gamma
irradiation should be included in educational programs so as to introduce consumers
to it. Rogers [123] mentioned the importance of including characteristics such as
advantages, compatibility, complexity, trialability, and observability in the diffusion
and adaptation of new technology. Complexity is defined as the degree to which an
innovation is perceived as difficult to understand and use. Observability is defined
as the degree to which the applied technology is visible to others; this will allow
consumers to observe the process and the results of the irradiated food. Trialability
is expressed by Rogers [123] as the degree to which the innovated product can be
experienced on a limited basis. For instance, when a consumer has his first contact
with irradiated food, his organoleptic senses will allow him to compare and determine
his opinion with respect to irradiated and non-irradiated foods.

8.2. Labeling; RADURA Symbol

In order to identify an irradiated product and alert consumers to its quality, the Pilot
Plant for Food Irradiation at Wageningen, The Netherlands, created the symbol
RADURA (Figure 1). This word is related to “radurization,” a word derived from
radiation and the Latin word “durus” for “lasting.” The term is used for the process of
exposing food to ionizing radiation to enhance and extend the shelf life. Thus, the
product is irradiated at doses in the range from 0.4 to 1 kGy to decreased number of
spoilage bacteria [124]. Due to the fact that external microorganisms can also
encounter the irradiated product, food packaging is part of the process. The
“RADURA” symbol was established to represent the irradiation treatment. The
symbol presents a plant (dot and two leaves), in a closed package (circle), irradiated
with ionizing rays passing through the package to the food (dashed lines). Despite
the fact that the use of the RADURA symbol is optional, according to the Codex
Alimentarius standard [125], if the food or an ingredient product is treated with
ionizing radiation a written statement shall be placed in proximity to the food to
indicate that the treatment was done. This last requirement might change from
country to country. For instance, in the United States, labeling is only required if the
whole food has been irradiated and labeling is not required at restaurants/catering
establishments [126]. Canada requires labels and written statements such as
“irradiated,” “treated with radiation,” or “treated by irradiation” when the whole
product was irradiated or more than 10% of the ingredients that compose the final
product [127]. On the other hand, New Zealand demands labeling for even minor
ingredients and in restaurant/catering establishments [32].

Figure 1 The RADURA symbol.

Labeling is an important step that assures consumers whether they are deciding to
buy or not buy irradiated products. Indeed, once consumers know about the
identification of products, it is easy for them to accept the risks of purchasing food
derived from a new technology [128].

According to Junqueira-Gonçalves et al. [113] the RADURA symbol is not frequently

present on food labels in Chile. However, studies mentioned that 55.8% of people
would buy irradiated food because the symbol transmits the sensation of confidence
and safety. Similar discussions were held by Rollin et al. [3], who said that labeling
products treated with new technology will raise awareness and improve
transparency. However, there are still ambiguous reactions demonstrated in the
study performed by He et al. [128], who reported that over 30% of people in the USA
would consider the term “irradiated” beef product as a warning and only 21% would
consider it safe and buy it.

9. Food Retailers

Since food irradiation is not a familiar technology for everybody and does not show
any detectable changes in food, consumers’ confidence relies on food processors
who might inform them whether the product was irradiated or not [110]. Thus, the
role of food retailers is as important as that of educators, giving a sense of assurance
of food quality to people who still are scared and avoid the technology [129].

However, even if trust between food retailers and consumers is gained, it can be
easily lost with a single mistake [129]. Trust-destroying mistakes such as withholding
of information or scientific tests indicating that a product is less safe than originally
conceived might affect consumer perceptions about gamma irradiation.

Thus, in order to be transparent and maintain consumers’ confidence, food retailers

should firstly give consumers the freedom to choose between irradiated and non-
irradiated products. For this, positive disclosure is required. Secondly, because
irradiation is a process that does not exhibit any modification on food, retailers must
indicate that the irradiation process has been done in an appropriate manner.

Moreover, food retailers are the first facilitators in developing irradiation acceptability
by merchandising irradiated goods to the consumer. As the first contact of
consumers with such product is in the market, food retailers should give consumers
the opportunity to observe and judge for themselves about the new technology.
Thus, consumer confidence in the food supply chain will be enhanced.

10. Future Directions for Gamma Irradiation

To encourage consumers to accept gamma irradiation, some authors have

considered future strategies for increasing retail of irradiated food [69,110,130]:

 Highlight the advantages of the technology rather than pointing out the
technology. Consumers value “freshness” more than increased shelf life,
which can be seen as “unnatural.”
 Take into account the positive and negative aspects that will coexist in any
food debate.
 Use labels to show advantage information, thus offsetting the warnings that
labels are perceived to bring with them. A labeled product will be assured. ,
thus decreasing consumer opposition to irradiated food. This fact was
observed in Australia and New Zealand.
 Create a partnership with food retailers so they can promote the marketing of
irradiated food, especially to those who are small or medium-sized.
 It can be worthwhile to have stakeholders that believe in the value of food
irradiation, thus food retailers will be seen as less biased and consumer trust
will increase.
11. Conclusions

Foods processed by novel and emerging technologies, e.g., biotechnology, ionizing

radiation, pulsed electric field, ultra violet laser treatment, etc. pose a serious
challenge to factors such as consumer choice, purchasing behavior, and acceptance
of irradiated foods. Future research on novel food processing and preservation
technologies should focus more directly on questions and issues related to
consumers’ expectations. Developing a better understanding of the variables that
directly influence the acceptance of these products and more effective marketing
and informational strategies could improve the acceptance of these novel
technologies in tomorrow’s marketplace.

Evidence has shown that gamma irradiation, especially related to food, represents
a hazardous technology to consumers. This is the reason why commercialization
has encountered several barriers to being adopted and accepted by consumers. It
has been demonstrated that communication, labels, and education about new
technology will enhance consumers’ perception of irradiated food. Thus, showing the
possible problems carried by food-borne diseases can create consciousness of the
importance of food safety. On the other hand, due to the fact that gamma irradiation
is a process that does not affect physical aspects of the product, the role of the food
industry and food retailers should be to inform the public either by labeling products
or by telling consumers about the benefits of the irradiation.

New strategies based on positive messages of gamma irradiation marketplace will

thus encourage consumers to be more receptive to safety-enhanced and high-
quality irradiated foods.

Based on knowledge derived from over half a century of research, irradiated foods
are safe and wholesome at a specified radiation dose. The irradiated foods are
generally nutritionally equivalent to non-irradiated foods subjected to normal
processing, even better in some cases; radiation processing can lead to an
increment in the nutritional value of irradiated fruits and vegetables such as vitamin
C content and phenolic compounds.

Furthermore, many studies have shown that irradiation technology in combination

with other treatments can be used as an innovative and effective method to add
value to food products. As previously detailed, people are still confused and fail to
differentiate irradiated foods from radioactive foods. When well informed, a reduced
number of consumers will reject irradiated food. What a consumer is looking for is a
product with good quality and a competitive price. When consumers are aware of
the short- and long-term dangers of chemical additives, they accept more irradiation
treatments being applied to food products. However, companies should update their
quality system and implement new procedures to support risk management and the
supply and distribution chains.

Acknowledgments
This research was supported by the Natural Sciences and Engineering Council of
Canada (NSERC) through the discovery program.

Author Contributions

All persons who meet authorship criteria are listed as authors, and all authors certify
that they have participated sufficiently in the work to take public responsibility for the
content, including participation in the concept, design, writing, or revision of the
manuscript. Every author contributed extensively to prepare one section of the work
presented in this paper.

Conflicts of Interest

The authors declare no conflict of interest.

References

1. Brody A.L., Labuza T.P. MIT Food Technology: The major driver for food
technology for 50 years. J. Food Sci. 2014;79:iv–v. [PubMed]
2. Heldman D.R., Lund D.B. Food Engineering Interfaces. Springer; New York, NY,
USA: 2010. The Beginning, current, and future of food engineering: A perspective;
pp. 3–18.
3. Rollin F., Kennedy J., Wills J. Consumers and new food technologies. Trends
Food Sci. Technol. 2011;22:99–111. doi: 10.1016/j.tifs.2010.09.001. [Cross Ref]
4. Agriculture and Agri-Food Canada Emerging Food Products, Technologies and
Processes: Insights for Regulators. [(accessed on 12 August 2015)]. Available
online: http://www.agr.gc.ca/eng/industry-markets-and-trade/statistics-and-market-
information/by-product-sector/processed-food-and-beverages/reports-and-
resources-food-processing-innovation-and-regulations/emerging-food-products-
technologies-and-processes-insights-for-regulators/?id=1339618362423.
5. Frewer L., Bergmann K., Brennan M., Lion R., Meertens R., Rowe G., Siegrist M.,
Vereijken C. Consumer response to novel agri-food technologies: Implications for
predicting consumer acceptance of emerging food technologies. Trends Food Sci.
Technol. 2011;22:442–456. doi: 10.1016/j.tifs.2011.05.005. [Cross Ref]
6. Lusk J.L., Roosen J., Bieberstein A. Consumer acceptance of new food
technologies: Causes and roots of controversies. Annu. Rev. Resour. Econ.
2014;6:381–405. doi: 10.1146/annurev-resource-100913-012735. [Cross Ref]
7. Alston J., Norton G., Pardey P. Science under Scarcity: Principles and Practice
for Agricultural Research Priority Setting and Evaluation. Cornell University; Ithaca,
NY, USA: 1995.
8. Slovic P., Finucane M.L., Peters E., MacGregor D.G. Risk as analysis and risk as
feelings: Some thoughts about affect, reason, risk, and rationality. Risk Anal.
2004;24:311–322. doi: 10.1111/j.0272-4332.2004.00433.x. [PubMed] [Cross Ref]
9. Erdem S., Rigby D., Wossink A. Using best–worst scaling to explore perceptions
of relative responsibility for ensuring food safety. Food Policy. 2012;37:661–670. doi:
10.1016/j.foodpol.2012.07.010. [Cross Ref]
10. World Health Organization (WHO) High-Dose Irradiation: Wholesomeness of
Food Irradiatied with Doses above 10 kGy. World Health Organization; Geneva,
Switzerland: 1999. [PubMed]
11. Cardello A.V., Schutz H.G., Lesher L.L. Consumer perceptions of foods
processed by innovative and emerging technologies: A conjoint analytic study.
Innov. Food Sci. Emerg. Technol. 2007;8:73–83. doi: 10.1016/j.ifset.2006.07.002.
[Cross Ref]
12. Nayga R.M., Aiew W., Nichols J.P. Information effects on consumers’ willingness
to purchase irradiated food products. Appl. Econ. Perspect. Policy. 2005;27:37–48.
doi: 10.1111/j.1467-9353.2004.00206.x. [Cross Ref]
13. Fischler C. Food, self and identity. Soc. Sci. Inf. 1988;27:275–292. doi:
10.1177/053901888027002005. [Cross Ref]
14. Pliner P., Hobden K. Development of a scale to measure the trait of food
neophobia in humans. Appetite. 1992;19:105–120. doi: 10.1016/0195-
6663(92)90014-W. [PubMed] [Cross Ref]
15. Cox D., Evans G. Construction and validation of a psychometric scale to
measure consumers’ fears of novel food technologies: The food technology
neophobia scale. Food Qual. Preference. 2008;19:704–710. doi:
10.1016/j.foodqual.2008.04.005. [Cross Ref]
16. Capiola A., Raudenbush B. The effects of food neophobia and food neophilia on
diet and metabolic processing. Food Nutr. Sci. 2012;3:1397–1403. doi:
10.4236/fns.2012.310183. [Cross Ref]
17. Sparks P., Shepherd R. Public perceptions of the potential hazards associated
with food production and food consumption: An empirical study. Risk Anal.
1994;14:799–806. doi: 10.1111/j.1539-6924.1994.tb00291.x. [PubMed] [Cross Ref]
18. Frewer L.J., Shepherd R. Ethical concerns and risk perceptions associated with
different applications of genetic engineering: Interrelationships with the perceived
need for regulation of the technology. Agric. Hum. Values. 1995;12:48–57. doi:
10.1007/BF02218074. [Cross Ref]
19. Grunert K.G., Lähteenmäki L., Nielsen N.A., Poulsen J.B., Ueland O., Åström A.
Consumer perceptions of food products involving genetic modification—Results
from a qualitative study in four Nordic countries. Food Qual. Preference.
2001;12:527–542. doi: 10.1016/S0950-3293(01)00049-0. [Cross Ref]
20. Wilcock A., Pun M., Khanona J., Aung M. Consumer attitudes, knowledge and
behaviour: A review of food safety issues. Trends Food Sci. Technol. 2004;15:56–
66. doi: 10.1016/j.tifs.2003.08.004. [Cross Ref]
21. Cardello A.V. Consumer concerns and expectations about novel food processing
technologies: Effects on product liking. Appetite. 2003;40:217–233. doi:
10.1016/S0195-6663(03)00008-4. [PubMed] [Cross Ref]
22. Bruhn C.M. Enhancing consumer acceptance of new processing technologies.
Innov. Food Sci. Emerg. Technol. 2007;8:555–558. doi: 10.1016/j.ifset.2007.04.006.
[Cross Ref]
23. Center for Media and Public Affairs Food for Thought V Research. [(accessed
on 23 May 2014)]. Available online: http://ific.org.
24. Food Marketing Institute . U.S. Grocery Shopper Trends. Food Marketing
Institute; Washington, DC, USA: 2015. [(accessed on 6 February 2015)]. Available
online: http://www.fmi.org/docs/default-source/document-share/fmitrends15-exec-
summ-06-02-15.pdf.
25. Schutz H.G., Cardello A.V. Information effects on acceptance of irradiated foods
in a military population. Dairy Food Environ. Sanit. 1997;17:470–481.
26. Campbell H., Fitzgerald R. Follow the fear: A multi-sited approach to GM. Rural
Soc. 2001;11:211–224. doi: 10.5172/rsj.11.3.211. [Cross Ref]
27. Li-Cohen A.E., Bruhn C.M. Safety of consumer handling of fresh produce from
the time of purchase to the plate: A comprehensive consumer survey. J. Food
Protect. 2002;65:1287–1296. [PubMed]
28. Cox D., Evans G., Lease H. The influence of information and beliefs about
technology on the acceptance of novel food technologies: A conjoint study of farmed
prawn concepts. Food Qual. Preference. 2007;18:813–823. doi:
10.1016/j.foodqual.2007.01.011. [Cross Ref]
29. Bell R., Paniesin R. Product Testing with Consumers for Research Guidance:
Special Consumer Groups. Volume 2. ASTM; Philadelphia, PA, USA: 1992. The
influence of sauce, spice, and name on the perceived ethnic origin of selected
culture-specific foods; pp. 22–36.
30. Mossel D., Drake D. Processing food for safety and reassuring the consumer.
Food Technol. 1990;44:63–67.
31. Pohlman A.J., Wood O., Mason A.C. Influence of audiovisuals and food samples
on consumer acceptance of food irradiation. Food Technol. 1994;48:48–49.
32. Costa-Font M., Gil J.M., Traill W.B. Consumer acceptance, valuation of and
attitudes towards genetically modified food: Review and implications for food policy.
Food Policy. 2008;33:99–111. doi: 10.1016/j.foodpol.2007.07.002. [Cross Ref]
33. Curtis K.R., McCluskey J.J., Wahl T.I. Consumer acceptance of genetically
modified food products in the developing world. AgBioForum. 2004;7:70–75.
34. Frewer L., Lassen J., Kettlitz B., Scholderer J., Beekman V., Berdal K.G. Societal
aspects of genetically modified foods. Food Chem. Toxicol. 2004;42:1181–1193.
doi: 10.1016/j.fct.2004.02.002. [PubMed] [Cross Ref]
35. Fox J.A. Influences on purchase of irradiated foods. Food Technol. 2002;56:34–
37.
36. Zienkewicz L.S.H., Penner K.P. Consumers’ perceptions of irradiated ground
beef after education and product exposure. Food Protect. Trends. 2004;24:740–745.
37. Wilson C., Evans G., Leppard P., Syrette J. Reactions to genetically modified
food crops and how perception of risks and benefits influences consumers’
information gathering. Risk Anal. 2004;24:1311–1321. doi: 10.1111/j.0272-
4332.2004.00528.x. [PubMed] [Cross Ref]
38. Patil B.S., Pike L.M., Howard L.R. Effect of gamma irradiation on quercetin on
onion. Subtrop. Plant Sci. 1999;51:16–22.
39. Lacroix M., Follett P. Combination irradiation treatments for food safety and
phytosanitary uses. Stewart Postharvest Rev. 2015;11:1–10. doi:
10.2212/spr.2015.3.4. [Cross Ref]
40. Crawford L.M., Ruff E.H. A review of the safety of cold pasteurization through
irradiation. Food Control. 1996;7:87–97. doi: 10.1016/0956-7135(96)00004-7.
[Cross Ref]
41. Mostafavi H.A., Mirmajlessi S.M., Fathollahi H. The potential of food irradiation:
Benefits and limitations. In: Eissa A.H.A., editor. Trends in Vital Food and
Engineering. INTECH; Rijeka, Croatia: 2012.
42. Follett P. Phytosanitary irradiation for fresh horticultural commodities: Generic
treatments, current issues, and next steps. Stewart Postharvest Rev. 2014;10:1–7.
43. Follett P.A., Neven L.G. Current trends in quarantine entomology. Annu. Rev.
Entomol. 2006;51:359–385. doi: 10.1146/annurev.ento.49.061802.123314.
[PubMed] [Cross Ref]
44. Ehlermann D.A.E. Wholesomeness of Irradiated Food. Radiat. Phys. Chem.
2016;125:24–29. doi: 10.1016/j.radphyschem.2016.08.014. [Cross Ref]
45. Komolprasert V. Packaging for foods treated with ionizing radiation. In: Han J.H.,
editor. Packaging for Non-Thermal Processing of Food. 1st ed. Blackwell Publishing
Ltd.; Ames, IA, USA: 2007. pp. 87–117.
46. Roberts P.B. Food Irradiation: Standards, Regulations and World-Wide Trade.
[(accessed on 05 June 2016)];Radiat. Phys. Chem. 2016 129:30–34. doi:
10.1016/j.radphyschem.2016.06.005. Available online:
http://dx.doi.org/10.1016/j.radphyschem.2016.06.005. [Cross Ref]
47. Codex Alimentarius, FAO/WHO . General Principles of Food Hygiene. Codex
Alimentarius Commission; Rome, Italy: 2003. CAC/RCP 1-1969, Rev 4-2003.
48. Codex Alimentarius, FAO/WHO . General Standard for Irradiated Foods. Codex
Alimentarius Commission; Rome, Italy: 1983. CODEX STAN 106-1983.
49. Codex Alimentarius, FAO/WHO . General Standard for Irradiated Foods. Codex
Alimentarius Commission; Rome, Italy: 2003. CODEX STAN 106-1983, Rev.1-2003.
50. Food and Drug Administration (FDA) Irradiation in the Production, Processing,
and Handling of Food. Final Rule. Fed. Regist. 2005;70:48057–48073. [PubMed]
51. Olson D.G. Scientific Status Summary Irradiation of Food—A Publication of The
Institute of Food Technologists’ Expert Panel on Food Safety and Nutrition. Food
Technol. Mag. 1998;52:56–62.
52. Pauli G.H., Tarantino L.M. FDA Regulatory Aspects of Food Irradiation. J. Food
Protect. 1995;58:209–212.
53. Keener K.M. Food Irradiation—To Zap or Not to Zap? Department of Food
Science, NCSU; Raleigh, NC, USA: 2001.
54. Mostafavi A.H., Fathollahi H., Motamedi F., Mirmajlessi S.M. Food Irradiation:
Applications, public acceptance and global trade. Afr. J. Biotechnol. 2010;9:2826–
2833.
55. Woodside J.V. Nutritional Aspects of Irradiated Food. Stewart Postharvest Rev.
2015;11 doi: 10.2212/spr.2015.3.2. [Cross Ref]
56. Khattak K.F., Rahman T.U. Effect of gamma irradiation on the vitamins,
phytochemicals, antimicrobial and antioxidant properties of Ziziphus mauritiana
Lam. Leaves. Radiat. Phys. Chem. 2016;127:243–248. doi:
10.1016/j.radphyschem.2016.07.001. [Cross Ref]
57. Swallow A.J. Wholesomeness and Safety of Irradiated Foods. Adv. Exp. Med.
Biol. 1991;1:11–31. [PubMed]
58. Morehouse K.M., Komolprasert V. Irradiation of food and packaging: An
overview. ACS Symp. 2004;875:1–11.
59. Rady A.H., Maxwell R.J., Wierbicki E., Phillips J.G. Effect of gamma irradiation
at various temperatures and packaging conditions on chicken tissues. I. Fatty Acid
profiles of neutral and polar lipids separated from muscle irradiated at −20°C. Int. J.
Radiat. Appl. Instrum. C Radiat. Phys. Chem. 1998;31:195–202. doi: 10.1016/1359-
0197(88)90125-7. [Cross Ref]
60. Thomas M.H. Use of ionizing radiation to preserve food. In: Karmas E., Harris
R.S., editors. Nutritional Evaluation of Food Processing. 1st ed. AVI Publications;
New York, NY, USA: 1998. pp. 457–490.
61. Murray T.K. Nutritional aspects of food irradiation. In: Elias P.S., Cohen A.J.,
editors. Recent Advances in Food Irradiation. 1st ed. Elsevier Biomedical Press;
Amesterdam, The Netherland: 1983. pp. 203–216.
62. De Groot A.P., Mijll Dekker V., Slump P., Vos H.J., Willems J.J.L. New
wholesomeness data on radiation-pasteurized chicken. Food Irradiat. Inf.
1973;2:71–72.
63. Lacroix M., Jobin M., Latreille B., Nouchpramool K., Gagnon M. The effect of
gamma irradiation on physical and nutritional quality of penaeus monodon shrimps.
Radiat. Phys. Chem. 1995;46:731–737. doi: 10.1016/0969-806X(95)00251-R.
[Cross Ref]
64. Diehl J.F. Nutritional adequacy of irradiated foods. In: Diehl J.F., editor. Safety
of Irradiated Foods. 2nd ed. Marcel Dekker, Inc.; New York, NY, USA: 1995. pp.
241–282.
65. Fanaro G.B., Hassimotto N.M.A., Bastos D.H.M., Villavicencio A.L.C.H. Effects
of γ-radiation on microbial load and antioxidant properties in green tea irradiated with
different water activities. Radiat. Phys. Chem. 2015;107:40–46. doi:
10.1016/j.radphyschem.2014.09.008. [Cross Ref]
66. Cetinkaya N., Ozyardımci B., Denli E., Ic E. Radiation processing as a post-
harvest quarantine control for raisins, dried figs and dried apricots. Radiat. Phys.
Chem. 2006;75:424–431. doi: 10.1016/j.radphyschem.2005.10.009. [Cross Ref]
67. World Health Organization (WHO) Food Irradiation. World Health Organization;
Geneva, Switzerland: 1997. Press Release WHO/68.
68. Roberts P.B. Food irradiation is safe: Half a century of studies. Radiat. Phys.
Chem. 2014;105:78–82. doi: 10.1016/j.radphyschem.2014.05.016. [Cross Ref]
69. Filho L.T., Lucia S.M.D., Limaa R.M., Scolforoa C.Z., Carneiroa J.C.Z., Pinheirob
C.J.G., Passamai J.L., Jr. Irradiation of strawberries: Influence of information
regarding preservation technology on consumer sensory acceptance. Innov. Food
Sci. Emerg. Technol. 2014;26:242–247. doi: 10.1016/j.ifset.2014.06.015. [Cross
Ref]
70. Chen Q., Caoc M., Chena H., Gaoa P., Fua Y., Liua M., Wanga Y., Huanga M.
Effects of gamma irradiation on microbial safety and quality of stir fry chicken dices
with hot chili during Storage. Radiat. Phys. Chem. 2016;127:122–126. doi:
10.1016/j.radphyschem.2016.06.022. [Cross Ref]
71. Xavier M.P., Duber C., Mussiob P., Delgadob E., Maquieirab A., Soriab A.,
Curuchetb A., Márquezb R., Méndeza C., López T. Use of mild irradiation doses to
control pathogenic bacteria on meat trimmings for production of patties aiming at
provoking minimal changes in quality attributes. Meat Sci. 2014;98:383–391. doi:
10.1016/j.meatsci.2014.06.037. [PubMed] [Cross Ref]
72. Park J.-N., Sung N.-K., Parka J.-N., Sungb N.-Y., Byunb E.-H., Byuna E.-B.,
Songa B.-S., Kima J.-H., Leec K.-A., Sond E.-J., et al. Microbial analysis and survey
test of gamma-irradiated freeze-dried fruits for patient’s food. Radiat. Phys. Chem.
2015;111:57–61. doi: 10.1016/j.radphyschem.2015.02.011. [Cross Ref]
73. Finten G., Garrido J.I., Agüero M.V., Jagus R.J. Irradiated ready-to-eat spinach
leaves: How information influences awareness towards irradiation treatment and
consumer’s purchase intention. Radiat. Phys. Chem. 2017;130:247–251. doi:
10.1016/j.radphyschem.2016.09.004. [Cross Ref]
74. Lacroix M., Lapointe M., Latreille B., Gagnon M. Effect of gamma irradiation
combined with hot water dip on the chemical and nutritional qualities of mangoes.
Microbiol. Aliments Nutr. 1991;9:247–256.
75. Lacroix M., Ouattara B. Combined industrial processes with irradiation to assure
innocuity and preservation of food products—A review. Food Res. Int. 2000;33:719–
724. doi: 10.1016/S0963-9969(00)00085-5. [Cross Ref]
76. Turgis M., Millette M., Salmieri S., Lacroix M. Elimination of Listeria inoculated in
ready-to-eat carrots by combination of antimicrobial coating and γ-irradiation. Radiat.
Phys. Chem. 2012;81:1170–1172. doi: 10.1016/j.radphyschem.2011.11.023. [Cross
Ref]
77. Turgis M., Han J., Borsa J., Lacroix M. Combined effect of natural essential oils,
modified atmosphere packaging, and gamma radiation on the microbial growth on
ground beef. J. Food Protect. 2008;71:1237–1243. [PubMed]
78. Lacroix M., Ayari S., Dussault M., Salmieri S., Turgis M., Takala P., Vu K.D.
Irradiation in combined treatments and food safety. J. Radioanal. Nucl. Chem.
2013;296:1065–1069.
79. Turgis M., Borsa J., Millette M., Salmieri S., Lacroix M. Effect of selected plant
essential oils or their constituents and modified atmosphere packaging on the
radiosensitivity of Escherichia coli O157:H7 and Salmonella typhi in ground beef. J.
Food Protect. 2008;71:516–521. [PubMed]
80. Lacroix M. Combined industrial processes with irradiation to assure innocuity
and preservation of fruits and vegetables; Proceedings of the International Nuclear
Atlantic Conference (INAC); Santos, Brazil. 28 August–2 September 2005.
81. Institute of Food Science and Technology Food irradiation. [(accessed on June
2015)]. Available online: http://www.ifst.org/print/118.
82. Kume T., Todoriki S. Food irradiation in Asia, the European Union, and the
United States: A status update. Radioisotopes. 2013;62:291–299. doi:
10.3769/radioisotopes.62.291. [Cross Ref]
83. Eustice R.F. Food Irradiation: A Global Perspective and Future Prospects.
[(accessed on 9 January 2011)]. Available online:
http://ansnuclearcafe.org/2011/06/09/food-irradiation-a-global-perspective-future-
prospects/#sthash.YyeeMQ5M.dpbs.
84. Follett P.A., Weinert E.D. Phytosanitary irradiation of fresh tropical commodities
in Hawaii: Generic treatments, commercial adoption, and current issues. Radiat.
Phys. Chem. 2012;81:1064–1067. doi: 10.1016/j.radphyschem.2011.12.007. [Cross
Ref]
85. Kume T., Furuta M., Todoriki S., Uenoyama N., Kobayashi Y. Status of food
irradiation in the world. Radiat. Phys. Chem. 2009;78:222–226. doi:
10.1016/j.radphyschem.2008.09.009. [Cross Ref]
86. International Atomic Energy Agency (IAEA) Consumer acceptance and market
development of irradiated food in Asia and the Pacific. IAEA TECDOC-1219;
Proceedings of the Final Research Co-Ordination Meeting; Leuven, Belgium. 24–28
September 2001; pp. 17–19.
87. Taubenfeld R. Irradiated food coming to a supermarket near you. Chain React.
2013;118:10.
88. European Commission List of Approved Facilities for the Treatment of Food and
Food Ingredients with Ionising Radiation in the Member States. [(accessed on 13
February 2015)]. Available online:
http://storage0.dms.mpinteractiv.ro/media/1/186/3927/8018052/3/instalatii-iradiere-
ue.pdf.
89. Heddle N., Lane S., Sholapur N., Arnold E., Newbold B., Eyles J., Webert K.
Implementation and public acceptability: Lessons from food irradiation and how they
might apply to pathogen reduction in blood products. Vox Sang. 2014;107:50–59.
doi: 10.1111/vox.12135. [PubMed] [Cross Ref]
90. Eustice R.F., Bruhn C.M. Food Irradiation Research and Technology. USDA,
ARS, Eastern Regional Research Center; Wyndmoor, PA, USA: 2006. Consumer
acceptance and marketing of irradiated foods; pp. 63–83.
91. Marcotte M. Consumer Acceptance of Irradiated Food. Nordion International Inc.;
Kanata, ON, Canada: 1989. p. 17.
92. Frenzen P.D., Majchrowicz T.A., Buzby J.C., Imhoff B. Consumer Acceptance of
Irradiated Meat and Poultry Products. United States Department of Agriculture,
Economic Research Service; Washington, DC, USA: 2000. p. 8.
93. Crowley O.V., Marquette J., Reddy D., Fleming R. Factors predicting likelihood
of eating irradiated meat. J. Appl. Soc. Psychol. 2013;43:95–105. doi:
10.1111/j.1559-1816.2012.00984.x. [Cross Ref]
94. Thompson B., Ribera K., Wingenbach G., Vestal T. The relationship between
attitudes, knowledge, and demographic variables of high school teachers regarding
food irradiation. J. Food Sci. Educ. 2007;6:24–29. doi: 10.1111/j.1541-
4329.2007.00020.x. [Cross Ref]
95. Hunter C. Changing attitudes to irradiation throughout the food chain. Radiat.
Phys. Chem. 2000;57:239–243. doi: 10.1016/S0969-806X(99)00387-4. [Cross Ref]
96. Eustice R.F., Bruhn C.M. Food Irradiation Research and Technology. Wiley; New
York, NY, USA: Consumer acceptance and marketing of irradiated foods; pp. 63–
83. Chapter 10.
97. Scott-Thomas C. Survey: Most Candians Support Food Irradiation. [(accessed
on 25 April 2012)]. Available online: http://www.foodqualitynews.com/Industry-
news/Survey-Most-Canadians-support-food-irradiation-once-they-ve-heard-of-it.
98. Stevenson M.H., Stewart E.M., McAteer N.J. A consumer trial to assess the
acceptability of an irradiated chilled ready meal. Radiat. Phys. Chem. 1995;46:785–
788. doi: 10.1016/0969-806X(95)00261-U. [Cross Ref]
99. MacRitchie L., Hunter C.J., Strachan N. Consumer acceptability of interventions
to reduce Campylobacter in the poultry food chain. Food Control. 2014;35:260–266.
doi: 10.1016/j.foodcont.2013.06.005. [PMC free article] [PubMed] [Cross Ref]
100. Gunes G., Tekin M.D. Consumer awareness and acceptance of irradiated
foods: Results of a survey conducted on Turkish consumers. LWT-Food Sci.
Technol. 2006;39:444–448. doi: 10.1016/j.lwt.2005.03.001. [Cross Ref]
101. Mehmetoglu A.C., Demirkol O. Preferences of Turkish people for irradiated, GM
or organic foods. J. Food Agric. Environ. 2007;5:74–80.
102. Parlato A., Giacomarra M., Galati A., Crescimanno M. ISO 14470: 2011 and
EU legislative background on food irradiation technology: The Italian attitude. Trends
Food Sci. Technol. 2014;38:60–74. doi: 10.1016/j.tifs.2014.04.001. [Cross Ref]
103. Byun M.W., Oh S.H., Kim J.H., Yoon Y., Park S.C., Kim H.S., Kim S.B., Han
S.B., Lee J.W. Information channel effects on women intention to purchase irradiated
food in Korea. Radiat. Phys. Chem. 2009;78:675–677. doi:
10.1016/j.radphyschem.2009.03.018. [Cross Ref]
104. Furuta M. Current status of information transfer activity on food irradiation and
consumer attitudes in Japan. Radiat. Phys. Chem. 2004;71:501–504. doi:
10.1016/j.radphyschem.2004.03.085. [Cross Ref]
105. Furuta M., Hayashi T., Hosokawa Y., Kakefu T., Nishihara H. Consumer
attitudes to radiation and irradiated potatoes at “radiation fair” in Osaka, Japan.
Radiat. Phys. Chem. 1998;52:67–71. doi: 10.1016/S0969-806X(98)00076-0. [Cross
Ref]
106. Lynch M., Nalder K. Australia’s export program of irradiated fresh produce,
Opportunities in Phytosanitary Irradiation for Fresh Produce; Proceedings of the
USDA-Chapman University Workshop; Orange, CA, USA. 25–26 March 2015; pp.
25–26.
107. Fischhoff B., Slovic P., Lichtenstein S., Read S., Combs B. How safe is safe
enough? A psychometric study of attitudes towards technological risks and benefits.
Policy Sci. 1978;9:127–152. doi: 10.1007/BF00143739. [Cross Ref]
108. Slovic P., Fischhoff B., Lichtenstein S. Rating the Risks. Environ. Sci. Policy
Sustain. Dev. 1979;21:14–39. doi: 10.1080/00139157.1979.9933091. [Cross Ref]
109. Henson S. Demand-side constraints on the introduction of new food
technologies: The case of food irradiation. Food Policy. 1995;20:111–127. doi:
10.1016/0306-9192(95)00020-F. [Cross Ref]
110. Resurreccion A.V.A., Galvez F.C.F., Fletcher S.M., Misra S.K. Consumer
attitudes toward irradiated food: Results of a new study. J. Food Protect.
1995;58:193–196.
111. Ornellas C.B.D., Gonçalves M.P.J., Silva P.R., Martins R.T. Atitude do
consumidor frente à irradiação de alimentos. Ciênc. Tecnol. Aliment. 2006;26:211–
213. doi: 10.1590/S0101-20612006000100033. [Cross Ref]
112. Junqueira-Gonçalves M.P., Galotto M.J., Valenzuela X., Dinten C.M., Aguirre
P., Miltz J. Perception and view of consumers on food irradiation and the Radura
symbol. Radiat. Phys. Chem. 2011;80:119–122. doi:
10.1016/j.radphyschem.2010.08.001. [Cross Ref]
113. Mead P.S., Slutsker L., Dietz V., McCaig L.F., Bresee J.S., Shapiro C., Griffin
P.M., Tauxe R.V. Food-related illness and death in the United States. Emerg. Infect.
Dis. 1999;5:607–625. doi: 10.3201/eid0505.990502. [PMC free article] [PubMed]
[Cross Ref]
114. Government of Canada HC Technical Summary—Health Canada’s Safety
Evaluation of Irradiation of Fresh and Frozen Raw Ground Beef—Health Canada.
[(accessed on 9 August 2016)]; Available online: http://www.hc-sc.gc.ca/fn-
an/securit/irridation/tech_sum_food_irradiation_aliment_som_tech-
eng.php#toxicology.
115. Zimmerman L., Kendall P., Stone M., Hoban T. Consumer knowledge and
concern about biotechnology and food safety. Food Technol. 1994;48:71–77.
116. Edwards K. The interplay of affect and cognition in attitude formation and
change. J. Personal. Soc. Psychol. 1990;59:202–216. doi: 10.1037/0022-
3514.59.2.202. [Cross Ref]
117. Fazio R.H., Powell M.C., Williams C.J. The role of attitude accessibility in the
attitude-to-behavior process. J. Consum. Res. 1989;16:280–288. doi:
10.1086/209214. [Cross Ref]
118. Sanbonmatsu D.M., Fazio R.H. The role of attitudes in memory-based decision
making. J. Personal. Soc. Psychol. 1990;59:614–622. doi: 10.1037/0022-
3514.59.4.614. [PubMed] [Cross Ref]
119. Bruhn C.M. Strategies for communicating the facts on food Irradiation to
consumers. J. Food Protect. 1995;58:213–216.
120. Laminack J., Dainello F., Vestal T.A., Wingenbach G. Experiential education
employed to demystify food irradiation as a viable technology for food industry
professionals. Hort Technol. 2006;16:318–323.
121. Groth E.I. Communicating with consumers about food safety and risk issues.
Food Technol. 1991;45:248–253.
122. Rogers E.M. Diffusion of Innovations. 4th ed. Free Press; New York, NY, USA:
2003. pp. 206–208.
123. Ehlermann D.A.E. The RADURA-terminology and food irradiation. Food
Control. 2009;20:526–528. doi: 10.1016/j.foodcont.2008.07.023. [Cross Ref]
124. Codex Stan. Codex General Standard for the Labelling of Prepackaged Foods.
[(accessed on 4 October 2016)]. Available online:
http://www.fao.org/docrep/005/y2770e/y2770e02.htm.
125. US Environmental Protection Agency Food Irradiation: Food Labeling.
[(accessed on 2 October 2016)]; Available online: https://www.epa.gov/radiation.
126. Government of Canada, Canadian food inspection agency Irradiated Foods.
[(accessed on 2 October 2016)];2012 Available online:
http://www.inspection.gc.ca/food/labelling/food-labelling-for-industry/irradiated-
foods/eng/1334594151161/1334596074872#tc_req2.
127. Ministry of Primary Industries Ministry for Primary Industries—Food Safety.
[(accessed on 2 October 2016)];2013 Available online:
http://www.foodsafety.govt.nz/
128. He S., Fletcher S., Rimal A. Attitudes, acceptance and consumption: The case
of beef irradiation. J. Food Distrib. Res. 2011;36:65–70.
129. Slovic P. Perceived risk, trust, and democracy. Risk Anal. 1993;13:675–682.
doi: 10.1111/j.1539-6924.1993.tb01329.x. [Cross Ref]
130. Roberts P., Hénon Y.M. Consumer response to irradiated food: Purchase
versus perception. Stewart Postharvest Rev. 2015;3:1–6. doi: 10.2212/spr.2015.3.5.
[Cross Ref]

XXVI. Unmasking of Olive Oil Adulteration Via a Multi-

Sensor Platform
Sensors (Basel). 2015 Sep; 15(9): 21660–21672.
Marco Santonico,1 Simone Grasso,1,* Francesco Genova,2 Alessandro Zompanti,1
Francesca Romana Parente,3 and Giorgio Pennazza1
Arun Bhunia, Academic Editor

Abstract

Methods for the chemical and sensorial evaluation of olive oil are frequently changed
and tuned to oppose the increasingly sophisticated frauds. Although a plethora of
promising alternatives has been developed, chromatographic techniques remain the
more reliable yet, even at the expense of their related execution time and costs. In
perspective of a continuous increment in the number of the analyses as a result of
the global market, more rapid and effective methods to guarantee the safety of the
olive oil trade are required. In this study, a novel artificial sensorial system, based on
gas and liquid analysis, has been employed to deal with olive oil genuineness and
authenticity issues. Despite these sensors having been widely used in the field of
food science, the innovative electronic interface of the device is able to provide a
higher reproducibility and sensitivity of the analysis. The multi-parametric platform
demonstrated the capability to evaluate the organoleptic properties of extra-virgin
olive oils as well as to highlight the presence of adulterants at blending
concentrations usually not detectable through other methods.

Keywords: olive oil authentication, olive oil adulteration, artificial sensorial system,
food quality control, gas analysis, liquid analysis, BIONOTE (BIOsensor-based
multisensorial system for mimicking Nose, Tongue and Eyes)
1. Introduction

Olive oil is the most popular vegetable oil produced and consumed in Mediterranean
countries. According to international standards [1], olive oils have to be obtained
exclusively from the fruit of the olive tree (Olea europaea) using cold pressing
techniques and in conditions that do not alter the organoleptic properties of the oil at
all. Current European Union regulation [2] and the International Olive Committee
(IOC) require olive oils to be graded in function of sensory assessment and three
fundamental chemical parameters: free acidity, peroxide value, and UV absorbance
[2]. By comparing oils scores with threshold values, these are classified as extra
virgin olive oil (EVOO), virgin olive oil, and other low-quality olive oil typologies. Olive
oil is a very complex matrix [3,4]. The main compounds are triacylglycerols and fatty
acids contributing to 94%–96% of their total weight. However, triacylglycerols and
fatty acid contents show a broad variability in olive oils chemical composition and
this is largely dependent on both cultivar and geographical origin [5]. Recently, the
authentication of products labeled as olive oil has become a fundamental issue for
either commercial or health aspects [6,7]. In fact, the high price of olive oil and its
increased popularity as a potential health food have made it an ideal target for frauds
[8]. Common olive oil adulterations include accidental contaminations during
production stages, deliberate mislabeling of less expensive oil categories and, more
often, the admixtures of expensive olive oils with low quality oils. Although advances
in knowledge and technology have undoubtedly led to greater success over frauds,
even more complex forms of adulteration have been developed to invalidate the
usefulness of official methods, thus leaving the authenticity verification still an
unsolved matter [9]. Actually, no rapid and universal method exists that is officially
recognized for all the authenticity issues [10]. Liquid and gas chromatographic
techniques represent the elective methods for the authentication and
characterization of individual olive oil compounds [11,12,13,14,15]. Nevertheless,
these analytic verifications require valuable instrumentation and highly-qualified
staff. All of these features together make authentication a time consuming and
expensive process which is not applicable as routine analysis. In this context the
BIONOTE (BIOsensor-based multisensorial system for mimicking Nose, Tongue
and Eyes), a recently developed sensor platform [16], has been employed. The
system, which embeds gas and liquid sensors having a common biologically-derived
sensing interface, allows the simultaneous analysis of the vapor and liquid phase of
the samples. As a consequence, the integrated multi-sensorial platform led different
sensors to catch more comprehensive information which, in turn, requires a further
elaboration through multivariate data analysis techniques. At the end of the analytical
procedure, similarities and differences between the samples are highlighted. In this
multi-parametric study, the correct discrimination of twelve EVOOs made up of
dissimilar olive cultivars and having different geographical origin has been achieved.
Furthermore, the high sensitivity and reproducibility of the analysis, which were
guaranteed by the innovative electronic interface of the system, permitted the
detection of fraudulent admixing of extraneous vegetable oils (pomace, soybean,
sunflower seeds, and peanut oils) up to concentrations lower than 5%. These
promising results altogether present BIONOTE as a rapid and economic tool for
high-throughput screening analysis.
2. Materials & Methods

2.1. Oil Samples

Twelve EVOO samples, indicated in the paper as EVOO #1, #2, #3, and so on, were
obtained from twelve different Italian orchards. Several characteristics of the oils are
reported (Table 1). The commercial EVOO as well as the pomace, soybean,
sunflower seeds, and peanut oils were bought at a local market.

Table 1 General EVOOs specifications.

Oil Geographical Year of

Oil Variety
Sample Origin Production
EVOO #1 Laterba 2013/2014 Picoline
EVOO #2 Castellaneta 2013/2014 Leccino
EVOO #3 Laterba 2013/2014 Picoline (organic)
EVOO #4 Laterba 2013/2014 Arbequina (organic)
Grottaglie and Picoline (50%), Nociara (35%),
EVOO #5 2013/2014
Crispiano Leccino (15%)
EVOO #6 Crispiano 2013/2014 Leccino
EVOO #7 Grottaglie 2013/2014 Ogliarola
EVOO #8 Grottaglie 2013/2014 Picoline
EVOO #9 Grottaglie 2012/2013 Cellina di Nardò
EVOO
Laterba 2013/2014 Leccino
#10
EVOO
Crispiano 2012/2013 Cellina di Nardò
#11
EVOO
Crispiano 2012/2013 Cima di Melfi
#12

2.2. Gas Analysis

Quartz Micro Balances (QMBs) with six functionalized piezoelectric sensors were
used as transducers for the gas sensor array as already described [16]. In order to
perform homogeneous gas measurements the following experimental set-up was
used. A volume of 2 mL for each olive oil sample was placed in a 50 mL glass flask
and kept for 10 min at room temperature to obtain an adequate headspace.
Dehumidified reference air was pumped into the sensors chamber at a flow rate of
3 L/min for 10 min to desorb any volatile trace from sensors surface before every
measure. Oil samples were analyzed five times, setting a sampling interval of 90 s.

2.3. Liquid Analysis

Electronic interface and sensors employed in the liquid analyses were the same
described in Santonico et al. Cyclic voltammetry in the range from −1 to 1 V was
performed using a triangular function at 10 mHz and a sampling interval of 1 second.
Olive oil samples for liquid sensor analysis were prepared following the procedure
reported below. Briefly, a volume of 1 mL of oil was poured into a tube with 3 mL of
methanol 70% (v/v) and mixed vigorously for 1 min. The vial containing the oil-
alcohol emulsion was centrifuged for 5 min at 1000 RCF and 4 °C to separate the
two phases efficiently. Finally, the methanol phase was collected and stocked in ice
up until the analysis.

2.4. Chemical Quality Control Analyses

Polyphenol content, free acidity, peroxide value, ∆K, and refractive index of olive oil
samples have been assessed following the standard chemical testing methods [15].
Briefly, polyphenol content was evaluated by means of Folin-Ciocalteu method,
according to the procedure reported by Singleton and Rossi [17]. Free acidity content
[18] was evaluated, dissolving the samples in a mixture of equal parts by volume of
ethyl ether (95%) and ethyl alcohol, thus titrating with an ethanolic solution of
potassium hydroxide, using phenolphthalein as indicator. Results were reported as
grams of oleic acid per 100 g of oil. To determine the peroxide value [19], oil samples
were dissolved in chloroform and glacial acetic acid, then a solution of potassium
iodide was added, leaving the mixture incubating for five minutes in the dark, and
finally a titration of the generated iodine with a standard sodium thiosulphate
solution, using starch solution as indicator, was performed. The peroxide value was
expressed in terms of milliequivalents of active oxygen per kilogram able to oxidize
potassium iodide under the operating conditions. The quality of the olive oils
employed in this study was also assessed measuring the absorption bands between
200 and 300 nm [20]. Samples were dissolved in iso-octane to obtain 1% (w/v)
solutions and the specific absorbance at 232 and 270 nm with reference to pure
solvent was determined. These absorptions were expressed as specific extinctions,
conventionally indicated by K. Finally, a ∆K value was calculated relating the
maximum recorded absorbance at 270 nm against the absorption of surrounding
spectral region (±4 nm). The refractometric index of olive oils was determined using
the Abbé refractometer, paying attention to correct the recorded value on a
temperature basis. Three independent parameter’s determinations were carried out
for each test sample. All the reagents used in this study were of certified analytical
quality.

2.5. Data Analysis

Multivariate data analysis: Principal Component Analysis (PCA) and Partial Least
Square Discriminant Analysis (PLS-DA), was performed using PLS-Toolbox
(Eigenvector Research Inc., Manson, WA, USA) in the Matlab Environment (The
MathWorks, Natick, MA, USA). PLS-DA models have been calculated in order to
detect EVO adulteration and investigate BIONOTE relevance to the chemical
parameters.
3. Results

3.1. Olive Oil BIONOTE Characterization

Twelve Italian EVOOs having different geographical origin and olive variety
compositions have been characterized through the BIONOTE system, performing
five measuring cycles each. Gas analysis was performed on EVOOs without any
modification of the samples. Volatile compounds released in the system headspace
at room temperature were characterized through their interaction with the
functionalized sensors, resulting in a reproducible pattern response (Figure 1). Olive
oil as such is not applicable for electrochemical analysis due to the absence of
conductivity and the high viscosity of the media. Therefore, oil samples underwent
liquid extraction with methanol and the deriving alcoholic fractions were analyzed by
the liquid sensor (Figure 1). Cyclic voltammetry in the range from −1 to 1 V was
performed using a triangular function at 10 mHz and a sampling interval of 1 second.
By means of this setup, an array of 100 virtual sensor responses has been obtained
from one physical sensor for each voltammetric measuring cycle. Finally, a data
fusion of the information deriving from the last three measuring cycles of gas and
liquid sensors was accomplished. The obtained data set has been evaluated by
Principal Component Analysis (PCA) and the ability of the system to sharply
discriminate the twelve EVOOs was demonstrated. The score plot of the first two
Principal Components (PCs), accounting for 76.94% of the explained variance, is
reported (Figure 2). Ten of the twelve oil samples clustered in three separate regions
along the Principal Component 2 (PC2). EVOOs #1, #6, and #12 formed a group in
the bottom part of the plane. EVOOs #5, #8, #10, and #11 distributed in a second
area at the interception of the two PCs. EVOOs #2, #4, and #9 clustered in the upper
portion of the plane (Figure 2). Nevertheless, within the groups almost every oil
sample can be discriminated from the others along the Principal Component 1 (PC1).
EVOOs #3 and #7 were distinguished from the rest of the analyzed samples by
positioning at the upper end and at the left edge of the plane, respectively (Figure
2). Additionally, a Partial Least Square Discriminant Analysis (PLS-DA) model using
the leave one out criterion has been calculated showing a correct classification rate
of 100% for the twelve different EVOOs (five independent repetitions each).
BIONOTE characterization of different EVOO samples. Liquid (left panels) and gas
(right panels) fingerprints.

Figure 1 BIONOTE characterization of different EVOO samples. Liquid (left panels)

and gas (right panels) fingerprints.

Score Plot of the first two principal components deriving from the data fusion of the
BIONOTE liquid and gas sensors responses.

Figure 2 Score Plot of the first two principal components deriving from the data fusion
of the BIONOTE liquid and gas sensors responses.

3.2. Olive Oil Chemical Characterization

To assess the quality of the EVOOs, common chemical analyses were also
performed. All the EVOOs got parameters satisfying the imposed normative limits,
even though some slight differences between the samples were found (Table 2),
thus supporting BIONOTE discrimination evidence. Free acidity and ∆K values were
significantly lower than normative standard ones being, however, slightly different
among each other. The refractive index of the twelve oil samples was almost the
same, while the peroxide parameter showed the greatest variability. The obtained
results confirmed the excellent quality of the oil samples, highlighting the absence
(in terms of usual parameters) of significant differences between the EVOOs
themselves.

Table 2 EVOO purity and quality characteristics according to the International

Olive Council [1].

Oil Free Acidity (mg/100 g Peroxide Value (mEq Refractive

∆K
Sample Oleic Acid) O2/Kg) Index

EVOO #1 3.4 ± 0.1 15.0 ± 0.4 0.0020 1.469

EVOO #2 3.4 ± 0.1 12.2 ± 0.1 0.0045 1.468

EVOO #3 4.9 ± 0.2 6.0 ± 0.1 0.0065 1.468

EVOO #4 2.0 ± 0.1 6.9 ± 0.1 0.0015 1.467

EVOO #5 7.3 ± 0.1 8.7 ± 0.1 0.0015 1.468

EVOO #6 6.0 ± 0.1 9.5 ± 0.3 0.0005 1.467

EVOO #7 5.3 ± 0.1 7.2 ± 0.4 0.0030 1.467

EVOO #8 4.3 ± 0.2 18.1 ± 0.2 0.0045 1.468

EVOO #9 2.8 ± 0.1 9.9 ± 0.2 0.0035 1.468

EVOO
3.9 ± 0.1 13.5 ± 0.4 0.0015 1.467
#10

EVOO
6.1 ± 0.2 9.4 ± 0.5 0.0030 1.467
#11

EVOO
3.1 ± 0.2 9.9 ± 0.3 0.0160 1.467
#12

3.3. Olive Oil Adulteration

A commercial EVOO was bought at local market and mixed with four vegetable oils
(pomace, soybean, sunflower seeds, and peanut oils) at different blending
concentrations (1.25%, 5%, 10%, and 25% (v/v)). The prepared EVOO’s admixtures
were characterized through the BIONOTE system, performing five measuring cycles
each. Sophisticated EVOO samples were treated as already described (see
Materials & Methods section) before being analyzed through either the liquid or the
gas sensors. A comprehensive array containing the overall sensors’ responses was
built for each EVOO sophistication independently and the collected data were further
analyzed using multivariate data analysis techniques. The calculated PLS-DA
models highlighted the ability of the system to distinguish an authentic EVOO from
an adulterated one in all the tested cases, showing also a rather high degree of
efficiency in the concentration discrimination (Figure 3). BIONOTE was able to
predict the presence of contaminating lower-grade oils up to concentration values
lower than 10% (v/v). The Root Mean Square Error in Cross Validation (RMSECV),
using the Leave One Out criterion, was slightly different among the four kinds of
sophistication. System performance was almost the same for the soybean, sunflower
seeds, and peanut oils with RMSECV ranging from 2.1% to 4.4%, while the
discrimination of the pomace oil sophistications resulted less precise accounting for
an error of 8.3% (v/v) (Figure 3).

Calculated PLS-DA model for the prediction of contaminating oils concentration.

Calibration model has been built using a commercial EVOO sophisticated with 0%–
25% (v/v) of (a) soybean oil; (b) sunflower seeds oil; (c) peanut oil; and (d) pomace
oil. RMSECV associated with the models are reported.

Figure 3 Calculated PLS-DA model for the prediction of contaminating oils

concentration. Calibration model has been built using a commercial EVOO
sophisticated with 0%–25% (v/v) of (a) soybean oil; (b) sunflower seeds oil; (c)
peanut oil; and (d) pomace ...

3.4. BIONOTE Relevance to the Chemical Parameters

BIONOTE relevance to the measured chemical parameters have been investigated

by calculating four different models to predict polyphenols content, free acidity,
peroxide value, and TEAC on the gas and liquid sensor array data. The results
obtained are very promising (see Figure 4, panel a: polyphenols; panel b: free
acidity; panel c: peroxide value; and panel d: TEAC).

Measured versus predicted (PLS-DA model based on BIONOTE data) values of (a)
polyphenols; (b) free acidity; (c) peroxide value; and (d) TEAC.

Figure 4 Measured versus predicted (PLS-DA model based on BIONOTE data)

values of (a) polyphenols; (b) free acidity; (c) peroxide value; and (d) TEAC.

4. Discussion

Adulteration is a common problem usually related to high-value products. As a

consequence of the fundamental role in the Mediterranean diet and the documented
nutraceutical effect [6], EVOO represents a clear target for sophistication aimed to
trade. According to recent studies, adulteration is becoming an escalating issue for
olive oil in the market with consequences undermining the quality attributes of the
product and sometimes even its safety consumption [21]. Although reliable and
accurate analyses intended to guarantee olive oil quality in the broadest sense
already exist, these are not routinely used. While chemical parameters as free
acidity, peroxide value, ∆K, and refractive index are necessary to define if an olive
oil fulfills the requirements to be labeled and marketed as EVOO, these constraints
are not sufficient for authenticity verification in the most of cases [1,22]. Fraudulent
olive oil admixtures are usually chemically corrected to meet international standards,
thus requiring more complex analyses to be recognized as adulterations.
Nevertheless, even when official analytical methods are applied to screen olive oil
samples, olives’ biological differences, due to geographical origin and genetic
aspects, sometimes generate problems to distinguish between sophistications and
authentic EVOOs [23]. So far, numerous modern techniques have been proposed to
support or replace official standard methods in the task of olive oil authentication
[10,24,25,26,27,28]. However, those do not offer clear advantages yet, because
their adulteration detection limits, being usually greater than 10% of contamination,
are worse in comparison with chromatographic techniques’ ones. In this study, a
novel system able to characterize EVOOs in terms of genuineness and authenticity
has been presented. The BIONOTE platform takes advantage of either liquid and
gas analysis to accomplish a multi-parametric characterization, giving
comprehensive information about the sample [17]. The overall sensors’ responses
are elaborated through multivariate data analysis techniques to highlight similarities
and differences, resulting in a correct classification rate of 100%, even when similar
EVOOs have been analyzed. Hence, BIONOTE showed the ability to discriminate
between twelve Italian EVOOs originating from different Apulian neighboring olive
tree orchards. The result highlighted the capability of BIONOTE not only to identify
EVOOs against lower grade olive oils, but also to discriminate between EVOOs
obtained from different olive cultivars. This is a notable outcome because this issue
is usually addressed via more complex genetic approaches. The innovative
electronic interface, providing to the system a higher reproducibility and sensitivity
comparable to similar devices [29,30,31,32], allowed BIONOTE to be also
successfully employed in the authenticity verification process, with admixtures
percentage thresholds below the best levels reported by literature. BIONOTE was
challenged with different kind of EVOO sophistications, covering concentrations
lower than 10% (v/v), and in all cases it was able to distinguish authentic oil from an
adulterated one. The system detected the presence of fraudulent admixing of
extraneous vegetable oils (soybean, sunflower seeds and peanut oils) up to
concentrations lower than 5%. However, when the pomace oil was used, system
performance decreased. This discrepancy, leading to an increment of the detection
limit to about 8%, could be probably explained by the shared origin between EVOO
and pomace oil. Considering the demand of EVOO traceability and safety claimed
by both producers and consumers, BIONOTE represents a potential solution. In fact,
the BIONOTE system is able to address the EVOO authenticity issue focusing not
only on the labeling control but also the genuineness of the oil, accounting for
geographical origin and olive varieties composition at the same time.

5. Conclusions

Nowadays, global markets and international regulations have increased significantly

the number of samples that require validation, raising the necessity of rapid
analytical methods. In this context, BIONOTE could represent a real opportunity
thanks to its reduced time of analysis. However, due to the profiling approach on
which the system is based on, BIONOTE has not been intended to replace the high
specificity of the official chromatographic methods. Hence, it is proposed as a rapid
tool for preliminary high-throughput screening, aimed to detect samples that require
further analytical verifications. This workflow has been designed to reduce the
employment of high-value instrumentation and qualified personnel only to specific
cases, thus decreasing the costs, while maintaining the elevated number of samples
analyzed.

Author Contributions

Marco Santonico and Giorgio Pennazza conceived and designed the experiments
and analyzed the data; Simone Grasso and Francesco Genova performed the
experiments and wrote the paper; Alessandro Zompanti and Francesca Romana
Parente contributed analysis tools.

Conflicts of Interest

The authors declare no conflicts of interest.

References

1. International Olive Council (IOC) Trade Standards Applying to Olive Oils and
Olive-Pomace Oils. International Olive Council (IOC); Madrid, Spain: 2013.
COI/T.15/NC No 3/Rev. 7.
2. European Communities (EC) Commission Implementing Regulation (EU) No
1348/2013. Official Journal of the European Union (Regulation No 1348/2013
L338/31, 16 December 2013) [(accessed on 28 August 2015)]. Available online:
http://www.beuth.de/en/regulation/euv-1348-2013/200370127.
3. Aparicio R., Harwood J. Handbook of Olive Oil Analysis and Properties. 2nd ed.
Springer; New York, NY, USA: 2013.
4. Angerosa F., Campestre C., Giansante G. Olive Oil: Chemistry and Technology.
2nd ed. AOCS Publishing; New York, NY, USA: 2006. Analysis and authentication.
5. Gómez-Rico A., Salvador M.D., Moriana A., Pérez D., Olmedilla N., Ribas F.,
Fregapane G. Influence of different irrigation strategies in a traditional cornicabra cv.
Olive orchard on virgin olive oil composition and quality. Food Chem. 2007;100:568–
578. doi: 10.1016/j.foodchem.2005.09.075. [Cross Ref]
6. Harwood J.L., Yaqoob P. Nutritional and health aspects of olive oil. Eur. J. Lipid
Sci. Technol. 2002;104:685–697. doi: 10.1002/1438-
9312(200210)104:9/10<685::AID-EJLT685>3.0.CO;2-Q. [Cross Ref]
7. Kamal-Eldin A., Moreau R.A. Gourmet and Health-Promoting Specialty Oils.
AOCS Publishing; New York, NY, USA: 2009.
8. Zhang J., Zhang X., Dediu L., Victor C. Review of the current application of
fingerprinting allowing detection of food adulteration and fraud in china. Food
Control. 2011;22:1126–1135. doi: 10.1016/j.foodcont.2011.01.019. [Cross Ref]
9. Aparicio R., Aparicio-Ruiz R., García-González D., van Amerongen A., Barug D.,
Lauwars M. Rapid methods for testing of oil authenticity: The case of olive oil. Rapid
Methods. 2007;10:163–188.
10. Aparicio R., Morales M.T., Aparicio-Ruiz R., Tena N., García-González D.L.
Authenticity of olive oil: Mapping and comparing official methods and promising
alternatives. Food Res. Int. 2013;54:2025–2038. doi:
10.1016/j.foodres.2013.07.039. [Cross Ref]
11. American Oils Chemists’ Society (AOCS) Determination of Fatty Acids in Olive
Oils by Capillary GLC. American Oils Chemists’ Society (AOCS); New York, NY,
USA: 2009. Official Method Ch 2–91.
12. International Olive Council (IOC) Determination of Trans-Unsaturated Fatty
Acids by Capillary Column Gas Chromatography. International Olive Council (IOC);
Madrid, Spain: 2001. COI/T.20/Doc. No 17/Rev. 1.
13. International Olive Council (IOC) Determination of Wax Content by Capillary-
Column Gas Chromatography. International Olive Council (IOC); Madrid, Spain:
2003. COI/T.20/Doc. No 18.
14. International Olive Council (IOC) Determination of Aliphatic Alcohols Content by
Capillary Gas Chromatography. International Olive Council (IOC); Madrid, Spain:
2003. COI/T.20/Doc. No 26.
15. International Olive Council (IOC) Determination of the Composition and Content
of Sterols and Triterpene Dialcohols by Capillary Column Gas Chromatography.
International Olive Council (IOC); Madrid, Spain: 2013. COI/T.20/Doc. No 30.
16. Santonico M., Pennazza G., Grasso S., Amico A., Bizzarri M. Design and test of
a biosensor-based multisensorial system: A proof of concept study. Sensors.
2013;13:16625–16640. doi: 10.3390/s131216625. [PMC free article] [PubMed]
[Cross Ref]
17. Singleton V.L., Rossi J.A. Colorimetry of total phenolics with phosphomolybdic-
phosphotungstic acid reagents. Am. J. Enol. Vitic. 1965;16:144–158.
18. American Oils Chemists’ Society (AOCS) Acid Value. American Oils Chemists’
Society (AOCS); New York, NY, USA: 2009. Official Method Cd 3d-63.
19. American Oils Chemists’ Society (AOCS) Peroxide Value Acetic Acid-Isooctane
Method. American Oils Chemists’ Society (AOCS); New York, NY, USA: 2011.
Official Method Ch Cd 8b-90.
20. American Oils Chemists’ Society (AOCS) Determination of Specific Extinction of
Oils and Fats, Ultraviolet Absorption. American Oils Chemists’ Society (AOCS); New
York, NY, USA: 2009. Official Method Ch 5–91.
21. García-González D.L., Aparicio R. Research in olive oil: Challenges for the near
future. J. Agric. Food Chem. 2010;58:12569–12577. doi: 10.1021/jf102735n.
[PubMed] [Cross Ref]
22. International Olive Council (IOC) Global Method for the Detection of Extraneous
Oils in Olive Oils. International Olive Council (IOC); Madrid, Spain: 2006.
COI/T.20/Doc. No. 25.
23. Mignani A.G., García-Allende P.B., Ciaccheri L., Conde O.M., Cimato A., Attilio
C., Tura D. Comparative Analysis of Quality Parameters of Italian Extra Virgin Olive
Oils According to Their Region of Origin; Proceedings of the SPIE; San Francisco,
CA, USA. 3 March 2008.
24. Agiomyrgianaki A., Petrakis P.V., Dais P. Detection of refined olive oil
adulteration with refined hazelnut oil by employing nmr spectroscopy and
multivariate statistical analysis. Talanta. 2010;80:2165–2171. doi:
10.1016/j.talanta.2009.11.024. [PubMed] [Cross Ref]
25. Calvano C.D., Ceglie C.D., D’Accolti L., Zambonin C.G. Maldi-tof mass
spectrometry detection of extra-virgin olive oil adulteration with hazelnut oil by
analysis of phospholipids using an ionic liquid as matrix and extraction solvent. Food
Chem. 2012;134:1192–1198. doi: 10.1016/j.foodchem.2012.02.154. [PubMed]
[Cross Ref]
26. Chiavaro E., Vittadini E., Rodriguez-Estrada M.T., Cerretani L., Bendini A.
Differential scanning calorimeter application to the detectionof refined hazelnut oil in
extra virgin olive oil. Food Chem. 2008;110:248–256. doi:
10.1016/j.foodchem.2008.01.044. [PubMed] [Cross Ref]
27. Maggio R.M., Cerretani L., Chiavaro E., Kaufman T.S., Bendini A. A novel
chemometric strategy for the estimation of extra virgin olive oil adulteration with
edible oils. Food Control. 2010;21:890–895. doi: 10.1016/j.foodcont.2009.12.006.
[Cross Ref]
28. Poulli K.I., Mousdis G.A., Georgiou C.A. Rapid synchronous fluorescence
method for virgin olive oil adulteration assessment. Food Chem. 2007;105:369–375.
doi: 10.1016/j.foodchem.2006.12.021. [Cross Ref]
29. Apetrei C., Apetrei I.M., Villanueva S., de Saja J.A., Gutierrez-Rosales F.,
Rodriguez-Mendez M.L. Combination of an e-nose, an e-tongue and an e-eye for
the characterization of olive oils with different degree of bitterness. Anal. Chim. Acta.
2010;663:91–97. doi: 10.1016/j.aca.2010.01.034. [PubMed] [Cross Ref]
30. Cosio M.S., Ballabio D., Benedetti S., Gigliotti C. Geographical origin and
authentication of extra virgin olive oils by an electronic nose in combination with
artificial neural networks. Anal. Chim. Acta. 2006;567:202–210. doi:
10.1016/j.aca.2006.03.035. [Cross Ref]
31. Escuderos M.E., Sánchez S., Jiménez A. Virgin olive oil sensory evaluation by
an artificial olfactory system, based on quartz crystal microbalance (QCM) sensors.
Sens. Actuators B Chem. 2010;147:159–164. doi: 10.1016/j.snb.2010.01.071.
[Cross Ref]
32. Haddi Z., Alami H., El Bari N., Tounsi M., Barhoumi H., Maaref A., Jaffrezic-
Renault N., Bouchikhi B. Electronic nose and tongue combination for improved
classification of moroccan virgin olive oil profiles. Food Res. Int. 2013;54:1488–
1498. doi: 10.1016/j.foodres.2013.09.036. [Cross Ref]

XXVII. Gas Chromatography Analysis with

Olfactometric Detection (GC-O) as a Useful
 Methodology for Chemical Characterization of
Odorous Compounds
Sensors (Basel). 2013 Dec; 13(12): 16759–16800.
Magda Brattoli, Ezia Cisternino, Paolo Rosario Dambruoso, Gianluigi de Gennaro,*
Pasquale Giungato, Antonio Mazzone, Jolanda Palmisani, and Maria Tutino

Abstract

The gas chromatography-olfactometry (GC-O) technique couples traditional gas

chromatographic analysis with sensory detection in order to study complex mixtures
of odorous substances and to identify odor active compounds. The GC-O technique
is already widely used for the evaluation of food aromas and its application in
environmental fields is increasing, thus moving the odor emission assessment from
the solely olfactometric evaluations to the characterization of the volatile
components responsible for odor nuisance. The aim of this paper is to describe the
state of the art of gas chromatography-olfactometry methodology, considering the
different approaches regarding the operational conditions and the different methods
for evaluating the olfactometric detection of odor compounds. The potentials of GC-
O are described highlighting the improvements in this methodology relative to other
conventional approaches used for odor detection, such as sensoristic, sensorial and
the traditional gas chromatographic methods. The paper also provides an
examination of the different fields of application of the GC-O, principally related to
fragrances and food aromas, odor nuisance produced by anthropic activities and
odorous compounds emitted by materials and medical applications.

Keywords: gas chromatography-olfactometry, odors, fragrances, food aromas,

sensorial methodology, materials, odor detection port, volatile compounds

1. Introduction

Odors have a direct effect on human behaviors and can significantly affect the quality
of life. Evolutionary history has demonstrated the importance of a good sense of
smell as it protected our primitive ancestors from predators and helped them to find
food. Nowadays, it may not have the same connotation for human survival, but it can
most certainly play an important role in human attractions, memories, and emotions.
People associate odors with past experiences and, from those experiences, they will
involuntarily assess the odor as likable, dislikable or indifferent. These responses
are individual and may vary from person to person. Through breathing, humans are
continuously testing air quality to obtain relevant information such as potential
dangers (e.g., smoke), the presence of food, another individual, and so on. In order
to detect an odor, several factors and properties can contribute to generate an
olfactory perception [1]:
  - Odor threshold (OT). This is the minimum concentration at which 50% of a
human panel can detect the presence of an odor or odorant without
characterizing the stimulus. This is different from the recognition threshold
which is the concentration that 50% of a human panel is able to detect and
describe qualitatively.
 - Physical and chemical properties. These include the appreciable volatility of
a substance at ordinary temperatures (less than 300–400 relative molecular
mass) to permeate the air near the sensory area, as well as the slight water-
solubility which allows an odor to pass through the mucous layer to the
olfactory cells and the lipid-solubility which is necessary since olfactory cilia
are composed primarily of lipid material.
 - Intensity. This is the relative strength of the odor above the recognition
threshold. It is logarithmically related to odorant concentration (Stevens' law
or the power law) which can be calculated with the following equation [2]:

I=KlogC

(1)

where I is the intensity, C is the concentration and K is a constant.

 - Hedonic tone. This is a measure of the pleasantness or unpleasantness of

an odor mixture.
 - Quality. This property identifies an odor and differentiates it from another
odor of equal intensity.
 - Molecular structure. Molecular geometry, in particular, the composition and
structure of the functional groups within a molecule, can deeply affect the
quality and features of an odor [3].

It should be noted that odors are complex mixtures of many volatile chemicals which
are present in different concentrations.
These chemicals can interact synergistically or additively in the mixtures according
to unpredictable rules [4–6], and are at the basis of the overall sensation of smell.

In the recent years, various research activities have been developed in different
scientific disciplines to investigate odors and odor perception with the aim of
producing direct applications in industry. Some applications focus on improving the
quality of scented products such as perfumes [7] and food products through the use
of natural or synthetic odorous compounds. For example, the results of an aroma
food analysis can help assess the quality of foods and make them more pleasing
and desirable [8]. Others examine the impact of odorous compounds on the quality
of human life and the environment. In this case, odors may derive from industrial
activities [9–13] (landfills, wastewater treatment plants, refineries, tanneries, and so
on) or be emitted from materials that people frequently use [14,15].
The main purpose of odor research is to identify the odor active compounds and to
relate them to human perception. Instrumental approaches to the characterization of
odorants using Gas Chromatography coupled with Mass Spectrometry (GC/MS)
have been widely used to produce lists of the substances present and their
concentrations [16,17]. The main limitation of this technique is connected to the
complexity of the odor under investigation. Since many volatile chemicals are often
present at concentrations lower than the instrumental detection limit and since
information about human perception is not provided, a linear correlation between a
quantified substance and an olfactory stimulus cannot be made [18].
Notwithstanding the usefulness of GC/MS analyses, the mammalian olfactory
system is the most sensitive and inclusive odor detector. Hence, the sensory
evaluation of smells by panels of sensory trained evaluators represents a valid
approach to odor assessment.

A notable improvement in odor identification consists of coupling GC-MS with

olfactometric detection (GC-MS/O). The gas chromatographic separation of an
odorous air sample can be useful for identifying specific odorant components. GC-
MS/O, thus, allows a better understanding of odorant composition through the
identification and quantification of its compounds while offering a partial correlation
between the chemical nature of an odorant and its perceived smell [19,20].

This paper will examine the state of the art of the gas chromatography-olfactometry
methodology (GC-O), different approaches regarding operational conditions, and
different methods for evaluating the olfactometric detection of odor compounds. The
potentials of GC-O will be described and its different fields of application such as
fragrance and food aromas, industrial and material emissions, and medical
applications will be examined.

2. GC-O Analysis

2.1. GC-O Configuration and Parameters Affecting the Analysis

GC-Olfactometry (GC-O) is a valuable method for the selection of odor components

from a complex mixture. A properly trained human assessor or a team of them is
employed as a detector and can sniff the eluate in order to detect the presence of
odor—active compounds via a specifically designed odor port (ODP) connected in
parallel to conventional detectors [21] (thermal conductivity, photo-ionization or
flame-ionization), as shown in Figure 1.
Scheme of the gas chromatograph equipped with an olfactometric detector
(reprinted from [22] with
6666666666666666666666666666666
permission from Elsevier).

Figure 1. Scheme of the gas chromatograph equipped with an olfactometric detector

(reprinted from [22] with permission from Elsevier).

Each separated compound, eluted by the GC, can be detected by a human assessor
(odor present or not), who is able to measure the duration of the odor activity (start
to end), to describe the quality of the odor perceived and to quantify its intensity. GC-
O in combination with a mass spectrometer (GC-O/MS) not only enables the
evaluation of odor compounds, but also their identification with mass spectral
information.

In particular, the flow of the eluate is split so that the analytes reach both detectors
simultaneously, permitting a comparison of both signals. The retention times of the
analytes might differ for the two detectors (typically shorter for the mass
spectrometer), due to the fact that the mass spectrometer works under vacuum
conditions while the olfactometric detector works under atmospheric pressure
conditions. This difficulty can be overcome by installing a restrictor, a narrow bore
capillary, before the mass spectrometer to increase the pressure drop between the
interface and the flow splitter, as well as through careful selection of the flows of the
carrier and auxiliary gases [23].

The design of all commercially available olfactometric ports is very similar. The
eluate reaches the port through an uncoated transfer line (deactivated silica
capillaries) and is sniffed in a glass or a PTFE conical port, fitted to the shape of a
nose. The transfer line is heated to prevent the condensation of semi volatile
analytes on the walls of the capillary. Auxiliary gas (moist air) is added to the eluate
to prevent the drying of the assessors' nose mucous membranes, which could cause
discomfort especially in longer analyses. The transfer line length can vary, but it must
be long enough to ensure a comfortable sitting position and to avoid discomfort due
to the vicinity of hot chromatograph components during detection. Each port is also
equipped with an electric push-button to generate a signal of 1 V when pressed. If
the extract analyzed is sufficiently concentrated, the eluate stream can sometimes
be separated into several streams and delivered to more olfactometric ports, for the
simultaneous detection by several assessors (Figure 2). This approach can yield
more representative results since different sniffers simultaneously smell the sample
providing an average value for each analysis [24,25].
Figure 2. Scheme of the GC/MS-O multi-sniffing system (reprinted from [25] with
permission from Elsevier).

A multi-gas chromatography-olfactometry device (eight way gas chromatography

olfactometry 8W-GC-O) has recently been developed [26]. This device consists of a
chromatograph coupled with a divider that synchronously distributes the volatile
component outflow to eight transfer lines which are connected to eight sniffing ports
located in separate booths. Flow rates can be set to ensure the best compromise
between chromatographic resolution and sufficient effluent at the sniffing ports so
that the odorous fractions can be adequately sensed.

There are several factors that determine the quality of the data collected by GC-O.
The method used to extract volatile compounds from the samples determines the
composition of the extract, and therefore the quality of the eluate available for
perception. The set-up of the GC instrument and of the separation conditions affects
the quality of chromatography and the response of the human detector. The peak
shape affects the perception of odor intensity and the calculation of detection
thresholds. Chromatographic behavior of odor substances varies with the
compounds and the stationary phases of the GC column. Non polar stationary
phases enable the elution of odor-active volatiles at the lowest possible temperature
[21]. However, very polar molecules, such as fatty acids, result in poor peak shapes
using nonpolar phases. Polar phases demonstrate greater selectivity, although the
overall quality of the separation will depend upon the composition of the sample [27].

Moreover, it should be noted that some volatile compounds are labile and will readily
decompose in heated injector blocks, forming artifacts; for example, sulfur
compounds are particularly susceptible to heat-induced decomposition. The odor
character of some compounds depends strongly on their concentration which can
become important in the case of poor chromatography or poor chromatographic
separation (co-elution). Many key odor compounds often occur at very low
concentrations in complex matrixes. Therefore, the identification of odor compounds
remains a hard task even with GC-O/MS because some compounds co-elute with
other analytes making the correlation between the detected aroma and the correct
compound difficult.

Comprehensive two-dimensional gas chromatography (GC × GC) appears to be the

most appropriate choice to meet the need for enhanced separation and better
sensitivity. This technique is based on the continuous collection of an effluent from
a GC column and the periodic reinjection of small portions of the effluent to a second
column of different polarity. This novel technique, which associates the resolution
power of GC × GC with the selectivity and sensitivity of the human olfactory system,
can enable the olfactive analysis of congested chromatographic areas [28–30].
However, GC × GC coupled with ODP is an extremely demanding technique for the
operator because peaks are eluted very quickly and the panelist may not have
enough time to both recognize odors and provide their descriptors. Respiratory rate
can also be critical and an analysis time of more than ten minutes is difficult to
achieve.

A direct gas chromatography–olfactometry (D-GC-O) method can be used to

perform a representativeness test on the global odor of a sample and select the best
conditions for analysis. This recent technique consists of connecting a deactivated
capillary column between the GC injector and the sniffing port to avoid
chromatographic separation. In this way, the aroma compounds arrive
simultaneously at the ODP, where the assessor perceives, evaluates and compares
the resulting odor of the starting sample [31,32].

The behavior of the human detector should also be considered during the
assessment of the data quality collected when using GC-O. There are very
significant differences in olfactory ability between humans; odor thresholds can vary
significantly among individuals and some people, with an otherwise normal sense of
smell, are unable to detect families of similar smelling compounds (anosmia) [21]. In
addition, the olfactory response of an individual is known to vary over time, even
during the course of a single day, and with the speed of breathing [33,34]. Sensitivity
may also fluctuate due to health status and mood.

In order to obtain reproducible data, potential assessors should be screened for

sensitivity, motivation, ability to concentrate, and ability to recall and recognize odor
qualities [21]. They should be asked to refrain from smoking and eating/drinking
strongly flavored foods for 1 h prior to performing GC-O. They also should not to
wear aftershave, perfume or strong deodorants on the day of assessment [21,35].
Moreover, the assessor's comfort and ability to sniff free of distraction should be
considered. GC-O instrument should be located in a dedicated laboratory with
temperature and pressure control. Finally, a maximum sniff time of 25–30 min is
recommended since GC-O sniff duration can impact human detector performance
[21].

2.2. GC-O Olfactometric Detection Methods

In recent decades, several techniques have been developed to collect and process
GC-O data and to estimate the sensory contribution of a single odor active
compound, rated as intensity, in order to evaluate the relative influence on the total
odor of the sample [36]. These methods can be categorized into three groups:
frequency detection methods, dilution to threshold methods, and direct intensity
methods.

2.2.1. Frequency Detection Methods

The frequency detection method involves a team of 6–12 people who analyze the
same sample in order to provide the percentage of people who sensed the odor
compound at a given retention time [37,38]. Each odor can be evaluated using Nasal
Impact Frequency (NIF) or Surface of Nasal Impact Frequency (SNIF) values. The
NIF value is set at a value of one when each of the evaluators sensed a given odor,
and zero when no-one sensed any odor at a given retention time [39]. Therefore, the
NIF value corresponds to the peak height of the olfactometric signal. The SNIF
values describes the peak areas obtained multiplying the frequency percentage by
duration, and enables the production of an aromagram (see Figure 3) [24,38,40].
Simplicity is the main advantage of detection frequency-based methods which do
not necessitate qualified evaluators. The methods are repeatable and the results
reflect the differences in sensitivity between the evaluators, which can also be
related to differences within a given population. Hence, the impact of inattention,
specific anosmia, etc. on the aromagram is minimized. However, as the detection
frequency is related to the intensity of the odor perceived by the assessors, the
correlation of the peak height with the real concentration of the odor compound in
the sample cannot be obtained. In particular, odorous compounds present in
different concentrations, all above the detection threshold, will produce an
aromagram with peaks of equal intensity [21,22,41].

Figure 3. Scheme of an aromagram obtained using detection frequency methods,

with four evaluators (reprinted from [22] with permission from Elsevier).
2.2.2. Dilution to Threshold Methods

Dilution to threshold methods provide a quantitative description of the odor potential

of a given compound based on the ratio between its concentration in the sample and
its sensory threshold in air [21]. These methods consist of preparing a dilution series
of an extract, usually using twofold, threefold, fivefold or 10-fold dilution levels (R)
and then analyzing them with GC-O [42]. The assessors state under which dilution
the compound analyzed can still be sensed, and usually describe the type of smell.
Odor potency is equivalent to the concept of “aroma values”, “odor values”, “odor
units”, “flavor units”, and “odor activity values” (OAV) [21]. OAV is the most
commonly used index and represents the ratio of the concentration of a given
compound on its sensory detection threshold [43–46].

The most frequently reported dilution methods are “Aroma Extract Dilution Analysis”
(AEDA) and “Combined Hedonic Aroma Response Measurement”
(CharmAnalysis™) [24,47–57]. AEDA measures the highest sample dilution at which
the odor of the analyzed compound is still detectable and reports this as the flavor
dilution factor (FD) [36]. If the last dilution under which the analyte is still detectable
is equal to p (p = 0, 1, 2, 3, …), then its dilution factor is RP, where R is the dilution
level [42]. The overall results obtained with this method are reported in an
aromagram presenting the FD value, or its logarithm, against the retention index (RI)
[58,59]. CharmAnalysis™ records the duration of odors (start and end) and
generates chromatographic peaks. The assessors record the start and the end of
each detected odor, hence the aromagram is obtained by plotting the duration of the
odor sensation against the dilution value. The peak areas are expressed in
dimensionless “Charm” values (C), which are proportional to the amount of the
analyte in the sample, and inversely proportional to the sensory detection threshold
[22]. Charm value can be calculated with the formula:

C=Rn−1
(2)

where n is the number of coincident odor responses detected at a single retention

index and R is the dilution level [60]. CharmAnalysis™ considers both the peak width
and the peak shape, thus a short and broad peak may have the same Charm value
as a tall and narrow peak that is perceived at a higher dilution. This gives
CharmAnalysis™ more discriminating power than AEDA, but it also results in greater
variation than AEDA [21].

A drawback of the dilution methods is the length of the total analysis due to the large
number of dilutions for each extract and evaluator. Therefore, these methods are
normally performed by only one or two evaluators. For example, in a series of 10
dilutions, 30 GC injections are needed, which requires about 2 weeks [61]. Dilution
to threshold methods are also criticized for the underlying false assumption that the
odor intensity increases in parallel with the concentration for all odor components in
a sample [62]. In order to attain a complete analysis of key odorants, recombination
models and omission experiments can be performed. In the former, the aroma model
system for a specific sample is prepared based on the combination of previously
achieved AEDA or CHARM values, and/or OAVs. Odorants showing higher values
are used to formulate a recombined model, which is then compared to the real
sample for similarity or difference [25]. The omission experiments, on the other hand,
deal with the preparation of an aroma model for a specific sample in which one or
more odorants are omitted. In this experiment, the panelists are asked to perform
duo and triangle tests to compare the reduced model with the complete one and
indicate the perceived sensorial differences [25,31,63].
2.2.3. Direct Intensity Methods

The odor intensity and its duration can be measured with direct intensity methods
using different kinds of quantitative scales: category scales or unstructured line
scales [21]. These methods include a single time-averaged measurement registered
after the elution of the analyte (posterior intensity evaluation methods) or a dynamic
measurement, where the appearance of an odor, its maximum intensity and decline
are registered in a continuous manner (OSME, the Greek word for odor, and Finger
span method) [64–66]. In the first case, the assessor assigns an appropriate value,
from a previously defined intensity scale, to each detected compound while in the
second case, the olfactogram obtained is similar to conventional chromatograms in
which the height of the peak corresponds to the maximum odor intensity and the
width corresponds to odor duration. Depending on the method, the measurement
can be performed in different ways, best results will be obtained if a panel of
assessors is used, and the average panel result is treated as one signal. There is a
correlation between the logarithm of the odor intensity obtained using the OSME
method and the logarithm of the analyte concentration, as described by Steven's
Law [29,62]. The intensity is recorded as a function of time by moving the cursor of
a variable resistor [25,66]. In the finger span method, the olfactograms are built
moving the potentiometer slider using the thumb and the index or middle finger (195
mm) [21,34,67]. The distance between the two is proportional to the intensity of the
odor, and the time of sliding corresponds to the duration of the odor in the
olfactometric port. Some studies demonstrated that even a completely unprepared
team of evaluators is able to repeatedly perform olfactometric measurements [22].
Another device, based on the use of a spring-loaded button can be used to build the
aromagram. This device can relate a perceived odor intensity to the physical
stimulus of hand pressure, thereby improving the reliability of recorded odor intensity
data [41,68]. Using a 3- or 7-point category scale with half values, the data are
processed using the modified frequency MF (%), that can determine the most
important odorous compounds present in the sample. It can be calculated with the
following formula:

MF(%)=F(%)×I(%)−−−−−−−−−−√
(3)

where F (%) is the detection frequency of an aromatic attribute expressed as

percentage and I (%) is the average intensity expressed as a percentage of the
maximum intensity. Usually, odorous stimuli detected with a MF (%) higher than 50
represent the most important compounds present in each sample [31,69]. This group
of methods requires expert evaluators in order to obtain fast, repeatable and
generally consistent results even in a single run [21].

2.3. Sample Preparation Methods

The choice of an appropriate sample preparation method is crucial in GC-O

analyses. The flavor profile is closely related to the isolation procedure which should
prevent decomposition of labile compounds, loss of highly volatile compounds and
heat-induced artifact formation. Therefore, the selected method should yield a
product which is as representative as possible of the sample. According to the
properties of the investigated sample, the preparation may include mincing,
homogenization, centrifugation, steam distillation (SD), solvent extraction (SE),
simultaneous distillation-extraction (SDE), solid phase extraction (SPE), supercritical
fluid extraction (SFE), Soxhlet extraction, solvent assisted flavor evaporation
(SAFE), microwave-assisted hydrodistillation (MAHD), headspace (HS) techniques,
solid-phase microextraction (SPME), matrix solid-phase dispersion (MSPD) and/or
methylation, direct thermal desorption (DTD), among others. Conventional solvent
extraction and/or distillation methods are widely used to isolate volatile organic
compounds present in food, beverages and materials [32,49,51,52,59,70–72].
However, these procedures yield extracts of a sample that do not always reflect the
composition of the odor that is perceived by a subject when smelling or eating the
sample. In particular, highly volatile compounds that most contribute to the original
odor of foods, plants, flowers or materials can be lost during these procedures [73].

Solid phase extraction is carried out by shaking the sample with resin particles or
more simply by eluting the sample in SPE columns [45,46]. The SPE resins are
usually washed and conditioned with different solvents before sample extraction or
clean up [28,74]. Another very popular method is SAFE, which may be applied after
SE techniques or be used as an individual extraction method for aqueous samples
such as milk, fruit and urine [49]. The technique removes volatile compounds under
low temperature and high vacuum conditions. Headspace techniques are used for
the sampling of volatiles, using either static or dynamic methods. Sorption traps with
porous polymers such as Tenax® TA or Porapak™ Q, and resins such as Lichrolut®
EN, are most often used to concentrate collected volatiles [63,66,75,76]. The volatile
components are then chemically or thermally desorbed from the trap and analyzed
with GC-O [66,75]. Compared to conventional extraction techniques, headspace
methods have the benefit of usually not causing the loss of the most volatile
compounds and enable chromatographic analysis of these compounds without
interferences due to the solvent peak. Headspace volatiles can be also concentrated
exposing SPME fibers coated with specific extraction phases such as
Carboxen®/PDMS or divinylbenzene/carboxen/polydimethylsiloxane
[30,31,47,48,50,77–82]. The chemical profile of the collected volatiles depends upon
the type, thickness and length of the fiber, as well as on the sampling time and
temperature.

Another technique, worthy of note is the direct thermal desorption (DTD), a widely
applied solvent-free method [83–87]. For this method, volatile compounds are
collected onto adsorption tubes filled with specific resins by using an air pump. The
resins are selected according to the type of matrix (Tenax® GR, Carbograph™,
Carboxen® GR). Sampled tubes are then thermally desorbed and then analyzed with
GC-O [84–87]. Alternately, bags made of inert materials (Nalophan®, Tedlar®,) are
used to collect air samples for analysis with TD-GC-O [83].

3. Applications
Because of the its unique features, GC-O is already well established in areas
involving fragrance and food aromas and is becoming more and more common in
areas related to the environment, medicine and materials. In this section, these
principal areas of application are explored and a sampling of results are presented
in order to highlight the potentials of this technique.

3.1. Food Application

Consumers select and consume food based on three principal properties: flavor,
appearance (color) and texture. Flavor is usually divided into the subsets of taste
and smell, which are perceived in the mouth and the nose, respectively [60]. It is
also defined as the sensation arising from the integration or interplay of signals
produced as a consequence of sensing smell, taste, and irritating stimuli from food
or beverage [88]. Instead, odor usually refers to the smell of food before it is put into
the mouth (nasal perception) while aroma is the retronasal smell of food in the mouth.

The application of GC-MS in this area has marked a real turning point for flavor
research. Indeed, the number of known flavors has increased to over 7,000
compounds [89], however, there is no information about odor active components.
Gas chromatography in combination with olfactometric techniques (GC-O) can help
to detect potent odorants, without knowing their chemical structures, even at very
low concentrations and this makes it the only viable method for the selection of
aroma-active components from a complex mixture.

GC-O studies on food products focus essentially on three main issues:

 (1)

The “aroma profile” of various foods and beverages and the dependence
between the odor and the chemical composition of the volatile fraction on
these products;

 (2)

The odor changes in food due to processing techniques (fermentation,

cooking, the addition of preservatives and flavorings);

 (3)

The discrimination among a family of foodstuffs (cheese types, coffee).

3.1.1. The Study of “Aroma Profiles”

In recent years, intensive food “aroma profiles” studies have been carried out with
the aim of characterizing the substances responsible for odor. In particular, the
characterization of the odor-active compounds in different kinds of fruit has been the
topic of many papers reported in literature [90–97]. In several studies, Pino et al.
have investigated the odor-active compounds in fruits such as banana, guava and
pineapple [47,98,99]. In these investigations, volatile compounds were extracted
from the fresh fruit homogenate headspace using SPME fiber coatings and then
introduced in successive sequences into the GC port [100–106]. Moreover, the
volatile compounds were examined by isolating the volatile compounds using
Simultaneous Distillation-Extraction (SDE) [107–109]. A trained panel of three
assessors perceived and evaluated the global odor of the fruits by performing SPME
direct gas chromatography (GC-O). The combination of SPME-GC-O and SDE-GC-
O detected thirty-one odor-active compounds; eleven of which were reported for the
first time as important odorants of banana fruit. Guava fruit volatiles included more
than 100 compounds that had been reported in previous studies [110–115], as well
as ethyl acetate which had not been previously reported as a major compound.
Pineapple fruit volatiles included esters (51), aldehydes (7), alcohols (5), acids (3),
terpenes (2), furans (2) and miscellaneous compounds (9) which had all been
reported in previous studies, with the exception of methyl 2-methylbutanoate [116–
120].

This work reveals two emerging tendencies in GC-O applications: the improvement
of repeatability and reliability of the obtained results. These developments were
achieved by unifying, simplifying and shortening procedures, especially those
involving sample preparation (SPME avoids the use of solvents and the resulting
artifacts) and the integration with other extracting techniques, such as SDE. SPME
allows the isolation of high and medium volatile compounds whereas SDE can cause
their losses and, consequently, an underestimation of their aroma contribution.
Hence, the combination of SPME-GC-O and SDE-GC-O can be a way of overcoming
this discrepancy in the evaluation of the contributions of volatiles.

Camembert cheese has been studied using aroma extract concentration analysis
(AECA) and headspace gas chromatography-olfactometry (HGC-O). This approach
revealed the complexity of an odorant matrix in which the most potent odorants are
2,3-butanedione, 3-methylbutanal, methional, 1-octen-3-ol, 1-octen-3-one,
phenethyl acetate, 2-undecanone, decalactone, butyric acid and isovaleric acid
[121]. The work identified the neutral odorants with the highest OAVs as
methanethiol, methional and dimethyl sulphide which contributed to the garlic-like
sensory attribute in the odor profile of Camembert. Instead, 1-octen-3-ol and the
corresponding ketone were found to be responsible for the mushroom-like sensory
attribute while acetic, butyric and capric acid were associated with the acidic sensory
attribute.

Using two strategies, a recent work [122] has applied GC-O to the characterization
of the aroma active compounds in black truffles (Tuber melanosporum) and summer
truffles (Tuber aestivum). The first approach used aroma extract dilution analysis
(AEDA), while the second involved a GC-O technique combining measurements of
both intensity and modified detection frequency [123–125]. Eighteen different odor
zones, with a MF higher than 15%, were detected in the GC-O experiments. The
newly identified components were relevant compounds in the aroma composition of
the two truffles, particularly in the case of the summer truffles. In fact, 1-hexen-3-one
was amongst the five most important aroma compounds. The olfactogram obtained
from black truffle was more intense and complex than that of the summer truffle. In
addition, the results suggested that there were relevant differences between the
aroma profiles of both varieties. In particular, while DMS, DMDS and 3-methyl-1-
butanol were among the five most important aroma compounds in both cases, black
truffle aroma was rich in 2,3-butanedione and ethyl butyrate while summer truffle
aroma contained methional. The comparison between the GC-O profiles of both
varieties is shown in Figure 4 where a spider web diagram of the data (normalized
so that the maximum = 100%) reveals that the most important differences are related
to methional (c.11) and 3-ethylphenol (c.17) which were much richer in the GC-O
profile of summer truffle, and 2,3-butanedione (c.3), ethyl butyrate (c.4), ethyl 3-
methylbutyrate (c.5) and 3-ethyl-5-methylphenol (c.12), which were particularly
important in the aroma profile of black truffle.

Figure 4. Spider web diagram comparing the GC-O olfactometric profiles

(normalized so that the odorant showing maximum MF (%) = 100) obtained from
black and summer truffles (reprinted from [122] with permission from Elsevier).

Black truffle was particularly rich in phenols (3-ethyl-5-methylphenol, 5-methyl-2-

propylphenol, given as mass of 3-propylphenol and 3-ethylphenol, respectively) and
in β-phenylethanol, while the emissions of summer truffle was mostly a product of β-
phenylethanol, DMS and 3-ethylphenol.
The characterization of aroma-impact compounds in yerba mate (YM) using GC-O
and GC-MS was carried out to clarify consumer preferences of major commercial
brands of YM sold in Uruguay and its neighboring countries (Argentina, Brazil and
Paraguay) [69,126]. For this purpose, all the samples studied were extracted with an
identical extraction system and experiments were performed with a system that
represented an “artificial mouth”. Approximately 50 odorants were detected during
the GC-O experiments but, for simplicity, those not reaching a maximum GC-O
modified detection frequency (MF) of 50%, were considered as noise. Sixteen odor-
active compounds presenting MF ≥ 50 [127] were detected (Table 1).

Table 1. Odor-active compounds in yerba mate detected by GC–O with MF ≥

50 (reprinted from [69] with permission from Elsevier).

LRIa %MFb Descriptor Compound

1154 53 Herbaceous, sweet Myrcene

1278 76 Citrus Octanal

1294 53 Mushroom 1-Octen-3-one

1340 74 Moss 6-Methyl-5-hepten-2-one

1450 68 Flower (Z)-linalool oxide (furanoid)

1478 76 Nuts (E,Z)-2,4-heptadienal

1490 68 Nuts (E,E)-2,4-heptadienal

1510 84 Mushroom (E,Z)-3,5-octadien-2-one

1528 66 Mushroom (E,E)-3,5-octadien-2-one

1540 65 Flower Linalool

1745 61 Flower Geranial

1798 76 Sweet Nerol

1832 71 Apple β-Damascenone

1843 76 Flower α-Ionone

1936 79 Sweet β-Ionone

1987 53 Oxidized, metallic (E)-4,5-epoxy-(E)-2-decenal

aLRI in Carbowax™;
bMF, modified frequency.

Intensive studies have been carried out using gas chromatography with olfactometric
detection (GC-O) to evaluate the sensory activity of the individual odorous
components of different alcoholic beverages. In these cases, sample preparation
represents a critical step since exhaustive extraction methods, such as solvent
extraction and distillation, are time consuming, involve many steps and do not always
reflect the composition of the odor reaching the receptors during their actual
consumption. Moreover, during the concentration step, oxidation of volatiles may
occur without the use of antioxidants. Isolation methods are the most used, as static
and dynamic headspace with purge and trap (on Tenax® TA or Porapak™ Q, as well
as resins, such as Lichrolut® EN), followed by thermal desorption or solvent elution,
or Solid Phase Microextraction (SPME) [22]. In alcoholic beverages, GC-O is most
commonly used to investigate odor compounds in order to reconstruct alcoholic
beverage odors, check the quality of the raw materials used in the production
processes and identify the compounds responsible for the aftertaste [128].

3.1.2. The Study of Odor Changes in Food after Technological Treatments

Another interesting area of research concerns the generation of different aroma

compounds in foods as the result of technological treatment and the consequent
changes of sensory characteristics. In one study, Feng et al. set out to evaluate the
crucial impact of fermentation on soy sauce aroma [77]. An aroma extract obtained
with SPME of a harvested koji sample was subject to GC-O analysis. The results
detected 2-phenylpropenal and di-epi-α-cedrene for the first time and concurred with
previous findings that had noted the presence of volatile [129–131] or aromatic
compounds [132–134] as well as.

GC-O combined with the aroma extract dilution analysis (AEDA) approach [59,135]
was used to determine key odorants after 1 and 5 days of fermentation and the
subsequent frying of soy tempeh. The volatile compounds isolated from the tempeh
represented different chemical classes, mainly aldehydes and ketones,
hydrocarbons, mono and sesquiterpenes, sulfur containing compounds, nitrogen
containing compounds, alcohols, and furans.

A recent study has investigated a new alternative antioxidant in meat originating from
a natural plant source [48]. In this work, Kim et al. studied the effects of the addition
of two commercial rosemary extracts (RE), namely RMD (oil-soluble type) and RMP
(water-soluble type), on potent odorants in cooked beef extracts (BE) using solid-
phase-micro-extraction-gas chromatography-olfactometry (SPME-GC-O) [136]. The
sensory evaluation results indicated that the addition of RE influenced the odor
character of cooked BE and thus analysis using SPME-GC-O coupled with AEDA
was subsequently conducted to identify the odorants responsible for this trend. A
combined total of 57 odorants including 10 unknowns were detected in the
treatments. Kim et al. concluded that two effects, namely odor-supplementation and
odor-suppression, were triggered by the addition of RE. The odor-supplementation
consisted in an addition of sweet and floral notes to BE; these odorants were mainly
esters, terpenes and phenolic compounds which had originated from the RE, and
most notably from the RMD. This effect was thought to have either caused pleasant
flavors or masked certain off-flavors in the cooked beef. Instead, the odor
suppression effect was mainly observed for the odorants generated from lipid
oxidation or the Maillard reaction [137–143]. The study highlighted that CG-O could
be a useful tool to monitor and optimize some treatments.

Another interesting study addressing the use of GC-O in evaluating the effect of
technological treatments on odor characteristics was carried out on cooked pork
products [75]. This study set out to obtain new knowledge about how nitrite helps to
develop aroma in cooked pork products and to describe the reaction mechanisms
involved [144–151]. For this purpose, two complementary GC-MS/O instruments
were used to identify odorant compounds (GC-MS/8O and GC × GC-MS/O)
[26,152]. Using GC-MS/8O, the authors identified 22 odorant zones that
subsequently were explored in detail with GC × GC-MS/O for a reliable identification
of odorant compounds in the nitrited or nitrite-free cooked ham headspaces. Among
the detected compounds, several oxidation products were identified. Indeed, meat
fatty acid oxidation explained the presence of the numerous saturated or unsaturated
aldehydes with 6–10 atoms of carbon, ketones and alcohols [153,154]. In addition,
the authors revealed sulfur compounds resulting from the breakdown of sulfur-
containing precursors during cooking [155]. A comparison of the aromagrams of
nitrite-free (Figure 5a) and nitrite-cured cooked hams (Figure 5b) is reported,
showing that the odor intensity of most of the olfactory peaks was weakened by the
added nitrite. Moreover, the authors concluded that an analysis of the differences
observed indicates that the zones of the aromagrams most significantly changed (p
< 0.01) by nitrite correspond to fatty acid oxidation products (green peaks in Figure
5).
Figure 5. GC-MS/8O aromagrams of cooked hams without (a) and with (b) nitrite expressed
in mean intensities of perception, each calculated from 16 individual sniffing sessions (one
type of ham × 8 sniffers × 2 repeats). The breakdown of the signal into three classes of
chemical origin shows the odorant zones originating from: lipid oxidation (in green), sulfur
compound degradation (in red) and unspecified origins (in grey). (reprinted from [75] with
permission from Elsevier).

In conclusion, the authors demonstrated that the addition of nitrited salt in ham
production is not directly involved in the production of the odorous substances that
give nitrite cured pork products their specific aroma. Indeed, the absence of nitrite
simply promotes the oxidation of fatty acids and, in particular, the production of
aldehydes which will mask the odor of the sulfur-containing compounds responsible
for the aromatic note typical of nitrite-cured pork products. Thus, the odor of nitrite-
cured pork products is merely the outcome of a balanced perception of certain sulfur-
containing compounds and fatty acid oxidation products, important among which are
aldehydes.

GC-O analysis, based on a detection frequency method has also been used to
describe the aroma attributes of beef like flavors (BFs). This has been carried out to
determine aroma-active compounds and identify volatile compounds in oxidized
tallow samples [156] as well as to clarify the influence of enzymatic hydrolysis-mild
thermal oxidation on the odor produced to obtain oxidized tallow. GC-MS profiles of
oxidized tallow were analyzed together with quantitative descriptive sensory data
and GC-O responses of BFs to understand which compounds had significant effects
on aroma-active compounds and sensory attributes of BFs. Through the analyses,
the characteristic flavor precursors from enzymatic hydrolysis thermal oxidation
tallow were identified and the main differences between enzymatic hydrolysis-
thermal oxidation tallow and simple thermal oxidation tallow were elucidated.
Compounds with detection frequencies greater than 50% were considered
characteristic flavor components; a total of 34 aroma-active compounds were
identified, mainly consisting of heterocyclic sulphur or nitrogen compounds and
aldehydes. Among these compounds, the most potent odorants were acetic acid,
nonanal, 3-(methylthio)propionaldehyde, 2-methyl-3-furanthiol and bis(2-methyl-3-
furyl)disulphide. The study also identified bis(2-methyl-3-furyl)disulphide and 2-
methyl-3-furanthiol, the latter which is responsible for beef like aroma had also been
detected in previous studies [154,157–164].

A similar study was carried out for the evaluation of the changes in the aroma
characteristics of mutton process flavors (MPFs) prepared from sheep bone protein
hydrolysates (SBPHs) with different DHs (degrees of hydrolysis) using descriptive
sensory analysis (DSA) and analyzing the corresponding volatile odor-active
compounds with GC-MS/O [165]. The results showed 58 odor-active compounds
while on the basis of the detection frequency method only 36 of these possessed an
odor activity in the MPFs. The developed odorous compounds varied according to
the starting sample and depending on the degree of hydrolysis. DH was an important
index in the preparation of meat flavors; in particular, compounds with a DH range
of 25.92%–30.89% produced a wider range of odor-active compounds through
thermal reaction.

GC-O analysis was also used to discriminate between fresh and frozen lamb meat
in a study attempting not only to evaluate the chemical basis of the aroma of grilled
lamb but also to determine the major aroma changes linked to meat freezing [166].
Experiments were performed using a multidimensional gas chromatography–
olfactometry–mass spectrometry (GC × GC-O/MS). The experimental design
included the GC-O analysis of nine different extracts: two blanks, two samples of
fresh grilled meat, two of fresh grilled muscle, two of previously frozen grilled meat
and one of fresh grilled fat. The main advantage of this methodology compared with
others presented in recent years [167] is that it was performed in vivo with the help
of volunteers; therefore, the relevant processes occurring inside the mouth such as
chewing and enzyme action from saliva were taken into account. This work was able
to detect the most important compounds in the aroma of lamb during grilling and
consumption and for the first time 2-isopropyl-3-methoxypyrazine, 2-
methylbenzaldehyde and vanillin were detected in lamb.

GC-O with statistical analysis was applied to aromatic caramel which is widely used
in the food industry especially as a food flavoring. Caramel can be liquid or solid and
brown to dark brown. It is soluble in water and obtained by the controlled action of
heat on sugars [168,169]. Products resulting from the thermal degradation of sugars,
like coffee, and odorant properties are closely linked to the volatile fraction which
represent 5%–10% of their total mass; GC-O and GC-MS could be used to identify
and structurally characterize those compounds. The relationship between
physicochemical and sensory data sets was studied by means of multivariate
statistical tools such as Partial Least Square (PLS) regression. The odorant
compounds detected belonged to several chemical classes: oxygenated
heterocycles, carbocyclic compounds, carboxylic acids, phenolic compounds,
esters, aldehydes and carbonyled compounds. Among them, 13 odor zones were
related to oxygenated heterocycles, carbocyclic compounds and carboxylic acids.
These results suggested the importance of these three chemical classes to the
odorant properties of burnt sugars.

Extensive reviews dealing with the application of GC-O on dairy products, including
tables of classified odorants according to their chemical classes are present in
literature [25,170]. In one study, the change in aroma composition of cow's milk
during heating and fermentation was investigated [19]. Seven common odorants
were found in four different types of raw milk: dimethylsulfone, ethyl butanoate, ethyl
hexanoate, heptanal, indole, nonanal, and 1-octen-3-ol. Heating brought about the
formation of four common odor-potent compounds: hexanal, 2-nonanone,
benzothiazole, and δ-decalactone. Instead, fermentation resulted in the formation of
1-octen-3-one, methional, 3-methylbutanal, and butyric acid.
3.1.3. GC-O as a Discrimination Tool among a Family of Foodstuffs

Discriminant analysis between foods and among varieties of the same class of foods
is another of the applications for GC-O. Several studies have used GC-O to verify
odorant compounds in coffee and to compare the highly volatile odorants of the
powders and brews prepared from roasted Arabica and Robusta coffees. A brew
was also obtained from a soluble coffee powder to detect odorants with a boiling
point lower than that of the extraction solvent [171]. Furthermore, GC-O was
successfully used for the aroma analysis of coffee flavor by evaluating defects in
green coffee beans due to microorganisms responsible for the formation of off-
flavors [172]. In this study, the authors compared green Arabica coffee beans from
Mexico, obtained by a dry post-harvest treatment, with a coffee bean of identical
origin having no noticeable organoleptic defect and then a moldy/earthy character
defined. A frequency detection method was applied for obtaining the olfactogram.
GC sniffing profiles made it possible to locate zones with a typical moldy/earthy
character and to attribute moldy/earthy off-flavor to six substances.

In another study, GC × GC-O was extensively used to check odor compounds in

complex samples such as brewed coffee. This led to the detection of numerous odor
compounds which could be well resolved and identified in both roasted and brewed
coffee [28]. The authors clearly demonstrated that certain odor regions correspond
to several overlapping compounds in 1D GC, however, these co-eluting peaks were
well resolved in the 2D axis of GC × GC analysis. Moreover, it was found that a short
15 m 1D column could exacerbate the resolution problem in 1D analysis and a better
resolution could be obtained on a longer column. However, the latter would add time
and aggravate the problems of assessor's fatigue when an olfactometric port is used.

An interesting study has recently been carried out [173] on 7 different semi-hard
French cheeses. The work investigated the relationship between physicochemical
and sensory data sets using multivariate statistical tools. It began by identifying
thirteen odor attributes. A characterization was made using a conventional AFNOR
sensory profile to correlate the perceived orthonasal aroma compounds with those
extracted from the headspace and analyzed with GC-O [174,175]. Only 9 of the 13
odor attributes showed any significant difference among the seven cheeses
(analyzed with ANOVA). As illustrated in Figure 6, the “Smoked”, “Dairy”, “Nutty” and
“Melted Cheese” attributes did not have enough discriminant power to differentiate
the 7 cheeses.
Figure 6. Mean ratings of the 13 odor attributes for the seven (C1–C7) semi-hard cheeses
(13 judges; 3 repetitions). Significant differences are shown: * significant at p < 5%; **
significant at p < 1%; *** significant at p < 0.1% (reprinted from [173] with permission from
Elsevier).

In the second phase, the odor-active compounds were detected using 8-way gas
chromatography-olfactometry (GC-O/8). The single-port GC-MS/O chromatogram
and the GC-O/8 aromagrams were aligned, by means of standards, and processed
with AcquiSniff software [26,173,176]. It was found that some odor zones were
present in all of the cheeses while some were specific to only one or a few cheeses.
ANOVA was carried out on the intensity ratings from the eight judges for each odor-
active compound and it was noted that a total of 15 compounds significantly
discriminated the cheeses. Evaluation of the correlations between the sensory
profile and GC-O was performed using PLS with a nonlinear iterative partial least
squares (NIPALS) algorithm to explain sensory ratings from the GC-O intensities.
The total intensity of GC-O data was considered as the X matrix and the mean
sensory scores as the Y matrix. The optimal number of dimensions for model
prediction was determined by cross-validation. The results of the bi-plot correlation
between the sensory profile and the GC-O data are reported in Figure 7. The sensory
attributes, “Buttery”, “Cream”, “La vache qui rit” (cheese type), and “Raw Milk”, were
correlated with odorant compounds having green (E-2-nonenal and 1-nonen-3-one)
and roasted (2,3-dimethylpyrazine) characteristics. The attributes, “Musty”, “Animal”,
“Sweaty” and “Acidic”, were mostly correlated with odorant compounds having
cheesy, sulphury, rancid and chocolate characteristics as butanoic acid and 3-
methyl-1-butanol.

Figure 7. Bi-plot of the two first components as a result of PLS analysis of the sensory
profiles (Y matrix, black) and the GC-O intensity measurements for the odor-active
compounds (X matrix, grey) (reprinted from [173] with permission from Elsevier).

The work underlines that the use of a simultaneous multi-sniffing detection system
aims to overcome a well-known drawback of the GC-O technique, that is the lack of
repeatability and reliability of some measurements with respect to instrumental ones,
especially when only one assessor is employed.

3.2. Fragrance Applications

Chemical characterization methods coupled with olfactometric detection can be a

powerful tool for the sensorial characterization of odor. In the field of fragrances, this
method can be used for research purposes or for improving industrial manufacturing
processes. The use of fragrance materials dates back to antiquity, when spices and
resins from animal and plant sources were used in perfumery. Today, perfumers
work with several thousand natural or synthetically manufactured ingredients to
create different fragrance compositions [177]. The combination of commercial
demands with the development of monodimensional gas chromatography (GC),
GC/mass spectrometry (GC/MS), and GC-olfactometry (GC-O) has produced an
explosive acceleration of the evolution of flavor and fragrance materials [29].
Essential oils, extracted from the source array, or perfumes used as such, are
subject to quality control as well as chemical and odorous characterization in industry
where the main goal is to make these products as pleasant as possible and to confer
a characteristic odor. Hence, GC-O can be readily used because of its ability to
efficiently separate and characterize the principal molecules constituting these
matrix such as terpenes, aldehydes and alcohols.

For fragrance recognition, innovative FFNSC (flavor and fragrance natural and
synthetic compounds) libraries are used; the best matche can be found between an
investigated compound and the target one based on similarities in fragmentation and
the closeness of the RI values [25,178]. In the case of particularly complex mixtures
or when compounds are present at trace-level concentrations, multidimensional gas
chromatography techniques (GC × GC) are preferred in order to obtain a better
separation of the different compounds. The main advantage of using GC × GC-O is
shown in Figure 8. Indeed, the comparison of a GC-O chromatogram and a GC ×
GC-O 2D plot of a commercial perfume has revealed that these type of matrixes are
very complex and that conventional GC-O analyses cannot adequately record the
presence of all constituents [29].
Figure 8. GC-O chromatogram (A) and GC × GC-O 2D plot (B) of a commercial
perfume achieved without (A) and with (B) cryogenic modulation (reprinted from [29]
with permission from Elsevier).

Using GC-O analysis, detailed reports of the chemical and aroma components of
some essential oils have been compiled. For example, GC-O analyses revealed
several compounds related to the camphoraceous, herbaceous and fresh odor that
characterizes the essential oil of Tarchonanthus camphoratus L. It is notable that the
majority of the compounds having the highest intensity score belonged to the class
of oxygenated monoterpenes which are considered the most expressive class of
terpenes used in perfumery [179].

While studying P. mirifica, a commercially available Thai leguminosae plant

considered to be a rejuvenating drug, α-necrodol, a terpene having anti-insect
activity in the Pueraria genus, was detected for the first time by Yagi et al.
[57,180,181]. Moreover, odor analysis allowed to distinguish green odor (C9
aldehydes group, such as phenylacetaldehyde and (2E)-nonenal) and sweet odor
(monoterpene alcohols, such as geraniol). Hence, the main advantage of GC-O
analysis lies in the possibility of assigning a characteristic odor to the various
compounds present in the same species of plants, in order to obtain information for
a possible use of these aromatic samples in food products or in medicinal or
cosmetic applications [182].

AEDA analysis has been performed for odor characterization in essential oils.
Relevant AEDA results were found in the GC-O characterization of Scutellaria
laeteviolacea essential oil: relative flavor activity (RFA) was calculated using the
equation reported by Song et al., starting with the FD-factor:

RFA=logFDfactorS0,5
(4)

In the formula reported above, the FD-factor is the dilution factor while S is the weight
percentage of the component present in the mixture [183–185]. In GC and GC-MS
analyses, 100 compounds were characterized and seventeen peaks were confirmed
by sniffing with GC-O. The work demostrated that the compound germacrene D and
Scutellaria laeteviolacea essential oil had the most similar smell, even if the latter
was characterized by a relatively low (0.5) flavor activity and by the highest FD-factor
(7). These findings suggest that the relative flavor activity, defined as a new odor
unit, and the FD-factor often have no relation to the aroma character of a compound.
In other words, even if the RFA of one compound is not comparatively high, it often
contributes significantly to the original odor and can be used when considering flavor
activity [55].

Clinopodium tomentosum (Kunth) Govaerts essential oil was studied with GC-O in
order to calculate threshold odor concentration (TOC) values of the main odorants
in a mixture [53]. As displayed in Table 2, Benzo et al. found a good agreement
between the calculated and measured TOC values of a few odorants in the essential
oil. They established a relationship between the odorant concentration at the sniffing
port and that in the injected solution using the TOC value of limonene as a reference
compound.

Table 2. TOC values calculated using AEDA method in GC-O technique and
measured by dynamic dilution olfactometry (reprinted from [53] with
permission from Elsevier).

Calculated TOC Measured TOC

No. Compound Descriptor
(μg/m3) (μg/m3)

1 1-Octen-3-ol + 6-methyl- 0.223 2.36 and 18.89 Mushrooms

5-hepten-3one

2 1,8-Cineole 2.67 5.08 Balsamic

3 Isomenthone 40.451 n.d. Wine bottle

stopper

4 Isopulegone 0.076 n.d. Minty

5 Pulegone 0.884 1.87 Minty

6 cis-Piperitone oxide 22.427 n.d. Minty

3.3. Environmental Odor Applications

Odors produced by anthropic sources are a complex issue because they directly
affect both the environment and the human quality of life. Industrial plants and farms
are often a source of bad odor, hence, their close proximity to residential zones can
lead to complaints by local residents [4,186,187]. Furthermore, odors can strongly
affect people's daily life and wellbeing since they may provoke both physiological
symptoms (respiratory problems, nausea, headaches) and psychological stress
[188,189].

The official methodology for odor emissions assessment is dynamic olfactometry, a

sensorial technique standardized by international technical laws [190,191]. It is
based on the use of a dilution instrument, called an olfactometer. This device
releases the odor sample diluted with odor-free air at precise ratios to a panel of
human assessors who have been selected according to their perception threshold
for a reference gas. The odor concentration, usually expressed in odor units (ou/m 3),
is numerically equal to the dilution factor required to reach an odor threshold, which
is the minimum concentration perceived by 50% of the population [190,191].
However, it is not sufficient to completely evaluate a case of olfactory nuisances for
various reasons [192]:
  -Continuous and field measurements, which are useful for monitoring
industrial processes causing odor emissions, cannot be performed;
 -Odor concentration refers to the whole odor sample, without discriminating
between single chemical compounds and their contribution to that
concentration;
 -Odor samples require rapid analysis since they are instable and difficult to
store;
 -Olfactometry can be quite time-consuming and expensive, and the frequency
and duration of the analyses are limited.

In order to overcome these limitations, sensoristic and analytical methodologies as

well as others are widely employed and their information are often integrated to
achieve a more complete understanding of olfactory nuisance cases. The use of a
hybrid instrumentation such as GC-O has been shown to provide interesting
information in the environmental field due to the coupling of the chemical
characterization with sensorial perception related to the single compounds eluted by
the column.

A considerable number of scientific works dealing with the use of GC-O have
focused on odor assessment produced by different types of animal farms. The
activities connected with high density livestock operations can produce several
hundred volatile organic compounds (acids, alcohols, aldehydes, amines, volatile
fatty acids, hydrocarbons, ketones, indoles, phenols, nitrogen and sulfur
compounds) yet relatively few of them are responsible for the typical odor of these
environments. Hence, the aim is to the extract compounds that are actually
responsible for the primary odor impacting livestock environments. Several factors
can complicate this task; for instance, the variability among species, manure
management systems and animal production practices [193].

Using GC-O to profile odor is a functional approach for defining, prioritizing and
tracking livestock odorants. Wright et al. [193] have studied the odor profiles
produced by swine and beef cattle operations by optimizing the collection time and/or
increasing the distance from the odor source. The increased distance showed a
significant reduction in the total number of detectable odors, as shown in Figures 9,
,10.10. The authors found that p-cresol was the only significant olfactory response
even 2,000 m away from the odor site, meaning that it could be considered as a
surrogate parameter correlated to odor.
Figure 9. Aromagram for 4 h SPME fiber collection 20 m downwind (“near” site) from
commercial beef cattle feed yard (reprinted from [193] with permission from
Elsevier).

Figure 10. Aromagram for 4 h SPME fiber collection 2,000 m downwind (“distant”
site) from commercial beef cattle feed yard (reprinted from [193] with permission
from Elsevier).
A lot of studies have reported that most of the odor produced by swine barns is
carried on dust [194–196] and that its reduction is possible using dust filters for
Particulate Matter (PM) [197,198]. To further understand this aspect, GC-MS/O was
used to identify odorous VOCs adsorbed/absorbed on different size swine barn dust
(PM1, PM2,5, PM10 and TSP). The study of the aromagrams of the VOCs extracted
from TSP filters at different SPME sampling times allowed Cai et al. [81] to identify
the key odorants for the different granulometric fractions of the particulate matter as
well as identify what is actually carrying the odorous compounds.

GC-MS/O is also widely employed for determining the efficiency of different

treatment systems for odor emission reduction and for developing a suitable and
cost effective strategy for implementation. In particular, the sensorial data associated
to each single compound are able to indicate what compounds, among those that
really contribute to odor, have to be mitigated in order to reduce odor emissions. This
aspect has a practical implication in the development of opportune abatement
systems specific for the compounds causing odor emissions [78,83]. For example,
GC-MS/O was used to evaluate the effectiveness of topical zeolite applications to
mitigate VOCs and odor from simulated poultry manure storage [199]. It was also
used to characterize the odorants either before and after the application of
abatement products (activated carbon, silica gel and zeolite) in order to reduce odor
compounds emitted by broiler litter materials [200] or prior to and after waste gas
treatment from a fat refinery [80], as shown in Figure 11. An analogous approach
was used by Chen et al. [78] for examining two types of wood chip-based biofilters
as common abatement systems.
Figure 11. FID/O-chromatograms of waste gas from a fat refinery obtained prior to
and after waste gas treatment: (a) untreated waste gas; (b) after bioscrubber; (c)
after biofilter; a–p odor signals (reprinted from [80] with permission from Elsevier).

In the same way, Agus et al. [74,201] set up a noteworthy application of GC-O in
environmental matrixes. Their study identified trace amounts of odorous organic
compounds in drinking water that was obtained from highly treated wastewater
during the different phases of the treatment.

Some authors have tried to combine the GC-O results with those acquired from other
approaches. The goal of this integration may be to discover useful correlations in
order to obtain a better understanding of a case study or to test the effectiveness of
GC-O in determining the principal odorants in a mixture. The work of Sohn et al.
[202] can exemplify the former objective. Their study combines GC-O with a real
time monitoring system (an artificial olfaction system) to measure in shed odor
concentrations at two different poultry farms. The results were combined with
ventilation rates and weather data to calculate the odor emission rate (OER)
throughout the batches, observing that OERs varied significantly between farms.
This variation was linked to changes in ventilation rates, bird activity and other
management and environmental factors. Similarly, GC-MS/O results showed
different chemical profiles during the poultry production cycle. The matrix was first
dominated by terpenes originating from the bedding material of young birds while
later it became populated by aldehydes, ketones and sulphides as the bedding
became soiled with manure.

In other studies, the results of GC-O were integrated with the odor concentrations
measured by dynamic olfactometry. In these cases, correlations between the
concentrations of odorous VOCs and measured odor units were investigated in order
to define the main compounds contributing to the entire perception of odorous VOCs
[83,84]. In another case, Zhang et al. [87] characterized the odor emissions
produced at swine and dairy sites and associated odor intensity and hedonic tone to
compounds. They were also able to demonstrate that concentrations of odorous
compounds correlated well with the measured log stimulus intensity.

For the purpose of demonstrating the effectiveness of GC-O in identifying the

principal odorants in a mixture, Trabue et al. [86] compared the results of GC-O with
odor activity values (OAV) obtained in open cattle feedlots. Here, the sampling was
carried out at the source and far from it. Based on OAVs, the chemical
characterization of the source revealed that Volatile Fatty Acids (VFA) were the
principal contributors to odor, followed by phenols and indoles. At 250 m downwind
from the source, the total OAV declined by almost 95% while at 3.2 km downwind it
decreased to a value of less than one, meaning that no odor should be present,
unlike to what was described. On the other hand, GC-O results revealed that, at 250
m downwind of the feedlot, a panelist could not perceive pentanoic acid odor, despite
its concentration was above its OT while perceived 4-ethylphenol with an OAV less
than one. These results indicate that an approach based solely on OAVs is not
sufficient to describe odor characteristics because of the OT value uncertainty, as
revealed in literature.

3.4. Material Applications

The GC/MS-O method has proven to be a useful and reliable tool for the detection
and identification of odor active VOCs responsible for off-flavors coming from a wide
range of materials. Material applications have two main goals. One is to inform
industry of manufacturing processes and any possible improvements in that practice.
The other is to find replacements for raw materials that generate odorous
compounds with other odorless compounds in order to avoid or, at least, reduce odor
nuisance. This research is important for food packaging materials since it can search
for ways of avoiding the possible migration of materials into food or other susceptible
products where they may cause unexpected and unsightly changes. Tyapkova et al.
used the GC-O approach to characterize the volatile chemicals that forms in
sterilization processes, with 60Co γ-irradiation in the presence of oxygen, of
polypropylene (PP) packaging materials, used in food, pharmaceutical or cosmetic
fields [52]. The authors compared VOCs emission from irradiated (rays at 10 and 20
kGy) and not irradiated polypropylene, including a sensory evaluation with a panelist.
Besides compositional changes in volatile odorous substances from PP during
treatment, the results showed a shift towards a different odor descriptor (fatty, sweet,
sour, burnt, stinging, metallic, wax-like, plastic) depending on the γ-irradiation
condition. The experimental results are reported in Figure 12 as orthonasal
comparative Flavor Profile Analyses (cFPA).

Figure 12. Orthonasal comparative flavor profile analysis (cFPA) of three powdered
PP samples. The data are displayed as mean numerical values of the sensory
evaluations (three sessions with six panelists each) (Reprinted from [52] with
permission from Elsevier).

Nowadays, GC-O methodology is widely applied to evaluate VOCs and odor

emissions from waste-recycled innovative materials, employed both in the
automotive industry and for construction, furniture and consumer products. Although
the recycling of waste materials is well established and widely applied, it has to
satisfy chemical requirements in order to obtain safety certification.

Felix et al. applied HS-SPME extraction and GC-O/MS analysis to wood-plastic

composites (WPC) produced with landfill derived plastic and sawdust in order to
characterize VOC emissions and to assess the impact of the odor on the end-users
[82]. This study found that the WPC prototype had a characteristic odor profile and
that many of the compounds observed are related to reprocessed materials,
confirming the hypothesis that repeated recycling can generate thermo-oxidative
degradation (Table 3). In addition to the findings reported by a previous study [203],
Felix et al. stressed that the degradation of a polymeric matrix produces a
characteristic odor associated with aldehydes, ketones and carboxylic acids,
whereas the degradation of the lignocellulosic component releases acetic acid,
formaldehyde, formic acids, aldehydes and other acids [82].

Table 3.
Olfactory description, chemical identity and modified frequency percentage MF (%) for
each odorant identified in WPC prototype (Adapted from [82] with permission from
Elsevier).

RT KIexp MF
a KIrefb Odor Descriptor Compound
(min) (%)

Diacetyl, cream, sweet, yogurt, curd,

7.268 1011 970 83 Diacetyl (2,3-Butanedione)
wood, fruity

Grass, herb, green, flower, solvent,

9.188 1120 1084 76 Hexanal
chemical

Fruity, ester, candies, jelly, plastic,

10.468 1177 1150 53 m-Xylene
varnish

Aldehyde, medicine, chemical, herb,

13.849 1321 1280 82 Octanal
flower, field, lemon, grapefruit, orange

Aldehyde, bleach or lemon cleaner,

16.314 1424 1385 77 Nonanal
unpleasant

Acid, unpleasant, solvent, glue, sweat,

17.668 1481 1450 79 Acetic acid
sunflower seeds

18.732 1527 1484 Aldehyde, powdered sugar, acid 51 Decanal

Gas, burnt, green shield bug, fresh wood,

19.029 1540 1490 71 Acetylfuran
fried, oily

19.316 1553 1491 Dry fruit, nut, almond, mold, dense 58 Camphor

Cheese, rancid cheese, butiric or

21.655 1658 - 77 Unknown
propanoic acid
 RT KIexp MF
a KIrefb Odor Descriptor Compound
(min) (%)

22.611 1702 - Acid, cheese, butiric acid 76 Unknown

Bug, nail polish remover, naphthalene

23.080 1724 1720 53 α-Terpineol
balls

Acid, trash, waste, foot, wood, hair

26.123 1872 1829 74 Hexanoic acid (caproic acid)
removal wax, licorice

Phenol, shoeshine, medicine, sweat, bug, 2-Methoxyphenol

26.508 1891 1859 65
vanilla (Guaiacol)

Phenol, aromatic, sweet, zinc oxide

28.105 1973 - 74 Unknown
adhesive plaster, opium, hospital

30.857 2075 - Unpleasant, acid, wood, manure 65 Terpin hydrate

2-Methoxy-4-vinylphenol
32.727 2130 2198 Lactone, burnt, car tire, rubber 35 c
(4-Vinylguaiacol)

2358
35.655 2204 d Flower, salt water, hair removal wax 38 c Diethyl phthalate (DEP)

Car tire, vanilla, soluble chocolate

39.374 2270 2569 53 Vanillin
powder, burnt
a
Kovats retention index calculated from BP-20, 30 m column;
b
Kovats retention index reported in the Flavornet Database (Carbowax™ 20 m column);
c
Compounds with MF < 50% but relevant to the sample;
d
Kovats retention index calculated from DB-Wax 60 m column.

Table 3.
Olfactory description, chemical identity and modified frequency percentage MF (%)
for each odorant identified in WPC prototype (Adapted from [82] with permission
from Elsevier).

The applied combination of sensory assessment and GC/MS analysis also seems
to be a useful approach in the effort to eliminate unwanted odors from building
products. In this regard, Knudsen et al. focused attention on VOCs and odor
emissions derived from environmentally friendly products with linseed oil (e.g.,
linoleum, wall paint) that influence the perceived air quality more negatively than
similar synthetic products and for a longer period of time [204]. The primary goal of
this work was to test if the odor of linseed oil influenced the odor of the final building
product. Experimental data obtained indicated that the undesirable odor-active
VOCs, having low odor thresholds, had probably originated from the degradation of
the linseed oil, due to ozone-oxidation processes. As a result, the authors suggested
the use of less odorous linseed oils and gave useful instruction to manufacturers to
improve the acceptability of odorous emissions from these building products. Hence,
once the objective perception of an odor, for example, by conducting a preliminary
olfactometry analysis, is verified, the identification of the principal volatile compounds
contributing to the perceived overall odor derived from the object of investigation can
be obtained with the application of GC-O/MS.

Another interesting application of GC-O/MS investigation has evidenced the link

between the physical and chemical properties of oak wood with the chemical
composition, olfactory and gustatory qualities of wines fermented and/or aged in oak
barrels. The aim of a study carried out by Díaz-Maroto et al. was to investigate the
sensory importance of oak wood VOCs in order to evaluate the contribution of wood-
derived volatile compounds to the overall aroma of oak-aged wines [51]. This study
demonstrated that oak wood treatments such as seasoning and toasting, as well as
other factors like tree species and geographic location, can modify both the physical
and chemical qualities of the wood, resulting in the characteristic aromas (fruity,
fresh/green/grassy and floral) of wines. Moreover, trans-2-nonenal and decanal
which can transmit unpleasant aromas to wine were detected in non-toasted oak
woods suggesting that the toasting treatment of the wood could reduce the problem.

Taking into account the aforementioned applications of GC-O/MS methodology for

the optimization of production processes and the improvement of the quality and
odor acceptability of a final product, it can be stated that GC-O/MS methodology has
revealed potentials not associated to other analytical techniques.

3.5. Medical Applications

One of the more recent and promising applications of GC-O is in medical research
where the study of volatile and odorous profiles of biological matrixes is mainly used
to aid in the diagnoses of diseases and dysfunctions. In this area, interesting results
have been obtained from the characterization of volatile and odorous profiles in
human urine. Indeed, the study of urine could yield a wealth of physiological
information and increase the understanding of metabolization and excretion
processes of low molecular weight compounds originating from dietary or
endogenous sources. Moreover, changes in individual profiles can be potential
indicators and mechanistic clues of deviations or even misbalances in physiological
conditions induced by diseases or hormonal changes. However, since the
application of modern analytical tools in volatile analysis in urine has been limited,
the diagnostic potential of urinary volatile fraction is not yet fully understood.

Recently, Wagenstaller and Buettner [49] have applied GC-O methods to evaluate
a combination of comprehensive chemo-analytical and human-sensory methods for
the characterization of human urine odorants. A total of 14 odorants were detected
in most of the untreated urine samples and 24 odorants in the glucuronidase-treated
samples. Most of these potent odor compounds had not previously been detected in
human urine when traditional methodologies were used. The study highlighted the
high analytical potential of the combined approach in elucidating the fate of
volatile/odorous food constituents or substances originating from other origins (e.g.,
pharmaceuticals) and in characterizing the excretion of endogenous substances.
This approach may very well pave the way for a better understanding of the
diagnostic potentials of the odorous and volatile fraction in human urine.

GC-O was also used to study the influence of thermal reaction and microbial
transformation on the odor of human urine after being boiled or fermented [72]. The
headspace of the samples was analyzed to isolate and identify the malodor-
generating bacteria present in the urine of healthy people. The study showed that
urine composition may certainly influence pH, bacterial composition and urine odors.
Incubation of freshly collected urine revealed that the maximum concentration of total
bacterial counts was reached after 4 days at room temperature (20–25 °C) and after
1 day at 37 °C. Indeed, most of the volatile compounds were generated during the
first hours of incubation. Analytical comparisons between boiled and fermented urine
revealed that the incubation of sterile urine with a bacterial mixture of E. fergusonii,
Enterococcus faecalis, Citrobacter koseri, S. agalactiae and M. morganii produced
a characteristic aged urine odor.

Shirasu et al. identified dimethyl trisulfide (DMTS) as the main odorant responsible
for severe malodors in some advanced cancer patients by performing gas
chromatography-mass spectrometry-olfactometry analysis of volatiles from
fungating cancer wounds [205]. The intensity and quality of the body odors emitted
from the fungating wounds of three female patients with breast cancer and of two
male patients with head and neck cancer were examined. In particular, sterile gauze
pads were placed on the wounds for 6–12 h and headspace volatiles were extracted
using SPME fiber and then analyzed with GC-O. The study produced useful data for
the development of a strategy to prevent or reduce the DMTS odor which helped to
improve the quality of life of the patients.

In another study, the strong “maple-syrup” odor which accompanies fenugreek

ingestion was investigated by Mebazaa et al. [206]. The odor active compounds
present in HS-SPME armpit sweat extracts collected before and during fenugreek
ingestion periods were analyzed with GC-O and evaluated by a panel of eight
experienced assessors. Collected data were treated using the frequency of detection
methodology. Among the 44 compounds identified, 10 were detected by assessors
before and during fenugreek ingestion such as a-pinene, 6-methyl-5-hepten-2-one,
nonanal, 1-octen-3-ol, 2-phenylethyl alcohol and benzenemethanol. Although these
compounds had already been identified in human armpit sweat collected from
different male and female subjects, this is the first study of their effect in the overall
odor of human armpit sweat.

4. Conclusions
In recent years, researchers involved in the study of odorous substances have
recognized the great potential of gas-chromatography/olfactometry in their work as
it can simultaneously provide analytical and sensorial information about an odor
mixture. However, the complexity of this matter has limited the application of GC-O
whose first application dates back to 1964. The many factors linked with
chromatographic parameters as well as the variables related to human perception
and panel assessment represent a challenge for researchers. The goal is to find the
best use of these components in order to improve both the practical experimental
use of GC-O and the quality of the data obtained.

GC-O has been extensively employed in food aromas and fragrance studies. This
application has been primarily industrial and driven by commercial desires to
improve of the quality of products in order to make them more pleasing and desirable
to consumers. In the recent years, researchers have also begun exploring GC-O
applications in environmental, material and medical areas. One remarkable
application of GC-O is in the medical field where the characterization of volatile and
odorous profiles of biological matrixes could be useful for carrying out screening
analyses of human diseases and dysfunctions.

GC-O is becoming a reliable and reproducible method for characterizing the odor
footprint of complex mixtures of chemicals of diverse origin. Noteworthy application
improvements include the introduction of new and efficient interfaces between the
GC and olfactometric port, the reduction of subjective error perception and anosmia
by increasing the number of simultaneous panelists (up to 8), and the use of bi-
dimensional GC techniques and sophisticated statistical programs for data
processing and interpretation. In conclusion, GC-O may be considered one of the
most valuable methods in odor research today. Future improvements in the
effectiveness of GC-O application should include a standardization of the different
approaches being used. This would be valuable in order to estimate the sensory
contribution of a single odor active compound in a complex mixture as well as in the
resulting aromagrams.

Conflicts of Interest

The authors declare no conflict of interest.

References

1. Ruijten M.W.M.M., van Doorn R., van Harreveld A.P. Assessment of Odor
Annoyance in Chemical Emergency Management. RIVM; Bilthoven, The
Netherlands: 2009. RIVM Report 609200001.
2. Both R., Sucker K., Winneke G., Koch E. Odor intensity and hedonic tone—
important parameters to describe odor annoyance to residents? Water Sci. Technol.
2004;50:83–92. [PubMed]
3. Czerny M., Brueckner R., Kirchhoff E., Schmitt R., Buettner A. The influence of
molecular structure on odor qualities and odor detection thresholds of volatile
alkylated phenols. Chem. Senses. 2011;36:539–553. [PubMed]
4. Yuwono A.S., Lammers P.S. Odor pollution in the environment and the detection
instrumentation. Int. Agric. Eng. J. 2004;6:1–33.
5. Vincent A., Hobson J. Odor Control. Terence Dalton Publishing Ltd.; London, UK:
1998.
6. Craven M.A., Gardner J.W., Bartlett P.N. Electronic noses—development and
future prospects. Trends Anal. Chem. 1996;15:486–493.
7. Schilling B., Kaiser R., Natsch A., Gautschi M. Investigation of odors in the
fragrance industry. Chemoecology. 2010;20:135–147.
8. Yeomans M.R. Olfactory influences on appetite and satiety in humans. Physiol.
Behav. 2006;87:800–804. [PubMed]
9. Beghi S.P., Santos J.M., Reis N.C., de Sá L.M., Goulart E.V., de Abreu E. Impact
assessment of odours emitted by a wastewater treatment plant. Water Sci. Technol.
2012;66:2223–2228. [PubMed]
10. Li L., Han C., Guo B., Zhou Y. Comprehensive monitoring and assessment of
odour emission sources from refinery I. monitoring of odorous sources. Pet. Process.
Petrochem. 2013;44:95–101.
11. Aatamila M., Verkasalo P.K., Korhonen M.J., Suominen A.L., Hirvonen M.-R.,
Viluksela M.K., Nevalainen A. Odour annoyance and physical symptoms among
residents living near waste treatment centres. Environ. Res. 2011;111:164–170.
[PubMed]
12. Romain A.-C., Nicolas J., Cobut P., Delva J., Nicks B., Philippe F.-X. Continuous
odour measurement from fattening pig units. Atmos. Environ. 2013;77:935–942.
13. González C.R.-N., Björklund E., Forteza R., Cerdà V. Volatile organic
compounds in landfill odorant emissions on the island of Mallorca. Int. J. Environ.
Anal. Chem. 2013;93:434–449.
14. Peng C.Y., Lan C.H., Wu T.J. Investigation of indoor chemical pollutants and
perceived odor in an area with complaints of unpleasant odors. Build. Environ.
2009;44:2106–2113.
15. Bitter F., Muller B., Muller D. Estimation of odour intensity of indoor air pollutants
from building materials with a multi-gas sensor system. Build. Environ. 2010;45:197–
204.
16. Davoli E., Gangai M.L., Morselli L., Tonelli D. Characterisation of odorants
emissions from landfills by SPME and GC/MS. Chemosphere. 2003;51:357–368.
[PubMed]
17. Dincer F., Odabasi M., Muezzinoglu A. Chemical characterization of odorous
gases at a landfill site by gas chromatography–mass spectrometry. J. Chromatogr.
A. 2006;1122:222–229. [PubMed]
18. Di Francesco F., Lazzerini B., Marcelloni F., Pioggia G. An electronic nose for
odor annoyance assessment. Atmos. Environ. 2001;35:1225–1234.
19. Friedrich J.E., Acree T.E. Gas Chromatography Olfactometry (GC/O) of dairy
product. Int. Dairy J. 1998;8:235–241.
20. Lo Y.C.M., Koziel J.A., Cai L., Hoff S.J., Jenks W.S., Xin H. Simultaneous
chemical and sensory characterization of volatile organic compounds and semi-
volatile organic compounds emitted from swine manure using solid phase
microextraction and multidimensional Gas Chromatography-Mass Spectrometry-
Olfactometry. J. Environ. Qual. 2008;37:521–534. [PubMed]
21. Delahunty C.M., Eyres G., Dufour J.P. Gas chromatography-olfactometry. J.
Sep. Sci. 2006;29:2107–2125. [PubMed]
22. Plutowska B., Wardencki W. Application of gas chromatography-olfactometry
(GC-O) in analysis and quality assessment of alcoholic beverages—A review. Food
Chem. 2008;107:449–463.
23. Hochereau C., Bruchet A. Design and application of a GCSNIFF/MS system for
solving taste and odor episodes in drinking water. Water Sci. Technol. 2004;49:81–
87. [PubMed]
24. Debonneville C., Orsier B., Flament I., Chaintreau A. Improved hardware and
software for quick gas chromatography: Olfactometry using CHARM and GC-“SNIF”
analysis. Anal. Chem. 2002;74:2345–2351. [PubMed]
25. D'Acampora Zellner B., Dugo P., Dugo G., Mondello L. Gas chromatography-
olfactometry in food flavour analysis. J. Chromatogr. A. 2008;1186:123–143.
[PubMed]
26. Berdagùe J.L., Tournayre P., Cambou S. Novel multi-gas chromatography-
olfactometry device and software for the identification of odor-active compounds. J.
Chromatogr. A. 2007;1146:85–92. [PubMed]
27. Ferreira V., Lopez R., Aznar M. Olfactometry and Aroma Extract Dilution Analysis
of Wines. In: Jackson J., Linskens H., editors. Analysis of Taste and Aroma.
Springer-Verlag; Berlin, Germany: 2002. pp. 90–122.
28. Chin S.T., Eyres G.T., Marriotta P.J. Identification of potent odorants in wine and
brewed coffee using gas chromatography-olfactometry and comprehensive two-
dimensional gas chromatography. J. Chromatogr. A. 2011;1218:7487–7498.
[PubMed]
29. D'Acampora Zellner B., Casilli A., Dugo P., Dugo G., Mondello L. Odour
fingerprint acquisition by means of comprehensive two-dimensional gas
chromatography-olfactometry and comprehensive two-dimensional gas
chromatography/mass spectrometry. J. Chromatogr. A. 2007;1141:279–286.
[PubMed]
30. Koziel J.A., Lo Y.C., Cai L., Wright D.W. Simultaneous characterization of VOCs
and livestock odors using solid-phase microestraction- multidimensional Gas
Cromatography-mass spettrometry-olfactometry. Chem. Eng. Trans. 2010;23:73–
78.
31. Vera P., Uliaque B., Canellas E., Escudero A., Nerína C. Identification and
quantification of odorous compounds from adhesives used in food packaging
materials by headspace solid phase extraction and headspace solid phase
microextraction coupled to gas chromatography-olfactometry-mass spectrometry.
Anal. Chim. Acta. 2012;745:53–63. [PubMed]
32. Naudé Y., Rohwer E.R. Investigating the coffee flavour in South African Pinotage
wine using novel offline olfactometry and comprehensive gas chromatography with
time of flight mass spectrometry. J. Chromatogr. A. 2013;1271:176–180. [PubMed]
33. Friedrich J.E., Acree T.E. Flavour Chemistry: Industrial and Academic Research.
In: Risch S.J., Ho C.T., editors. ACS Symposium Series. Volume 756 American
Chemical Society; Washington, DC, USA: 2000.
34. Hanaoka K., Sieffermann J.M., Giampaoli P. Effects of the sniffing port air
makeup in gas chromatography-olfactometry. J. Agric. Food Chem. 2000;48:2368–
2371. [PubMed]
35. Etiévant P.X., Callement G., Langlois D., Issanchou S., Coquibus J.N. Odor
intensity evaluation in gas chromatography-olfactometry by finger span method.
Agric. Food Chem. 1999;47:1673–1680. [PubMed]
36. Ruth S.M. Methods for gas chromatography–olfactometry: A review. Biomol.
Eng. 2001;17:121–128. [PubMed]
37. Fur Y., Mercurio V., Moio L., Blanquet J., Meunier J.M. A new approach to
examine the relationships between sensory and gas chromatography-olfactometry
data using generalized procrustes analysis applied to six French Chardonnay wines.
J. Agric. Food Chem. 2003;51:443–452. [PubMed]
38. Pollien P., Ott A., Montigon F., Baumgartner M. Hyphenated headspace-gas
chromatography-sniffing technique: Screening of impact odorants and quantitative
aromagram comparisons. J. Agric. Food Chem. 1997;45:2630–2637.
39. Ferrari G., Lablanquie O., Cantagrel R., Ledauphin J., Payot T., Fournier N.
Determination of key odorant compounds in freshly distilled cognac using GC-O,
GC-MS, and sensory evaluation. J. Agric. Food Chem. 2004;52:5670–5676.
[PubMed]
40. Villberg K., Veijanen A., Gustafsson I., WickstrÖm K. Analysis of odor and taste
problems in high-density polyethene. J. Chromatogr. A. 1997;791:213–219.
[PubMed]
41. Etiévant P.X. Odour Intensity Evaluation in GC-Olfactometry by Finger-Span
Method. In: Jackson J., Linskens H., editors. Analysis of Taste and Aroma. Springer-
Verlag; Berlin, Germany: 2002. pp. 223–237.
42. Ferreira V., Petka J., Aznar M. Aroma extract dilution analysis. Precision and
optimal experimental design. J. Agric. Food Chem. 2002;50:1508–1514. [PubMed]
43. Audouin V., Bonnet F., Vickers Z.M., Reineccius G.A. Limitation in the Use of
Odor Activity Values to Determine Important Odorants in Foods. In: Leland V.,
Scheiberle P., Buettner A., Acree T.E., editors. Gas Chromatography-Olfactometry:
The State of the Art. American Chemical Society; Washington, DC, USA: 2001. pp.
156–171.
44. Aznar M., Lopez R., Cacho J.F., Ferreira V. Identification and quantification of
impact odorants of aged red wines from Rioja. GC-olfactometry, quantitative GC-
MS, and odor evaluation of HPLC fractions. J. Agric. Food Chem. 2001;49:2924–
2929. [PubMed]
45. Cullere L., Escudero A., Cacho J., Ferreira V. Gas chromatography-olfactometry
and chemical quantitative study of the aroma of six premium quality Spanish aged
red wines. J. Agric. Food Chem. 2004;52:1653–1660. [PubMed]
46. Lopez R., Ortin N., Perez-Trujillo J.P., Cacho J., Ferreira V. Impact odorants of
different young white wines from the Canary Islands. J. Agric. Food Chem.
2003;51:3419–3425. [PubMed]
47. Pino J.A. Odor-active compounds in pineapple (Ananas comosus [L.] Merril cv.
Red Spanish) Int. J. Food Sci. Technol. 2013;48:564–570.
48. Kim H., Cadwallader K.R., Kido H., Watanabe Y. Effect of addition of commercial
rosemary extracts on potent odorants in cooked beef. Meat Sci. 2013;94:170–176.
[PubMed]
49. Wagenstaller M., Buettner A. Characterization of odorants in human urine using
a combined chemo-analytical and human-sensory approach: A potential diagnostic
strategy. Metabolomics. 2013;9:9–20.
50. Mallia S., Piccinali P., Rehberger B., Badertscher R., Escher F., Schlichtherle-
Cerny H. Determination of storage stability of butter enriched with unsaturated fatty
acids/conjugated linoleic acids (UFA/CLA) using instrumental and sensory methods.
Int. Dairy J. 2008;18:983–993.
51. Díaz-Maroto M.C., Guchu E., Castro-Vázquez L., de Torres C., Pérez-Coello
M.S. Aroma-active compounds of American, French, Hungarian and Russian oak
woods, studied by GC-MS and GC-O. Flavour Fragr. J. 2008;23:93–98.
52. Tyapkova O., Czerny M., Buettner A. Characterisation of flavour compounds
formed by γ-irradiation of polypropylene. Polym. Degrad. Stab. 2009;94:757–769.
53. Benzo M., Gilardoni G., Gandini C., Caccialanza G., Finzi P.V., Vidari G., Abdo
S., Layedra P. Determination of the threshold odor concentration of main odorants
in essential oils using gas chromatography–olfactometry incremental dilution
technique. J. Chromatogr. A. 2007;1150:131–135. [PubMed]
54. Eyres G., Marriott P.J., Dufour J.P. The combination of gas chromatography-
olfactometry and multidimensional gas chromatography for the characterization of
essential oils. J. Chromatogr. A. 2007;1150:70–77. [PubMed]
55. Miyazawa M., Nomura M., Marumoto S., Mori K. Characteristic odor components
of essential oil from Scutellaria Iaeteviolacea. J. Oleo Sci. 2013;62:51–56. [PubMed]
56. Clery R. High-impact odorants in essential oils. Flavour Fragr. J. 2010;25:117–
120.
57. Yagi N., Nakahashi H., Kobayashi T., Miyazawa M. Characteristic chemical
components of the essential oil from white kwao krua (Pueraria mirifica) J. Oleo Sci.
2013;62:175–179. [PubMed]
58. D'Acampora Zellner B., Bicchi C., Dugo P., Rubiolo P., Dugo G., Mondello L.
Linear retention indices in gas chromatographic analysis: A review. Flavour Fragr. J.
2008;23:297–314.
59. Jeleń H., Majcher M., Ginja A., Kuligowski M. Determination of compounds
responsible for tempeh aroma. Food Chem. 2013;141:459–465. [PubMed]
60. Blank I. Gas Chromatography-Olfactometry in Food Aroma Analysis. In: Marsili
R., editor. Techniques for Analyzing Food Aroma. CRC Press; Rockford, IL, USA:
1996.
61. Le Guen S., Prost C., Demaimay M.J. Critical comparisons of three olfactometric
methods for the identification of the most potent odorants in cooked mussels (Mytilus
edulis) J. Agric. Food Chem. 2000;48:1307–1314. [PubMed]
62. Petka J., Ferreira V., Cacho J. Posterior evaluation of odor intensity in gas
chromatography-olfactometry: Comparison of methods for calculation of panel
intensity and their consequences. Flavour Fragr. J. 2005;20:278–287.
63. Hallier A., Courcoux P., Sérota T., Prosta C. New gas chromatography-
olfactometric investigative method, and its application to cooked Silurus glanis
(European catfish) odor characterization. J. Chromatogr. A. 2004;1056:201–208.
[PubMed]
64. Fan W., Qian M.C. Identification of aroma compounds in Chinese “Yanghe Daqu”
liquor by normal phase chromatography fractionation followed by gas
chromatography/olfactometry. Flavour Fragr. J. 2006;21:333–342.
65. Fu S.G., Yoon Y., Bazemore R. Aroma-active components in fermented bamboo
shoots. J. Agric. Food Chem. 2002;50:549–554. [PubMed]
66. Garruti D.S., Franco M.R.B., da Silva A.M.A.P., Janzantti N.S., Alves G.L.
Assessment of aroma impact compounds in a cashew apple-based alcoholic
beverage by GC-MS and GC-olfactometry. LWT Food Sci. Technol. 2006;39:372–
377.
67. Gurbuz O., Rouseff J.M., Rouseff R.L. Comparison of aroma volatiles in
commercial Merlot and Cabernet Sauvignon wines using gas chromatography-
olfactometry and gas chromatography mass spectrometry. J. Agric. Food Chem.
2006;54:3990–3996. [PubMed]
68. O'Riordan P.J., Delahunty C.M. Comparison of volatile compounds released
during the consumption of Cheddar cheese with compounds extracted by vacuum
distillation using gas chromatography-olfactometry. Flavour Fragr. J. 2001;16:425–
434.
69. Márquez V., Martínez M., Guerra M., Fariña L., Boido E., Dellacassa E.
Characterization of aroma-impact compounds in yerba mate (Ilex paraguariensis)
using GC-olfactometry and GC-MS. Food Res. Int. 2013;53:808–815.
70. Engel W., Bahr W., Schieberle P. Solvent assisted flavour evaporation—a new
and versatile technique for the careful and direct isolation of aroma compounds from
complex food matrices. Eur. Food Res. Technol. 1999;209:237–241.
71. Yao S.S., Guo W.F., Lu Y., Jiang Y.X. Flavour characteristics of lapsang
souchong and smoked lapsang souchong, a special chinese black tea with pine
smoking process. J. Agric. Food Chem. 2005;53:8688–8693. [PubMed]
72. Troccaz M., Niclass Y., Anziani P., Starkenmann C. The influence of thermal
reaction and microbial transformation on the odor of human urine. Flavour Fragr. J.
2013;28:200–211.
73. Grosch W. Evaluation of the key odorants of foods by dilution experiments,
aroma models and omission. Chem. Senses. 2001;26:533–545. [PubMed]
74. Agus E., Zhang L., Sedlak D.L. A framework for identifying characteristic odor
compounds in municipal wastewater effluent. Water Res. 2012;46:5970–5980.
[PubMed]
75. Thomas C., Mercier F., Tournayre P., Martin J.L., Berdagué J.L. Effect of nitrite
on the odourant volatile fraction of cooked ham. Food Chem. 2013;139:432–438.
[PubMed]
76. Campo E., Ferreira V., Escudero A., Cacho J. Prediction of the wine sensory
properties related to grape variety from dynamic headspace gas chromatography-
olfactometry data. J. Agric. Food Chem. 2005;53:5682–5690. [PubMed]
77. Feng Y., Cui C., Zhao H., Gao X., Zhao M., Sun W. Effect of koji fermentation on
generation of volatile compounds in soy sauce production. Int. J. Food Sci. Technol.
2013;48:609–619.
78. Chen L., Hoff S.J., Koziel J.A., Cai L., Zelle B., Sun G. Performance evaluation
of a wood-chip based biofilter using solid-phase microextraction and gas
chromatography-mass spectroscopy-olfactometry. Bioresour. Technol.
2008;99:7767–7780. [PubMed]
79. Heitmann K., Wichmann H., Bahadir M. Chemical causes of the typical burnt
smell after accidental fires. Anal. Bioanal. Chem. 2009;395:1853–1865. [PubMed]
80. Kleeberg K.K., Liu Y., Jans M., Schlegelmilch M., Streese J., Stegmann R.
Development of a simple and sensitive method for the characterization of odorous
waste gas emissions by means of solid-phase microextraction (SPME) and GC–
MS/olfactometry. Waste Manag. 2005;25:872–879. [PubMed]
81. Cai L., Koziel J.A., Lo Y.C., Hoff S.J. Characterization of volatile organic
compounds and odorants associated with swine barn particulate matter using solid-
phase microextraction and gas chromatography-mass spectrometry-olfactometry. J.
Chromatogr. A. 2006;1102:60–72. [PubMed]
82. Félix J.S., Domeño C., Nerín C. Characterization of wood plastic composites
made from landfill-derived plastic and sawdust: Volatile compounds and
olfactometric analysis. Waste Manag. 2013;33:645–655. [PubMed]
83. Ranau R., Kleeberg K.K., Schlegelmilch M., Streese J., Stegmann R., Steinhart
H. Analytical determination of the suitability of different processes for the treatment
of odorous waste gas. Waste Manag. 2005;25:908–916. [PubMed]
84. Lehtinen J., Veijanen A. Odor Monitoring by combined TD-GC-MS-Sniff
technique and dynamic olfactometry at the wastewater treatment plant of low H 2S
concentration. Water Air Soil Pollut. 2011;218:185–196.
85. Rabaud N.E., Ebeler S.E., Ashbaugh L.L., Flocchini R.G. Characterization and
quantification of odorous and non-odorous volatile organic compounds near a
commercial dairy in California. Atmos. Environ. 2003;37:933–940.
86. Trabue S., Scoggin K., McConnell L., Maghirang R., Razote E., Hatfield J.
Identifying and tracking key odorants from cattle feedlots. Atmos. Environ.
2011;45:4243–4251.
87. Zhang S., Cai L., Koziel J.A., Hoff S.J., Schmidt D.R., Clanton G.J., Jacobson
L.D., Parker D.B., Heber A.J. Field air sampling and simultaneous chemical and
sensory analysis of livestock odorants with sorbent tubes and GC-MS/olfactometry.
Sens. Actuators B: Chem. 2010;146:427–432.
88. Laing D.G., Jinks A. Flavor perception mechanism. Trends Food Sci. Technol.
1996;7:387–389.
89. Rowe D.J., editor. Chemistry and Technology of Flavours and Fragrances. CRC
Press; Boca Raton, FL, USA: 2005.
90. Pino J.A., Febles Y. Odor-active compounds in banana fruit cv. Giant Cavendish.
Food Chem. 2013;141:795–801. [PubMed]
91. Pino J.A., Bent L. Odor-active compounds in guava (Psidium guajava L. cv. Red
Suprema) J. Sci. Food Agric. 2013;93:3114–3120. [PubMed]
92. Schieberle P. Recent Developments in Methods for Analysis of Flavour
Compounds and Their Precursors. In: Goankar A., editor. Characterization of Food:
Emerging Methods. Elsevier; Amsterdam, The Netherlands: 1995. pp. 403–431.
93. Aprea E., Carlin S., Giongo L., Grisenti M., Gasperi F. Characterization of 14
raspberry cultivars by solid-phase microextraction and relationship with gray mold
susceptibility. J. Agric. Food Chem. 2010;58:1100–1105. [PubMed]
94. Marković K., Vahčić N., Ganić K.K., Banović M. Aroma volatiles of tomatoes and
tomato products evaluated by solid-phase microextraction. Flavour Fragr. J.
2007;22:395–400.
95. Quijano C.E., Pino J. Analysis of volatile compounds of camu-camu (Myrciaria
dubia (HBK) McVaugh) fruit isolated by different methods. J. Essent. Oil Res.
2007;19:527–533.
96. Quijano C.E., Pino J. Analysis of volatile compounds of cacao maraca
(Theobroma bicolor Humb. et Bonpl.) fruit. J. Essent. Oil Res. 2009;21:211–215.
97. Thaiphanit S., Anprung P. Physicochemical and flavour changes of fragrant
banana (Musa acuminata AAA group “Gross Michel”) during ripening. J. Food
Process. Preserv. 2010;34:366–382.
98. Wang J., Zhi L.Y., Chen R.R., Bao J.Y., Yang G.M. Comparison of volatiles of
banana powder dehydrated by vacuum belt drying, freeze-drying and airdrying. Food
Chem. 2007;104:1516–1521.
99. Nogueira J.M.F., Fernandes P.J.P., Nascimento A.M.D. Composition of volatiles
of banana cultivars from Madeira Island. Phytochem. Anal. 2003;14:87–90.
[PubMed]
100. Pino J., Quijano C. Study of the volatile compounds from plum (Prunus
domestica L. cv. Horvin) and estimation of their contribution to the fruit aroma.
Ciência e Tecnologia de Alimentos (Campinas) 2012;32:76–83.
101. Tamura H., Boonburung S., Yoshizawa T., Varayanond W. The volatile
constituents in the peel and pulp of a green Thai mango, Khieo Sawoei cultivar
(Mangifera indica L.) Food Sci. Technol. Res. 2001;7:72–77.
102. Steinhaus M., Sinuco D., Polster J., Osorio C., Schieberle P. Characterization
of the aroma-active compounds in pink guava (Psidium guajava, L.) by application
of the aroma extract dilution analysis. J. Agric. Food Chem. 2008;56:4120–4127.
[PubMed]
103. Hashinaga F., Shima Y., Itoo S. Production of volatile components of guava
during maturation. Bull. Fac. Agric. Kagoshima Univ. 1987;37:59–64.
104. Vernin G., Verin E., Vernin C., Metzger J. Extraction and GC-MSSPECMA data
bank analysis of the aroma of Psidium guajava L. fruit from Egypt. Flavour Fragr. J.
1991;6:143–148.
105. Chyau C.C., Chen S.Y., Wu C.M. Differences of volatile and nonvolatile
constituents between mature and ripe guava (Psidiumguajava Linn) fruits. J. Agric.
Food Chem. 1992;40:846–849.
106. Pino J., Ortega A., Rosado A. Volatile constituents of guava (Psidium guajava
L.) fruits from Cuba. J. Essent. Oil Res. 1999;11:623–628.
107. Porat R., Tietel Z., Zippori I., Dag A. Aroma volatile compositions of high- and
low-aromatic guava varieties. J. Sci. Food Agric. 2011;91:2794–2798. [PubMed]
108. Takeoka G.R., Buttery R.G., Teranishi R., Flath R.A., Guntert M. Identification
of additional pineapple volatiles. J. Agric. Food Chem. 1991;39:1848–1851.
109. Wu P., Kuo M.C., Hartman T.G., Rosen R.T., Ho C.T. Free and glycosidically
bound aroma compounds in pineapple (Ananas comosus L. Merr.) J. Agric. Food
Chem. 1991;29:170–172.
110. Umano K., Hagi Y., Nakahara K., Shoji A., Shibamoto T. Volatile constituents
of green and ripened pineapple (Ananas comosus [L.] Merr.) J. Agric. Food Chem.
1992;39:170–172.
111. Brat P., Thi Hoang L.N., Soler A., Reynes M., Brillouet J.M. Physicochemical
characterization of a new pineapple hybrid (FLHORAN41 Cv.) J. Agric. Food Chem.
2004;52:6170–6177. [PubMed]
112. Elss S., Preston C., Hertzing F., Heckel F., Richling E., Schreier P. Aroma
profiles of pineapple fruit (Ananas comosus [L.] Merr.) and pineapple products.
Lebensmittel-Wissenschaft und-Technologie. 2005;28:263–274.
113. Beever D.J., Hopkirk G. Fruit Development and Fruit Physiology. In: Warrington
I., Weston G.C., editors. Kiwifruit: Science and Management. Ray Richards
Publisher; Auckland, New Zealand: 1990. pp. 97–126.
114. Atkinson R.G., Gunaseelan K., Wang M.Y., Luo L., Wang T., Norling C.L.
Dissecting the role of climacteric ethylene in kiwifruit (Actinidia chinensis) ripening
using an 1-aminocyclopropane-1-carboxylic acid oxidase knockdown line. J. Exp.
Bot. 2011;62:3821–3835. [PubMed]
115. Sarry J.E., Günata Z. Plant and microbial glycoside hydrolases: Volatile release
from glycosidic aroma precursors. Food Chem. 2004;87:509–521.
116. Reineccius G. Flavour Chemistry and Technology. 2nd ed. Taylor and Francis
Group; Boca Raton, FL, USA: 2006.
117. Garcia C., Stevenson R.J., Atkinson R.G., Winz R.A., Quek S.-Y. Changes in
the bound aroma profiles of “Hayward” and “Hort16A” kiwifruit (Actinidia spp.) during
ripening and GC-olfactometry analysis. Food Chem. 2013;137:45–54. [PubMed]
118. Visscher C.A., Maarse H., Willemsens L.C., Boelens M.H., Nijssen L.M.,
editors. Volatile Compounds in Food: Qualitative and Quantitative Data. TNO
Nutrition and Food Research Institute; Zeist, The Netherlands: 1996. TNO Nutrition
and Food Research Institute. Mango (Mangifera indica L.) pp. 1–10.
119. Chauhan O.P., Raju P.S., Bawa A.S. Mango Flavour. In: Hui Y.H., editor.
Handbook of Fruit and Vegetable Flavours. John Wiley & Sons, Inc.; Hoboken, NJ,
USA: 2010. pp. 319–344.
120. Lalel H.J.D., Singh Z., Tan S.C. Glycosidically-bound aroma volatile
compounds in the skin and pulp of “Kensington Pride” mango fruit at different stages
of maturity. Postharvest Biol. Technol. 2003;29:205–218.
121. Kubikova J., Grosch W. Evaluation of potent odorants of camembert cheese by
dilution and concentration techniques. Int. Dairy J. 1997;7:65–70.
122. Culleré L., Ferreira V., Chevret B., Venturini M.E., Sánchez-Gimeno A.C.,
Blanco D. Characterisation of aroma active compounds in black truffles (Tuber
melanosporum) and summer truffles (Tuber aestivum) by gas chromatography–
olfactometry. Food Chem. 2010;122:300–306.
123. Campo E., Cacho J., Ferreira V. The chemical characterization of the aroma of
dessert and sparkling white wines (Pedro Ximenez, Fino, Sauternes, and Cava) by
gas chromatography-olfactometry and chemical quantitative analysis. J. Agric. Food
Chem. 2008;56:2477–2484. [PubMed]
124. Cullere L., Cacho J., Ferreira V. Comparative study of the aromatic profile of
different kinds of wine cork stoppers. Food Chem. 2009;112:381–387.
125. Escudero A., Campo E., Farina L., Cacho J., Ferreira V. Analytical
characterization of the aroma of five premium red wines. Insights into the role of odor
families and the concept of fruitiness of wines. J. Agric. Food Chem. 2007;55:4501–
4510. [PubMed]
126. Filip R., Ferraro G.E., Bandoni A.L., Bracesco N., Nunes E., Gugliucci A. Mate
(Ilex Paraguariensis) In: Imperato F., editor. Recent Advances in Phytochemistry.
Research Signpost; Kerala, India: 2009. pp. 113–131.
127. Dravnieks A. Atlas of Odor Character Profiles. ASTM; Philadelphia, PA, USA:
1985. p. p. 354.
128. Plutowska B., Wardencki W. Gas Chromatography-Olfactometry of Alcoholic
Beverages. In: St. Angelo A.J., editor. Alcoholic Beverages. Sensory Evaluation and
Consumer Research. Woodhead Publishing Limited; Cambridge, UK: 2012. pp.
101–130.
129. Lee S.M., Seo B.C., Kim Y.S. Volatile compounds in fermented and acid-
hydrolyzed soy sauces. J. Food Sci. 2006;71:C146–C156.
130. Yan L., Zhang Y., Tao W., Wang L., Wu S. Rapid determination of volatile
flavour components in soy sauce using head space solid-phase microextraction and
gas chromatography-mass spectrometry. Chin. J. Chromatogr. 2008;26:285–291.
[PubMed]
131. Gao X.L., Cui C., Zhao H.F., Zhao M.M., Yang L., Ren J.Y. Changes in volatile
aroma compounds of traditional Chinese-type soy sauce during moromi
fermentation and heat treatment. Food Sci. Biotechnol. 2010;19:889–898.
132. Baek H.H., Kim H.J. Solid phase microextraction-gas chromatography-
olfactometry of soy sauce based on sample dilution analysis. Food Sci. Biotechnol.
2004;13:90–95.
133. Petra S., Peter S. Characterization of the key aroma compounds in soy sauce
using approaches of molecular sensory science. J. Agric. Food Chem.
2007;55:6262–6269. [PubMed]
134. Moy Y.S., Lu T.J., Chou C.C. Volatile components of the enzyme-ripened sufu,
a Chinese traditional fermented product of soy bean. J. Biosci. Bioeng.
2011;113:196–201. [PubMed]
135. Grosch W. Detection of potent odorants in foods by aroma extract dilution
analysis. Trends Food Sci. Technol. 1993;4:68–73.
136. Martí M.P., Mestres M., Sala C., Busto O., Gausch C. Solip-phase
microextraction and gas chromatography olfactometry analysis of successively
diluted samples. A new approach of the aroma extract dilution analysis applied to
the characterization of wine aroma. J. Agric. Food Chem. 2003;51:7861–7865.
[PubMed]
137. Blank I., Fay L.B. Formation of 4-hydroxy-2,5-dimethyl-3(2H)-fruanone and 4-
hydroxy-2(or 5)- ethyl-5 (or 2)-methyl-3(2H)-fruanone through Maillard reaction
based on pentose sugars. J. Agric. Food Chem. 1996;44:531–536.
138. Hofmann T., Schieberle P. Identification of potent aroma compounds in
thermally treated mixtures of glucose/cysteine and rhamnose/cysteine using aroma
extract dilution technique. J. Agric. Food Chem. 1997;45:898–906.
139. Lalas S., Dourtoglou V. Use of rosemary extract in preventing oxidation during
deep-fat frying of potato chips. J. Am. Oil Chem. Soc. 2003;80:579–583.
140. Grosch W., Konopka U.C., Guth H. Characterization of Off-Flavours by Aroma
Extract Dilution Analysis. In: St. Angelo A.J., editor. Lipid Oxidation in Food.
American Chemical Society; Washington, DC, USA: 1992. pp. 266–278.
141. Konopka U.C., Guth H., Grosch W. Potent odorants formed by lipid peroxidation
as indicators of the warmed-over flavour (WOF) of cooked meat. Zeitschrift für
Lebensmittel-Untersuchung und -Forschung A. 1995;201:339–343.
142. Christlbauer M., Schieberle P. Characterization of the key aroma compounds
in beef and pork vegetable gravies á la chef by application of the aroma extract
dilution analysis. J. Agric. Food Chem. 2009;57:9114–9122. [PubMed]
143. Guth H., Grosch W. Identification of the character impact odorants of stewed
beef juice by instrumental analysis and sensory studies. J. Agric. Food Chem.
1994;42:2862–2866.
144. Gray J.I., Pearson A.M. Lipid-Derived Off-Flavours in Meat—Formation and
Inhibition. In: Shaididi F., editor. Flavour of Meat and Meat Products. Blackie
Academic and Professional; London, UK: 1994. pp. 129–131.
145. Honikel K.O. The use and control of nitrate and nitrite for the processing of meat
products. Meat Sci. 2008;78:68–76. [PubMed]
146. Sebranek J.G., Bacus J.N. Cured meat products without direct addition of
nitrate or nitrite: What are the issues? Meat Sci. 2007;77:136–147. [PubMed]
147. Lücke F.K. Manufacture of Meat Products without Added Nitrite or Nitrate—
Quality and Safety Aspects. Foodbalt—2008; Proceedings of the 3rd Baltic
Conference on Food Science and Technology; Jelgava, Germany. 2008. pp. 41–45.
148. Sindelar J.J., Milkowski A.L. White Paper, Series No. 3. American Meat Science
Association; Champaign, IL, USA: 2011. Sodium Nitrite in Processed Meat and
Poultry Meats: A Review of Curin and Examining the Risk and Benefit of Its Use; pp.
1–14.
149. Sindelar J.J., Cordray J.C., Sebranek J.G., Love J.A., Ahn D.U. Effects of
varying levels of vegetable juice powder and incubation time on colour, residual
nitrate and nitrite, pigment, pH, and trained sensory attributes of ready-to-eat
uncured ham. J. Food Sci. 2007;72:S388–S395. [PubMed]
150. Corpet D.E. Red meat and colon cancer: Should we become vegetarians, or
can we make meat safer? Meat Sci. 2011;89:310–316. [PubMed]
151. Demeyer D., Honikel K., de Smet S. The World Cancer Research Fund report
2007: A challenge for the meat processing industry. Meat Sci. 2008;80:953–959.
[PubMed]
152. Berdagué J.L., Tournayre P. Gas Chromatography-Olfaction Analysis Device
and Method. Patent WO 2005/001470 A2. 2005 Jan 6;
153. Brunton N.P., Cronin D.A., Monahan F.J. Volatile components associated with
freshly cooked and oxidized off-flavours in turkey breast meat. Flavour Fragr. J.
2002;17:327–334.
154. Mottram D.S. Flavour formation in meat and meat products: A review. Food
Chem. 1998;62:415–424.
155. Nanamori M., Kanatani R., Kihara M., Kawahara K., Hayashi K., Watanabe T.
Effects of nitrogen application on malt modification and dimethyl sulfide precursor
production in two Japanese barley cultivars. J. Sci. Food Agric. 2008;88:1464–1471.
156. Shi X., Zhang X., Song S., Tan C., Jia C., Xia S. Identification of characteristic
flavour precursors from enzymatic hydrolysis-mild thermal oxidation tallow by
descriptive sensory analysis and gas chromatography-olfactometry and partial least
squares regression. J. Chromatogr. B. 2013;913–914:69–76. [PubMed]
157. Song S.Q., Zhang X.M., Hayat K., Huang M.G., Liu P., Karangwa E., Gu F.L.,
Jia C.S., Xia S.Q., Xiao Z.B., et al. Contribution of beef base to aroma characteristics
of beeflike process flavour assessed by descriptive sensory analysis and gas
chromatography olfactometry and partial least squares regression. J. Chromatogr.
A. 2010;1217:7788–7799. [PubMed]
158. Baek H.E., Kim C.J., Ahn B.H., Nam H.S., Cadwallader K.R. Aroma extract
dilution analysis of a beef-like process flavour from extruded enzyme-hydrolyzed
soybean protein. J. Agric. Food Chem. 2001;49:790–793. [PubMed]
159. Kerscher R., Grosch W. Comparative evaluation of potent odorants of boiled
beef by aroma extract dilution and concentration analysis. Zeitschrift für
Lebensmitteluntersuchung und -Forschung A. 1997;204:3–6.
160. Moon S.Y., Cliff M.A., Li-Chan E.C.Y. Odor-active components of simulated
beef flavour analysed by solid phase microextraction and gas chromatography-mass
spectrometry and -olfactometry. Food Res. Int. 2006;39:294–308.
161. Wilson R.A., Mussinan C.J., Katz I., Sanderson A. Isolation and identification
of some sulfur chemicals present in pressure-cooked beef. J. Agric. Food Chem.
1973;21:873–876. [PubMed]
162. Leod G.M., Ames J.M. Capillary gas chromatography-mass spectrometric
analysis of cooked ground beef aroma. J. Food Sci. 1986;51:1427–1434.
163. Peterson R.J., Chang S.S. Identification of volatile flavour compounds of freshly
frozen beef stew and a comparison of these with those of canned beef stew. J. Food
Sci. 1982;47:1444–1448.
164. Wu Y.-F.G., Cadwallader K.R. Characterization of the aroma of a meat-like
process flavouring from soybean-based enzyme-hydrolyzed vegetable protein. J.
Agric. Food Chem. 2002;50:2900–2907. [PubMed]
165. Song S.Q., Zhang X.M., Hayat K., Liu P., Jia C.S., Xia S.Q., Xiao Z.B., Tian
H.X., Niu Y.W. Formation of the beef flavour precursors and their correlation with
chemical parameters during the controlled thermal oxidation of tallow. Food Chem.
2011;124:203–209.
166. Bueno M., Resconi V.C., Mar Campo M., Cacho J., Ferreira V., Escudero A.
Gas chromatographic-olfactometric characterisation of headspace and mouthspace
key aroma compounds in fresh and frozen lamb meat. Food Chem. 2011;129:1909–
1918.
167. Resconi V.C. Ph.D. Thesis. Universidad de Zaragoza; Zaragoza, Spain: 2010.
Efecto de la alimentación sobre la calidad sensorial y los compuestos aromáticos en
carne de cordero y de vacuno (in Spanish)
168. ISO/TC 93—Starch (Including Derivatives and by-Products) [(accessed on 27
November 2013)]. Available online:
http://wwwiso.org/iso/home/store/catalogue_tc/catalogue_tc_browse.htm?commid=
50564&development=on.
169. Paravisini L., Gourrat-Pernin K., Gouttefangeas C., Moretton C., Nigay H.,
Dacremonta C., Guicharda E. Identification of compounds responsible for the
odorant properties of aromatic caramel. Flavour Fragr. J. 2012;27:424–432.
170. Curioni P.M.G., Bosset J.O. Key odorants in various cheese types as
determined by gas chromatography-olfactometry. Int. Dairy J. 2002;12:959–984.
171. Semmelroch P., Grosch W. Analysis of roasted coffee powders and brews by
gas chromatography-olfactometry of headspace samples. Lebensm.-Wiss. u.-
Technol. 1995;28:310–313.
172. Cantergiani E., Brevard H., Krebs Y., Feria-Morales A., Amadò R., Yeretzian
C. Characterization of the aroma of green Mexican coffee and identification of
mouldy/earthy defect. Eur. Food Res. Technol. 2001;212:648–657.
173. Thomsen M., Martin C., Mercier F., Tournayre P., Berdagué J.-L., Thomas-
Danguin T., Guichard E. Investigating semi-hard cheese aroma: Relationship
between sensory profiles and gas chromatography-olfactometry data. Int. Dairy J.
2012;26:41–49.
174. Analyse sensorielle. AFNOR; Paris, France: 1995. Recherche et Sélection de
Descripteurs Pour l'élaboration d'un Profil Sensoriel, Par Approche
Multidimensionnelle (in French) p. p. 26. NF ISO 11035 Juillet 1995.
175. Analyse Sensorielle. AFNOR; Paris, France: 2003. Méthodologie. Directives
Générales pour l'établissement d'un Profil Sensoriel (in French) p. p. 24. NF EN ISO
13299 Mai 2010.
176. AcquiSniff. a software for a complete analysis of GCO data; Institut National de
la Recherche Agronomique (INRA) UR QuaPA T2A; Saint Genes Champanelle,
France: 2007.
177. Curtis T., Williams D.G. Introduction to Perfumery. Micelle Press; New York,
NY, USA: 2001.
178. Costa R., d'Acampora Zellner B., Crupi M.L., de Fina M.R., Valentino M.R.,
Dugo P., Dugo G., Mondello L. GC-MS, GC-O and enantio-GC investigation of the
essential oil of Tarchonanthus camphoratus L. Flavour Fragr. J. 2008;23:40–48.
179. Sell C. Perfumery Materials of Natural Origin. In: Pybus D.H., Sell C.S., editors.
The Chemistry of Fragrances. Royal Society of Chemistry; Cambridge, UK: 1999.
pp. 24–50.
180. Ito F., Iwasaki M., Watanabe T., Ishikawa T., Higuchi Y. The first total synthesis
of kwakhurin, a characteristic component of a rejuvenating plant, “kwao keur”:
Toward an efficient synthetic route to phytoestrogenic isoflavones. Org. Biomol.
Chem. 2005;3:674–681. [PubMed]
181. Sucontphunt A., de-Eknamkul W., Nimmannit U., Dimitrijevich D.S., Gracy W.R.
Protection of HT22 neuronal cells against glutamate toxicity mediated by the
antioxidant activity of Pueraria candollei var. mirifica extracts. J. Natl. Med.
2011;65:1–8. [PubMed]
182. Ngassoum M.B., Ousmaila H., Ngamo L.T., Maponmetsem P.M., Jirovetz L.,
Buchbauer G. Aroma compounds of essential oils of two varieties of the spice plant
Ocimum canum Sims from northern Cameroon. J. Food Compos. Anal.
2004;17:197–204.
183. Tu N.T.M., Onishi Y., Son U.S., Ogawa E., Ukeda H., Sawamura M.
Characteristic odor components of Citrus inflata Hort. ex Tanaka (Mochiyu) cold-
pressed peel oil. Flavour Fragr. J. 2003;18:454–459.
184. Choi H.S., Kondo Y., Sawamura M. Characterization of the odor-active volatiles
in citrus hyuganatsu (Citrus tamurana Hort.ex Tanaka) J. Agric. Food Chem.
2001;49:2404–2408. [PubMed]
185. Miyazawa M., Kawauchi Y., Utsumi Y., Takagashi T. Character impact odorants
of wild edible plant Cacalia hastate L. ver. orientalis-used in Japanese traditional
food. J. Oleo Sci. 2010;59:527–533. [PubMed]
186. Stuetz R., Frechen F.B., editors. Odors in Wastewater Treatment:
Measurement, Modelling and Control. IWA Publishing; London, UK: 2001.
187. Nicell J.A. Assessment and regulation of odor impacts. Atmos. Environ.
2009;43:196–206.
188. Brennan B. Odor nuisance. Water Waste Treat. 1993;36:30–33.
189. Schiffman S.S. Livestock odors: Implications for human health and well-being.
J. Anim. Sci. 1998;76:1343–1355. [PubMed]
190. Air Quality—Determination of Odor Concentration by Dynamic Olfactometry.
Committee for European Normalization (CEN); Brussels, Belgium: 2003. EN13725.
191. Standard Practice for Determination of Odor and Taste Thresholds by a Forced-
Choice Ascending Concentration Series Method of Limits. American Society for
Testing and Materials; Philadelphia, PA, USA: 2004. ASTM E679–04.
192. Brattoli M., de Gennaro G., de Pinto V., de Marinis Loiotile A., Lovascio S.,
Penza M. Odour detection methods: Olfactometry and chemical sensors. Sensors.
2011;11:5290–5322. [PMC free article] [PubMed]
193. Wright D.W., Eaton D.K., Nielsen L.T., Kuhrt F.W., Koziel J.A., Spinhirne J.P.,
Parker D.B. Multidimensional gas chromatography-olfactometry for the identification
and prioritization of malodors from confined animal feeding operations. J. Agric.
Food Chem. 2005;53:8663–8672. [PubMed]
194. Hammond E.G., Fedler C., Smith R.J. Analysis of particle-borne swine house
odors. Agric. Environ. 1981;6:395–401.
195. Honey L.F., McQuitty J.B. Some physical factors affecting dust concentrations
in a pig facility. Can. J. Agric. Eng. 1979;21:9–14.
196. Razote E.B., Maghirang R.G., Seitz L.M., Jeon I.J. Characterization of volatile
organic compounds on airborne dust in a swine finishing barn. Trans. ASAE.
2004;47:1231–1238.
197. Hammond E., Fedler C., Junk G. Identification of dust-borne odors in swine
confinement facilities. Trans. ASAE. 1979;22:1186–1192.
198. Hoff S.J., Bundy D.S., Li X.W. Dust Effects on Odor and Odor Compounds.
Proceedings of the International Symposium Ammonia and Odor Control from
Animal Production Facilities; Vinkeloord, The Netherlands. 6–10 October 1997; pp.
101–110.
199. Cai L., Koziel J.A., Liang Y., Nguyen A., Xin H. Evaluation of zeolite for control
of odorants emissions from simulated poultry manure storage. J. Environ. Qual.
2007;36:184–193. [PubMed]
200. Pillai S.M., Parcsi G., Wang X., Stuetz R.M. Odor abatement of poultry litter
using odor control products. Chem. Eng. Trans. 2012;30:247–252.
201. Agus E., Lim M.H., Zhang L., Sedlak D. Odorous compounds in municipal
wastewater effluent and potable water reuse systems. Environ. Sci. Technol.
2011;45:9347–9355. [PubMed]
202. Sohn J.H., Parcsi G., Dunlop M.W., Gallagher E.M., Stuetz R.M. Air quality
monitoring and characterisation at two commercial broiler farms. Chem. Eng. Trans.
2010;23:141–146.
203. Espert A., de las Heras L.A., Karlsson S. Emission of possible odorous low
molecular weight compounds in recycled biofiber/polypropylene composites
monitored by head-space SPME-GC-MS. Polym. Degrad. Stab. 2005;90:555–562.
204. Knudsen H.N., Clausen P.A., Wilkins C.K., Wolkoff P. Sensory and chemical
evaluation of odorous emissions from building products with and without linseed oil.
Build. Environ. 2007;42:4059–4067.
205. Shirasu M., Nagai S., Hayashi R., Ochiai A., Touhara K. Dimethyl trisulfide as
a characteristic odor associated with fungating cancer wounds. Biosci. Biotechnol.
Biochem. 2009;73:2117–2120. [PubMed]
206. Mebazaa R., Rega B., Camel V. Analysis of human male armpit sweat after
fenugreek ingestion: Characterisation of odor active compounds by gas
chromatography coupled to mass spectrometry and olfactometry. Food Chem.
2011;128:227–235. [PubMed]

PLoS One. 2015; 10(9): e0137805.

XXVIII. Gradual Reduction in Sodium Content in Cooked Ham, with
Corresponding Change in Sensorial Properties Measured by Sensory
Evaluation and a Multimodal Machine Vision System

Kirsti Greiff,1,2,* John Reidar Mathiassen,1 Ekrem Misimi,1 Margrethe Hersleth,3 and
Ida G. Aursand1
Ramona Natacha PENA i SUBIRÀ, Editor

Abstract

The European diet today generally contains too much sodium (Na +). A partial
substitution of NaCl by KCl has shown to be a promising method for reducing sodium
content. The aim of this work was to investigate the sensorial changes of cooked
ham with reduced sodium content. Traditional sensorial evaluation and objective
multimodal machine vision were used. The salt content in the hams was decreased
from 3.4% to 1.4%, and 25% of the Na+ was replaced by K+. The salt reduction had
highest influence on the sensory attributes salty taste, after taste, tenderness,
hardness and color hue. The multimodal machine vision system showed changes in
lightness, as a function of reduced salt content. Compared to the reference ham
(3.4% salt), a replacement of Na+-ions by K+-ions of 25% gave no significant
changes in WHC, moisture, pH, expressed moisture, the sensory profile attributes
or the surface lightness and shininess. A further reduction of salt down to 1.7–1.4%
salt, led to a decrease in WHC and an increase in expressible moisture.

Introduction

A high consumption of sodium has been directly associated with a greater likelihood
of increased blood pressure, which in turn has been directly related to the
development of cardiovascular and renal diseases [1]. For these reasons, national
and international bodies have set targets for a reduction in sodium consumption
down to 2 g/day [2–4].

Due to the low fat content, cooked ham is a good source of animal protein in the diet.
Usually when manufacturing cooked ham, the pork meat is coarsely ground into
pieces that are tumbled with brine (containing water, salt, nitrite and other ingredients
such as phosphate and polysaccharides). During the tumbling with brine, the salt-
soluble myofibrillar proteins are extracted and they form a network. The network gels
during the heating and chilling process [5], thereby solidifying the ham and
minimizing cooking loss. Reducing salt contents in cooked ham increases cooking
loss [6] and thereby the production costs.

Sodium chloride is added to a wide variety of foods to increase shelf life and
palatability (e.g. in soups, meat products, bread, sauces and snacks). When it comes
to sensory quality, sodium chloride contributes to saltiness and the overall flavor,
while also suppressing bitterness [7]. Accordingly, the development of palatable
sodium-reduced products is important in order to guide consumers towards more
healthy food choices. Partial substitution of NaCl by KCl has shown to be one of the
best alternatives for reducing sodium content [5, 8]. Studies have shown that KCl
induces a bitter taste at high concentrations, in dry-cured loins [9], fermented
sausage[10] and cooked ham [11]. Those results show that the maximum level of
KCl replacement may vary among different type of product.

It is essential to characterize and understand the sensory effects of sodium reduction

in foods. Sensory analysis is the most appropriate approach to fully describe the
sensory perception of foods and descriptive profiling has been a popular sensory
technique for cognitive descriptions of products in many years. These techniques
give a complete sensory description of products and make it possible to identify
underlying ingredient and process variables [12]. For measurement of flavor, no
instruments currently exist, that can replace the human senses and describe the
sensory perception. For appearance and texture attributes, the situation is different,
as there are several examples of studies where instrumental data representing
appearance and texture attributes, show a strong correlation with data from trained
sensory panels [13–16].

In particular, classification of ham qualities (pork and turkey) is possible based on

color and textural features extracted from digital color images. Logistic regression,
used on these features, could to a large degree explain consumers' responses for
visually-based sensorial attributes [13]. The qualities of pork ham could be
distinguished based on visual texture characteristics extracted from fractal analysis
of digital color images [16]. This showed that the fractal-based features are able to
quantify quality-specific visual texture characteristics. A combination of geometric
and texture features, extracted from digital color images, were input to multivariate
classifiers, including artificial neural networks (ANNs), and were capable of
predicting meat tenderness–as measured by shear force [17]. The eating quality of
beef could be predicted from wavelet surface texture and other texture features
extracted from digital color images. For the optical setup, a polarizer filter was placed
on the lens and the meat pieces were surface-dried to mitigate some specular
reflection [14]. Other machine vision approaches have been designed specifically to
improve the objective analysis of color and texture in transclucent or semi-
transparent muscle foods [18]. These works and others [15, 19] show that computer
vision, with optimal illumination in combination with appropriate feature extraction
and classifiers, can be used to predict sensorial properties from digital color images.

The effect of reduced salt/sodium content on cooked ham has previously been
investigated by different authors [5, 20, 21], but nobody, to the best of our
knowledge, has used the combination of sensorial Descriptive Analysis (DA) and
multimodal machine vision system to investigate the changes in the sensorial
properties in cooked ham with effect of reduced salt/sodium content. In this paper, a
proof-of-concept novel dual-polarization multimodal machine vision system is used,
that is new to meat imaging [22].

The aim of this study was to investigate sensory quality of cooked pork ham
containing gradually reduced salt/sodium content and a partial replacement of
sodium (Na+) by (K+). The product quality parameters addressed were
physicochemical and sensory properties obtained by sensory descriptive analyses
and a multimodal machine vision system.

Materials and Methods

Samples and Chemicals

Cooked ham preparation

Eight different formulas of cooked hams were manufactured by Espeland AS, Ålgård,
Norway 26–28 February 2013. Each of the formulas contained 20 kg of fresh
coarsely grounded (perforated disc with hole size 30 mm), lean pork meat (M.
semimembranosus, M. adductor, M. semitendinosus, M. biceps femoris and M.
psoas major), obtained from a local meat processing company (PrimaGruppen,
Norway) four days after slaughtering. The samples are identified with sample IDs
having the following codes: Na100, NaK100, NaK80, NaK60, NaK40, NaK20, which
vary in the % of the amount of NaCl relative to the reference (Na100) and addition
of potassium (NaK). See Table 1 for details. Reference cooked ham (Na100) had no
added potassium, containing only potassium naturally occurring in the meat
(approximately 330 mg K+/100g). The amount of sodium in the reference (Na100) in
the formulation was at same level as in a commercially cooked ham, 1.24 g
Na+/100g, corresponding to 3.1% salt. The mole ratio of sodium (Na+) and potassium
(K+) was kept constant (Na+: K+ = 3:1) in all formulas containing potassium (NaK),
were the contribution of all sodium and potassium in all ingredients were taken into
account. The salt (NaCl) content (g/100g) was calculated as: added Na+ (including
natural sodium content in the meat) x 2.5 (recalculation factor for NaCl from Na +),
according on the provision of food information to consumers [23]. The amount of
total salt content in the formulation was gradually decreased to 20% of the amount
of the salt in the reference cooked ham, corresponding to 1.3% salt (NaK20). The
brine solution for each formulation was prepared by dissolving 200 g commercial
phosphate mixture (57% P2O5)(A. B. Corneliussen AS, Norway), 150 g gelatin
(Gelita Sweden AB, Sweden), 60 g carrageenan (CEAMSA, Pontverda, Spain), 450
g Na-lactate (Purac Biochem, Gorichem, The Netherlands) and 150 g maltodextrin
(Avebe U.A., The Netherlands) in 2.5 L tap water (t = 2–4°C). All formulas, excluding
the reference ham (Na100), were added 3.3 g sodium nitrite (Merck, KGaA
Darmstadt, Germany). Commercial food salt containing nitrite salt (Hoff Norske
Potetindustrier, Gjøvik, Norway) was added to the reference ham (Na100). Sodium
chloride (Hoff Norske Potetindustrier, Gjøvik, Norway) and potassium chloride
(Culinar Lyckeby, Sweden) were added to the brine to achieve the expected Na + and
K+ content as shown in Table 1.

Table 1. Sample ID, description of the sample, estimated sodium, potassium and
NaCl content in the preparation of the eight cooked hams.
Each formulation varied in its amount of total salt. The reference cooked ham
contained nitrite salt (Na 100 Reference). In the rest of the formulas (NaK100,
NaK80, NaK60, NaK40 and NaK20) the amount of total salt content was gradually
decreased to 20% of the amount of the salt in the reference cooked ham. To keep
the mole ratio of sodium (Na+) and potassium (K+) constant (Na+: K+ = 3:1), all
contribution of sodium and potassium from the raw material and the ingredients was
added when calculation of the mole ratio.

Estimated Estimated % reduction of

Sample sodium potassium sodium
Description NaCl**(%)
ID content content (g/100 compared to
(g/100 g ham) g ham) the reference

Na 100 Reference ham 1.24 0.28 3.1 0

100% of the
NaK
amount of salt 1.06 0.61 2.6 14
100_1*
in the reference

100% of the
NaK
amount of salt 1.06 0.61 2.6 14
100_2*
in the reference

80% of the
NaK 80 amount of salt 0.93 0.53 2.3 25
in the reference

60% of the
NaK
amount of salt 0.80 0.46 2.0 35
60_1*
in the reference

60% of the
NaK
amount of salt 0.80 0.46 2.0 35
60_2*
in the reference

40% of the
NaK 40 amount of salt 0.67 0.38 1.7 46
in the reference

20% of the
NaK 20 amount of salt 0.53 0.31 1.3 57
in the reference

* Produced twice to control the reproducibility of the production

** The salt content (%) is calculated by using the formula: added sodium (Na +),
including natural Na+ content in the meat x 2.5 (recalculation factor for NaCl from
Na+)

The brine was added to the pork meat in a tumbler (Fatosa, Barcelona, Spain)
without vacuum and massaged for 2 x 25 min, stored 20 h at 2°C, and then tumbled
for 15 more min. The meat matrix was stuffed in to a plastic casing (160 mm in
diameter, Viscofan S. A., Navarra, Spain). Two parallel production batches of the
hams with formulation NaK100 and NaK60 were prepared to control the
reproducibility of the production. Seven hams (3.0 kg) from each formulation were
produced. The hams were pressed in molds and kept at 0–2°C for 20 h before
cooking. The hams were cooked in a cooking chamber (Fessmann, Germany) until
they reached a core temperature of 78°C, chilled down in tap water and kept in a
cold-storage chamber at 0–2°C. Two of the hams per formula were sliced in 20 slices
(16 ± 1 g) 12 days after production, and alternate of the slices were subjected to
sensorial evaluation and analysis of machine vision, respectively. Sensorial
evaluation and machine vision analyses were carried out directly after slicing. The
remaining cooked hams were stored at 4°C for 15 days after production before
chemical analysis. In addition, one slice (approximately 100 g) was used for further
analysis.

Chemicals

Ammonium chloride (NH4Cl), ammonium hydroxide (NH4OH) and ammonium

hydrogen fluoride (NH4FHF < 1%, LD50 mg/kg not found) of analytical-reagent grade
(Thermo Fisher Scientific, USA). Chemicals of food grade: sodium tripoly phosphate
(Na5P3O10) (A. B. Corneliussen AS, Norway), Na-lactate (C3H5NaO3) (Purac
Biochem, Gorichem, The Netherlands) and sodium nitrite (NNaO2)(Merck, KGaA
Darmstadt, Germany).

Imaging Setup and Image Acquisition

The imaging system consisted of a ColorRanger multimodal line-scan camera (SICK

IVP AB, Linköping, Sweden), two high-intensity white LED linear array lights (Banner
Engineering, Minneapolis, MN, USA), with polarizers on the LED arrays and in front
of the camera lens. Additionally, a nematic liquid crystal industrial-grade polarization
rotator (ARCoptix SA, Neuchâtel, Switzerland) was placed in front of the polarizer
on the camera lens. The polarization rotator is controlled by the image acquisition
PC and enables effectively to rotate the polarizer on the lens by means of an electric
signal. The purpose of rotating the polarizer direction on the lens polarizer relative
to the polarizer on the LEDs, was to be able to capture light reflected from the object
for two polarization states.

The experiment sought to separately image the subsurface color and the surface
color of the ham slices, in order to separately image the light that was scattered in
the subsurface from the light that was directly reflected off the surface. Subsurface
imaging revealed the bulk color of the ham near the surface, whereas surface
imaging revealed the surface roughness, shininess and other effects such as
"mother-of-pearl" appearance commonly seen in some hams. To separate the
subsurface from the surface image, two images were acquired–an image I∥ with LED
and camera polarizers oriented parallel to each other, and an image I⊥ with the
polarizers oriented perpendicular to each other. Image I∥ will image both the
subsurface and the surface components, whereas 0049⊥ will image the subsurface
components only, and hence the difference I∥—I⊥ between the two images will image
only the surface components of the light. This principle of light interactions as a
function of polarization state is illustrated in images in Fig 1.

Fig 1. Illustration of the components of light interacting with the imaged raw
material, imaged using parallel polarizers (left), crossed polarizers (middle)
and the difference between the two (right).

The extracted imaging features were simply the mean of the red, green and blue
(r,g,b) values, as acquired with the ColorRanger, over the entire ham slice and for
both polarizer orientations. Thus, for each ham slice the (r∥,g∥,b∥) and (r⊥,g⊥,b ⊥)
values were obtained. The sum of (r⊥,g⊥,b⊥) indicated the lightness of the sample
and an increase either in r⊥,g⊥ or b⊥, indicated increased lightness. An average of
nine slices per formulation of cooked hams, were used in the image acquisition
experiment.

Supplementary Image Acquisition

Image acquisition of the hams used for the sensory evaluation was done with a
polycarbonate window in the imaging setup, which unfortunately affected the
polarization of the imaged light. This resulted in suboptimal images. Therefore, only
the blue channel images (b∥ and b⊥) were used for analysis and a supplementary
image acquisition was done to obtain more optimal images. Here, the polycarbonate
window was replaced with an optical grade AR-coated BK7 glass window. A
supplementary production of cooked hams was conducted in the same manner as
for the main experiment, and three formulation of cooked ham (Na100, NaK80 and
NaK60) were manufactured for the supplementary image acquisition, solely for
validation of the multimodal imaging acquisition system. The formulation, production
process, and storage were similar to the cooked ham preparation as described
above. An example set of images is seen in Fig 2. The images in Fig 2 are included
in order to provide the reader with a visual understanding of the possibilities of using
multi-modal imaging of cooked ham with dual polarization states.

Fig 2. Images of a slice of ham, from supplementary Image Acquisition using

parallel polarizers (left), crossed polarizers (middle) and the difference
between the two (right).

Chemical analysis

Water activity (aW) was determined in cooked hams with a fast water activity-meter
(GBX FAst/lab, Romans sur Isère Cedex, France). The pH measurements on light
and dark muscle, in brine and cooked ham were carried out using a digital pH-meter
WTW pH3110 (Weilheim, Germany) with a puncture electrode (WTW A 120513078,
Weilheim, Germany). The moisture content was determined by drying three parallel
samples of 5 g of minced cooked ham at 105°C for 24 hours [24]. Water holding
capacity (WHC) was determined on minced cooked ham by low-speed
centrifugation as described by Eide, Børresen [25] with a centrifugation force of 210
g. The WHC is expressed as the percentage of water retained in the mince after
centrifugation for 5 min. The analyses were run in quadruplicate. Sodium contents
in cooked hams were measured in an extract of the ham samples, using a Dual
StarTM pH/ISE Meter (Thermo Fisher Scientific, Waltham, MA, USA) with a Na-
selective electrode (Ross® Sodium Ion Selective Electrode, Thermo Fisher
Scientific, Waltham, MA USA). For preparing the extract, 7.5g of ham was
homogenized in ultra-pure water using an Ultra-turrax T-25 D (IKA, Labortechnik,
Staufen, Germany) at 9000 rpm for 1 min. Then, samples were warmed up to 90°C
for 30 min, cooled down to room temperature, transferred to a volumetric flask and
diluted to 200 mL with ultra-pure water. Finally, samples were filtered through a
cellulose filter paper (Whatman n° 1, Whatman International Ltd., Maidstone, UK).
The Na-selective electrode method was a modification of the Kivikari [26] method.
In this study the direct calibration method was used, whereas Kivikari, used the
known addition method. A calibration curve was made by using three standards of
analytical-grade NaCl from (Merck KGaA, Darmstadt, Germany) and Sodium ionic
strength adjustor (Thermo Fisher Scientific, Waltham, MA, USA) was added to all
solutions to ensure that samples and standards had similar ionic strength.

The modified method of Grau and Hamm [27] was used to measure expressible
moisture (EM) for cooked hams. Samples were punched out with a hollow drill (25
mm in diameter) from cooked slices of hams (15 mm thick). Each sample was placed
in the middle of ten filter papers (Whatman No. 1) and pressed down with a flat-
ended cylindrical plunger (80 mm diameter), by single compression test mode and
a test speed of 0.8 mm/s until 50% compression of total height, using a Texture
Analyzer T.A.XT2 (Stable Micro System, Surrey, U.K.). Expressible moisture was
determined as the amount of water released per gram of meat and was expressed
in percentage.

Sensory evaluation

Descriptive Analysis (DA) was conducted by a trained sensory panel according to

Generic Descriptive Analysis as described by [12]. All assessors (n = 9) were
selected and trained in accordance with ISO 8586–1 (ISO, 1993) and the test was
done in a sensory laboratory designed in accordance with ISO 8589 (ISO, 2007).
Each assessor evaluated all samples using EyeQuestion for direct recording of data
(v3.8.7, Logic8, Elst(Gld), The Netherlands). A list of 21 attributes was generated
from a brainstorming session with the assessors. This list included attributes
representing appearance (color-hue, color intensity, whiteness, color evenness,
shiny, marbling, cohesiveness, visible moist on surface), odor (sour, pork meat,
metal), taste/flavor (sourness, sweetness, saltiness, bitterness, pork meat, metal,
after flavor) and texture (hardness, juiciness, tenderness). Attributes were evaluated
using a continuous, non-structured scale ranging from no intensity (1) on the left to
high intensity (9) on the right. The assessors had previous experience with analyses
of meat products including ham and were calibrated on selected attributes in a pre-
test. DA was performed during 6 sessions on one day, in which 8 different hams
were served in two replicates (totally 16 samples). The hams were sliced by a
machine to a weight of 16 ± 1 g, and each assessor got one slice served per sample.
Appearance attributes were evaluated on the surface of the ham slice, while the
texture was evaluated biting over a rolled slice. The serving order was randomized
across all sessions. All samples were expectorated and unsalted crackers and
lukewarm water was available for rinsing.

Data analysis

The influence of the different levels of sodium on physicochemical properties was

studied through one-way analysis of variance (ANOVA), (Minitab 16, Minitab Inc.) In
cases where the effect was defined as significant (p<0.05), the means were
compared using Tukeys test to find the significant difference between the samples
containing different amounts of salt.

PanelCheck 1.3.2 (www.panelcheck.com) was used to evaluate the panel

performance in the pre-test. For determination of sensory attributes discriminating
between samples, a two-way ANOVA with product as a fixed factor, panelist as a
random factor and product x panelist as an interaction factor was performed. Tukey’s
Multiple Comparisons Test was applied to determine which products that was
significantly different. The significance level was defined to p<0.05. (The ANOVA
was run by SAS 9.2, SAS Institute, Inc., Cary, NS, USA).

Results and Discussion

Chemical analysis

The results of the chemical analyses of cooked ham at different salt/sodium levels
are summarized in Table 2.

Table 2. Water holding capasity (WHC), moisture, pH, sodium, salt and expressible
moisture of cooked ham prepared with different levels of salt at constant Na +: K+
mole ratio 3:1.

The reference cooked ham contained nitrite salt (Na 100 Reference). In the rest of
the formulas (NaK100, NaK80, NaK60, NaK40 and NaK20) the amount of salt was
gradually decreased to 20% of the amount of salt in the reference ham. Mean values
± SD (n = 3).

WHC Moisture Salt Expressible

Group pH Sodium(g/100g)
(%) (%) (g/100g) moisture(%)

78.45 ± 74.58 ± 6.16 ± 3.38 ±

Na 100 1.35 ± 0.17a 0.96 ± 0.30abc
1.03abc 0.23bc 0.02 ab 0.43a

NaK 82.57 ± 73.63 ± 6.15 ± 2.70 ±

1.08 ± 0.01b 0.86 ± 0.16ab
100_1 1.68a 0.26c 0.04 ab 0.03b

NaK 80.59 ± 73.67 ± 6.14 ± 2.58 ±

1.03 ± 0.04bc 0.84 ± 0.16a
100_2 0.50ab 0.15c 0.04a 0.10bc

81.55 ± 74.40 ± 6.19 ± 2.25 ±

NaK 80 0.92 ± 0.01cd 1.07 ± 0.23de
1.13ab 0.14bc 0.02 abcd 0.03cd

NaK 80.83 ± 73.83 ± 6.21 ± 2.05 ±

0.82 ± 0.01de 1.55 ± 0.39de
60_1 0.16ab 0.19a 0.02 bcde 0.03de
 WHC Moisture Salt Expressible
Group pH Sodium(g/100g)
(%) (%) (g/100g) moisture(%)

NaK 79.54 ± 75.14 ± 6.21 ± 2.03 ±

0.81 ± 0.01de 1.12 ± 0.17abcd
60_2 0.88abc 0.18ab 0.02bcd 0.03de

76.31 ± 75.43 ± 6.24 ± 1.73 ±

NaK 40 0.69 ± 0.01ef 1.49 ± 0.23cde
0.25cde 0.28ab 0.02cdef 0.03ef

72.71 ± 75.40 ± 6.25 ± 1.38 ±

NaK 20 0.55 ± 0.01f 1.96 ± 0.35e
1.83de 0.46ab 0.01 def 0.03f

p-value 0.000 0.000 0.000 0.000 0.000 0.000

a-gDifferent upper-case letters within a column indicate significant differences

(p<0.005) between different groups. Means that do not share a common letter are
significantly different.

Sodium content

The salt contents (%) in the cooked ham were calculated as measured Na+-content
multiplied with 2.5. The measured sodium content in the reference ham (Na 100
added as Na+—ions only) was 1.35 g Na+/100 g, this result was slightly higher than
estimated in the formulation of the sample, 1.24 g Na +/100g. The rest of the hams,
were prepared with different levels of total salt including a constant Na +:K+ mol ratio
of 3:1, and the measured sodium content were in good agreement with the estimated
in the formulation.

Water holding capacity (WHC)

As expected, a decrease in WHC was observed in hams where the salt content was
reduced by 60% (NaK40) and 80% (NaK20) (on molar basis), corresponding to a
measured salt content of 1.73 and 1.38% salt, respectively. These results are in
accordance with Hamm (1972) and Offer & Knight (1988) who found that sodium
chloride increases the water-binding of meat [28, 29]. Albarracín, Sánchez [30]
explained that increased WHC in meat, resulting from increasing salt content, is due
to the anions (Cl-) preferentially binding to protein molecules. The moisture content
in hams added a mixture of sodium and potassium salt without reduction of the total
salt, 2.70% salt (NaK100) was significant lower than in cooked ham containing
1.38% salt (NaK20).

pH

The pH in the raw material was within the range of 5.5–5.8. These pH values are
similar to typical ultimate pH values in Norwegian pork [31]. The pH of the hams
increased from pH 6.15 ± 0.02 to 6.25 ± 0.01 with decreasing sodium content. The
pH in hams with the lowest levels of sodium, 1.73% salt (NaK40) and 1.38% salt
(NaK20), were significant higher than the reference containing 3.38% salt (Na100)
and the cooked ham with the mixture of sodium and potassium salt (NaK100). These
findings are in accordance with Puolanne, Ruusunen [32], who found an average
decrease, about 0.1 pH-units/%-units of salt, in cooked sausages.

Expressible moisture

The expressible moisture increased with decreasing salt/sodium content, and those
hams containing 1.38% salt (NaK20) had significantly higher expressible moisture
than those containing 3.38% salt (Na 100).

Water activity (aw)

The aw decreased slightly from 0.97 to 0.96 with increasing salt content, although
the differences were not significant (data not shown). According to Mossel, Corry
[33], such levels of aw in cooked hams were too high to represent a hurdle for
unwanted spoilage- and pathogenic microorganisms.

Compared to the reference ham (Na100), a replacement of Na+-ions by 25% of K+-

ions gave no significant changes in WHC, moisture, pH or expressible moisture. This
replacement results in a total sodium content of approximately 1.1 g /100g cooked
ham. This finding is in accordance with Zanardi, Ghidini [34] who found no effects
on pH and water activity in reduced sodium when replacing (Na +) by a mixture of
KCl, CaCl2 and MgCl2 in Cacciatore salami compared to the traditional recipe.

Descriptive sensory profile

Table 3 gives an overview of the mean values for the significant sensory attributes
plus the attributes whiteness and shiny, as appearance attributes were most relevant
for data from vision analysis. The table shows significance for 8 out of a total of 21
sensory attributes.

Table 3. Sensory properties in the cooked hams.

Apperance Taste and odour Texture

Meta Meta
Colo
Gro Cohesiven Whiten Shi Salty l l After Tendern Hardne
ur
up ess ess ny taste flavo odo taste ess ss
hue
ur ur

Na 6.67a 6.18ab 4.74a 4.94a 5.92ab

6.97ab 5.61ab 3.71 5.36a 4.78ab
100 bc cd bc bc c
 Apperance Taste and odour Texture

Meta Meta
Colo
Gro Cohesiven Whiten Shi Salty l l After Tendern Hardne
ur
up ess ess ny taste flavo odo taste ess ss
hue
ur ur

NaK
6.84a
100_ bc 7.12ab 5.59ab 4.21 6.53ab 5.12a 5.19a 6.21ab 5.02a 5.03a
1

NaK
6.97a 4.83a 4.91a
100_ b 7.42a 5.49ab 3.76 6.58a bc bc 6.36a 5.11a 4.91ab
2

NaK 6.97a 6.18ab 4.92a 4.84a 6.14ab

7.18a 5.61ab 4.09 5.09a 5.02a
80 b cd b bc c

NaK 6.77a 5.55bc 4.76a 4.94a

6.94ab 5.79ab 3.32 5.52bc 5.40a 4.73ab
60_1 bc de bc bc

NaK 6.49a 4.64a 5.08a

7.27a 5.72ab 3.98 5.26de 5.54bc 5.88abc 4.50abc
60_2 bcd bc b

NaK 6.77a 4.34a 4.63a 5.73ab

6.66abc 5.91ab 3.76 4.90ef 6.13abc 4.06cd
40 bc bc bc c

NaK 4.03b 4.28b

5.99d 6.01bc 6.20a 3.36 4.06f c c 4.73d 6.83bc 3.58de
20

p-
0.001 <0.00 0.006 0.00 <0.00 <0.000
valu 0.0013 n.s. n.s. 0.0001
0 01 5 98 01 1
e

a-fDifferent upper-case letters within a column indicate significant differences

(p<0.005) between different groups. Means that do not share a common letter are
significantly different. Non-significant: n.s.

Appearance

Compared to the reference ham (Na100), a replacement of Na+ ions by K+ ions of

25% and reduction to 60% (NaK40) of total salt had no effect on the color hue.
However, a reduction of total salt content by 80% and replacement of Na + ions by
K+ ions of 25% (NaK20) showed a significant lower color hue. There were significant
differences between samples for cohesiveness, but there were no clear correlation
between the salt content and the cohesiveness score. A possible explanation might
be the test production. In small scale test production without vacuum, it can be
difficult to avoid air and small gaps between the muscle pieces.

Taste, flavor and odor

A reduction to 40% (NaK60) of total salt, corresponding to 2.04% salt, was possible
without significantly influencing the salty taste. The changes in saltiness in this
experiment correspond to another low-salt ham study, Ruusunen, Särkkä-Tirkkonen
[5] found that hams with 1.4% salt were less salty than those containing 1.7–2.6%
salt. The reduced saltiness can be explained by the reduced content of Cl -—anions
in the cooked ham since Cl- anions have an effect on the receptor cells and
consequently on the perception of salt taste [35]. There were no significant
differences in metal flavor and metal odor except for a higher value for metal flavor
for one of the parallels of cooked ham containing the highest amount of potassium
chloride (NaK100_1) compared to the cooked ham containing the lowest salt content
(NaK20). These results indicate that a three to one replacement of Na + ions with K+
ions might be an acceptable level, regarding the issue of bitter (no significant
differences between samples) and metallic flavor (NaK20 significant different from
the other samples). Previous studies have shown that KCl may induce a bitter taste
at high concentrations. In dry-cured loins [9] and fermented sausages [10], the
maximum level was 40% replacement of sodium with potassium. In a study of
cooked ham they showed that 2.0% NaCl had a higher score than a 50%
replacement with KCl [11]. Another study found that it was possible to replace one-
third of NaCl with KCl without altering the sensory properties of smoked salmon [36].
Those results show that the maximum level of KCl replacement may be dependent
on product type.

Table 3 shows that a reduction to 60% (NaK40) of total salt had no effect on the after
taste. However, a reduction of total salt content by 80% and replacement of Na + ions
by K+ ions by 25% (NaK20) gave a less pronounced aftertaste compared to all other
hams. This can be explained by the fact that salt has a flavor enhancing effect in
meat products [37], and the salt content can also effect on biochemical and
enzymatic processes that in turn may impact flavor and/or structure of a product [30].
These findings are in accordance with Ruusunen, Vainionpää [38], who found that
ground meat patties with salt content down to 0.6% NaCl, had weak flavor intensity.

Texture

Compared to the reference ham (Na100), a reduction down to 60% (NaK40) of total
salt, corresponding to 1.73% salt, had no effect on tenderness or hardness in the
cooked hams. However, a reduction of total salt content by 80% and replacement of
Na+ ions by K+ ions of ¼ (NaK20), corresponding to a salt content of 1.38%, gave a
more tender and less hardness on cooked ham. These results were as expected
and in accordance with Ruusunen, Vainionpää [39], who found that more added salt
increased the firmness in Bologna type sausages at 1.10, 1.35 to 1.60% added salt,
Sofos [40] found that a reduction in salt content of more than 20% (<2.0% NaCl) in
Frankfurters resulted in softer and less firm texture. It is important to keep in mind
that the effects of proteins may differ in cooked ham where small pieces of lean meat
are connected together whereas sausages have an emulsified fat protein network.
In both cases the solubility and swelling of the myosin are nevertheless important.
In cooked ham, molecular bonds are able to form inside the exudate matrix (gel
cohesion) and between the exudate and the muscle [20], and when the salt
concentration is reduced, the protein extractability is limited due to the solubility of
the myosin [28, 29].

Compared to the reference ham (Na100), a replacement of Na+-ions by K+-ions of

25% gave no significant changes in sensory profile attributes. This replacement
corresponds to a total sodium content of approximately 1.1 g /100 g cooked ham.
These findings are in accordance with other studies [9–11, 36].

Multimodal machine vision system

The main imaging experiment, using the multimodal machine vision system, was
done using the same hams as with the chemical and sensorial measurements.

Machine vision was used to objectively study changes in color and texture in the
surface. The change in lightness, as a function of reducing the salt content, is clearly
seen in Fig 3A.
Fig 3. A, Mean subsurface reflectance b_⊥ for the hams with varying salt
content. B, Mean surface reflectance b_∥−b_⊥ for the hams with varying salt
content.

The intensity of the subsurface backscattered reflected blue light b⊥ shows a clear
increase as the salt content is reduced. This is confirmed by sensorial evaluation,
indicating an increase in whiteness and decrease in color-hue with decreased salt
content. A possible explanation for this behavior might be the decreased swelling of
the salt soluble proteins and a weaker heat induced gel in cooked ham with 1.38%
salt (NaK20) as discussed in above in section 3.3. Due to this theory, the surface of
the cooked ham with less salt might be rougher, and the interaction of light will
change, as has been shown previously [13]. Another explanation for the increased
whiteness due to decreased salt content may be that salt affects the formation of
heat stable nitric oxide myoglobin, the "cured meat color" [41].

A reduction in surface shininess, as measured by b∥−b⊥, using imaging, was only

weakly linked to reduction in salt content, Fig 3B. Despite the challenges in the
imaging setup in the main experiment, the improvements from the supplementary
experiment in Fig 2 show that imaging with two polarization orientations can provide
images that potentially can quantify subtle changes in visual appearance.

Since only simple features were used, namely the (r∥,g∥,b∥) and (r⊥,g⊥,b⊥) values
computed as the mean over each slice, there is a greater potential in dual-
polarization imaging if more advanced features such as those described in the
literature [13–15, 17, 19] would be applied to dual-polarization images using the
multimodal machine vision system described in this paper. Automatic segmentation,
between the dark and light portions of the meat, may also help improve the
quantification of the visual appearance characteristics, similar to how that human
sensorial evaluators observe using their combined vision and decision processes.

Conclusions

Compared to the reference ham (3.4% salt), a replacement of 25% Na +-ions by K+-
ions gave no significant changes in WHC, moisture, pH, expressed moisture, the
sensory profile attributes or the surface lightness and shininess. However, a
reduction of salt content by 60% and 80% on molar basis or more (down to 1.7–1.4%
salt) led to a decrease in WHC and an increase in expressible moisture. The salt
reduction had highest influence on the sensory attributes salty taste, after taste,
tenderness, hardness and color hue. To summarize, a reduction to 40% of total salt,
corresponding to 2.04% salt, was possible without significantly influencing the salty
taste. Further reduction down to 1.4% salt led to increased softness and reduced
hardness in the cooked hams. When reduction of salt by more than 60% (down to
1.7% salt) together with a replacement of 25% Na+-ions by K+-ions, the low-salt ham
manufacture procedure has to be modified. The multimodal machine vision system
showed changes in lightness, as a function of reduced salt content.

Acknowledgments

The authors thank Marte Schei and Anne Blikra for their contribution in production
and analyzing of the hams, MSc Aleksander Eilertsen and research scientist Morten
Bondø for their assistance in setting up the electronics and programming of the
machine vision system.

Funding Statement

The funding for this study is as follows: Research Council of Norway 210431/O10 to
JRM EM IGA MH; Research Council of Norway 185063/O10 to KG. KG, JRM, EM,
and IGA are employed by the commercial company SINTEF Fisheries and
Aquaculture. This funder provided support in the form of salaries for authors KG,
JRM, EM, and IGA but did not have any additional role in the study design, data
collection and analysis, decision to publish, or preparation of the manuscript.
Additionally, MH is employed by the commercial company NOFIMA AS. This funder
provided support in the form of salary for author MH but did not have any additional
role in the study design, data collection and analysis, decision to publish, or
preparation of the manuscript. The specific roles of these authors are articulated in
the ‘author contributions’ section.

Data Availability

All relevant data are within the paper.

References

1. He FJ, MacGregor GA. Effect of modest salt reduction on blood pressure: a meta-
analysis of randomized trials. Implications for public health. J Hum Hypertens.
2002;16(11):761–70. [PubMed]
2. WHO. Reducing salt intake in populations. Report of a WHO Forum and Technical
Meeting. World Health Organization, Paris, 2006.
3. The Norwegian Directorate of Health. Strategy for the reduction of salt intake in
the population Recommendation from the National Nutrition Council. In: The
Norwegian Directorate of Health, editor. Available:
http://helsedirektoratet.no/folkehelse/ernering/strategier-og-
satsninger/Documents/saltstrategi.pdf: The Norwegian Directorate of Health; 2011.
4. FSAI. Food Safety Authority of Ireland, Salt and health: review of the scientific
evidence and recommendationsfor public policy in Ireland. 2005.
5. Ruusunen M, Särkkä-Tirkkonen M, Puolanne E. Saltiness of coarsely ground
cooked ham with reduced salt content. Agricultural and food science in Finland.
2001;10:27–31.
6. Pietrasik Z, Gaudette NJ. The impact of salt replacers and flavor enhancer on the
processing characteristics and consumer acceptance of restructured cooked hams.
Meat Sci. 2014;96(3):1165–70. doi: 10.1016/j.meatsci.2013.11.005 [PubMed]
7. Breslin PAS, Beauchamp GK. Suppression of Bitterness by Sodium: Variation
Among Bitter Taste Stimuli. Chem Senses. 1995;20(6):609–23. [PubMed]
8. Olson D. Salt for processing probably can be cut by only one quarter. The National
Provisioner. 1982;7–10.
9. Gou P, Guerrero L, Gelabert J, Arnau J. Potassium chloride, potassium lactate
and glycine as sodium chloride substitutes in fermented sausages and in dry-cured
pork loin. 1996(0309–1740 (Print)). [PubMed]
10. Gelabert J, Gou P, Guerrero L, Arnau J. Effect of sodium chloride replacement
on some characteristics of fermented sausages. Meat Sci. 2003;65(2):833–9. doi:
10.1016/S0309-1740(02)00288-7 [PubMed]
11. Frye CB, Hand LW, Calkins CR, Mandigo RW. Reduction or Replacement of
Sodium Chloride in a Tumbled Ham Product. J Food Sci. 1986;51(3):836–7.
12. Lawless HT, Heymann H. Sensory evaluation of food—principles and practices
(2nd ed.). New York: Springer; 2010.
13. Iqbal A, Valous NA, Mendoza F, Sun D-W, Allen P. Classification of pre-sliced
pork and Turkey ham qualities based on image colour and textural features and their
relationships with consumer responses. Meat Sci. 2010;84(3):455–65. doi:
10.1016/j.meatsci.2009.09.016 [PubMed]
14. Jackman P, Sun DW, Du CJ, Allen P. Prediction of beef eating qualities from
colour, marbling and wavelet surface texture features using homogenous carcass
treatment. Pattern Recognition. 2009;42(5):751–63.
15. Jackman P, Sun DW, Allen P. Prediction of beef palatability from colour, marbling
and surface texture features of longissimus dorsi. J Food Eng. 2010;96(1):151–65.
16. Valous NA, Mendoza F, Sun D-W, Allen P. Texture appearance characterization
of pre-sliced pork ham images using fractal metrics: Fourier analysis dimension and
lacunarity. Food Res Int. 2009;42(3):353–62.
17. Chandraratne MR, Samarasinghe S, Kulasiri D, Bickerstaffe R. Prediction of
lamb tenderness using image surface texture features. J Food Eng. 2006;77(3):492–
9.
18. Alçiçek Z, Balaban MÖ. Development and application of “The Two Image”
method for accurate object recognition and color analysis. J Food Eng.
2012;111(1):46–51.
19. Mendoza F, Valous NA, Allen P, Kenny TA, Ward P, Sun D-W. Analysis and
classification of commercial ham slice images using directional fractal dimension
features. Meat Sci. 2009;81(2):313–20. doi: 10.1016/j.meatsci.2008.08.009
[PubMed]
20. Bombrun L, Gatellier P, Carlier M, Kondjoyan A. The effects of low salt
concentrations on the mechanism of adhesion between two pieces of pork
semimembranosus muscle following tumbling and cooking. Meat Sci. 2014;96(1):5–
13. doi: 10.1016/j.meatsci.2013.06.029 [PubMed]
21. Aaslyng MD, Vestergaard C, Koch AG. The effect of salt reduction on sensory
quality and microbial growth in hotdog sausages, bacon, ham and salami. Meat Sci.
2014;96(1):47–55. doi: 10.1016/j.meatsci.2013.06.004 [PubMed]
22. Mathiassen JR, Misimi E, Bondø M, Veliyulin E, Østvik SO. Trends in application
of imaging technologies to inspection of fish and fish products. Trends in Food
Science & Technology. 2011;22(6):257–75.
23. Regulation EU no 1169/2011 of the European Parliament and the council of 25 th
October 2011. In Official Journal of the European Union L 304/18 22.11.2011.
24. AOAC. Official Methods of Analysis of the Association of Official Analytical
Chemists. 15th ed. Arlington, Virgina: AOAC; 1990.
25. Eide O, Børresen T, Strøm T. Minced Fish Production From Capelin (Mallotus
villosus). A New Method for Gutting, Skinning and Removal of Fat from Small Fatty
Fish Species. J Food Sci. 1982;47(2):347–9.
26. Kivikari R. Analysis of sodium in meat products using an Na-selective eletrode.
(In Finnish). Proc. of Meat Day Seminar 1996, Nr, 536. Proc. of Meat Day Seminar
1996. 1996.
27. Grau R, Hamm R. Eine einfache Methode zur Bestimmung der Wasserbindung
im Muskel. Die Naturwissenschaften. 1953;40(1):29–30.
28. Hamm R. Koloidchemie des Fleisches: Parey, Berlin; 1972.
29. Offer G, Knight P. The structural basis of water-holding in meat DimS- R.A.
Lawrie, editor. Capters 3–4: London: Elsevier Applied Science; 1988.
30. Albarracín W, Sánchez IC, Grau R, Barat JM. Salt in food processing; usage and
reduction: a review. Int J Food Sci Technol. 2011;46(7):1329–36.
31. Gjerlaug-Enger E, Aas L, Ødegård J, Vangen O. Genetic parameters of meat
quality traits in two pig breeds measured by rapid methods. Animal.
2010;4(11):1832–43. doi: 10.1017/S175173111000114X [PubMed]
32. Puolanne E, Ruusunen MH, Vainionpää JI. Combined effects of NaCl and raw
meat pH on water-holding in cooked sausage with and without added phosphate.
Meat Sci. 2001;58(1):1–7. [PubMed]
33. Mossel DAA, Corry JET, Struijk CB, Bair RM. Essentials of the Microbiology of
Foods A Textbook for Advanced Studies: John Wiley & Sons, Chichester, England;
1995.
34. Zanardi E, Ghidini S, Conter M, Ianieri A. Mineral composition of Italian salami
and effect of NaCl partial replacement on compositional, physico-chemical and
sensory parameters. Meat Sci. 2010;86(3):742–7. doi:
10.1016/j.meatsci.2010.06.015 [PubMed]
 35. Murphy C, Cardello AV, Brand JG. Tastes of fifteen halide salts following water
and NaCl: Anion and cation effects. Physiology & Behavior. 1981;26(6):1083–95.
[PubMed]
36. Almli V, Hersleth M. Salt replacement and injection salting in smoked salmon
evaluated from descriptive and hedonic sensory perspectives. Aquaculture
International. 2013;21(5):1091–108.
37. Desmond E. Reducing salt: A challenge for the meat industry. Meat Sci.
2006;74(1):188–96. doi: 10.1016/j.meatsci.2006.04.014 [PubMed]
38. Ruusunen M, Vainionpää J, Lyly M, Lähteenmäki L, Niemistö M, Ahvenainen R,
et al. Reducing the sodium content in meat products: The effect of the formulation in
low-sodium ground meat patties. Meat Sci. 2005;69(1):53–60. doi:
10.1016/j.meatsci.2004.06.005 [PubMed]
39. Ruusunen M, Vainionpää J, Puolanne E, Lyly M, Lähteenmäki L, Niemistö M, et
al. Effect of sodium citrate, carboxymethyl cellulose and carrageenan levels on
quality characteristics of low-salt and low-fat bologna type sausages. Meat Sci.
2003;64(4):371–81. doi: 10.1016/S0309-1740(02)00178-X [PubMed]
40. Sofos JN. Effects of Reduced Salt (NaCl) Levels on Sensory and Instrumental
Evaluation of Frankfurters. J Food Sci. 1983;48:1692–9.
41. Honikel K-O. The use and control of nitrate and nitrite for the processing of meat
products. Meat Sci. 2008;78(1–2):68–76. doi: 10.1016/j.meatsci.2007.05.030
[PubMed]

Foods. 2015 Dec; 4(4): 665–677.

XXIX. Sensory Profile and Consumers’ Liking of Functional Ovine Cheese

Antonella Santillo* and Marzia Albenzio

Felix Barron, Academic Editor

Abstract

The present research was undertaken to evaluate the sensory profile and
consumers’ liking of functional ovine cheese containing probiotic cultures. Ovine
cheese was made from ewe’s milk by animals reared in extensive conditions;
cheesemaking trials were performed by using rennet paste containing probiotic cells.
Experimental cheeses were denoted: cheese manufactured using lamb rennet paste
without probiotic (C), cheese manufactured using lamb rennet paste containing a
mix of Bifidobacterium lactis and Bifidobacterium longum (BB), and cheese
manufactured using lamb rennet paste containing Lactobacillus acidophilus (LA).
Ovine cheese containing probiotic strains highlighted a more intense proteolysis and
a greater level of short chain free fatty acids and conjugated linoleic acid due to the
metabolic activity of the adjunct microflora. The sensorial profile of ovine cheese
showed lower humidity and gumminess in cheeses containing probiotics as a
consequence of differences in the maturing process; furthermore, probiotic cheeses
scored higher ratings for salty and pungent attributes. An interaction effect of
probiotic, gender, and age of the consumers was detected in the perceived and the
expected liking. The higher rate of expected liking in all experimental cheeses is
attributed to the information given, regarding not only the presence of probiotic
strains but also the farming conditions and cheesemaking technology.

Keywords: probiotic, ovine cheese, sensory profile, cheese liking

1. Introduction

Today, food is not merely viewed as a vehicle for essential nutrients to ensure proper
growth and development, but as a route to optimal wellness. Functional foods are
those which contain some health-promoting components that go beyond the
traditional nutrients; one way in which foods can be modified to become functional
is by adding probiotics [1]. The development of probiotic cheeses implies knowledge
of the processing steps as well as a level of influence (positive or negative) on the
survival of microorganisms, sensory acceptance, chemical stability, and
microbiological conditions throughout shelf life [2].

The manufacture of probiotic cheese should have minimum changes when

compared to traditional products [3] and must provide an adequate probiotic cell load
upon cheese consumption. Cheeses could offer certain advantages as a delivery
system of live probiotics to the gastrointestinal tract, having a higher pH than
fermented milk and a high fat content that may protect the organisms during passage
through the gastrointestinal tract [4]. The potential of different typologies of ovine
cheese, e.g., semi-hard cheese and pasta filata stretched curd as a functional food
delivering different probiotic bacteria has been previously reported [5,6,7]. Those
reports showed that probiotics yielded a complex outfit of proteolytic and lipolytic
enzymes able to influence not only cheese microbiology but also the maturing
process. Furthermore, probiotics in ovine cheese were involved in the production of
molecules such as essential amino acids [8], bioactive peptides [9], and
polyunsaturated fatty acids and conjugated linoleic acid (CLA) [10].

The commercial success of any functional food, especially those containing probiotic
strains, ultimately depends on taste, appearance, price, and health claim appeal to
consumers [2]. Thus, in the development of probiotic ovine cheese, the sensory
evaluation by consumers is a crucial and essential step that rules product innovation.
Furthermore, the information about the characteristics of the innovative dairy product
must be provided to the consumers, it being a factor able to increase consumers’
awareness and willingness to choose the new food. Napolitano et al. [11] reported
that expectation induced by the information can affect the quality perception and
oriented the consumers in their choice. The application of sensory methodology
allows researchers and developers to obtain important results on the formulated
product with respect to its descriptive sensory profile and acceptance on the
consumer market [12]. Developing a sensory profile of the product helps to identify
the principal sensory features of functional products, their negative and positive
attributes compared to the conventional product. The results from a descriptive
analysis test also provides a basis for determining those sensory attributes that are
important to acceptance. Some authors [13,14] provided a very interesting guide for
the sensory evaluation of ewe’s milk cheese with a trained panel. However, it is
crucial to remember that a trained panel must not measure liking, acceptance, or
preference. Once panelists are trained to identify and quantify attributes in products
(or grades and defects as in product judging), they are no longer typical consumers
[12]. In the light of these considerations, the aim of the present research was to study
the sensory profile and consumer liking of functional ovine cheese containing
probiotic cells.

2. Experimental Section

2.1. Lamb Rennet Paste and Ovine Cheese Production

Gentile di Puglia ewe’s milk was produced on a dairy farm located in the province of
Foggia, Southern Italy; the experiment was conducted in early spring (March) when
animals were reared in extensive conditions, fed mainly on fresh pasture with
concentrate integration at the milking parlor. Rennet paste was produced according
to the following protocol: briefly, abomasa were extracted from suckling lambs and
the perivisceral fat was removed; the abomasa were stored in salt at 6 °C and 70%
relative humidity for 3 months and then ground to obtain a paste.

The experimental rennet pastes used were: control lamb rennet paste (C), rennet
paste containing Lactobacillus acidophilus (LA), and lamb rennet paste containing a
mix of Bifidobacterium lactis and Bifidobacterium longum (BB). The ripened rennet
paste was inoculated with fresh probiotic cells to obtain a final concentration of 10 9
cfu/g of rennet one day before ovine cheese manufacturing. Three experimental
cheesemaking trials were performed for each type of lamb rennet paste. Ewe’s milk
from morning and afternoon milking was collected for two consecutive days, stored
at 4 °C, and processed on the third day for cheese production. Three experimental
cheesemaking trials were performed for each type of lamb rennet paste. Each
cheesemaking trial was performed with the same batch of milk. Briefly, raw ewes’
milk was thermized (65 °C for 5 min) and then cooled to 38 °C. Rennet paste was
added (30 mL of an aqueous solution 60% w/v) to obtain coagulation in 30 min, then
the curd was cut to grain size. After molding and pressing, the curds were held at 42
°C for 5 h, then salted in brine for 12 h (20% w/v NaCl) and ripened for 45 day (12
°C; R.H. 90%). Cheese weight was about 1.5 kg. At 45 day of ripening, counts of
probiotic bacteria in cheese were 7.4 × 107 cfu/g and 7.1 × 107 cfu/g of cheese for L.
acidophilus and the mix of bifidobacteria, respectively. After 45 day of ripening,
experimental cheeses were stored under vacuum at 4 °C.

2.2. Analyses on Cheese

2.2.1. Determination of Proteolysis and Lipolysis

Total nitrogen (TN) and phosphotungstic acid-soluble nitrogen (PASN) were

determined as described by Gripon et al. [15], and water-soluble nitrogen (WSN)
was measured as described by Stadhouders [16]. The TN minus WSN gave the
casein nitrogen [17]. The derivatized free amino acids (FAA) were separated,
identified, and quantified by reversed phase-HPLC (Agilent 1260 Infinity, Santa
Clara, CA, USA, equipped with a binary pump G1312A, automatic sampler G1313A,
degassing system, and column oven thermostatized at 40 °C) on a Zorbax Eclipse
AAA column (4.6 × 150 mm, 3.5 μm film thickness; Agilent PN9634000-9029). The
mobile phases were (A) 40 nM NaH2PO4 (pH 7.8) and (B)
acetonitrile:methanol:H2O (45:45:10, v/v/v). Quantization was done using the area
under each peak with the Agilent software ChemStation (Agilent Technologies,
Santa Clara, CA, USA). Detection was performed on an Agilent diode-array detector
G1315B (Agilent Technologies, Santa Clara, CA, USA) and a fluorescence detector
G1321A (Agilent Technologies, Santa Clara, CA, USA).

Total lipids from cheeses were extracted according to the method of de Jong and
Badings [18]. Free fatty acid (FFA) derivatization was performed according to the
method of Morrison and Smith [19]. FFA and CLA were separated on a capillary
column (HP88; 100 m × 0.25 mm i.d., 0.20-μm film thickness; Agilent Technologies
Inc. Santa Clara, CA, USA). The injector and flame ionization detector temperatures
were 250 °C. The temperature was held at 70 °C for 4 min and then increased to
175 °C (13 °C/min, held for 27 min), then increased to a final temperature 215 °C (4
°C/min, held for 45 min). The split ratio was 1:20 and helium was the carrier gas,
with a pressure of 227.5 kPa. Pure CLA isomers were purchased as FA methyl
esters from Matreya Inc. (Pleasant Gap, PA, USA). All solvents were analytical grade
from J. T. Baker Inc. (Milan, Italy). Analyses were performed in duplicate.

2.2.2. Determination of Textural Parameters

Samples for cheese texture analysis were obtained by cutting a 1-cm-thick slice from
the central diameter of the cheese wheel after 45 day of ripening. Then, 6 rectangular
parallelepipeds, 1 × 1 cm thick and 2 cm long, were obtained from the slide. The
cheese samples were left at 20 °C for 10 min before testing. Texture profile analysis
was evaluated with an Instron 4301 tensile tester (Instron Ltd., High Wycombe, UK),
using a modified compression device that avoids transversal elongation of the
samples. Each sample underwent 2 cycles of 80% compression; force × time data
were used to calculate the following parameters: hardness, cohesiveness,
springiness, gumminess, and chewiness. A typical texture profile of cheese obtained
using the Instron equipment is interpreted to obtain the parameters reported in the
study. Hardness is the height of the second peak (H2) in the first bite; cohesiveness
is the ratio of area on second bite to that of the first bite (A2/A1). Gumminess is the
hardness × cohesiveness × 100 (A1 × (A2/A1) × 100); Elasticity is the difference
between distance B and distance C (C is the same measurement made using an
inelastic material); Chewiness is the hardness × cohesiveness × springiness (A1 ×
(A2/A1) × (B − C)).

2.2.3. Sensory Analysis

Subjects were recruited among students and personnel of the University of Foggia.
The consumer panel consisted of 80 subjects; the panel was homogeneous on the
basis of age (20–45 years) and gender. In addition subjects were selected using
predetermined screening criteria based on consumption of dairy products made from
ewe’s milk and their awareness of probiotic food products; subjects were asked to
give an example of a food containing probiotics.

The sensory analysis of ovine cheese was carried out on 45 day ripened cheese
using the descriptive model of Coppola et al. [20] with a few modifications [21].
Before the sensory analysis, the untrained panel was preliminarily briefed on the use
of the sensory attributes on a 9-point intensity scale (0–9) of the scorecard. The
scorecard contained descriptive attributes according to Santillo et al. [22] adapted
for ovine cheese, namely 2 for odor/flavor (overall intensity, acid, rancid), 8 for taste
(overall intensity, salty, acid, pungent, bitter, sweet, mold, rancid), color (overall
intensity, uniformity), appearance, and texture (overall uniformity, moisture, chalky,
rubbery, grainy). Cheeses were taken out of the refrigerator 1 h before serving. Each
sample was assigned a 3-digit random number, and cheese slices (1.5 mm thick)
from the 3 replications of the same batch were mixed randomly so that all replications
from the same batch were presented an equal number of times. A glass of water and
unsalted crispy bread were provided and consumers were instructed to take a small
bite of bread and a sip of water after each cheese tasting. Duplicate trays of samples
were presented to the panel at 10-min intervals.

2.2.4. Determination of Functional Ovine Cheese Liking

The perceived, expected, and actual liking was measured in accordance with the
protocol proposed by Napolitano et al. [11]. The protocol provided for three sessions:
in the first session subjects were asked to taste the ovine cheeses and rate their
liking in blind conditions without receiving any information on the production protocol
of cheese (perceived liking). In the second session the subjects received the
information and were asked to rate their liking on the basis only of the information
given (expected liking). In the third session consumers were given the cheeses along
with the information and expressed their liking (actual liking). Consumers rated their
liking in a 9-point hedonic scale anchored with “like extremely” and “dislike
extremely” and with a neutral center point of “neither like nor dislike”.
In sessions 2 and 3 the following information concerning ovine cheese production
was given to the consumers.

Ovine cheese—cheese made using milk from ewes reared in extensive conditions
based on grazing on natural pasture. The production of cheese involved the use of
whole milk heated to a temperature of 65 °C for few minutes and the use of rennet
paste as coagulant. Rennet paste is a type of coagulant used for typical dairy
productions; it is a white paste obtained from the abomasa of suckling lambs. The
cheese is salted in brine and then subjected to aging at controlled temperature and
humidity for up to 45 days.

Functional ovine cheese containing Lactobacillus acidophilus—cheese made using

milk from ewes reared in extensive conditions based on grazing on natural pasture.
The production of cheese involved the use of whole milk heated to a temperature of
65 °C for few minutes and the use of rennet paste as coagulant. Rennet paste is a
type of coagulant used for typical dairy productions; it is a white paste obtained from
the abomasa of suckling lambs. The cheese is salted in brine and then subjected to
aging at controlled temperature and humidity for up to 45 days. The rennet paste
contains live cells of Lactobacillus acidophilus recognized as probiotic able to exert
beneficial effects on human health. The cheese contains high level of probiotic cells
in accordance with the current guidelines for food products.

Functional ovine cheese containing bifidobacteria—cheese made using milk from

ewes reared in extensive conditions based on grazing on natural pasture. The
production of cheese involved the use of whole milk heated to a temperature of 65
°C for few minutes and the use of rennet paste as coagulant. Rennet paste is a type
of coagulant used for typical dairy productions; it is a white paste obtained from the
abomasa of suckling lambs. The cheese is salted in brine and then subjected to
aging at controlled temperature and humidity for up to 45 days. The rennet paste
contains live cells of Bifidobacterium longum and Bifidobacterium lactis recognized
as probiotic able to exert beneficial effects on human health. The cheese contains
high level of probiotic cells in accordance with the current guidelines for food
products.

2.2.5. Statistical Analysis

All the variables were tested for a normal distribution using the Shapiro-Wilk test
[23]. Data were analyzed by ANOVA using the GLM procedure of SAS [24], and the
effect of probiotic strain was tested on chemical and sensory attributes of ovine
cheese. The effect of probiotic, age, and gender of the panelist, and the interaction
of these variables, was studied for the perceived, expected and actual liking of the
ovine cheese. When significant effects were found (p < 0.05), Student’s t-test was
used to locate significant differences between means.

3. Results
Chemical composition of probiotic ovine cheese is presented in Table 1. Casein
content at 45 day of ripening was lower in probiotic cheeses with respect to the
control. PASN/TN and FAA showed the same behavior among experimental
cheeses: the mentioned parameters were lower in control cheese, intermediate in
LA and higher in BB cheese. Fat content was highest in BB cheese. Among the free
fatty acids, butyric acid (C4:0) showed the highest levels in cheeses containing
probiotics and total CLA was the highest in LA, intermediate in BB, and the lowest in
control cheese.

Table 1. Chemical composition of probiotic ovine cheese at 45 day of ripening (n =

18).

Cheese Effect, p
Parameter SEM
C BB LA Probiotic
Casein, % 29.4 b 23.8 a 24.3 a 0.4 **
PASN/TN, % 6.4 a 10.9 c 8.0 b 0.5 *
FAA, μg/g cheese 3598.5 a 7424.6 c 6340.8 b 169.9 **
Fat, % 26.7 a 31.8 b 29.9 a 0.7 ***
C 4:0, % 14.2 a 24.1 b 19.6 b 1.2 *
CLA, % 0.9 a 2.0 b 2.3 c 0.1 *

Cheese: C = cheese made with traditional lamb rennet paste; BB = cheese made
with lamb rennet paste containing a mix of B. longum and B. lactis; LA = cheese
made with lamb rennet paste containing L. acidophilus; SEM = standard error; * p <
0.05; ** p < 0.01; *** p < 0.001; a,b,c Mean values followed with different superscripts
differ significantly; PASN = phosphotungstic acid-soluble nitrogen; TN = total
nitrogen; FAA = free amino acids; CLA = conjugated linoleic acids.

Textural parameters of experimental cheeses at 45 day of ripening are reported in

Table 2. Cohesiveness and gumminess were affected by the addition of probiotic
cells, being higher in BB cheese, intermediate in LA, and lower in C cheese.

Table 2. Rheological parameters of probiotic ovine cheese at 45 day of ripening (n

= 18).

Cheese Effect, p
Parameter SEM
C BB LA Probiotic
Hardness, N 25.6 29.4 26.0 2.1 NS
Cohesiveness 0.1 a 0.2 b 0.2 a,b 0.1 *
Gumminess, N 0.4 a 0.9 b 0.6 a,b 0.1 *
Chewiness, Nxmm 4.5 5.8 4.5 1.0 NS
Elasticity, mm 7.9 7.9 7.9 0.8 NS
Cheese: C = cheese made with traditional lamb rennet paste; BB = cheese made
with lamb rennet paste containing a mix of B. longum and B. lactis; LA = cheese
made with lamb rennet paste containing L. acidophilus; SEM = standard error; * p <
0.05; NS = not significant; a,b Mean values followed with different superscripts differ
significantly.

Sensorial attributes of ovine cheese at 45 day of ripening are reported in Table 3.

Among appearance attributes, differences emerged for humidity and gumminess,
which turned out to be the lowest in both cheeses containing probiotics whereas
graininess was the highest in LA cheese. No differences were reported for uniformity
and chalky appearance. All the attributes ascribed to color and odor were not
affected by the experimental treatment. Regarding taste attributes, salty scored the
highest value in BB and pungent was higher in both probiotic cheeses. The overall
taste intensity and acid, bitter, sweet, mold, and rancid attributes were not affected
by treatment.

Table 3. Sensorial attributes of probiotic ovine cheese at 45 day of ripening.

Cheese Effect, p
Parameter SEM
C BB LA Probiotic
Uniformity 5.1 5.4 5.1 0.3 NS
Humidity 6.4 b 4.8 a 5.0 a 0.3 ***
Appearance Chalky 3.8 4.0 3.7 0.4 NS
Gumminess 5.1 b 4.0 a 3.8 a 0.3 **
Graininess 2.7 a 2.8 a 2.9 b 0.3 NS
Uniformity 6.3 5.9 5.8 0.2 NS
Color
Intensity 6.1 5.9 5.6 0.3 NS
Intensity 6.1 5.7 5.6 0.3 NS
Odor Acid 3.4 3.4 3.1 0.3 NS
Rancid 1.8 1.7 1.8 0.3 NS
Intensity 5.9 a 6.8 b 6.1 a,b 0.3 *
Salty 4.3 a 5.7 b 4.8 a 0.3 ***
Acid 2.7 3.3 3.2 0.3 NS
Pungent 2.5 a 4.6 b 4.1 b 0.3 ***
Taste
Bitter 2.7 2.5 2.8 0.3 NS
Sweet 3.5 2.7 3.1 0.3 NS
Mould 1.1 0.9 1.2 0.3 NS
Rancid 1.0 1.3 1.9 0.4 NS

Cheese: C = cheese made with traditional lamb rennet paste; BB = cheese made
with lamb rennet paste containing a mix of B. longum and B. lactis; LA = cheese
made with lamb rennet paste containing L. acidophilus; SEM = standard error; * p <
0.05; ** p < 0.01; *** p < 0.001; NS = not significant; a,b Mean values followed with
different superscripts differ significantly.

The rating given by the consumer panel during the three hedonic tests was not
affected by the use of probiotic strains in ovine cheese production, scoring a mean
value of 7.01 ± 0.2, 7.64 ± 0.25, and 7.34 ± 0.3 for the perceived, the expected, and
the actual liking, respectively. The rating of ovine cheese liking is presented in Table
4. Cheese liking was not affected both by probiotic addition and consumers’ group,
whereas an interaction effect of probiotic, gender, and age of consumers was
detected in the perceived and the expected liking. A higher rate was expressed for
the perceived liking of the BB and LA cheeses by female consumers over 30 years.
On the contrary, females less than 30 years old scored lower rates for the probiotic
cheeses. BB cheese received an intermediate score from males less than 30 years
old, whereas LA cheese received intermediate scores from males over 30 years old.
For expected liking, higher rates were found in female over 30 years for all the
experimental cheeses, whereas LA cheese scored higher rates in males over 30
years. The actual liking of experimental cheeses was not influenced by age and
gender. All experimental cheeses obtained a judgment from “moderately pleasant”
to “very pleasant”.

Table 4. Rating of probiotic ovine cheese liking.

Cheese Effect, p
Gender
Parameter SEM Gender x Probiotic ×
and Age C BB LA Probiotic
Age Gender × Age
M < 30 (n = 6.5 7.1 6.6
19) a a,b a

M > 30 (n = 6.5 6.5 7.1

21) a a a,b
Perceived
0.2 NS NS *
liking F < 30 (n = 7.1 6.5 6.6
21) a,b a a

F > 30 (n = 6.6 7.3 7.3

19) a b b

M < 30 (n = 7.5 7.3 7.7

19) a a b

M > 30 (n = 7.7 7.6 7.8

21) a a b
Expected
0.2 NS NS *
liking F < 30 (n = 7.5 7.1 7.1
21) a a a

F > 30 (n = 7.8 8.2 8.1

19) b b b
 Cheese Effect, p
Gender
Parameter SEM Gender x Probiotic ×
and Age C BB LA Probiotic
Age Gender × Age
M < 30 (n =
7.1 7.5 7.3
19)
M > 30 (n =
7.2 7.5 7.1
21)
Actual liking 0.4 NS NS NS
F < 30 (n =
7.2 7.1 7.1
21)
F > 30 (n =
7.7 7.6 7.5
19)

Cheese: C = cheese made with traditional lamb rennet paste; BB = cheese made
with lamb rennet paste containing a mix of B. longum and B. lactis; LA = cheese
made with lamb rennet paste containing L. acidophilus; SEM = standard error; * p <
0.05; NS = not significant; a,b Mean values followed with different superscripts differ
significantly.

4. Discussion

Cheese ripening involves different biochemical pathways, such as proteolysis and

lipolysis ascribed to endogenous and exogenous factors, mainly enzymes yielded
by rennet and microflora. Proteolysis is the main process in cheese ripening;
determining changes in the texture due to the breakdown of the protein network, and
in flavor formation through the release of peptides, free amino acids, and catabolic
products [25]. Cheese containing a mix of Bifidobacterium longum and
Bifidobacterium lactis, and cheese containing Lactobacillus acidophilus highlighted
a more intense proteolysis than cheese without probiotic as evidenced by a greater
hydrolysis of the intact casein fractions. The disruption of the casein matrix led to a
major accumulation of nitrogen fractions comprising low molecular weight peptides
and free amino acids. However, differences emerged among probiotic cheeses, with
cheese containing bifidobacteria showing a higher proteolytic potential. Previous
studies reported that ovine cheese containing a mix of B. longum and B. lactis
showed greater caseins hydrolysis, evidencing that probiotic strains have a different
impact on the proteolysis of cheese [4,25]. BB cheese showed a more complex FAA
profile as an outcome of the proteinase and peptidase activities, evidencing the real
contribution of adjunct microflora to cheese ripening [26,27]. Lipolysis is another
important biochemical event that leads to the formation of FFAs which also
contribute to cheese flavour with volatile compounds [28]. Lipases associated with
lactic acid microflora are acknowledged to be selective for short chain fatty acid
release [29]. The greater level of C4:0 in probiotic cheese could be attributed to the
metabolic activity of the adjunct microflora; in particular, the ability of L. acidophilus
to convert linoleic acid in to CLA is a consequence of the detoxification mechanism
enacted by this probiotic strain. CLA enrichment led to an ameliorated composition
of the fat fraction of ovine cheese; indeed the involvement of these polyunsaturated
fatty acids has been reported in the prevention of many human diseases [30].

Cheese texture may be defined as a composite sensory attribute resulting from a

combination of physical properties perceived by the senses of sight, touch, and
hearing [31]. Cheese texture could be measured indirectly by using instrumental
rheological techniques and is related to the composition, structure, and strength of
the attractions between the structural elements of the cheese. The behavior of the
texture profile revealed that cheeses containing bifidobacteria were less brittle, in
accordance with Buriti et al. [32], probably as a consequence of the greater
proteolysis associated with the cheese maturing process.

Evaluation of sensory attributes permits the definition of the perceived profile and
overall acceptability by consumers of dairy products with innovative characteristics.
The analysis of the sensorial profile of the cheeses permits the identification of
specific attributes that could be preference drivers and the evaluation of the impact
of health information on consumer preference, expectation, and choice. Higher
scores for humidity and gumminess were in accordance with the minor proteolytic
process observed in the control cheese; cheese without probiotic was judged as the
cheese with an overall lower intensity for taste attributes. Regarding taste attributes,
the score for the Pungent attribute doubled in probiotic cheese with respect to the
control cheese as an outcome of the major accumulation of C4:0. Butyric acid is one
the major odorants in Cheddar and Camembert [33], and is also an important
component of Feta cheese, contributing greatly to its flavor and piquant taste [34].
The salty taste, together with the pungent attribute, led to higher overall intensity
perceived upon consumption of cheese containing bifidobacteria compared to
control cheese. It is worth noting that the drivers of cheese acceptability are different
depending on the consumer’s gender; a previous study on Caciocavallo cheese
reported that salty and piquant attributes were mostly enjoyed by male rather than
female panelists [22].

Some food attributes are directly perceived as price and sensory properties; others
must be communication objects as healthy and/or ethical aspects. Indeed, the
perception of a healthier product (e.g., cheese containing high levels of probiotic
microorganisms) could increase the actual acceptability of cheese [35]. The absence
of differences among cheeses in the rating of ovine cheese liking is an encouraging
result because the adjunct of probiotic cultures in cheese did not lead to the
development of aroma and flavor defects that permit consumers to distinguish
between a traditional and an innovative cheese. Overall, the manufacture of probiotic
cheese should have minimum changes when compared to traditional products [2].
The addition of bifidobacteria in Gouda cheese [36] and in Cottage cheese [37] led
to negative effects on cheese flavor, reducing the acceptability of probiotic cheese
with respect to a traditional one. In Cheddar cheese, Ong et al. [38] reported that
cheese acceptance depended on the probiotic microorganism used in
cheesemaking, evidencing the advisability of consumer acceptance testing
whenever selected probiotic strains are used in the cheesemaking process.
The perceived liking of ovine cheese determined in blind conditions represented the
baseline for the evaluation of the impact of information on consumers’ expectation;
in particular, grouping the panel on the base of gender and age allowed for the
evaluation of information on different segments of consumers. The expected liking
was different from the perceived liking expressed in blind conditions, indicating that
a negative disconfirmation occurred: higher rates of expected liking highlighted that
the products are worse than expected although the overall rate referred to an
acceptable product with a score over 6.5. In general, the higher rates of expected
liking in all experimental cheeses was attributed to the information given to
consumers concerning not only the presence of probiotic strains in cheese but also
the farming conditions of the ewes, and the particular cheesemaking technology. It
can be argued that consumers are prone to consider the farming conditions and the
production of ovine cheese using traditional protocols as an added value to the
quality of ovine cheese. Napolitano et al. [11] reported that information about organic
farming can be a major determinant of cheese liking when comparing organic and
conventional cheese. When the cluster of females over 30 years was considered, a
higher weight was ascribed to the information related to the healthy effect of cheese
when compared to male consumers and females less than 30 years of age. The
higher expected liking for LA cheese in males over 30 years old could be ascribed
to the fact that L. acidophilus is one of the most commonly used probiotic strains
used in dairy foods. Probiotic products seemed to create a higher expectation in
females over 30 years, probably due to an elevated consciousness of the benefits
of healthy food. Indeed, a choice experiment on semi-hard cheese highlighted that
female, rather than male participants, were affected by health information on their
diet choices, evidencing clear diet-health awareness [39]. When actual liking was
measured, scores moved towards the expectation liking rate, evidencing that
assimilation of the information given contributed to the increase of ovine cheese
liking.

Probiotic cheese combines nutritional and functional properties and is promising for
the expansion of dairy products from ovine milk. The experimental cheeses
containing probiotics were judged as the cheeses with an overall higher intensity for
taste attributes in accordance with the more intense proteolysis and a greater level
of short chain free fatty acids and conjugated linoleic acid. The results from the
present research also highlight that the design and development of functional ovine
cheese represents a technological innovation that needs to be supplied with
adequate information in order to be able to orient consumers in their food choice.
Nowadays, consumers are aware of the great impact of nutrition on health and
reward ovine cheese as a food product associated with a sustainable production
system. Special attention must be paid to the sensory profile of probiotic ovine
cheese in order to provide information on the cheese attributes able to influence
consumers’ liking.

Acknowledgments

This work was partly supported by Fondazione Banca del Monte (Foggia, Italy).
Author Contributions

Antonella Santillo and Marzia Albenzio gave equal contribution to the design and
conduction of the experimental plans, to the analyses of the experimental data and
to the writing of the manuscript. The authors read and approved the final manuscript.

Conflicts of Interest

The authors declare no conflict of interest.

References

1. Castro J.M., Tornadijo M.E., Fresno J.M., Sandoval H. Biocheese: A food probiotic
carrier. BioMed Res. Int. 2015 doi: 10.1155/2015/723056. [PMC free article]
[PubMed] [Cross Ref]
2. Granato D., Branco G.F., Cruz A.G., de Assis Fonseca Faria J., Shah N.P.
Probiotic dairy products as functional foods. Compr. Rev. Food Sci. Food Saf.
2010;9:455–470. doi: 10.1111/j.1541-4337.2010.00120.x. [Cross Ref]
3. Cruz A.G., Buriti F.C.A., Souza C.H.B., Faria J.A.F., Saad S.M.I. Probiotic cheese:
Health benefits, technological and stability aspects. Trends Food Sci. Technol.
2009;20:344–354. doi: 10.1016/j.tifs.2009.05.001. [Cross Ref]
4. Stanton C, Gardiner G., Lynch P.B., Fitzgerald G., Ross R.P. Probiotic cheese.
Int. Dairy J. 1998;8:491–496.
5. Santillo A., Albenzio M. Influence of lamb rennet paste containing probiotic on
proteolysis and rheological properties of pecorino cheese. J. Dairy Sci.
2008;91:1733–1742. doi: 10.3168/jds.2007-0735. [PubMed] [Cross Ref]
6. Albenzio M., Santillo A., Caroprese M., Ruggieri D., Napolitano F., Sevi A.
Physicochemical properties of Scamorza ovine cheese manufactured with different
probiotic cultures. J. Dairy Sci. 2013;96:2781–2791. doi: 10.3168/jds.2012-6218.
[PubMed] [Cross Ref]
7. Albenzio M., Santillo A., Caroprese M., Braghieri A., Sevi A., Napolitano F.
Composition and sensory profiling of probiotic Scamorza ewe’s milk cheese. J. Dairy
Sci. 2013;96:2792–2800. doi: 10.3168/jds.2012-6273. [PubMed] [Cross Ref]
8. Santillo A., Caroprese M., Marino R., Muscio A., Sevi A., Albenzio M. Influence of
lamb rennet paste on chemical and enzymatic characteristics of Pecorino Foggiano
cheese. Int. Dairy J. 2007;17:535–546. doi: 10.1016/j.idairyj.2006.06.024. [Cross
Ref]
9. Albenzio M., Santillo A., Marino R., della Malva A., Caroprese M., Sevi A.
Identification of peptides in functional Scamorza ovine milk cheese. J. Dairy Sci.
2015 doi: 10.3168/jds.2015-9844. [PubMed] [Cross Ref]
10. Santillo A., Albenzio M., Quinto M., Caroprese M., Marino R., Sevi A. Probiotic
in lamb rennet paste enhances rennet lipolytic activity, and conjugated linoleic acid
and linoleic acid content in pecorino cheese. J. Dairy Sci. 2009;92:1330–1337. doi:
10.3168/jds.2008-1598. [PubMed] [Cross Ref]
11. Napolitano F., Braghieri A., Piasentier E., Favotto S., Naspetti S., Zanoli R.
Cheese liking and consumer willingness to pay as affected by information about
organic production. J. Dairy Res. 2010;77:280–286. doi:
10.1017/S0022029910000130. [PubMed] [Cross Ref]
12. Cruz A.G., Cadena R.S., Walter E.H.M., Mortazavian A.M., Granato D., Faria
J.A.F., Bolini H.M.A. Sensory analysis: Relevance for prebiotic, probiotic, and
symbiotic product development. Compr. Rev. Food Sci. Food Saf. 2010;9:358–373.
doi: 10.1111/j.1541-4337.2010.00115.x. [Cross Ref]
13. Barcenas P., Elortondo F.J.P., Albisu M., Mège J., Roseiro L.B., Scintu M.F.,
Torre P., Loygorri S., Lavanchy P. An international ring trial for the sensory
evaluation of raw ewes’ milk cheese texture. Int. Dairy J. 2007;17:1139–1147. doi:
10.1016/j.idairyj.2006.12.012. [Cross Ref]
14. Lavanchy P., Mège J., Elortondo F.J.P., Roseiro L.B., Scintu M.F., Torre P.,
Barcenas P., Loygorri S. A Guide to the Sensory Evaluation of the Texture of Hard
and Semi-Hard Ewes’ Milk Cheese. Luxembourg Office for Official Publications of
the European Communities; Liebefeld, Switzerland: 1999.
15. Gripon J.C., Desmazeaud M.J., le Bars D., Bergere J.L. Etude du role des micro-
organismens et des enzymes au cours de la maturation des fromages. Le Lait.
1975;548:502–515. doi: 10.1051/lait:197554828. [Cross Ref]
16. Stadhouders J. The hydrolysis of protein during ripening of Dutch cheese. The
enzymes of the bacteria involved. Neth. Milk Dairy J. 1960;14:106–110.
17. Prieto B., Franco I., Prieto J.G., Bernardo A., Carballo J. Proteolytic and lipolytic
changes during the ripening of Leon raw cow’s milk cheese, a Spanish traditional
variety. Int. J. Food Sci. Technol. 2002;37:661–671. doi: 10.1046/j.1365-
2621.2002.00596.x. [Cross Ref]
18. De Jong C., Badings H.T. Determination of free fatty acids in milk and cheese.
Procedures of extraction, clean up, and capillary gas chromatographic analysis. J.
High Resol. Chromatogr. 1990;13:94–98. doi: 10.1002/jhrc.1240130204. [Cross Ref]
19. Morrison W.R., Smith L.M. Preparation of fatty acids methyl esters and
dimethylacetals from lipids with boron fluoridemethanol. J. Lipid Res. 1964;5:600–
608. [PubMed]
20. Coppola S., Villani F., Coppola R., Parente E. Comparison of different starter
systems for water-buffalo Mozzarella cheese manufacture. Le Lait. 1990;70:411–
423. doi: 10.1051/lait:19905-631. [Cross Ref]
21. De Angelis M., de Candia S., Calasso M.P., Faccia M., Guinee T.P., Simonetti
M.C., Gobbetti M. Selection and use of autochthonous multiple strain cultures for the
manufacture of high-moisture traditional Mozzarella cheese. Int. J. Food Microbiol.
2008;125:123–132. doi: 10.1016/j.ijfoodmicro.2008.03.043. [PubMed] [Cross Ref]
22. Santillom A., Caroprese M., Ruggieri D., Marino R., Sevi A., Albenzio M.
Consumer acceptance and sensory evaluation of Monti Dauni Meridionali
Caciocavallo cheese. J. Dairy Sci. 2012;95:4203–4208. doi: 10.3168/jds.2011-4777.
[PubMed] [Cross Ref]
23. Shapiro S.S., Wilk M. An analysis of variance test for normality. Biometrika.
1965;52:591–601. doi: 10.1093/biomet/52.3-4.591. [Cross Ref]
24. SAS Institute . SAS User’s Guide: Statistics. version 9.2 ed. SAS Institute Inc.;
Cary, NC, USA: 2011.
25. Albenzio M., Santillo A., Caroprese M., Marino R., Trani A., Faccia M.
Biochemical patterns in ovine cheese: Influence of probiotic strains. J. Dairy Sci.
2010;93:3487–3496. doi: 10.3168/jds.2009-3036. [PubMed] [Cross Ref]
26. Corbo M.R., Albenzio M., de Angelis M., Sevi A., Gobbetti M. Microbiological and
biochemical properties of Pecorino hard cheese supplemented with bifidobacteria.
J. Dairy Sci. 2001;84:551–561. doi: 10.3168/jds.S0022-0302(01)74507-9. [PubMed]
[Cross Ref]
27. Santillo A., Albenzio M., Bevilacqua A., Corbo M.R., Sevi A. Encapsulation of
probiotic bacteria in lamb rennet paste: Effects on the quality of Pecorino cheese. J.
Dairy Sci. 2012;95:3489–3500. doi: 10.3168/jds.2011-4814. [PubMed] [Cross Ref]
28. Caporaso N., Armento V., Sacchi R. Volatile profile of Conciato Romano cheese,
a traditional Italian cheese, during ripening. Eur. J. Lipid Sci. Technol.
2015;117:1422–1431. doi: 10.1002/ejlt.201400153. [Cross Ref]
29. De la Fuente M.A., Fontecha J., Juarez M. Fatty acid composition of the
triglyceride and free fatty acid fractions in different cows-, ewe- and goats-milk
cheeses. Z. Lebensm. Unters. Forsch. 1993;196:155–158. doi:
10.1007/BF01185577. [Cross Ref]
30. Williams C.M. Dietary fatty acids and human health. Ann. Zootech. 2000;49:165–
180. doi: 10.1051/animres:2000116. [Cross Ref]
31. Pinho O., Mendez E., Halves M.M., Ferreira I.M.P.L.V.O. Chemical, physical,
and sensorial characteristics of “Terrincho” ewe cheese: Changes during ripening
and intravarietal comparison. J. Dairy Sci. 2004;87:249–257. doi:
10.3168/jds.S0022-0302(04)73163-X. [PubMed] [Cross Ref]
32. Buriti F., da Rocha J.S., Assis E.G., Saad S.M.I. Probiotic potential of Minas
fresh cheese prepared with the addition of Lactobacillus paracasei. Lebensm.-Wiss.
Technol. 2005;38:173–180. doi: 10.1016/j.lwt.2004.05.012. [Cross Ref]
33. Yvon M., Rijnen L. Cheese flavor formation by amino acid catabolism. Int. Dairy
J. 2001;11:185–201. doi: 10.1016/S0958-6946(01)00049-8. [Cross Ref]
34. Georgala A, Moschopoulou E., Aktypis A., Massouras T., Zoidou E., Kandarakis
I., Anifantakis E. Evolution of lipolysis during the ripening of traditional Feta cheese.
Food Chem. 2005;93:73–80.
35. Napolitano F., Caporale G., Carlucci A., Monteleone E. Effect of information
about animal welfare and product nutritional properties on acceptability of meat from
Podolian cattle. Food Qual. Pref. 2007;2:305–312. doi:
10.1016/j.foodqual.2006.02.002. [Cross Ref]
36. Gomes A.M.P., Malcata F.X., Klaver F.A.M., Grande H.J. Incorporation and
survival of Bifidobacterium spp. Strain Bo and Lactobacillus acidophilus strain Ki in
a cheese product. Neth. Milk Dairy J. 1995;49:71–95.
37. Blanchette L., Roy D., Bèlanger G., Gauthier S.F. Production of Cottage cheese
using dressing fermented by bifidobacteria. J. Dairy Sci. 1996;79:8–15. doi:
10.3168/jds.S0022-0302(96)76327-0. [Cross Ref]
38. Ong L., Henriksson A., Shah N.P. Chemical analysis and sensory evaluation of
Cheddar cheese produced with Lactobacillus acidophilus, Lb. casei, Lb. paracasei
or Bifidobacterium sp. Int. Dairy J. 2007;17:937–945. doi:
10.1016/j.idairyj.2007.01.002. [Cross Ref]
39. Øvrum A., Alfnes F., Almli V.L., Rickersten K. Health information and diet
choices: Results from a cheese experiment. Food Policy. 2012;37:520–529. doi:
10.1016/j.foodpol.2012.05.005. [Cross Ref]

J Food Sci Technol. 2015 Jun; 52(6): 3681–3688.

Published online 2014 Jun 10. doi: 10.1007/s13197-014-1427-2
PMCID: PMC4444897

XIX. Evaluation of functional properties of composite flours and sensorial

attributes of composite flour biscuits

Suresh Chandra, Samsher Singh, and Durvesh Kumari

Abstract
The present study was undertaken to develop biscuits from the composite flours.
Composite flours were prepared by blending wheat flour with rice flour, green gram
flour and potato flour in ratios of 100:0:0:0 (W 100), 85:5:5:5 (W 85), 70:10:10:10 (W 70)
and 55:15:15:15 (W 55), respectively. The functional properties of composite flours
such as swelling capacity, water absorption capacity, oil absorption capacity,
emulsion activity, emulsion stability, foam capacity, foam stability, gelatinization
temperature, least gelation concentration and bulk density were increased with
increase in the incorporation of other flours with wheat flour. Overall acceptability for
composite flour biscuits was awarded highest score for W 55 followed by W 70 and W 85
as compared to control biscuits. All biscuits coincided in the range of ‘like moderately’
to ‘like very much’ for composite flours biscuits while ‘like slightly’ to like moderately’
for control biscuits.

Keywords: Composite flour, Swelling capacity, Oil absorption, Gelatinization

temperature, Least gelation concentration, Sensorial attributes

Introduction

Composite flours may be considered firstly as blends of wheat and other flours for
the production of (i) leavened breads, (ii) unleavened baked products, (iii) pastas,
(iv) porridges, and (v) snack foods; or, secondly, wholly non-wheat blends of flours
or meals, for the same purpose. Sometimes, only flour is used as replacement-for
example, tortillas and wheat-less bread from sorghum, pastas from sorghum or
maize. The composite flour for staple foods such as baking items, it should be noted
that there are two reasons for mixing the wheat with other flours i.e. economic and
nutritional. Like that soy flour to increase the protein content of the baked products,
or adding vitamins, is of marginal economic relevance and of debatable use in the
health context. Using blends, now called composite flours (CF), of wheat and other
flours for biscuit making has always occurred in times of scarcity of wheat, from
whatever cause, climatic or economic. As an ingredients being blends in composite
flour may be cassava, maize, rice, sorghum, the millets, potato, barley, sweet potato
and yam. In selecting raw materials for use as alternatives one must consider such
as (a) Compatibility - that is to say, suitability for end use and (b) availability and cost
at point of use (Dendy 1993). Composite flours are quite different from the ready-
mixed flours familiar to millers and bakers. Whereas ready-mixed flours contain all
the non-perishable constituents of the recipe for certain baked products. Composite
flours are only a mixture of different vegetables flours rich in starch or protein, with
or without wheat flour, for certain groups of bakery products. This gives rise to the
following definition: “Composite flours are a mixture of flours from tubers rich in
starch (e.g. cassava, yam, sweet potato) and/or protein rich flours (e.g. soy, peanut)
and /or cereals (e.g. maize, rice, millet, buckwheat), with or without wheat flour.” In
another words, “A flour made by blending or mixing varying proportion of more than
one non-wheat flour with or without wheat flour and used for production of leavened
or unleavened baked or snack products that are traditionally made from wheat flour
and increase the essential nutrients in human diet is called composite flour.”
Functional properties are the fundamental physico-chemical properties that reflect
the complex interaction between the composition, structure, molecular conformation
and physico-chemical properties of food components together with the nature of
environment in which these are associated and measured (Kinsella 1976).
Functional characteristics are required to evaluate and possibly help to predict how
new proteins, fat, fibre and carbohydrates may behave in specific systems as well
as demonstrate whether or not such protein can be used to stimulate or replace
conventional protein (Mattil 1971).

The biscuits available in market are prepared from wheat flour (whole/refined) which
lacks in good quality protein because of its deficiency in lysine; and dietary fibre
contents. Rice flour, mung flour and potato flour which are highly nutritious in protein,
vitamin, minerals and lysine content has been found for its incorporation into
preparation of biscuit. The study provides the information about a commercially
viable application of increasing protein and fibre content in biscuits and also these
can be solve the problem of malnutrition and other essential macro and micro
nutrients deficiency among the population. The objective of the present study was
also to expand the utility of rice flour, green gram flour and potato flour by value
addition through incorporating with wheat flour to prepare the composite flour and
used to develop the biscuit and their characterization. Little work has been reported
on to study the functional properties of flours and biscuits made from composite flour
incorporating wheat flour, rice flour, green gram flour and potato flour. It is clear as
per cited literature that wheat and rice flours are superior source of carbohydrate
and starch content, green gram flour is rich in protein content while potato flour is
rich source of macro and micro mineral and vitamin contents. Present study was
conducted to assess the functional properties of wheat and composite flours and
also evaluate the consumer acceptability.

Material and methods

The experiments were conducted in Bakery Lab and Food Analysis Laboratory in
the Department of Agricultural Engineering and Food Technology, Sardar
Vallabhbhai Patel University of Agriculture and Technology, Meerut (India). Raw
materials viz., wheat flour (maida or refined flour), rice flour, green gram flour, potato
etc. were procured from the local market for the present study. Initial moisture
content of flours and biscuits were determined by hot air oven drying method as
recommended by AOAC (2000).

Evaluation of functional properties of flours

The functional properties of flours were analyzed i.e. Swelling capacity (ml), water
absorption capacity (WAC, %), oil absorption capacity (OAC, %), Emulsion activity
(EA, %), Emulsion stability (ES, %), Foam capacity (FC, %), Foam stability (FS, %),
Gelatinization temperature (GT, °C), Least gelatinization concentration (LGC, %)
and Bulk density (g/cc).

The swelling capacity was determined by the method described by Okaka and Potter
(1977). 100 mL graduated cylinder was filled with the sample to 10 mL mark. The
distilled water was added to give a total volume of 50 mL. The top of the graduated
cylinder was tightly covered and mixed by inverting the cylinder. The suspension
was inverted again after 2 min and left to stand for a further 8 min. The volume
occupied by the sample was taken after the 8th min.

The water absorption capacity of the flours was determine by the method of Sosulski
et al. (1976). One gram of sample mixed with 10 mL distilled water and allow to stand
at ambient temperature (30 ± 2 °C) for 30 min, the centrifuged for 30 min at
3,000 rpm or 2000 × g. Water absorption was examined as per cent water bound
per gram flour. The oil absorption capacity was also determine by the method of
Sosulski et al. (1976). One gram of sample mixed with 10 mL soybean oil (Sp.
Gravity: 0.9092) and allow to stand at ambient temperature (30 ± 2 °C) for 30 min,
the centrifuged for 30 min at 300 rpm or 2000 × g. Water absorption was examined
as percent water bound per gram flour.

The emulsion activity and stability by Yasumatsu et al. (1972) described and followed
as method for the emulsion in present study (1 g sample, 10 mL distilled water and
10 mL soybean oil) was prepared in calibrated centrifuge tube. The emulsion was
centrifuged at 2000 × g for 5 min. The ratio of the height of emulsion layer to the total
height of the mixture was calculated as emulsion activity in percentage. The
emulsion stability was estimated after heating the emulsion contained in calibrated
centrifuged tube at 80 °C for 30 min in a water-bath, cooling for 15 min under running
tap water and centrifuging at 2000 × g for 15 min. The emulsion stability expressed
as percentage was calculated as the ratio of the height of emulsified layer to the total
height of the mixture.

The foam capacity (FC) and Foam stability (FS) by (Narayana and Narsinga Rao
1982) were determined as described with slight modification. The 1.0 g flour sample
was added to 50 mL distilled water at 30 ± 2 °C in a graduated cylinder. The
suspension was mixed and shaken for 5 min to foam. The volume of foam at 30 s
after whipping was expressed as foam capacity using the formula:

Foamcapacity(%)=volumeoffoamAW−VolumeoffoamBWvolumeoffoamBW×100

Where, AW = after whipping, BW = before whipping

The volume of foam was recorded 1 hour after whipping to determine foam stability
as per percent of initial foam volume.

The least gelation concentration (LGC) was evaluated using a method of Coffman
and Garcia (1977) with modification. The flour dispersions of 2, 4, 6, 8, 10, 12, 14,
16, 18, 20, 22, and 30 % (w/v) prepared in 5 mL distilled water was heated at 90 °C
for 1 h in water bath. The contents were cooled under tap water and kept for 2 h at
10 ± 2 °C. The least gelation concentration was determined method given as that
concentration when the sample from inverted tube did not slip.
Gelatinization temperature was determined by Shinde (2001). One gram flour
sample was weighed accurately in triplicate and transferred to 20 mL screw capped
tubes. Ten mL of water was added to each sample. The samples were heated slowly
in a water bath until they formed a solid gel. At complete gel formation, the respective
temperature was measured and taken as gelatinization temperature.

The volume of 100 g of the flour was measured in a measuring cylinder (250 mL)
after tapping the cylinder on a wooden plank until no visible decrease in volume was
noticed, and based on the weight and volume, the apparent (bulk) density was
calculated (Jones et al. 2000).

Biscuit development

The composite flour biscuits were prepared from various combinations of wheat
flour, rice flour, green gram flour and potato flour in ratio of 100:0:0:0, 85:5:5:5,
70:10:10:10 and 55:15:15:15, respectively. The standardized formulations for biscuit
had ingredients as 100 g flour, 45 g sugar, 45 g hydrogenated fat, 1.25 g sodium
bicarbonate, 1.25 g baking powder and 1.0 g curry leave powder. Hot liquid
Hydrogenated fat and sugar were taken and creamed to a uniform consistency. The
flour, required amount of water, baking powder and sodium bicarbonate were added
to creamed mixture and mixed for 10 min at medium speed in dough mixer to obtain
a homogeneous mixture. The dough was rolled out into thin sheet of uniform
thickness and was cut into desired shape using mould. The cut pieces were placed
over perforated tray and transferred into convective baking oven at 180 °C for 10–
15 min till baked. The well baked biscuits were removed from the oven, cooled to
room temperature, packed and stored at room temperature for further studies.

Sensorial evaluation

A semi trained panel consisting of both gender more than 10 judges of different age
groups having different eating habits were constituted to evaluate the quality. The
judges were selected from the faculty staff and students of Department of
Agricultural Engineering & Food Technology, SVPUA&T, Meerut (U.P.). Samples
were served to the panelists and they were asked to rate the acceptability of the
product through sense of organs. The overall acceptability of biscuits were rated on
the basis of 9- point hedonic scale ranging from 1 (extremely dislike) to 9 (extremely
like).

The data obtained from the various experiments were recorded during the study and
were subjected to statistical analysis as per method of “Analysis of Variance” by
Factorial Randomized Block Design (factorial R.B.D.). The significant difference
between the means was tested against the critical difference at 5 % level of
significance (Gomez and Gomez 1984). STATPAC (OPSTAT) software was used
for analyze the recorded data.

Results and discussion

In this research, different functional properties of composite flours were analyzed
with using standard procedures (Table 1). Functional properties or characteristics
are the intrinsic physico-chemical properties that reflect the complex interaction
between the composition, structure, confirmation and physico-chemical properties
of protein and other food components and the nature of environment in which these
are associated and measured (Kinsella 1976). The effect of incorporation
proportions of different flours on the functional properties of composite flours are
discussed as follows.

Table 1. Functional properties of different flours

Functional Wheat flour Composite flours

CD0.05
Properties W100 W85 W70 W55
Moisture, % 13.28 ± 1.47 11.67 ± 0.43 11.34 ± 0.24 10.93 ± 0.09 0.831
SC (ml) 17.60 ± 1.85 16.00 ± 0.71 20.00 ± 0.71 22.30 ± 0.91 1.854
WAC, % 140.00 ± 12.25 132.00 ± 22.80 142.00 ± 14.83 176.00 ± 16.73 22.103
OAC, % 146.00 ± 08.94 130.0 ± 10.00 156.00 ± 26.08 156.00 ± 16.73 19.360
EA, % 43.88 ± 4.12 41.49 ± 1.96 44.00 ± 2.45 44.69 ± 0.96 3.627
ES, % 38.38 ± 4.78 47.27 ± 2.49 48.40 ± 2.97 48.65 ± 3.74 4.870
FC, % 12.92 ± 5.03 14.10 ± 0.29 16.40 ± 2.61 17.60 ± 1.67 3.260
FS, % 1.94 ± 0.05 4.00 ± 0.16 9.20 ± 1.79 13.40 ± 1.09 1.898
GT, °C 59.22 ± 0.15 56.22 ± 0.57 59.42 ± 0.11 60.56 ± 0.06 0.288
LGC, % 8 8 10 10 -
BD (g/cc) 0.762 ± 0.00 0.774 ± 0.00 0.786 ± 0.00 0.820 ± 0.00 0.010

Value = mean ± SE, n = 5

SC  = Swelling capacity, WAC = Water absorption capacity, OAC = Oil absorption

capacity, EA = Emulsion activity, ES = Emulsion stability, FC = Foam capacity,
FS = Foam stability, GT = Gelatinization temperature

LGC = Least gelation concentration, BD = Bulk density, SE = Standard error

W100 : Wheat flour (100 %) or control

W85 : Wheat flour (85 %) + rice flour (5 %) + green gram flour (5 %) + potato flour
(5 %)

W70 : Wheat flour (70 %) + rice flour (10 %) + green gram flour (10 %) + potato flour
(10 %)

W55 : Wheat flour (55 %) + rice flour (15 %) + green gram flour (15 %) + potato flour
(15 %)
Moisture content

Before preparation of composite flour, moisture content of wheat flour, rice flour,
green gram and potato flour were analyzed and found as 13.28 %, 11.22 %, 9.60 %
and 8.05 %, respectively. The moisture content (w.b.) for wheat and composite flours
are presented in Table 1 which ranged from 10.93 % to 13.28 % depending upon
the blending ratio. The moisture content of composite flours i.e. W 100 (13.28 %), W 85
(11.670 %), W 70 (11.342 %) and W 55 (10.928 %) were also determined. From the
Table 1, it is clear that the moisture content of composite flours decreased with
increase in proportions of other flours. The moisture content of composite flour was
highly affected by blending of green gram and potato flours. The highest moisture
content was observed for W 85 (11.67 %) and lowest for W 55 (10.928 %) in the
composite flours. The study revealed that moisture content of composite flours
decreased with decrease in proportions of wheat flour from 100 % to 55 %. Similar
trends were reported by Kaushal et al. (2012). They used the blends of taro, rice and
pigeon pea flour which resulted in reduction of moisture content of composite flours.

Swelling capacity

The swelling capacity of different flours ranged between 16.00 to 22.30 ml. From
Table 1, it is clear that lowest value of swelling capacity was observed in W 85
(16.00 ml) whereas the maximum in W 55 (22.30 ml). The value of swelling capacity
was found for W 100 (17.60 ml) and W 70 (20.00 ml). The swelling capacity of flours
depends on size of particles, types of variety and types of processing methods or
unit operations. As per literature, the flour of parboiled rice has more swelling
capacity as compared to raw rice. Swelling capacity of composite flours increased
with increase in the level of incorporation ratio of rice, green gram and potato flour
and decreased with level of wheat flour. The composite flour (W 55) had the highest
swelling capacity (22.30 ml) while W 85 (16.00 ml) had lowest values. It is explicit that
the swelling capacity of composite flours highly affected by the level of potato flour,
because potato flour was pre-gelatinized and rich source of starch content.

Water absorption capacity (WAC, %)

The water absorption capacity for composite flours is given in Table 1. The WAC
ranged between 132 to 176% for all flours. The WAC was observed highest in W 55
(176 %) and lowest in W 85 (132 %). While composite flours W 70 and wheat flour
(W 100) had 142 and 140% WAC. From the present study, potato flour had highest
WAC (752 %). The result suggests that addition of rice, green gram and potato flour
to wheat flour affected the amount of water absorption. This could be due to
molecular structure of the rice, green gram and potato starch which inhibited water
absorption, as could be seen from the lower values of WAC, with increase in
proportions of other flours to wheat flours. Similar observation was reported by
Kaushal et al. (2012). Kuntz (1971) reported that lower WAC in some flours may be
due to less availability of polar amino acids in flours. The increase in WAC of blends
after incorporating potato flour may be due to increase in the amylose leaching and
solubility and loss of starch crystalline structure.
High WAC of composite flours suggests that the flours can be used in formulation of
some foods such as sausage, dough, processed cheese and bakery products. The
increase in the WAC has always been associated with increase in the amylose
leaching and solubility, and loss of starch crystalline structure. The flour with high
water absorption may have more hydrophilic constituents such as polysaccharides.
Protein has both hydrophilic and hydrophobic nature and therefore they can interact
with water in foods. The good WAC of composite flour (W 55) may prove useful in
products where good viscosity is required such soups and gravies. The observed
variation in different flours may be due to different protein concentration, their degree
of interaction with water and conformational characteristics (Butt and Batool 2010).

Oil absorption capacity (OAC, %)

The OAC ranged between 130 to 156% among all the flours. The composite flours
(W 85 and W 55) had highest OAC (156 % for both) and lowest for W 70 (130 %) as
compared to wheat flour (146 %). It is clear that the OAC of composite flours (W 85
and W 55) increased with increase in the proportion of other flours. The presence of
high fat content in flours might have affected adversely the OAC of the composite
flours. The OAC was found to be insignificant to each other at p ≤ 0.05 level of
significance. Therefore, the possible reason for increase in the OAC of composite
flours after incorporation of potato and green gram flour and is the variations in the
presence of non-polar side chain, which might bind the hydrocarbon side chain of
the oil among the flours. Similar findings were observed by Kaushal et al. (2012).
However, the flours in the present study are potentially useful in structural interaction
in food specially in flavor retention, improvement of palatability and extension of shelf
life particularly in bakery or meet products where fat absorption is desired (Aremu et
al. 2007). The major chemical component affecting OAC is protein which is
composed of both hydrophilic and hydrophobic parts. Non-polar amino acid side
chains can form hydrophobic interaction with hydrocarbon chains of lipids
(Jitngarmkusol et al. 2008).

Emulsion activity and stability (%)

Protein being the surface active agents can form and stabilize the emulsion by
creating electrostatic repulsion on oil droplet surface (Kaushal et al. 2012). The EA
and ES of flours are shown in Table 1. EA of different flours ranged between 41.49
and 44.69%. The highest EA for W 55 flour (44.69 %) and lowest for W 85 (41.49 %)
were observed. Emulsion stability (ES) for different flours varied from 38.38 to
48.65% (Table 1). Highest ES was observed for W 55 flour (48.65 %) followed by W 70
flour (48.40 %), W 85 flour, (47.27 %) and lowest for wheat flour (38.38 %). The EA
and ES of composite flours were found to be significantly increased with decreasing
in the proportions of wheat flour up to 55%. Emulsion stability can be greatly
increased when highly cohesive films are formed by the absorption of rigid globular
protein molecules that are more resistant to mechanical deformation (Graham and
Phillips 1980). Increasing emulsion activity (EA), emulsion stability (ES) and fat
binding during processing are primary functional properties of protein in such foods
as comminuted meat products, salad dressing, frozen desserts and mayonnaise. All
composite flours showed relatively good capacity of emulsion activity.

Foam capacity (FC, %) and Foam stability (FS, %)

Foam capacity of protein refers to the amount of interfacial area that can be created
by the protein (Fennama 1996). Foam is a colloidal of many gas bubbles trapped in
a liquid or solid. Small air bubbles are surrounded by thin liquid films. Foam capacity
of different flours varied from 12.92 to 17.60%. The highest foam capacity was
observed for W 55 flour (17.60 %), W 70 flour (16.40 %), W 85 flour (14.10 %), and
lowest W 100 (12.92 %). The foam stability (FS) refers to the ability of protein to
stabilize against gravitational and mechanical stresses (Fennama 1996). The foam
stability varied from 1.94 to 13.40% among the flours. The highest FS was observed
for W 55 flour (13.40 %) followed by W 70 flour (9.20 %), W 85 flour (4.00 %), and lowest
for wheat flour (1.94 %). FC and FS of composite flours was increased with
increasing in the blending ratio of different flours. There was an inverse relationship
between foam capacity and foam stability. Flours with high foaming ability could form
large air bubbles surrounded by thinner a less flexible protein film. This air bubbles
might be easier to collapse and consequently lowered the foam stability
(Jitngarmkusol et al. 2008).

Gelatinization temperature (GT, °C)

The temperature at which gelatinization of starch take place is known as the

gelatinization temperature (Sahay and Singh 1996). GT of flours ranged 56.22 °C to
60.56 °C. Highest GT was found for W 55 flours (60.56 °C) followed by W 70 (59.42 °C)
and lowest for W 85 (56.22 °C). GT was increased with increase in the incorporation
of rice, green gram and potato flour with wheat flour. The study revealed that the
flour which was higher in starch content took lowest temperature for gelatinization.
As rice and potato flour was took less time for gelatinization due to higher starch
while green gram flour took more time due to lower starch content. The GT of
composite flours was significantly increased with increase in the incorporation of
different flours except wheat flour. Therefore, GT of composite flours increased with
decrease in the incorporation ratio of wheat flour. The gelatinization temperature of
W85 flour was lower than that from W 100 (wheat flour).

Least gelation concentration (LGC, %)

The least gelation concentration (LGC) which is defined as the lowest protein
concentration at which gel remained in the inverted tube was used as index of
gelation capacity. The data for LGC of different flours are given in Table 1.
Composite flours W 70 and W 55 formed a gel at a significantly higher concentration
(10 g/ 100 ml). W 100 and W 85 flour formed gel quickly at very lowest concentration
(8 g/100 ml). Pulse/legume flours contain high protein and starch content and the
gelation capacity of flours is influenced by physical competition for water between
protein gelation and starch gelatinization (Kaushal et al. 2012). The lower the LGC,
the better the gelating ability of the protein ingredient (Akintayo et al. 1999) and the
swelling ability of the flour was enhanced (Kaushal et al. 2012). (Vautsinas and Nakai
1983) reported that protein gelation was significantly affected by exposed
hydrophobicity and square of sulfhydryls of proteins. As the percentage of
incorporation of green gram, potato and rice flour in wheat flour (composite flour)
increased, gelling properties decreased.

The low gelation concentration of W 85 flour as composite flour may be added an

asset for the formation of curd or as an additive to other gel forming materials in food
products. The variation in the gelling properties may be ascribed to ratios of the
different constituents such as protein, carbohydrates and lipids in different
pulse/legume flours, suggesting that interaction between such components may also
have a significant role in functional properties (Aremu et al. 2007). Least gelation
concentration values were compared favorable with those reported for African yam
bean (16 to 20%) by Abbey and Ayuk (1991). However, lower values were recorded
for several Phaseolus species and Lablab bean by Chau and Cheung (1997), and
Deshpande et al. (1982). Sathe et al. (1983) also reported a LGC of 12 % for black
gram flour. The LGC for other flour such as lupin (Sathe et al. 1982), cowpea (Akubor
et al. 2003b), plantain (Akubor et al. 2003a), safflower and maize flour (Akubor 2007)
were 14, 6, 6, 8 and 6% (w/v). The composite flours (W 85, W 70 and W 55) would be
useful in food system such as puddings, sauce and other foods which require
thickening and gelling (Alobo 2003).

Bulk density

The bulk density (g/cm3) of flour is the density measured without the influence of any
compression. The bulk densities of flours ranged from 0.762 g/cc to 0.820 g/cc. The
highest highest bulk density was observed W 55 flour (0.820 g/cc) followed by W 70
flour (0.786 g/cc), W 85 flour (0.774 g/cc) and lowest for wheat flour (0.762 g/cc). The
present study revealed that bulk density depends on the particle size and initial
moisture content of flours. Bulk density of composite flour increased with increase in
the incorporation of different flours with wheat flour. It is clear that decreased the
proportion of wheat flour increase the bulk density of composite flours. The high bulk
density of flour suggests their suitability for use in food preparations. On contrast,
low bulk density would be an advantage in the formulation of complementary foods
(Akapata and Akubor 1999). Therefore, present study suggests that highest bulk
density of composite flour (W 55) suggests its suitability to be used as thickener in
food products and for use in food preparation since it help to reduce paste thickness
which is an important factor in convalescent and child feeding. Bulk density of
composite flours increased significantly with increase in the incorporation of rice,
green gram and potato flour with wheat flour. Similar findings were reported by
Eltayeb et al. (2011).

Sensorial attributes of composite flour biscuits

Biscuits were subjected to a panel of 10 semi-trained judges comprising male and

female (of teenager group) using 9-point Hedonic rating scale (ISI 1971;
Swaminathan 1987; and Ranganna 1995) of different eating habits (Goyal 2008).
Sensorial attributes viz., colour, flavour, taste, crispiness, texture and overall
acceptability are shown in Table 2. Overall acceptability was calculated by taking
average of all the scores of sensorial attributes. The order of presentation of sample
to the panel was place in random position.

Table 2. Sensorial attributes of composite flour biscuits

Overall
Biscuit Crispines
Colour Flavour Taste Texture Acceptabilit
s s
y
8.60 ± 0.09 7.70 ± 0.12 8.10 ± 0.10 8.00 ± 0.11 8.30 ± 0.13
W100 8.14 ± 0.212
9 0 2 0 2
8.10 ± 0.13 8.20 ± 0.09 8.20 ± 0.20 8.60 ± 0.21 8.20 ± 0.32
W85 8.26 ± 0.321
2 8 1 4 1
8.60 ± 0.23 7.70 ± 0.10 8.60 ± 0.22 8.60 ± 0.22 8.50 ± 0.21
W70 8.40 ± 0.214
4 2 0 0 1
8.90 ± 0.22 8.10 ± 0.32 8.90 ± 0.09 8.70 ± 0.23 8.40 ± 0.20
W55 8.60 ± 0.056
1 1 8 0 0

Value = mean ± SE, n = 5

The sensory data for colour scores of composite flours and wheat flour (as control)
biscuits are presented in Table 2. The highest colour score was awarded for fresh
biscuits made from W 55 (8.90) followed by W70 and control biscuits (8.60 for both)
and lowest for W 85 (8.10). The study revealed that colour scores were increased with
increasing incorporation of rice, green gram and potato flour with wheat flour. The
desired colour of biscuits is obtained mainly due to the mallard browning during
baking. Highest flavour scores were observed for W 85 biscuits (8.20) followed by W 55
(8.10) and lowest for W 70 (7.70) and control (7.70 %) just after baking and cooling.
The flavour scores decreased rapidly with increasing incorporation of flours with
wheat flour in the range of 5 to 15 %. With increase incorporation of rice, green gram
and potato flour with wheat flour in biscuits formulation, the sensory score for flavour
decreased. Similar research finding was reported by (Masur Shakuntala et al. 2009).

Results of the study revealed that fresh biscuits made from W 55 flour were observed
highest taste score of 8.90 followed by W 70 (8.60) and W 85 (8.20) while lowest for
W100 (8.10) toward between ‘like very much’ to ‘like extremely’ among all the
samples. Taste score decreased with increasing incorporation of different flours like
rice, green gram and potato flour with wheat flour in the formulation of biscuits.
Crispiness is perceived when food is chewed between molars, and is usually
expressed in terms of hardness and factorability (Noor Aziah and Komathi 2009).
Highest crispiness score was rated for fresh biscuits prepared from W 55 (8.70)
followed by W 70 (8.60), W 85 (8.60) and lowest for W 100 (8.00). Crispiness of biscuits
increased with increasing incorporation of different flours with wheat flour from 5 to
15 percent for each. The measurement of crispiness by sensory mean is not a
straight forward process. The difference between a sensory concept (i.e. the
collection of perception identified as relevant to the same class) and its label (i.e. the
word used by a community to refer to it) should be acknowledged. Thus, the use of
the same label in different studies, especially with trained panelists, is not a
guarantee that the same sensory concept is measured. So, crispiness is a complex
attributed resulting on the one hand from multiple sensations and on the other from
multiple physical parameters, combining molecular, structural and manufacturing
processes, as well as storage conditions (Roudaut et al. 2002).

Sensory score of texture was awarded highest for W 70 biscuits (8.50) followed by
W55 (8.40) and control biscuits (8.30) whereas lowest for W 85 biscuits (8.20) just after
preparation. From present study, it is clear that texture score of biscuits increased
with increasing the incorporation of rice, green gram and potato flour with wheat
flour. The sensorial data for effect on overall acceptability (OAA) of biscuits are given
in Table 2. Highest OAA was scored for W 55 biscuits (8.60) followed by W 70 (8.40)
and W 85 (8.26) while lowest for control biscuits (8.14) just after baking and cooling.
Sensorial data revealed that OAA of biscuits increased with increasing incorporation
of rice, green gram and potato flour with wheat flour in the formulation of biscuits. In
general, OAA of biscuits depends on the individual data of different sensory
attributes like colour, flavor, taste, crispiness and texture. In case of composite flour
biscuits, OAA was awarded highest for W 55 followed by W 70 and W85 as compared
to control biscuits. All biscuits coincided in the range of ‘like moderately’ to ‘like very
much’ for biscuits made from composite flours, while ‘like slightly’ to like moderately’
for control biscuits.

Conclusion

The functional properties of wheat flour and composite flours such as swelling
capacity, water absorption capacity, oil absorption capacity, emulsion activity,
emulsion stability, foam capacity, foam stability, gelatinization temperature, least
gelation concentration and bulk density were increased with increase in the
incorporation of other flours with wheat flour. The result showed that the addition of
rice flour, green gram flour and potato flour to wheat flour in the proportion of 5 to
15 % for each produced acceptable biscuits and also functionality of the flour was
not affected. Sensorial data revealed that Overall acceptability (OAA) of biscuits
increased with increasing in the incorporation of rice, green gram and potato flour
with wheat flour in the formulation of biscuits. The biscuits prepared with the flour
ratio of 55:15:15:15 liked most of the panelists. Incorporation of above flours to wheat
flour would therefore be an effective method of cost reduction of biscuits and other
allied products and solving malnutrition problems in children in India.

References

1. Abbey BW, Ayuk EJ. Functional properties of African yam bean flour
(Sphenasylis sternocarpa) Nig J Nutri Sci. 1991;12:44–46.
2. Akapata MI, Akubor PI. Chemical composition and selected functional
properties of sweet orange (Citrus sinensis) seed flour. Plant Food Hum Nutr.
1999;54:353–362. doi: 10.1023/A:1008153228280. [PubMed] [Cross Ref]
3. Akintayo ET, Oshadi AA, Esuoso KO. Effect of NaCl, ionic strength and pH
on the foaming and gelation of pigeon pea (Cajanus cajan) protein
concentrate. Food Chem. 1999;66:51–56. doi: 10.1016/S0308-
8146(98)00155-1. [Cross Ref]
4. Akubor PI. Chemical, functional and cookie baking properties of
soybean/maize flour blends. J Food Sci Technol. 2007;44(6):619–622.
5. Akubor PI, Adamolekun FA, Oba CO, Obari IT. Chemical composition and
functional properties of cowpea/plantain flour blends cookie production. Plant
Food Hum Nutr. 2003;58:1–9.
6. Akubor PI, Benje FN, Obiegbuna JE. Effects of cowpea supplementation on
the functional and biscuit making properties of wheat flour. J Sustainable
Agric Environ. 2003;5:247–253.
7. Alobo AP. Proximate composition and selected functional properties of
defatted papaya (Carica papaya) kernel papaya. Plant Food Hum Nutr.
2003;58:1–7.
8. AOAC . Official methods of analysis of AOAC International (17th ed.).
Gaitherburg. USA: AOAC International Inc.; 2000.
9. Aremu MO, Olaofe O, Akintayo ET. Functional properties of some Nigerian
varieties of legume seed flour concentration effect on foaming and gelation
properties. J Food Technol. 2007;5(2):109–115.
10. Butt MS, Batool R. Nutritional and functional properties of some promising
legumes proteins isolates. Pakistan J Nutr. 2010;9(4):373–379. doi:
10.3923/pjn.2010.373.379. [Cross Ref]
11. Chau CF, Cheung PCK. Functional properties of flours prepared from three
Chinese indigenous legume seeds. Food Chem. 1997;61:429–433. doi:
10.1016/S0308-8146(97)00091-5. [Cross Ref]
12. Coffman CV, Garcia VV. Functional properties and amino acid content of a
protein isolate from mung bean flour. J Food Technol. 1977;12:473–480. doi:
10.1111/j.1365-2621.1977.tb00132.x. [Cross Ref]
13. Dendy DAV (1993). Review of composite flour technology in the context of
Tanzania. Present a paper at the workshop entitled ‘Sorghum and millets
marketing and utilization’ held at Acrusha, Tanzania during 3–5 May, 1993.
14. Deshpande SS, Sathe SK, Cornforth D, Salunkhe DC. Effect of dehulling on
functional properties of dry bean (Phaseolus vulgaris L.) flours. Cereal Chem.
1982;59:396–401.
15. Eltayeb ARSM, Ali AO, Abou-Arab A-SFM. Chemical composition and
functional properties of flour and protein isolate extracted from Bambara
ground nut (Vigna subterranean) African J Food Sci. 2011;5(2):82–90.
16. Fennama RO. Food Chemistry (3rd eds) Basel, Hong Kong: Marcel Dekker
Inc. New York; 1996. pp. 36–39.
17. Gomez KA, Gomez AA. Statistical procedures for agricultural research (2nd
eds) John Wiley & Sons, New York: A Wiley International Pub; 1984. pp. 8–
12.
18. Goyal SK (2008). Studies on development of Safed Musli (Chlorophytum
borivilianum) and Stevia (Stevia rebaudiana) products and quality evaluation
during storage. Ph. D. Thesis submitted to Department of Agricultural
Engineering and Food Technology, SVPUAT Meerut (UP).
19. Graham DE, Phillips MC. The conformation of proteins at air-water interface
and their role in stabilizing foam. London: Academic; 1980.
20. ISI (1971). IS: 6723, Guide for sensory evaluation of foods. Part-II, Indian
Standards Institute, New Delhi
21. Jitngarmkusol S, Hongsuwankul J, Tananuwong K. Chemical composition,
functional properties and microstructure of defatted macademice flours. Food
Chem. 2008;110:23–30. doi: 10.1016/j.foodchem.2008.01.050. [PubMed]
[Cross Ref]
22. Jones D, Chinnaswamy R, Tan Y, Hanna M. Physiochemical properties of
ready-to-eat breakfast cereals. Cereal Foods World. 2000;45:164–168.
23. Kaushal P, Kumar V, Sharma HK. Comparative study of physico-chemical,
functional, anti-nutritional and pasting properties of taro (Colocasia
esculenta), rice (Oryza sativa), pegion pea (Cajanus cajan) flour and their
blends. LWT-Food Sci Technol. 2012;48:59–68. doi:
10.1016/j.lwt.2012.02.028. [Cross Ref]
24. Kinsella JE. Functional properties of protein in food- A survey. Crit Rev Food
Sci Nutr. 1976;5:219–225. doi: 10.1080/10408397609527208. [Cross Ref]
25. Kuntz ID. Hydration of macromolecules III. Hydration of polypeptides J Am
Chem Socie. 1971;93:514–515. doi: 10.1021/ja00731a036. [Cross Ref]
26. Masur Shakuntala B, Tarachand KC, Kulkarni Uma N. Development of high
protein biscuits from Bengal gram flour. Karnataka J Agric Sci.
2009;22(4):862–864.
27. Mattil KF. The functional requirement of protein in foods. J Am Oil Chem Soc.
1971;48:477. doi: 10.1007/BF02544664. [PubMed] [Cross Ref]
28. Narayana K, Narsinga Rao MS. Functional properties of war and heat
processed winged bean (Psophocarpus tetragonolobus) flour. J Food Sci.
1982;42:534–538.
29. Noor Aziah AA, Komathi CA (2009). Acceptability attributes of crackers made
from different types of composite flours. Intern Food Res 16: 479–482 (2009).
30. Okaka JC, Potter NN. Functional and storage properties of cow pea-wheat
flour blends in bread making. J Food Sci. 1977;42:828–833. doi:
10.1111/j.1365-2621.1977.tb12614.x. [Cross Ref]
31. Ranganna S (1995). Handbook of analysis and quality control for fruits and
vegetables products (2nd ed.). Tata McGraw Hill Publishing Co. Ltd. New
Delhi.
32. Roudaut G, Dacremont C, Valles Pamieb B, Colas B, LeMestle M. Crispiness:
A critical review on sensory and material science approaches. Trends in Food
Sci Technol. 2002;13:217–227. doi: 10.1016/S0924-2244(02)00139-5.
[Cross Ref]
33. Sahay KM, Singh KK (1996). Unit operations of Agricultural Processing. Vikas
Publishing House Pvt. Ltd. New Delhi. P. 237
 34. Sathe SK, Deshpade SS, Salunkhe DK. Functional properties of winged bean
(Psophoxarpus tetragonolobus, L) proteins. J Food Sci. 1982;47:503–506.
doi: 10.1111/j.1365-2621.1982.tb10112.x. [Cross Ref]
35. Sathe SK, Deshpade SS, Salunkhe DK. Functional properties of black gram
(Phaseolus mango L.) protein. LWT-Food Sci Technol. 1983;16:69–74.
36. Shinde BG. Isolation and characterization of starch horse grain. Rahuri
(India): Unpublished M.Sc. Thesis. Mahatma Phule Krishi Vidyapeeth; 2001.
pp. 16–17.
37. Sosulski FW, Garatt MO, Slinkard AE. Functional properties of ten legume
flours. Intern J Food Sci Technol. 1976;9:66–69.
38. Swaminathan M (1987). Sensory methods of analysis of foods. In: Food
Science, Chemistry and Experimental Foods. Bangalore Printing and
Publishing Co. Ltd., Bangalore. P-286-299.
39. Vautsinas LP, Nakai S. A simple turbidimetric method of determining the fat
binding capacity of protein. J Agric Food Chem. 1983;31:58–61. doi:
10.1021/jf00115a015. [PubMed] [Cross Ref]
40. Yasumatsu K, Sawada K, Maritaka S, Toda J, Wada T, Ishi K. Whipping and
emulsifying properties of soy bean products. Agri Biol Chem. 1972;36:719–
727. doi: 10.1271/bbb1961.36.719. [Cross Ref]

XXX. Processing of a novel powdered herbal coffee

(Pistacia Terebinthus L. Fruits Coffee) and its sensorial
properties
J Food Sci Technol. 2015 Jul; 52(7): 4625–4630.
S. S. Secilmis, D. Kocak Yanık, and F. Gogus

Abstract

In this study, the effects of roasting method, grinding and reduction in oil content on
the characteristics of Pistacia terebinthus fruit coffee were investigated. Pistacia
terebinthus fruit was roasted by microwave, pan and combined (microwave and
convection) methods. The degree of roasting was determined by L*, a*, b* color
values. The roasting times were 1,500, 1,900 and 1,620 s for microwave, pan and
combined roasting methods, respectively. Cold press was used to reduce the oil
content both prior to roasting and after the roasting. The oil content was reduced to
around 21.5 % in all roasting methods to approach to that of coffee beans. Powdered
Pistacia terebinthus fruit coffee brews were compared with each other and Turkish
coffee in terms of aroma, flavor, acidity aftertaste, and overall acceptability. Sensorial
analysis results showed that coffee brews prepared by pressing after the roasting
process were better than those pressing prior to roasting.
Keywords: Pistacia terebinthus fruit, Coffee, Roasting, Grinding, Oil removal,
Sensory properties

Introduction

Coffee is consumed in high amounts around the world and differs in terms of roasting
and brewing conditions with respect to different regions. However, many coffee
substations known as herbal coffee have been used to reduce negative side effects,
such as high caffeine content, of excessive coffee consumption. This allowed
obtaining herbal products which are produced and consumed similar to coffee
beans.

Pistacia Terebinthus fruit (PTF) coffee or “Menengiç” coffee is the most famous and
traditional herbal coffee in Turkey. The fruits have been obtained from Pistacia
terebinthus L. which is a member of the family Anacardiaceae. The PTF has been
used in folk medicine for gastralgia (internally), rheumatism and cough (externally)
and also as stimulant, diuretic and antitussive (Baytop 1984; Matthaus and Özcan
2006; Özcan 2004; Walheim and Stebbins 1981). In various regions of the world,
different organs of this tree are exploited for several purposes because of its
antioxidant effect (Topçu et al. 2007). Göğüş et al. (2011) reported that PTF contains
similar flavor compounds to coffee beans, suggesting that PTF may provide an
alternative for use in the coffee industry. Therefore, PTF have a high potential as an
alternative to coffee beans because of its unique flavor and aroma. Çiftçi et al. (2009)
reported that the PTF coffee was found to be rich in vitamins, unsaturated fatty acids
and trace elements, suggesting that they may be valuable for drink uses.

Roasting is a very important process for the coffee industry due to development of
aroma and flavor compounds. Buffo and Cardelli-Freire (2004) reported that roasting
methods and conditions have an important effect on the formation of the roasted
coffee flavors. Nebesny et al. (2007) showed that uniformly roasted coffee bean with
full aroma could be obtained by using microwave roasting. Göğüş et al. (2011)
reported that traditional pan roasting is the most common method for PTF coffee
processing. In contrast to Arabica coffee, PTF coffee has an oily structure like sludge
because of its high oil content. High oil content also results in a heavy oil flavor to
drink in final brewed coffee. This needs to reduce oil content to obtain a product in
powder form with a more desirable flavor.

The aim of this study was to investigate the effects of different roasting methods
before or after the removal of oil by cold pressing and particle size on sensorial
properties of powdered P. terebinthus fruit coffee brews.

Material and methods

Materials
The ripening PTF which were collected in September, 2011 season, were purchased
from a local market in Gaziantep, Turkey. The PTF were cleaned manually to remove
all undesirable materials such as dust, dirt, stone and broken kernels. They were
stored at 4 °C in polypropylene bag until used. Commercial roasted Turkish coffee
(TC) which was ground and packed in September, 2012 was purchased from a local
market.

Roasting, oil removal, grinding and fractionation

PTF were roasted by microwave, pan and combined roasting methods. 200 g
samples were used for each roasting process. PTF was stirred throughout roasting
to obtain homogen samples. The degree of roasting was followed by L*, a*, b* color
values. The roasted PTF was stored at 4 °C in polypropylene bags until used. A
programmable and the coupled convection-microwave domestic oven (Arçelik MD
583, Turkey), providing both convectional (convectional power of 1,250 W) and
microwave heating (maximum output of 1,300 W at 2,450 MHz) was used for
microwave roasting (MWR) and combined (convection and microwave) roasting
(CR). Pan roasting (PR) was carried out in a metal plate by using hot plate with a
magnetic stirrer.

Oil removal was carried out by cold pressing with a laboratory scale press (Ceselsan,
YP 0420, Turkey). Oil was removed both prior to roasting (Pressing-Roasting) and
after roasting (Roasting-Pressing). The oil content was reduced to approximately
21.5 %.

PTF were ground using laboratory blender (Waring HGB150, USA) for 30 s.
Fractionation was carried out by using the five sieves with mesh widths of 425, 450,
530, 710 and 800 μm. The particles were rotated all directions continuously to pass
through the sieve opening. The sieved particles were stored at 4 °C until sensory
analysis.

Determination of color and oil content

The color of samples was measured by a Hunter-Lab ColorFlex, A60-1010-615

model colorimeter (HunterLab, Reston, VA). The instrument (45°/0°geometry, D65
optical sensor, 10°observer) was standardized each time with a black and a white
(L* =93.76, a* = −1.05, b* = 0.74) tile. The L*, a*, b* color space was used to express
the color changes. The L* shows whiteness or brightness/darkness, a*
(redness/greenness) and b* (yellowness/blueness). The experiments were carried
out in triplicate.

The oil content of samples was determined by the Soxhlet extraction technique. The
oil from samples was extracted using hexane in a Soxhlet instrument (Velp
Scientifica, Usmate, Italy). The oil extract was collected in an extraction cup. The
extracted oil was weighed to determine the oil content. The experiments were carried
out in triplicate.
Determination of pH, titratable acidity and free fatty acids

2.00 g of sample was accurately weighed into a 200 ml glass bottle, and 100 ml of
deionized water was added in. The glass bottle was boiled for 10 min. Then, 50 ml
of the filtered extract was cooled to room temperature and used for pH determination
with a pH meter.

The titratable acidity was determined by AOAC 920.92. The results are expressed
as 0.1 ml alkali required neutralizing acidity of 100 g sample. Measurements were
taken in triplicate.

Free fatty acid content of extracted oil was analyzed according to official methods
AOCS Ca 5a-40. Measurements were taken in triplicate.

Coffee brewing and sensory analysis

PTF coffee and TC brews were prepared according to the standard procedure used
for Turkish coffee (5 g of ground coffee for each sample was gently mixed with 60 ml
water for each panelist and the mixtures were cooked in a coffee maker until first
boiling). The coffee brews were subjected to sensory analysis for aroma, flavor,
acidity, after taste and overall acceptability. The sensory evaluation was carried out
by 10 panelists (7 women and 3 men) selected from students and laboratory staff
using 9 point hedonic scale. Samples for analysis were served separately in a well
lit room on coded cups. The panelists washed their mouth with water before they
were served the subsequent coffee brewing.

Statistical analysis

The analysis results were compared with TC as a reference sample by one-way

analysis of variance (ANOVA). Means of the groups were compared using Duncan‘s
multiple range test using a SPSS statistical packet (Version 16, 2007, Polar
Engineering and Consulting, Nikiski, USA).

Result and discussion

The relationship between roasting methods and color values and oil content

Figure 1 shows the flow diagram for the production of powdered terebinthus fruit
coffee from the raw terebinthus fruit by roasting and cold pressing. Roasting
processes performed were pan, microwave and microwave-convection. Pressing
process has been applied to reduce the oil content of terebinthus fruit prior to
roasting and just after the roasting. It has been found that pressing prior to roasting
resulted in poor homogeneity in roasting process because of uneven particle size
distribution. Especially small particles produced during pressing burned in the first
period of the roasting process. So, pressing after the roasting process has been
used for further applications to obtain a homogenously roasted sample with lower
oil. In addition, the removal of oil was more successful in pressing after the roasting
applications. Dalgıç et al. (2011) also showed that PTF roasted at different
temperatures has a high oil yield when compared with unroasted ones.

Fig. 1. Flow diagram for Pistacia Terebinthus fruit coffee production

The time of roasting is a critical parameter on the characteristics of final product. The
time of roasting has been decided by the color values to make it comparable for
various roasting methods. The color values of the most desirable product have been
decided by many trials. The oil has been removed from the roasted samples to
observe the color, aroma and odor for each trial. The color values chosen as a
reference have been decided by the end of their pre-sensorial analysis. The L*, a*,
b* values chosen as a reference were in a range of 18.00–18.05, 1.0–1.5 and 1.5–
2.0, respectively. The roasting times needed to obtain those color values were 1,500,
1,900 and 1,620 s for microwave, pan and combined roasting methods, respectively.
Roasting time has been kept constant in all following experiments.

Table 1 shows the oil contents of Turkish coffee, P. terebinthus fruit and the
processed (microwave, pan and combined) terebinthus fruits after the removal of oil.
It has been observed that terebinthus fruits have an initial oil content of 40.05 %.
However, the terebinth fruit with very high oil content was not practical to obtain a
powder product after grinding. The trials showed that the critical point was 26 % to
obtain a product in powder form. So, it has been reduced to almost half of its initial
value after the roasting following with cold press oil extraction process. The oil
contents found for processed terebinthus fruits were quite similar to that of
commercial Turkish coffee. However, these values of oil content were still higher
than those found by Mazzafera (1999) for Arabica coffee.

Table 1. The effect of roasting conditions on color characteristic and oil content of
P.terebinthus fruits

Samples Oil Content (%) L* a* b*

TC 16.05 ± 0.13 36.68 ± 0.53 15.44 ± 0.22 27.44 ± 0.33a
a a

Raw PTF 40.50 ± 0.95 28.48 ± 0.95b −4.79 ± 0.11b 4.42 ± 0.17b

MWR 21.50 ± 0.72 18.37 ± 0.63c 1.16 ± 0.08c 1.84 ± 0.13c
CR 21.02 ± 0.88 18.34 ± 0.14c 1.14 ± 0.13c 1.63 ± 0.31c
PR 20.06 ± 0.93 18.12 ± 0.62c 1.20 ± 0.12c 1.78 ± 0.38c

Values are mean ± standard deviation (n = 3). Different letters indicate statistically
significant differences by Duncan’s multiple range test at p < 0.05. TC: Turkish
coffee, PTF: Pistacia Terebinthus fruits, MWR: Microwave roasting, CR: Combined
roasting, PR: Pan roasting

Table 1 presents the change in color values with respect to the applied process. The
L*, a*, b* color values of raw terebinth fruit describing brightness (L*), greenness or
redness (±a*) and yellowness (b*) were found to be 28.48, −4.79 and 4.42,
respectively. It has been observed that L*, a*, b* values found statistically different
when terebinthus fruits have been processed by three different methods to obtain a
powder product. L* values have been reduced sharply in all cases to around 18.00.
The roasting caused the loss of greenness and the final product in each case had a
red character. Yellowness followed an irregular path during roasting process.
However, it resulted in a decrease by the end of overall process. Final L*, a*, b*
values were almost the same for different roasting methods.

Particle size distribution

The size of coffee particles has an important affect on the final sensorial quality of
coffee because of solubility of the matter. Clarke (1987) claimed that in brew coffee,
when the grinding grade is finer, the extraction of soluble and volatile compounds is
higher. Illy and Viani (2005) also found that particle size is one of the important
factors which affects the sensory properties of coffee. Figure 2 shows the particle
size distribution of ground and pressed terebinthus fruits. The highest particle size
range was found to be 425–450 μm (26.67 %). The sizes in descending order with
regards to the amounts were found as < 425 μm (23.00 %), 450–530 μm (17.67 %),
over 800 μm (17.33 %) and 530–710 μm (15.33 %). The results found for particle
sizes of terebinth fruit coffee were in a good agreement with those of earlier studies
(Nebesny et al. 2007; Czerny et al. 1999). Even if the whole ground product had a
homogeneous distribution with a percentage over 80 % in a size range of 250–
800 μm, the size range of 425–450 μm was found as the best in pre-sensorial
analysis. This size range was used in the following chemical and sensorial analysis.

Fig. 2. Particle size distribution of ground powder Pistacia Terebinthus fruit coffee

Titratable acidity and pH values of terebinth fruit coffee roasted at different conditions
were presented in Table 2. Titratable acidity values were found to be 165.48, 237.05
and 220.83 ml NAOH/100 g coffee for MWR, PR and CR, respectively. Titratable
acidity for commercial TC used in this study was found to be 420.19 ml NAOH/100 g
coffee. Mazzafera (1999) presented the titratable acidity of different coffee samples
in the range of 228–267 (ml/100 g). Martin et al. (1999) were also reported the
titratable acidity of coffee samples in the range of 101–114 (ml/100 g). Titratable
acidity of commercial TC is found statistically different (p < 0.05). The different
roasting methods of terebinthus fruits caused significant difference in titratable
acidity statistically (p < 0.05). They are all found different to each other. The
differences may be due to roasting conditions, preparation of brew and age of coffee.

Table 2. The effect of roasting conditions on titratable acidity, pH, and free fatty acid

Sample Titratable Acidity (ml NaOH/100 g coffee) pH FFA (% Oleic)

TC 420.19 ± 0.48 a 5.60 ± 0.03 3.88 ± 0.06a
a

MWR 165.48 ± 0.53b 6.37 ± 0.01b 2.69 ± 0.04b

CR 220.83 ± 0.45c 6.57 ± 0.03c 2.70 ± 0.02b
PR 237.05 ± 0.32d 6.27 ± 0.04d 2.73 ± 0.01b

Values are mean ± standard deviation (n = 3). Different letters indicate statistically
significant differences by Duncan’s multiple range test at p < 0.05. TC: Turkish
coffee, PTF: Pistacia Terebinthus fruits, MWR: Microwave roasting, CR: Combined
roasting, PR: Pan roasting. FFA: Free fatty acid
The pH values were found to be 6.37, 6.27 and 6.57 for MWR, PR and CR,
respectively. The results found in this study are in a good agreement with those
reported in literature. Somporn et al. (2011) reported that pH values of Arabica coffee
were found to be 5.46, 5.49 and 5.48 for light, medium and dark roasting,
respectively. Santos and Oliveira (2001) were also reported the similar pH values for
different coffee samples. However, the pH value of TC was found to be statistically
different (p < 0.05) than those of pistacia terebinth coffee roasted at various
conditions. It has been discussed by Brollo et al. (2008) that the titratable acidity in
coffee brews could be a more reliable indicator for correlating coffee acidity than pH
values. Even if there is a good correlation between pH and titratable acidity values
in this study, especially higher pH values support the idea of more reliable values of
titratable acidity. Acidity basically is a result of the formation of formic, acetic, glycolic
and lactic acids during roasting (Ginz and Bradbury 2000). These are all degradation
products of oils, proteins and/or carbohydrates. Göğüş et al. (2011) identified some
of these acids as volatile components in pistacia terebinth coffee in various stages
of roasting. Various roasting conditions of terebinthus fruits caused significant
difference on the pH value. They are all found different to each other.

Table 2 also represents the free fatty acid of oil of terebinth fruit coffee as quality
parameter to show the effect of roasting on the degradation of triglycerides. It is seen
that the Turkish coffee which is used as a control has highest free fatty acid content.
In addition, free fatty acid content of Turkish coffee found statistically different (p 
< 0.05) than those of pistacia terebinth fruit coffees. They were found to be 2.69,
2.70 and 2.73 for MWR, CR and PR, respectively. Even if there is no statistically
difference among them, pan roasting has the highest value in free fatty acid content
compared to those of other roasting methods of terebinth fruit coffee. This might be
because of longer roasting time in pan roasting (1,900 s). Geçgel Ü and Arıcı (2008))
reported that free fatty acid changes between 0.76 and 2.40 in the raw oil of PTF.
Kocak et al. (2011) also found free fatty acid as 0.59 for raw pistacia terebinth fruit
oil. The increased value of free fatty acids compared to earlier studies is a clear
indication of high temperature roasting applications.

Evaluation of sensory properties

The average values of sensory properties, namely aroma, flavor, acidity, after taste
and overall acceptability were given for brews of terebinth fruit coffee roasted by
various methods and Turkish coffee in Table 3. Aroma scores were ranged from 5.6
to 6.2 points for powdered PTF coffees. It has been found that there is no difference
statistically (p  < 0.05) in aroma values of studied coffee samples. The highest
average aroma score was given to terebinth fruit coffee roasted by a combined
(microwave-convective) method. Turkish coffee which is used as a reference had an
aroma value of 5.5 with a very high standard deviation. These results show that the
panelists are in a good agreement for powdered pistacia terebinth fruits coffee
brews. Furans, furanones, benzene derivates and pyrazines are defined as typical
components of coffee volatiles. Göğüş et al. (2011) reported that the number of total
volatile compounds increased with increasing roasting time up to 20 min of pan
roasting then decreased slightly at 25 min. They also found that furans and
furanones produced in the first 5 min of roasting and they were the dominating
components by the end of the roasting of pistacia terebinth fruits. Nebesny et al.
(2007) also found similar results in terms of availability of furans and furanones in
Robusta coffee brews.

Table 3. Effect of roasting conditions on sensory attributes of powdered Pistacia

terebinthus fruits coffee brews

Control Roasting Conditions

Sensory Attributes
TC MWR CR PR
Aroma 5.5 ± 2.27a 5.6 ± 0.84a 6.2 ± 1.47a 5.6 ± 1.26a
Flavor 5.8 ± 2.09b 5.4 ± 0.97b 6.1 ± 1.59b 5.8 ± 1.23b
Acidity 5.2 ± 1.75c 5.3 ± 1.33c 6.0 ± 1.63c 6.3 ± 1.33c
After Taste 5.8 ± 1.55d 5.8 ± 1.47d 6.0 ± 1.88d 6.0 ± 1.56d
Overall Acceptability 5.9 ± 2.28e 5.6 ± 1.34e 6.2 ± 1.75e 6.3 ± 1.15e

Values are mean ± standard deviation (n = 10). Different letters indicate statistically
significant differences by Duncan’s multiple range test at p < 0.05. TC: Turkish
coffee, PTF: Pistacia Terebinthus fruits, MWR: Microwave roasting, CR: Combined
roasting, PR: Pan roasting

Flavor scores were ranged from 5.4 to 6.1 points for powdered PTF coffees. The
highest flavor score was given to powdered PTF coffee which is roasted in pan.
Turkish coffee flavor score was found as 5.8. This score is still lower that of combined
roasting method. However, ANOVA results showed that there is no difference
(p < 0.05) in flavor scores of studied coffee samples, statistically. Flavor is the most
important criterion for coffee quality evaluation and also one the major motivations
for consumer preference in coffee industry (Clarke 1987). The chemistry of flavor
development during coffee roasting is highly complex and not completely
understood. Even though, roasting process appears to be simple in terms of
processing conditions, it is quite complex from a chemistry point of view, since
hundreds of chemical reactions take place simultaneously. In roasted coffee, a large
proportion of macromolecules are composed of unidentified materials, including
melanoidins which is considered as final products of Maillard reaction (Andriot et al.
2004). Göğüş et al. (2011) found that the Maillard reaction products were dominant
components of volatiles of pistacia terebinthus coffee.

Acidity scores of powdered PTF coffee roasted by various methods were found in a
range of 5.3 to 6.3 points. The highest acidity score was given to the powdered PTF
coffee brewed from the roasted samples by combined method. Acidity score of
Turkish coffee was lower than those of powdered PTF coffees. Coffee acidity is the
bright and dry taste that adds life to a coffee. Some such researchers as Pangborn
(1982), Sivetz and Desrosier (1979) and Griffin and Blauch (1999)) claimed that the
pH of a coffee has been found to correlate with the perceived acidity in coffee.
Whereas, Voilley et al. (1981) suggested that titratable acidity produces a better
correlation to perceived coffee acidity. The results found for chemical acidity values
and sensorial acidity scores are in a good agreement. The chemical acidity value of
Turkish coffee found higher compared to those of powdered PTF coffees. Higher
acidity of Turkish coffee could be correlated with its very fine particles. International
Coffee Organization (1991) supported that amount of some organic acids (lactic,
malic, quinic, chlorogenic) in coffee brews increased from course grind to extra fine.
However, the panelists preferred the powdered PTF coffees with a moderate acidity.
Scores given for after taste were ranged from 5.8 to 6.0 points for powdered PTF
coffees. The samples brewed from powdered PTF coffees obtained from combined
and pan roasting processes had the highest aftertaste score. Aftertaste is the finish
of food or drink, and was defined as the sensation present in the mouth immediately
following the removal of whatever food or drink was being consumed (Jackson
2011). One of the crucial factors affecting aftertaste is the ability to properly develop
the flavor profile during roasting. Even if there is no difference (p < 0.05) in flavor
scores among the studied samples statistically, CR and PR caused the production
of most preferable flavor profile in this study.

The overall acceptability score of PR samples was found highest as 6.3. It was
lowest in samples brewed from roasted PTF by MWR. The score found for TC was
5.9. Overall acceptability includes all cupping vocabularies because of defining the
overall quality and acceptability of the product. Statistical analysis showed that there
is no significant difference in overall scores of samples. However, most of the
panelists have chosen the PR coffee as the best in terms of its overall acceptability.
This result has also been supported with the results of other sensorial and chemical
analysis.

Conclusion

High oil content in the PTF was reduced from 40.5 to about 21.5 % by cold press
before roasting process. PTF free from excessive oil content was roasted by
microwave, combined (microwave and convective) and pan. Roasted PTF were
ground and sieved through different screens to obtain powdered coffee. The
sensorial analysis has been performed for the brew of powdered PTF coffee with a
particle size range of 425–450 μm. The sensorial results showed that powdered
Pistacia terebinthus fruits coffee was not found to be different (p  < 0.05) from Turkish
coffee in terms of aroma, flavor, acidity, aftertaste and overall acceptability,
statistically. The results of this study showed that it is possible to produce a novel
powdered caffeine-free herbal coffee. It can be used as an alternative to Turkish
coffee in the coffee industry.

Acknowledgments

The financial support of the department of Food Engineering at the Gaziantep

University is gratefully acknowledged.

References
1. Andriot I, Quere JL, Guichard E. Interactions between coffee melanoidins and
flavor compounds: impact of freeze-drying (method and time) and roasting
degree of coffee on melanoidins retention capacity. Food Chem.
2004;85:289–294. doi: 10.1016/j.foodchem.2003.07.007. [Cross Ref]
2. Baytop T (1984) Treatment with plants in Turkey. Istanbul University
Publication. No: 3255. Istanbul University press, Istanbul.
3. Brollo G, Cappucci R, Navarini L (2008) Acidity in coffee: Bridging the gap
between chemistry and psychophysics. 22nd International Conference on
Coffee Science, Capinas, SP, Brazil, pp 270–280.
4. Buffo RA, Cardelli-Freire C (2004) Coffee flavour: an overview. Flavour Frag
J 19:99–104
5. Çiftçi H, Özkaya A, Kariptaş E. Determination of fatty acids, vitamins and trace
elements in Pistacia terebinthus coffee. J Food Agric Environ. 2009;7:72–74.
6. Clarke RJ. Grading, storage, pre-treatments and blending coffee. In: Clarke
RJ, Macrae R, editors. Coffee technology. Amsterdam: Elsevier Applied
Science; 1987. pp. 35–58.
7. Czerny M, Mayer F, Grosch W. Sensory study on the character impact
odorants of roasted arabica coffee. J Agr Food Chem. 1999;47:695–699. doi:
10.1021/jf980759i. [PubMed] [Cross Ref]
8. Dalgıç L, Sermet OS, Ozkan G. Farklı kavurma sıcaklıklarının menengiç yağ
kalite parametreleri üzerine etkisi. Akademik Gıda. 2011;9:26–36.
9. Geçgel Ü, Arıcı M (2008) Çıtlık (Pistacia terebinthus L.) çesitlerinin bazı
fizikokimyasal özellikleri ile yağ asitleri kompozisyonunun belirlenmesi.
Gidadernegi.org,
http://www.gidadernegi.org/TR/Genel/dg.ashx?DIL=1&BELGEANAH=1612&
DOSYAISIM=240934712.pdf. Accessed 01.09.2013
10. Ginz M, Bradbury HH. Formation of aliphatic acids by carbohydrate
degradation during roasting of coffee. Eur Food Res Technol. 2000;211:404–
410. doi: 10.1007/s002170000215. [Cross Ref]
11. Göğüş F, Ozel MZ, Kocak D, Hamilton JF, Lewis AC. Analysis of roasted and
unroasted Pistacia terebinthus volatiles using direct thermal desorption-
GCxGC–TOF/MS. Food Chem. 2011;129:1258–1264. doi:
10.1016/j.foodchem.2011.05.003. [PubMed] [Cross Ref]
12. Griffin M, Blauch J (1999) Determination of the relationship between
phosphate concentration and perceived acidity in coffee. In: Association for
science and information on coffee, Helsinki, pp 118–126
13. Illy A, Viani R. Espresso coffee: the science of quality. 2. London: Elsevier;
2005.
14. International Coffee Organization (1991) Sensory evaluation of coffee:
Technical Unit Quality Series, London, pp 209–243
15. Jackson R (2011) Look at aftertaste. Stocktongraham,
http://www.stocktongraham.com/a-look-at-aftertaste. Accessed 18.09.2013.
16. Kocak D, Keskin H, Fadıloglu S, Kowalski B, Göğüş F (2011) characterization
of terebinth fruit oil and optimization of acidolysis reaction with caprylic and
stearic acids. J Am Oil Chem Soc 88:1531–1538
17. Martin MJ, Pablos F, Gonzales AG. Characterization of arabica and robusta
roasted coffee varieties and mixture resolution according to their metal
 content. Food Chem. 1999;66:365–370. doi: 10.1016/S0308-8146(99)00092-
8. [Cross Ref]
18. Matthaus B, Özcan M. Quantitation of fatty acids, sterols, and tocopherols in
turpentine (Pistacia terebinthus Chia) growing wild in Turkey. J Agric Food
Chem. 2006;54:7667–7671. doi: 10.1021/jf060990t. [PubMed] [Cross Ref]
19. Mazzafera P. Chemical composition of defective coffee beans. Food Chem.
1999;64:365–370. doi: 10.1016/S0308-8146(98)00167-8. [Cross Ref]
20. Nebesny E, Budryn G, Kula J, Majda T. The effect of roasting method on
headspace composition of robusta coffee bean aroma. Eur Food Res
Technol. 2007;225:9–19. doi: 10.1007/s00217-006-0375-0. [Cross Ref]
21. Özcan M. Characteristics of fruit and oil of terebinth (Pistacia terebinthus L.)
growing wild in Turkey. J Sci Food Agr. 2004;84:517–552. doi:
10.1002/jsfa.1632. [Cross Ref]
22. Pangborn RM. Influence of water composition, extraction procedures and
holding time and temperature on quality of coffee beverage. Lebensm-Wiss
Technol. 1982;15:161–168.
23. Santos EJD, Oliveira ED. Determination of mineral nutrient and toxic
elements in Brazilian soluble coffee by ICP-AES. J Food Compos Anal.
2001;14:523–531. doi: 10.1006/jfca.2001.1012. [Cross Ref]
24. Sivetz M, Desrosier NW. Coffee technology. Westport: Avi Pub; 1979.
25. Somporn C, Kamtuo A, Theerakulpisu P. Effects of roasting degree on radical
scavenging activity, phenolics and volatile compounds of arabica coffee
beans (Coffea arabica L. cv. Catimor) Int J Food Sci Tech. 2011;46:2287–
2296. doi: 10.1111/j.1365-2621.2011.02748.x. [Cross Ref]
26. Topçu G, Bilici M, Sarıkürkcü C, Öztürk M. A new flavone from antioxidant
extracts of Pistacia terebinthus. Food Chem. 2007;103:816–822. doi:
10.1016/j.foodchem.2006.09.028. [Cross Ref]
27. Voilley A, Sauvageot F, Simatos D, Wojcik G. Influence of some processing
conditions on the quality of coffee brew. J Food Process Pres. 1981;5:135–
143. doi: 10.1111/j.1745-4549.1981.tb00629.x. [Cross Ref]
28. Walheim L, Stebbins RL. Western fruit, berries and nuts. Tucson: HP Books;
1981.
XXXI. Effect of NaCl/Monosodium Glutamate (MSG)
Mixture on the Sensorial Properties and Quality
Characteristics of Model Meat Products
Korean J Food Sci Anim Resour. 2014; 34(5): 576–581.
Ji-Yeon Chun,1 Byong-Soo Kim, Jung-Gyu Lee, Hyung-Yong Cho,2 Sang-Gi Min,1
and Mi-Jung Choi*

Abstract

Sodium chloride is an important ingredient added to most of foods which contributes

to flavor enhancement and food preservation but excess intake of sodium chloride
may also cause various diseases such as heart diseases, osteoporosis and so on.
Therefore, this study was carried out to investigate the effect of monosodium
glutamate (MSG) as a salty flavor enhancer on the quality and sensorial properties
of the NaCl/MSG complex and actual food system. For characterizing the spray-
dried NaCl/MSG complex, surface dimension, morphology, rheology, and saltiness
intensity were estimated by increasing MSG (0-2.0%) levels at a fixed NaCl
concentration (2.0%). MSG levels had no effect of the characteristics of the
NaCl/MSG complex, although the addition of MSG increased the surface dimension
of the NaCl/MSG complex significantly (p<0.05). Furthermore, the effect of MSG on
enhancing the salty flavor was not observed in the solution of the NaCl/MSG
complex. In the case of an actual food system, model meat products (pork patties)
were prepared by replacing NaCl with MSG. MSG enhanced the salty flavor, thereby
increasing overall acceptability of pork patties. Replacement of NaCl with MSG
(<1.0%) did not result in negative sensorial properties of pork patties, although
quality deterioration such as high cooking loss was found. Nevertheless, MSG had
a potential application in meat product formulation as a salty flavor enhancer or a
partial NaCl replacer when meat products were supplemented with binding agents.

Keywords: NaCl, MSG, spray-drying, saltiness, flavor enhancer

Introduction

It has been known that meat products are one of the foods containing high levels of
Na. In spite of the recommendation to reduce Na intake, meat products contain about
1.5-2% NaCl in the formulation owing to its functional properties (Song et al., 2013;
Takachi et al., 2013; WHO, 2012). NaCl plays key roles in meat products including
extraction of salt-soluble proteins, texturization, binding of water molecules, long-
term preservation of the products as well as salty flavor generation (Ahn et al., 2013;
Noort et al., 2010). Therefore, reduction of NaCl in the formulation is inevitably
manifested as a decrease in its qualities.

Processed meat industry has been facing the challenge of reducing Na in the product
formulation. Various strategies have been introduced to reduce Na in the meat
product formulations. Theoretically two strategies can be applied, i.e., the
substitution of NaCl with other ingredients (replacer) and the enhancement of salty
flavor of NaCl (enhancer). The former strategy is being attempted by substituting the
Na-salt with other salts including K, Ca and Mg-salts. However, these salts also
induce flavor or taste modification of the final meat products (Glabert et al., 2003;
Gou et al., 1996). The latter strategy is to add a salty taste enhancer such as
hydrolyzed amino acids or yeast extracts (Canto et al., 2014; Corral et al., 2013;
Horita et al., 2011).

Monosodium glutamate (MSG, C5H8NO4Na) is the most abundant naturally

occurring non-essential amino acid and the sodium salt of glutamic acid. As a
condiment, glutamate components produce a savory taste (umami taste) of foods
(Kurihara, 2008). Actually, MSG has been recognized as a food additive with
negative effects to consumers. In recent years, physiological and toxicological
evaluations were performed by the Food and Drug Administration (FDA) and World
Health Organization (WHO), and MSG is now classified as ‘Generally Recognized
As Safe’ (GRAS) (WHO, 2012). MSG affects the food taste by harmonizing and
providing a balance of tastes. It was identified that the savory taste enhanced overall
food flavor or created a synergistic effect among flavors (Ikeda, 2002; Manabe, 2008;
Yamaguchi and Takahashi, 1984).

There are five glutamate salts such as sodium glutamate, potassium glutamate,
ammonium glutamate, calcium glutamate, and magnesium glutamate which elicited
umami in addition to a certain metallic taste due to the other minerals. Among these
salts, MSG was the most soluble and palatable, and crystallized easily. The
palatability of MSG depends on the presence of NaCl. Manabe (2008) reported that
MSG enhanced the salty flavor although the NaCl concentration was reduced. It is
likely that the addition of MSG reduces the total amount of NaCl in the meat product
formulation (Kurihara, 2008; Yeomans et al., 2008). Therefore, to elucidate the effect
of MSG as a salty flavor enhancer, this study investigated sensorial properties and
quality characteristics of the NaCl/MSG mixture and model meat products after
partially replacing NaCl with MSG.

Materials and Methods

Materials

NaCl (Taepyungsalt Inc., Korea), maltodextrin (MD, 98%, dextrose equivalent: 15-
20, Weifang Codi Imp. & Exp. Co., Ltd., China), and MSG (DAE JUNG Co., Korea)
were purchased from a local market as food grade products. Pork legs were
purchased randomly from three carcasses at 48 h post-mortem. The meat and
backfat were separately ground using a 3 mm plate and vacuum packed in a
polyethylene pouch. Prior to patty preparation, meat and fat were stored at 4℃
(within 3 h).

Preparation of NaCl/MSG complexes

For the NaCl/MSG complex preparation, 20% (w/v) of NaCl and 10% (w/v) MD as a
carrier of NaCl matrix were completely dissolved in distilled water by using a
magnetic stirrer at a speed of 500 rpm for 30 min. For MSG treatment, 0.5-2.0%
(w/w) of MSG was added to the NaCl/MD matrix. Each mixture was dried using a
spray-dryer (SD-1000, Eyela, Japan). The selected spray-drying conditions were
inlet temperature of 150℃, atomization pressure of 180 kPa, blow rate of 0.6 m3/min,
and flow rate of 500 mL/h, which were optimized in a previous study.

Characterization of NaCl/MSG complexes

The surface dimension of the dried NaCl/MSG complex was assessed by a

microscope (Olympus CX31, Olympus Corp., Japan) and particle size was
measured using the UTHSCSA image tool (USA). More than 100 particles were
captured in one image for calculating the average particle size. For the
morphological observation, the NaCl and MSG complex was assessed by a
scanning electron microscope (FE-SEM, S-4700, Hitachi, Japan). All of the samples
were kept in a desiccator for 24 h and then coated for 40 s with platinum using an
ion sputter (E-1010, Hitachi, Japan) at a current of 15 mA.

For the determination of rheology, 1% (w/v) spray-dried NaCl/MSG complex was

dissolved in water and shear force of the solution was measured by linearly
increasing the shear rate from 1/s to 100/s for 500 s using a rheometer (MCR-302,
Anton-Paar, USA) fitted with a concentric cylinder for measuring the geometry
303316 (DG26.7/T200/AL). The measurement was conducted at a constant
temperature of 25℃.

To determine the saltiness of the complex, the complex solution (1%, w/v) was
tempered at an ambient temperature for 2 h. Saltiness intensity was measured using
a 5-point hedonic scale. The trained panel test was conducted by 8 panelists who
were already trained by testing 12 different concentrations (25 mM to 80 mM) of
NaCl solution.

Characteristics of model meat products prepared using MSG

Part of NaCl was substituted with MSG and the mixed powder was directly added in
the pork patty formulation. Pork patties were prepared by mixing 80 g of meat, 20 g
of fat, and 1.5 g of curing ingredient (NaCl or NaCl/MSG mixture) using a homomixer
(5K5SS, Kitchen Aid, USA). As a curing ingredient, powdered NaCl and MSG were
mixed in NaCl to MSG ratios of 3:0, 2:1, 1:2 and 0:3 (w/w), respectively. The mixture
was shaped manually in the form of a cylinder (2 cm thickness) and placed in a
plastic bag. All of the patties were cooked in a 95℃ water bath for 30 min and cooled
down prior to sensory testing.

For the sensory evaluation, pork patties were cut into 2 cm cubes and tempered at
an ambient temperature for 15 min in a plastic bag. The sample cubes were served
to 10 panelists trained to test saltiness, juiciness, tenderness, and overall
acceptability, and the sensorial intensity of each sample was rated by using the
ranking method (Kramer, 1974). The results were analyzed by the rank sum method
and differences among means were compared using the Kramer’s statistical chart.
In this study, the range of the rank sum test result from 17 to 38 indicated that there
was no significant difference among samples (p<0.05) (Kim and Lee, 2001; Kramer,
1974).

Cooking loss and compression test of pork patties were carried out as quality
indicators of the treatments. The cooking loss in pork patties was determined by
measuring the value of exudation after thermal treatment. Each pork patty was
weighed before and after cooking at 95℃ for 30 min, and cooking loss was
expressed as a percentage of the initial weight. After measuring the cooking loss,
double cycled compression test (texture profile analysis) was conducted using a
texture analyzer (CT3 Texture Analyzer, BROOKFIELD, USA) equipped with a probe
(TA43 sphere 25.4 mm D, BROOKFIELD, USA) under the conditions of target value
of 10 mm, trigger load of 20 g, test speed of 0.50 mm/s, target type of 10 mm
distance, and 2 cycles.

Statistical analysis

Completely randomized block design was adopted to estimate the effects of MSG
levels on the characteristics of the NaCl/MSG complex or on the quality
characteristics of pork patties. Analysis of variance was conducted and the means
were separated by Duncan’s test using the software SPSS 20.0 (SPSS Institute,
USA) when the main effect (MSG concentration) was significant (p<0.05).

Results and Discussion

Characteristics of the NaCl/MSG complex

Surface dimensions of the NaCl/MSG complex are presented in Fig. 1. Addition of

MSG increased the surface dimension of the NaCl/MSG complex. Mean surface
dimension of NaCl without MSG was 25 mm2 at 60% cumulative distribution.
Addition of MSG increased the surface dimension of the complex significantly
(p<0.05). The dimension of the NaCl/MSG complexes was largest with 0.5% MSG
addition and smallest with 1.0% MSG addition, however, the dimensions of all
NaCl/MSG complexes were not significantly different. As expected, MSG was
absorbed on the surface of NaCl/MD matrix, which made the NaCl/MSG complex
larger than the control (Frascareli et al., 2012; Kha et al., 2010; Rai et al., 1985;
Sivadas et al.,2008). With respect to morphology (Fig. 1), all of the complexes
exhibited a round shape with varying particle size distribution. The particle shape
appeared to be uniform with increasing MSG concentration, while aggregated
particles were observed with 0.5% MSG addition, reflecting a larger surface
dimension of 0.5% MSG as described above. The evidence of aggregation was also
observed after 1.0% MSG addition, although the intensity of aggregate formation
was considerably decreased compared to that after 0.5% MSG addition. Rheological
properties of the NaCl/MSG solution showed a Newtonian fluid behavior (Fig. 2).
Viscosity of the solution tended to increase with MSG addition, in particular, the
solution with 0.5% MSG addition had the highest viscosity among the others.
Nevertheless, the difference was not significant, and the viscosity of the solution
ranged from 2.0-2.2 mPa·s. The results indicated that the addition of MSG in meat
product formulation would not lead to rheological and textural modification.
Fig. 1. Surface area of NaCl complexes with the addition of various
concentrations of MSG and morphological image of NaCl complexes with the
addition of various concentrations of MSG by scanning electron microscope
×3000 magnification (A: 0.5%, B: 1.0%, C: 1.5%, and D: 2.0%).

Fig. 2. Saltiness evaluation of the NaCl/MSG complex with addition of various

concentrations of MSG.

Fig. 3 indicates the saltiness of the NaCl/MSG complex solution. In contrast to the
previous expectation, the NaCl/MSG complex with addition of 0.5% MSG showed
lower saltiness than that in the control (p<0.05). The saltiness of the NaCl/MSG
complex containing more than 1.0% MSG did not differ from that of the control.
Although, MSG was expected to be a salty flavor enhancer, the sensorial properties
of the NaCl/MSG complex were likely to depend on the food systems. Based on the
definition of umami (savory+salty), low level of MSG reduced the salty flavor because
of its savory taste (Ikeda, 2002; Manabe, 2008; Yamaguchi and Takahashi, 1984).
Therefore, it is essential to use MSG in the actual meat products such as pork
patties.
Fig. 3. Viscosity of the solution of NaCl mixed with various concentrations of
MSG.

Characteristics of pork patties prepared using MSG

Replacement of NaCl with MSG resulted in poor water binding properties of pork
patties. As depicted in Fig. 4A, replacing 0.5% NaCl with the corresponding amount
of MSG increased the cooking loss of pork patties from 28% to 39% (p<0.05). With
respect to MSG addition, there was no significant difference in cooking loss for pork
patties prepared using different MSG concentrations. The results were manifested
by less salt-soluble protein extraction caused due to reduction of NaCl in the
formulation (Keever, 2011). MSG did not seem to affect the water-binding properties
of meat products. Meanwhile, textural properties of pork patties were not affected by
MSG levels (Fig. 4B). As found in our preliminary study (data were not shown),
reducing the NaCl level resulted in moisture loss which might account for hard
texture of meat products. Consequently, it was clear that reducing the NaCl level
negatively affected the quality characteristics of meat products. Meanwhile, it may
be possible that the quality deterioration is minimized when the meat products are
finely ground to form emulsion-type products because of the presence of fat. In this
study, the pork patty was adopted as a model meat product, and a fat-free product
was prepared initially. The impact of NaCl reduction was more significant in this kind
of meat products (Keever, 2011).
Fig. 4. Cooking loss (A) or texture analysis (B) of pork patties after addition of
various MSG concentrations.

Sensorial properties of pork patties prepared using varying MSG concentrations are
presented in Fig. 5. Replacing NaCl with MSG up to 1.0% did not affect the saltiness
of pork patties and significantly low saltiness was found in the sample prepared using
1.5% MSG (no NaCl) (p<0.05). Because MSG was not classified as a salty taste
generator but as a flavor enhancer, MSG alone could not provide the salty flavor of
the product. Meanwhile, it was intriguing to note that 1.0% MSG addition caused no
difference in the saltiness intensity of pork patties compared to that in the control.
Unlike that in the NaCl/MSG complex, MSG enhanced the salty flavor in the actual
meat product. Therefore, it was confirmed that MSG has a potential to replace partial
amount of NaCl in the meat product formulation. However, juiciness and tenderness
of pork patties prepared using all MSG concentrations were lower than those of the
control (p<0.05). Juiciness and tenderness were not different among pork patties
prepared using MSG, reflecting that a strategy for improving the processing
characteristics has to be applied with the usage of MSG. Alternatively, the addition
of MSG up to 1.0% with reduction in NaCl level did not affect the overall acceptability
of pork patties. Low acceptability of a pork patty was observed when 1.5% MSG (no
NaCl) was added (p<0.05). Therefore, the results supported the possible application
of MSG as a salty flavor enhancer of meat products. Although, a technique to prevent
the quality loss caused due to reduction in the NaCl level is necessary, MSG (<1.0%)
enhanced the salty flavor thereby reducing the NaCl level without causing
deterioration of sensorial properties of meat products.

Fig. 5. Effect of MSG concentration on sensory properties of pork patties.

Conclusion

This study explored the effects of MSG as a salty flavor enhancer on the quality and
sensorial properties of the NaCl/MSG complex and the actual meat product. It was
found that MSG-induced enhancement of salty flavor was better in actual foods than
in the simple NaCl/MSG mixture. Although, the mechanisms involved in salty flavor
enhancement caused by MSG are still obscure, the present study demonstrated that
the addition of MSG was effective in reducing the NaCl level in the meat product
formulation.
Acknowledgments

This study was supported by Korea Institute of Planning & Evaluation for Technology
in Food, Agriculture Forestry & Fisheries (IPET Project No.2013-A008-013).

References

1. Ahn S., Park S., Kim J. M., Han S. N., Jeong S. B., Kim H. K. Salt content of school
meals and comparison of perception related to sodium intake in elementary, middle,
and high schools. Nutr. Res. Pract. (2013);7:59–65. doi: 10.4162/nrp.2013.7.1.59.
[PMC free article] [PubMed] [Cross Ref]
2. Canto A. C. V. C. S., Lima B. R. C. C., Suman S. P., Lazaro C. A., Monteiro M. L.
G., Conte-Junior C. A., Freitas M. Q., Cruz A. G., Santos E. B., Silva T. J. P.
Physicochemical and sensory attributes of low-sodium restructured caiman steaks
containing microbial transglutaminase and salt replacers. Meat Sci. (2014);96:623–
632. doi: 10.1016/j.meatsci.2013.08.003. [PubMed] [Cross Ref]
3. Corral S., Salvador A., Flores M. Salt reduction in slow fermented sausages
affects the generation of aroma active compounds. Meat Sci. (2013);93:776–785.
doi: 10.1016/j.meatsci.2012.11.040. [PubMed] [Cross Ref]
4. Frascareli E. C., Silva V. M., Tonon R. V., Hubingera M. D. Effect of process
conditions on the microencapsulation of coffee oil by spray drying. Food Bioprod.
Process. (2012);90:413–424. doi: 10.1016/j.fbp.2011.12.002. [Cross Ref]
5. Gelabert J., Gou P., Guerrero L., Arnau J. Effect of sodium chloride replacement
on some characteristics of fermented sausages. Meat Sci. (2003);65:833–839. doi:
10.1016/S0309-1740(02)00288-7. [PubMed] [Cross Ref]
6. Gou P., Guerrero L., Gelabert J., Arnau J. Potassium chloride, potassium lactate
and glycine as sodium substitutes in fermented sausages and in dry-cured pork
loins. Meat Sci. (1996);42:37–48. doi: 10.1016/0309-1740(95)00017-8. [PubMed]
[Cross Ref]
7. Horita C. N., Morgano M. A., Celeghini R. M. S., Pollonio M. A. R. Physico-
chemical and sensory properties of reduced-fat mortadella prepared with blends of
calcium, magnesium and potassium chloride as partial substitutes for sodium
chloride. Meat Sci. (2011);89:426–433. doi: 10.1016/j.meatsci.2011.05.010.
[PubMed] [Cross Ref]
8. Ikeda K. New Seasonings. Chem. Senses. (2002);27:847–849. doi:
10.1093/chemse/27.9.847. [PubMed] [Cross Ref]
9. Keever B. D. Salt concentration and species affects protein extractability and
processed meat characteristic. M.D. thesis. University of Illinois; Urbana
Champaign, United States: (2011).
10. Kha T. C., Nguyen M. H., Roach P. D. Effects of spray drying conditions on the
physicochemical and antioxidant properties of the Gac (Momordica cochinchinensis)
fruit aril powder. J. Food Eng. (2010);98:385–392. doi:
10.1016/j.jfoodeng.2010.01.016. [Cross Ref]
11. Kim G. O., Lee Y. C. Sensory evaluation of food products. HakYeonSa; Seoul,
Korea: (2001). pp. 326–329. (in Korean)
12. Kramer A., Cooper G. K., Papavasiliou D. A. A non-parametric ranking method
for the statistical evaluation of sensory data. Chem. Senses. (1974);1:121–133. doi:
10.1093/chemse/1.1.121. [Cross Ref]
13. Kurihara K. Glutamate: from discovery as a food fla-vor to role as a basic taste
(umami) Am. J. Clin. Nutr. (2009);90:719S–722S. doi: 10.3945/ajcn.2009.27462D.
[PubMed] [Cross Ref]
14. Manabe M. Saltiness enhancement by the characteristic flavor of dried bonito
stock. J. Food Sci. (2008);73:321–325. [PubMed]
15. Noort M. W. J., Bult J. H. F., Stieger M., Hamer R. J. Saltiness enhancement in
bread by inhomogeneous spatial distribution of sodium chloride. J. Cereal Sci.
(2010);52:378–386. doi: 10.1016/j.jcs.2010.06.018. [Cross Ref]
16. Rai G., Lavernia E., Grant N. J. Powder size and distribution in ultrasonic gas
atomization. JOM. (1985);37:22–26.
17. Sivadas N., O’Rourke D., Tobin A., Buckley V., Ramtoola Z., Kelly J. G., Hickey
A. J., Cryan S. A. A comparative study of a range of polymeric microspheres as
potential carriers for the inhalation of proteins. Int. J. Pharm. (2008);358:159–167.
doi: 10.1016/j.ijpharm.2008.03.024. [PubMed] [Cross Ref]
18. Song H. J., Cho Y. G., Lee H. J. Dietary sodium intake and prevalence of
overweight in adults. Metabolism. (2013);62:703–708. doi:
10.1016/j.metabol.2012.11.009. [PubMed] [Cross Ref]
19. Takachi R., Ishihara J., Iwasaki M., Ishii Y., Tsugane S. Self-reported taste
preference can be a proxy for daily sodium intake in middle-aged Japanese adults.
J. Acad. Nutr. Diet. (2013);144:781–787. [PubMed]
20. WHO. Guideline: Sodium intake for adults and children. W.H.O. (WHO) (Ed.);
(2012).
21. Yamaguchi S., Takahashi C. Interactions of monosodium glutamate and sodium
chloride on saltiness and palatability of a clear soup. J. Food Sci. (1984);49:82–85.
doi: 10.1111/j.1365-2621.1984.tb13675.x. [Cross Ref]
22. Yeomans M., Gould N., Mobini S., Prescott J. Acquired flavor acceptance and
intake facilitated by monosodium glutamate in humans. Physiol. Behav.
(2008);93:958–966. doi: 10.1016/j.physbeh.2007.12.009. [PubMed] [Cross Ref]
XXXII. The Impact of Hybridization on the Volatile and
Sensorial Profile of Ocimum basilicum L.
Andréa Santos da Costa, 1 Maria de Fátima Arrigoni-Blank, 1 Maria Aparecida
Azevedo Pereira da Silva, 2 Mércia Freitas Alves, 1 Darlisson de Alexandria Santos,
3 Péricles Barreto Alves, 3 and Arie Fitzgerald Blank 1 ,*

Abstract

The aim of the present study was to investigate the volatile and sensorial profile of
basil (Ocimum basilicum L.) by quantitative descriptive analysis (QDA) of the
essential oil of three hybrids (“Cinnamon” × “Maria Bonita,” “Sweet Dani” ×
“Cinnamon,” and “Sweet Dani” × “Maria Bonita”). Twelve descriptive terms were
developed by a selected panel that also generated the definition of each term and
the reference samples. The data were subjected to ANOVA, Tukey's test, and
principal component analysis. The hybrid “Cinnamon” × “Maria Bonita” exhibited a
stronger global aroma that was less citric than the other samples. Hybridization
favored the generation of novel compounds in the essential oil of the hybrid “Sweet
Dani” × “Maria Bonita,” such as canfora and (E)-caryophyllene; (E)-caryophyllene
also was a novel compound in the hybrid “Sweet Dani” × “Cinnamon”; this compound
was not present in the essential oils of the parents.

1. Introduction

Species of the genus Ocimum are widely used and appreciated due to their essential
oil that includes several components of interest. O. basilicum has the largest market
demand among the species and chemotypes and is cultivated for commercial
purposes in several countries, such as India, France, Morocco, and Italy [1].

Basil (Ocimum basilicum L.) is an herbaceous aromatic plant that can be grown as
an annual or perennial according to the region where the plant is cultivated. The
stalks grow up to 1 m tall and have greatly branched stems with quadrangular or
pubescent branches. The leaves are 1 to 4 cm long and 1.85 to 9.25 cm wide [2].
The leaves are petiolate, opposite, ovate, or toothed and either green or purple. As
a medicinal plant, basil is used to treat respiratory disorders, bacterial infections, and
bowel parasites; basil also improves digestion [3]. The quality of basil plants is
defined by the composition of their essential oil [4].

The essential oil can be extracted from fresh or dry leaves and flowers by three
methods: hydrodistillation, steam distillation, and solvent extraction [5]. Basil is used
in the food, pharmaceutical, and cosmetic industry [6] as well as for the control of
pests [7–9] and plant diseases [10]. A study conducted by [11] using hydrodistillation
and steam distillation found similar concentrations of chemical compounds in the
essential oils, whereas the amount of essential oil extracted was larger when steam
distillation was used.

Basil is classified according to its aroma into categories such as sweet, lemon,
cinnamate or cinnamon, camphor, anise, and clove. Variations in the chemical
composition of the essential oils are used to differentiate the basil as the following
types: European, French, or Sweet; Egyptian, Reunion, or Comoro; Bulgarian, Java,
or Methyl Cinnamate; and Eugenol. The European type of basil contains mostly
linalool and methyl chavicol [12, 13]. The Cinnamon type of basil is rich in methyl
cinnamate [14], the Sweet Dani cultivar in citral (68%) [15], and the Maria Bonita
cultivar in linalool (78.12%) [16]. Linalool has been widely studied as an acaricide
[17], bactericide, and fungicide [18], and this compound has been successfully used
in medicine as a sedative [19, 20] and anticonvulsant [21]. Studies conducted with
the Maria Bonita cultivar indicate antinociceptive activity of the essential oil [22],
whereas potential anti-Giardia activity has already been shown [23].
The chemical composition of the essential oil of O. basilicum was the focus of many
studies, and several chemotypes have been reported. Oils rich in 1.8-cineole (22%),
linalool (49.7%), methyl chavicol (47%), and methyl cinnamate (65.5%) were found
in Brazilian plants [24]. In plants cultivated in the Mongolian desert, the main
components were 1.8-cineole (8.54%), linalool (27.26%), (Z)-α-bergamotene
(10.00%), and methyl chavicol (19.77%) [1]. In 18 basil accessions, linalool and
methyl-chavicol were the main components [9]. In southern India, the main
components were (E)-methyl cinnamate (34.49%), linalool (28.44%), camphor
(13.08%), (Z)-methyl cinnamate (6.90%), and geraniol (3.84%) [25]. Another study
found the main components to be 1,8-cineole (7.23%), linalool (66.4%), and (Z)-α-
bergamotene (7.96%) [26]. In a study conducted with seven species of the genus
Ocimum, linalool (39.8%), estragol (20.5%), methyl cinnamate (12.9%), eugenol
(9.1%), and 1,8-cineole (2.9%) predominated [27]. Among 12 varieties of Ocimum,
ten exhibited a high percentage of methyl cinnamate (35–80%), one single variety
contained caryophyllene, and another variety contained linalool [28].

The production of hybrids by crossing cultivars contributes to the creation of new

essential oils for the world market. The most important decision for the hybridization
program is the choice of parents [29]; this decision depends on the characteristics
to be improved, the type of inheritance of the characteristics, and the available
source of germplasm [30]. Genetic variability, adaptability, and evolution of the
species are important for the success of any improvement program. An improvement
program was shown by [2] in which 55 accessions of Ocimum sp. were investigated.
These authors observed genotypic variation in the content and yield of essential oil
and were able to identify promising genotypes for improvement programs. Recently,
[31] showed that basil variability might be improved with hybrid combinations.
Studies of such combinations and their effects facilitate the selection of genotypes
for genetic improvement programs and the development of cultivars.

The aim of the present study was to assess the impact of hybridization on the volatile
and sensorial profile of Ocimum basilicum L.

2. Materials and Methods

2.1. Plant Material and Extraction of Essential Oil

Hybridization was performed in a greenhouse covered with white screen at the

Federal University of Sergipe. The cultivars used to perform the crosses were
“Cinnamon” (O. basilicum), “Sweet Dani” (Ocimum × citriodorum), and “Maria
Bonita” (O. basilicum) [31]. To realize hybridization, inflorescences of each plant
were selected to serve as pollen recipients (female) and marked with wool yarn;
different color was used for the male parent and pollen donor. Every morning (07:00–
09:00 am), the crosses were performed using collected inflorescences. The picked
flowers (containing pollen) were touched against the stigmas of emasculated
flowers. After hand pollination, the inflorescences functioning as female organs were
protected with paper bags to prevent the flowers from self-pollinating.
The leaves of three hybrids (“Cinnamon” × “Maria Bonita,” “Sweet Dani” ×
“Cinnamon,” “Sweet Dani” × “Maria Bonita”) and three cultivars (“Cinnamon,” “Sweet
Dani,” and “Maria Bonita”) were used in this work.

The assay was conducted in field of the Research Farm “Campus Rural da UFS”
using a randomized block design in triplicate. Each plot consisted of a row of five
plants. The spacing between rows was 0.50 m and between plants was 0.50 m. The
mean minimum and maximum temperature were 24 and 30°C, respectively.

After three months of field cultivation, when the plants were in full bloom, the aerial
parts were harvested and the leaves of the plants were dried in an oven dryer with
air renewal and circulation (Marconi model MA-037/5) at 40°C for five days [4]. The
extraction of the essential oils from the leaves was performed using a Clevenger-
type apparatus [5] for approximately 160 minutes [4].

2.2. Identification of the Volatile Compounds in the Essential Oils

The analysis of the essential oil chemical composition was performed in a gas
chromatograph coupled to a mass spectrometer (GC-MS) (Shimadzu, model QP
5050A) equipped with an AOC-20i auto injector (Shimadzu) and a fused-silica
capillary column (5%-phenyl-95%-dimethylpolysiloxane, 30 m × 0.25 mm id.,
0.25 μm film, J&W Scientific). Helium was used as the carrier gas at a flow rate of
1.2 mL/min. The temperature program was as follows: 50°C for 1.5 min, temperature
increase at 4°C/min until reaching 200°C, temperature increase at 15°C/min until
reaching 250°C and 250°C for 5 min. The injector temperature was 250°C, and the
detector (or interface) temperature was 280°C. The injection volume of ethyl acetate
was 0.5 μL, the partition rate of the injected volume was 1 : 87, and the column
pressure was 64.20 kPa. The mass spectrometer conditions were as follows: ionic
capture detector impact energy of 70 eV, scanning speed 0.85 scan/s from 40 to
550 Da.

Quantitative analysis of the chemical constituents was performed by flame ionization

gas chromatography (FID), using a Shimadzu GC-17A (Shimadzu Corporation,
Kyoto, Japan) instrument, under the following operational conditions: capillary ZB-
5MS column (5% phenyl-arylene-95%-dimethylpolysiloxane) fused silica capillary
column (30 m × 0.25 mm i.d. × 0.25 μm film thickness) from Phenomenex (Torrance,
CA, USA), under same conditions as reported for the GC-MS. Quantification of each
constituent was estimated by area normalization (%). Compound concentrations
were calculated from the GC peak areas and they were arranged in order of GC
elution.

The essential oil components were identified by comparing their mass spectra with
the available spectra in the equipment database (NIST05, NIST21, and WILEY8).
These libraries allowed for the comparison of spectral data and had a similarity index
of 80%. Additionally, the measured retention indices were compared with those in
the literature [32]. The relative retention indices (RRI) were determined using the
[33] equation and a homologous series of n-alkanes (C8–C18) injected under the
chromatography conditions described above.

For each volatile compound, the chromatographic peak area count was calculated,
and the value was used as an estimate of the concentration of the given compound
in the sample. The data were statistically analyzed by means of principal component
analysis (PCA) using the software SAS (Statistical Analysis System—SAS Institute
Inc., North Carolina, USA, 2010).

2.3. Sensorial Profile

The aroma profile of each essential oil sample was established using quantitative
descriptive analysis (QDA), as described by [34].

First, 10 volunteers were recruited among students and collaborators at the Federal
University of Sergipe and selected according to their sensitivity to the basic aromas,
olfactory memory, and discriminating power among aromas. Next, using the network
method [35] and other procedures described by [34, 36], we developed (i) descriptive
vocabulary for the essential oil samples, (ii) a consensual definition of each
descriptor, (iii) references for the panel training, and (iv) a form for the descriptive
assessment of the aromas perceived in the essential oils (Table 1). The descriptive
form, defined terms, and references are depicted in Tables Tables11 and and2.2.
These tools were used to train the panel in sensory assessment of essential oils.

Table 1. List of the definitions of the descriptive terms and their corresponding
references.

Attribute Definition References

Basil Natural aroma of fresh basil Sliced fresh basil leaves

Aroma characteristic of lemon, Fresh grated lemon and orange peel

orange peel, fresh and fresh sliced leaves of
Citric
Cymbopogon  densiflorus, or Cymbopogon  densiflorus and
Cymbopogon  winterianus Cymbopogon  winterianus

Aroma associated with caramel,

Sweet Honey, sugar
honey, or sugar

Table 2. Chemical composition of the Ocimum basilicum L. genotypes (mean

± standard deviation).
 % peak area

“Sweet
Pea RRI
RRI Compound Dani” × “Sweet Dani” “Cinnamon “Maria
k lit “Cinnamon “Swee
“Maria דCinnamon ” × “Maria Bonita
” t Dani”
Bonita ” Bonita” ”
”

103 102 3.93 ± 0.00 ± 5.35 ±

1 1,8-Cineole 4.49 ± 0.57 5.73 ± 0.54 4.38 ± 0.77
1 6 0.40 0.00 0.01

108 108 trans-Linalool 0.00 ± 0.00 ± 0.00 ±

2 0.30 ± 0.09 0.14 ± 0.05 0.31 ± 0.10
6 4 oxide 0.00 0.00 0.00

110 109 55.63 ± 0.00 ± 75.22 ±

3 Linalool 15.75 ± 1.32 35.33 ± 3.24 30.78 ± 0.35
0 5 2.61 0.00 1.05

114 114 0.26 ± 0.00 ± 0.00 ±

4 Camphor 0.50 ± 0.02 0.47 ± 0.05 0.55 ± 0.06
6 1 0.15 0.00 0.00

119 119 Methyl 0.00 ± 0.47 ± 0.00 ±

5 3.01 ± 0.31 1.30 ± 0.03 1.03 ± 0.13
7 5 chavicol 0.00 0.41 0.00

122 122 0.39 ± 3.13 ± 0.00 ±

6 Nerol 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00
2 7 0.05 1.68 0.00

123 123 14.57 ± 35.68 ± 0.00 ±

7 Neral 0.57 ± 0.12 0.00 ± 0.00 0.00 ± 0.00
6 5 1.29 1.30 0.00

124 124 0.82 ± 0.89 ± 14.66 ±

8 Geraniol 0.00 ± 0.00 0.30 ± 0.09 0.00 ± 0.00
9 9 0.08 0.23 0.94

126 126 19.35 ± 46.16 ± 0.00 ±

9 Geranial 0.84 ± 0.14 0.00 ± 0.00 0.00 ± 0.00
6 4 1.57 2.51 0.00

130 129 (Z)-Methyl 0.00 ± 0.00 ± 0.00 ±

10 6.41 ± 0.49 7.44 ± 0.55 5.75 ± 0.98
2 9 cinnamate 0.00 0.00 0.00

138 137 (E)-Methyl 0.00 ± 0.00 ± 0.00 ±

11 60.26 ± 0.96 43.72 ± 3.36 48.64 ± 1.72
3 6 cinnamate 0.00 0.00 0.00

(E)-
141 141 0.49 ± 1.65 ± 0.00 ±
12 Caryophyllen 0.72 ± 0.08 0.00 ± 0.00 0.00 ± 0.00
9 7 0.04 0.29 0.00
e

143 143 α-trans- 1.15 ± 0.79 ± 1.80 ±

13 0.27 ± 0.04 1.67 ± 0.10 0.00 ± 0.00
2 2 Bergamotene 0.06 0.08 0.27

148 147 0.39 ± 0.51 ± 0.32 ±

14 ϒ-Muurolene 0.92 ± 0.10 0.31 ± 0.03 1.15 ± 0.18
0 8 0.03 0.24 0.02

148 148 0.17 ± 1.49 ± 0.00 ±

15 β-Selinene 0.53 ± 0.01 0.00 ± 0.00 0.32 ± 0.12
8 9 0.01 0.06 0.00
 % peak area

“Sweet
Pea RRI
RRI Compound Dani” × “Sweet Dani” “Cinnamon “Maria
k lit “Cinnamon “Swee
“Maria דCinnamon ” × “Maria Bonita
” t Dani”
Bonita ” Bonita” ”
”

151 151 0.29 ± 0.00 ± 0.49 ±

16 ϒ-Cadinene 0.28 ± 0.02 0.84 ± 0.02 1.17 ± 0.09
1 3 0.01 0.00 0.01

164 163 0.41 ± 0.00 ± 0.58 ±

17 α-epi-Cadinol 0.73 ± 0.30 0.99 ± 0.22 2.90 ± 0.68
1 8 0.10 0.00 0.15

Total (%) 97.84 95.57 97.94 96.98 90.77 98.43

RRI: relative retention index.

After the training period was over, the trained panel was subjected to a selection
test. Each examiner was requested to assess three samples using the
abovementioned descriptive form. The experiment was performed in a fully
randomized block design in triplicate. According to the methods suggested by [37],
the data generated by each examiner were analyzed by ANOVA (variation sources:
samples and repetition) and Tukey's test. To compose the final panel, the examiners
who exhibited good discriminating power (PFsample ≤ 0.05), good repeatability
(PFrepetition ≥ 0.05), and consensus with the remainder of the panel were selected.

Next, the selected panel assessed all of the essential oil samples in triplicate using
the descriptive assessment form (Table 1). The samples were identified by a three-
digit code, and the panel members were requested to assess the intensity of each
descriptor listed in the descriptive form on a nine-cm scale with “none” on the left
end, “strong” on the right end, and “moderate” on the midpoint (Table 1). To avoid
sensory fatigue, only three essential oil samples were assessed at each session.
The order of the presentation of the samples was balanced among the panel
members according to the recommendations of [34, 36].

The data generated by the panel were analyzed by ANOVA (variation sources:
sampler, sample and interaction of sampler versus sample) and Tukey's test using
SAS software (Statistical Analysis System—SAS Institute Inc., North Carolina, USA,
2010).

3. Results and Discussion

The volatile compounds identified in each essential oil sample, their corresponding
linear retention indices, and the relative area expressed as percentage of the total
area under the chromatogram can be observed in Table 2. All of the volatile
compounds that exhibited a relative area that is equal to or higher than 0.25% in
weight in at least one sample also were included in Table 2. The sum of all of the
compounds was over 90% (Table 2), showing that they represent the main
components of each sample.

Linalool was the main volatile compound of the essential oils extracted from the
leaves of the basil genotype “Maria Bonita” and the hybrid “Sweet Dani” × “Maria
Bonita” (Table 2). The essential oil profile of the genotype “Maria Bonita” exhibits a
rather low number of components, and linalool corresponds to approximately 70%
of the total area under the chromatogram (Table 2).

Table 3 shows the description of the olfactory quality of each volatile component
listed in Table 2 and the chromatographic peak area count, which allows for the
estimation of the concentration of each volatile compound in each sample and the
calculation of significant differences among the essential oil samples in their
chemical components. The investigated genotypes exhibited significant differences
(P ≤ 0.05) in the concentration of each volatile compound in the investigated samples
(Table 3). These findings suggest that the hybridization of these genotypes induced
the generation of compounds in the hybrids that were not present in the parents and
an increase or decrease in some compounds present in the parents. The similarities
and differences among the essential oils in the volatile quantitative profile are
depicted in Table 1.

Table 3. Chemical composition and aroma of Ocimum basilicum L. genotypes.

Peak area count × 104

“Swee
t Dani” “Sweet
Compound Aroma* × “Cinnamo “Swee “Maria
Dani” × “Cinnamo
n” × “Maria t Bonita
“Maria “Cinnamo n”
Bonita” Dani” ”
Bonita n”
”

Mint,
1,8-Cineole 30.43a 26.65a 25.11a 14.99a 0.00a 32.94a
sweet

trans- Sweet,
Linalool floral, 0.00c 1.71a 0.67bc 1.08ab 0.00c 0.00c
oxide green

Flower,
431.99 461.74
Linalool lavende a 90.93b 146.03b 101.58b 0.00b a
r
 Peak area count × 104

“Swee
t Dani” “Sweet
Compound Aroma* × “Cinnamo “Swee “Maria
Dani” × “Cinnamo
n” × “Maria t Bonita
“Maria “Cinnamo n”
Bonita” Dani” ”
Bonita n”
”

Camph
Camphor 1.82a 1.82a 2.01a 1.83a 0.00a 0.00a
or

Fragran
t, sweet,
Methyl
cooling, 0.00b 17.82a 5.49b 3.45b 2.24b 0.00b
chavicol
fresh,
minty

Nerol Sweet 3.05b 0.00b 0.00b 0.00b 15.32a 0.00b

115.16 171.12
Neral Lemon b 3.21c 0.00c 0.00c a 0.00c

Rose,
Geraniol geraniu 6.37b 0.00b 1.81b 0.00b 4.29b 90.96a
m

Lemon, 152.93 221.31

Geranial b 4.76c 0.00c 0.00c a 0.00c
mint

(Z)-Methyl
— 0.00b 38.03a 31.74a 19.70ab 0.00b 0.00b
cinnamate

(E)-Methyl
— 0.00b 353.88a 171.26ab 159.71ab 0.00b 0.00b
cinnamate

(E)-
Caryophylle — 3.84a 4.12a 0.00b 0.00b 0.00b 0.00b
ne

α-trans- Fragran
Bergamoten t, sweet, 8.89a 1.55a 7.16a 0.00a 3.77a 37.69a
e fresh
 Peak area count × 104

“Swee
t Dani” “Sweet
Compound Aroma* × “Cinnamo “Swee “Maria
Dani” × “Cinnamo
n” × “Maria t Bonita
“Maria “Cinnamo n”
Bonita” Dani” ”
Bonita n”
”

Herb,
ϒ-
wood, 3.09a 5.31a 1.31a 3.72a 2.41a 7.87a
Muurolene
spice

β-Selinene Herb 1.29c 3.08b 0.00d 0.97c 7.11a 0.00d

ϒ-Cadinene Wood 2.31ab 1.62ab 3.54a 3.82a 0.00b 3.03ab

α-epi- Fragran
3.38b 4.00ab 4.55ab 9.12a 0.00b 3.70b
Cadinol t

Means followed by the same lowercase letters in a row do not differ according to
Tukey's test at P ≤ 0.05.

*Aroma according to [38].

Figure 1 depicts axes I and II generated by principal component analysis (PCA) of

the count values associated with the chromatographic peak areas of each volatile
compound. The first two axes explained 56.46% of the variation found among the
samples in their respective concentration of volatile compounds. The samples are
represented by triangles, where each vertex represents one analytical repetition, and
the assessed parameters (volatile compounds) are represented by vectors (Figure
1). To interpret the graphic generated by PCA (Figure 1), the vectors must first be
decomposed along axes I and II, and the “weight” of their components on each axis
must be analyzed. The volatile compounds 1,8-cineole, gamma-cadinene, alpha-
epi-cadinol, (Z)-methyl cinnamate, (E)-methyl cinnamate, camphor, gamma
muurolene, and linalool trans-oxide decomposed mostly on axis I (Figure 1). These
data suggest that the samples located on the positive side of axis I were unique in
composition compared to the other samples. In this case, the essential oil of the
hybrid “Cinnamon” × “Maria Bonita” had a greater concentration of volatile
compounds than “Sweet Dani” × “Maria Bonita”. The hybrid “Cinnamon” × “Maria
Bonita” exhibited a greater concentration of all of the compounds mentioned above
except for (E)-methyl cinnamate and gamma-muurolene (Table 3).
Figure 1. Projection of the sensory descriptors and essential oil samples on the first
two principal components.

The parents did not exhibit the abovementioned compounds in concentrations as

high as those found in the oil of hybrid “Cinnamon” × “Maria Bonita”, which suggests
the occurrence of heterosis in the hybrid compared to the parents. Therefore,
hybridization induced in the hybrid “Cinnamon” × “Maria Bonita” favored the
generation of compounds such as linalool trans-oxide, (E)-cinnamate and (Z)-methyl
cinnamate that were not present in the essential oils of the parents.

In the hybrids “Sweet Dani” × “Cinnamon” and “Sweet Dani” × “Maria Bonita,” we
noted that linalool trans-oxide was the single volatile compound present in the
hybrids that was not present in the parents (Figure 1 and Table 3).

The abovementioned instances of heterosis suggest several hypotheses. First, the

presence of codominance is suggested. For instance, the female parent exhibits
alleles A1A1, and the male parent exhibits alleles A2A2 of the same gene, whereas
alleles A1 and A2 by themselves are unable to induce the formation of a given
compound. When hybridization occurs, both alleles together might code for the
enzyme associated with the synthesis of a novel compound [38]. The structure of
some enzymes requires two or more polypeptide chains; thus, allele A 1 might code
for one polypeptide chain, which is not sufficient by itself to form the enzyme, and
allele A2 might code for another polypeptide chain. Conversely, the presence of both
alleles allows for the synthesis of two chains and the consequent formation of the
enzyme. A second hypothesis is related to the presence of genetic epistasis or
interaction, whereby complementary genetic action or a limiting reaction might occur
[39]. In addition, two different enzymes might be needed for the production of the
novel compound, whereas each enzyme is coded for by a different gene.

The essential oil extracted from the hybrid “Sweet Dani” and the hybrid “Sweet Dani”
× “Maria Bonita” had higher concentrations of nerol, neral, and geranial (Figure 1),
which is confirmed by Table 3. Indeed, previous studies [15] showed that geranial
was the main compound in the hybrid “Sweet Dani.” Consequently, the volatile profile
of the hybrid “Sweet Dani” predominated in its hybridization with the cultivar “Maria
Bonita.”

The cultivar “Maria Bonita” had higher geraniol and linalool concentrations (Figure 1
and Table 3). The cultivar “Maria Bonita” is distant from the other samples and
remarkably from its hybrids “Sweet Dani” × “Maria Bonita” and “Cinnamon” × “Maria
Bonita”, which suggests that its volatile profile differs significantly from the profile of
other samples (Figure 1). This finding indicates that upon hybridization, the cultivar
“Maria Bonita” exerted less influence than the cultivars “Sweet Dani” and “Cinnamon”
on the volatile profile of the hybrids.

The cultivar “Cinnamon” and the hybrid “Sweet Dani” × “Cinnamon” are located at
the center of Figure 1, which indicates that these samples not only exhibit similar
volatile profiles but also intermediate concentrations of all of the analyzed
compounds (Table 3).

Upon describing the essential oils extracted from the hybrids, the trained panel
identified aromatic notes such as basil, lemon, orange, lemongrass, lavender, lemon
balm, rosemary, cinnamon stick, wood, sweet, and pungent. Most of the descriptors
are consistent with the aroma of the volatile compounds identified in the samples
(Table 3).

We noted that the sample “Cinnamon” × “Maria Bonita” exhibited a global aroma
intensity that is significantly stronger (P ≤ 0.05) than that of the other samples
because the sample most likely exhibited a greater concentration of most of the
volatile compounds present in the hybrids (Table 4).

Table 4. Average intensity of the outstanding aromatic notes in the hybrids of

Ocimum basilicum.
 Aromatic notes1,2
Samples
Basil Sweet Citric Global aroma intensity

“Sweet Dani” × “Maria Bonita” 5.75a 4.44a 4.27ab 1.35c

“Sweet Dani” × “Cinnamon” 5.35a 4.83a 4.92a 3.86b

“Cinnamon” × “Maria Bonita” 5.68a 4.80a 2.97b 7.74a

1Means followed by the same letters in a column do not differ at P ≤ 0.05.

20: weak; 9: strong.

All of the samples exhibited moderate notes of basil and citric aromas and did not
differ in these attributes (Table 4). However, the samples exhibited significant
differences (P ≤ 0.05) in the intensity of the citric aroma, which was less intense in
the hybrid “Cinnamon” × “Maria Bonita.” One reason for this lessened intensity is
that the hybrid exhibited a high concentration of several noncitric aromatic volatile
compounds, such as 1,8-cineole, which the literature describes as having an aroma
of peppermint (http://www.flavornet.org/flavornet.html), gamma cadinene (wood)
(http://www.flavornet.org/flavornet.html), and camphor, among others. The aroma of
these volatile compounds most likely masked the citric aroma of the linalool present
in the hybrid.

The low intensity of the citric aroma exhibited by the hybrid “Cinnamon” × “Maria
Bonita” might also be explained by the lack of neral and geranial, which have a citric
aroma, in its composition (Table 4).

4. Conclusions

Hybridization favored the generation of novel compounds in the essential oil of the
hybrid “Sweet Dani” × “Maria Bonita,” such as canfora and (E)-caryophyllene; (E)-
caryophyllene also was a novel compound in the hybrid “Sweet Dani” × “Cinnamon”;
these compounds were not present in the essential oils of the parents. With
hybridization, also new aromas are produced, which were identified by sensory
analysis. The hybrid “Cinnamon” × “Maria Bonita” exhibited a stronger global aroma
intensity that was less citric than the other samples.

Acknowledgments

The authors thank FAPITEC/SE, CNPq, CAPES, and RENORBIO for their financial
support of this work.

Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of
this paper.

References

1. Shatar S, Altantsetseg S, Sarnai I, Zoljargal D, Thang TD, Dung NX. Chemical

composition of the essential oil of Ocimum basilicum cultivated in Mongolian Desert-
Gobi. Chemistry of Natural Compounds. 2007;43(6):726–727.
2. Blank AF, Carvalho Filho JLS, Santos Neto AL, et al. Caracterização morfológica
e agronômica de acessos de manjericão e alfavaca. Horticultura Brasileira.
2004;22:113–116.
3. Lorenzi H, Matos FJA. Plantas Medicinais do Brasil: Nativas e Exóticas
Cultivadas. Nova Odessa, Brazil: Instituto Plantarum; 2003.
4. Carvalho Filho JLS, Alves PB, Ehlert PAD, et al. Influence of the harvesting time,
temperature and drying period on basil (Ocimum basilicum L.) essential oil. Revista
Brasileira Farmacognosia. 2006;16:24–30.
5. Guenther E. The Essential Oils: Individual Essential Oils of the Plant Families
Rutaceae and Labiatae. Vol. 3. Malabar, Australia: Krieger; 1972.
6. Loughrin JH, Kasperbauer MJ. Aroma content of fresh basil (Ocimum basilicum
L.) leaves is affected by light reflected from colored mulches. Journal of Agricultural
and Food Chemistry. 2003;51(8):2272–2276. [PubMed]
7. Chang CL, Cho ILK, Li QX. Insecticidal activity of basil oil, trans-anethole,
estragole, and linalool to adult fruit flies of Ceratitis capitata, Bactrocera dorsalis, and
Bactrocera cucurbitae. Journal of Economic Entomology. 2009;102(1):203–209.
[PubMed]
8. Ehlert PAD, Blank AF, Arrigoni-Blank MF, Paula JWA, Campos DA, Alviano CS.
Hydrodistillation time for essential oil extraction of seven medicinal plant species.
Revista Brasileira de Plantas Medicinais. 2006;8(2):79–80.
9. Pascual-Villalobos MJ, Ballesta-Acosta MC. Chemical variation in an Ocimum
basilicum germplasm collection and activity of the essential oils on Callosobruchus
maculatus. Biochemical Systematics and Ecology. 2003;31(7):673–679.
10. Zhang J-W, Li S-K, Wu W-J. The main chemical composition and in vitro
antifungal activity of the essential oils of Ocimum basilicum Linn. var. pilosum (Willd.)
Benth. Molecules. 2009;14(1):273–278. [PubMed]
11. Charles DJ, Simon JE. Comparison of extraction methods for the rapid
determination of essential oil content and composition of basil (Ocimum spp.)
Journal of the American Society for Horticultural Science. 1990;115:458–462.
12. Hertwig IF. Plantas Aromáticas e Medicinais: Plantio, Colheita, Secagem e
Comercialização. São Paulo, Brazil: Icone; 1986.
13. Simon JE, Quinn J, Murray RG. Basil: a source of essential oils. In: Janick J,
Simon JE, editors. Advances in New Crops. Portland, Ore, USA: Timber Press;
1990. pp. 484–489.
14. Schulz H, Schrader B, Quilitzsch R, Pfeffer S, Krüger H. Rapid classification of
basil chemotypes by various vibrational spectroscopy methods. Journal of
Agricultural and Food Chemistry. 2003;51(9):2475–2481. [PubMed]
15. Morales MR, Simon JE. “Sweet Dani” a new culinary and ornamental lemon
basil. Hortscience. 1997;32(1):148–149.
16. Blank AF, De Souza EM, Arrigoni-Blank MDF, De Paula JWA, Alves PB. Maria
Bonita: a linalool type basil cultivar. Pesquisa Agropecuaria Brasileira.
2007;42(12):1811–1813.
17. Prates HT, Leite RC, Craveiro AA, Oliveira AB. Identification of some chemical
components of the essential oil from molasses grass (Melinis minutiflora Beauv.)
and their activity against cattle-tick (Boophilus microplus) Journal of the Brazilian
Chemical Society. 1998;9(2):193–197.
18. Belaiche T, Tantaoui-Elaraki A, Ibrahimy A. Application of a two levels factorial
design to the study of the antimicrobial activity of three terpenes. Sciences Aliments.
1995;15:571–578.
19. Elisabetsky E, Marschner J, Souza DO. Effects of Linalool on glutamatergic
system in the rat cerebral cortex. Neurochemical Research. 1995;20(4):461–465.
[PubMed]
20. Sugawara Y, Hara C, Tamura K, et al. Sedative effect on humans of inhalation
of essential oil of linalool: sensory evaluation and physiological measurements using
optically active linalools. Analytica Chimica Acta. 1998;365(1–3):293–299.
21. Elisabetsky E, Brum LFS, Souza DO. Anticonvulsant properties of linalool in
glutamate-related seizure models. Phytomedicine. 1999;6(2):107–113. [PubMed]
22. Venâncio AM, Onofre ASC, Lira AF, et al. Chemical composition, acute toxicity,
and antinociceptive activity of the essential oil of a plant breeding cultivar of Basil
(Ocimum basilicum L.) Planta Medica. 2011;77(8):825–829. [PubMed]
23. De Almeida I, Alviano DS, Vieira DP, et al. Antigiardial activity of Ocimum
basilicum essential oil. Parasitology Research. 2007;101(2):443–452. [PubMed]
24. Vieira RF, Simon JE. Chemical characterization of basil (Ocimum spp.) found in
the markets and used in traditional medicine in Brazil. Economic Botany.
2000;54(2):207–216.
25. Jirovetz L, Buchbauer G, Shafi MP, Kaniampady MM. Chemotaxonomical
analysis of the essential oil aroma compounds of four different Ocimum species from
southern India. European Food Research and Technology. 2003;217(2):120–124.
26. Carović-Stanko K, Orlić S, Politeo O, et al. Composition and antibacterial
activities of essential oils of seven Ocimum taxa. Food Chemistry. 2010;119(1):196–
201.
27. Lee S-J, Umano K, Shibamoto T, Lee K-G. Identification of volatile components
in basil (Ocimum basilicum L.) and thyme leaves (Thymus vulgaris L.) and their
antioxidant properties. Food Chemistry. 2005;91(1):131–137.
28. Viña A, Murillo E. Essential oil composition from twelve varieties of Basil
(Ocimum spp) grown in Colombia. Journal of the Brazilian Chemical Society.
2003;14(5):744–749.
29. Ramalho MP, Santos JB, Zimmermann MJO. Genética Quantitativa em Plantas
Autógamas: Aplicações ao Melhoramento do Feijoeiro. Goiânia, Brazil: UFG; 1993.
30. Fehr WR. Principles of Cultivar Development: Theories and Techniques. New
York, NY, USA: Macmillan; 1987.
31. Blank AF, Rosa YRS, Carvalho Filho JLSD, et al. A diallel study of yield
components and essential oil constituents in basil (Ocimum basilicum L.) Industrial
Crops and Products. 2012;38(1):93–98.
32. Adams RP. Identification of Essential Oil Components by Gas
Chromatograpy/Mass Spectroscopy. Carol Stream, Ill, USA: Allured Publishing
Corporation; 2007.
33. Vandendool H, Kratz PD. A generalization of retention index system including
linear temperature programmed gas-liquid partition chromatography. Journal of
Chromatography. 1963;11:463–471. [PubMed]
34. Stone H, Sidel JL. Sensory Evaluation Practices. 3rd edition. London, UK:
Academic Press; 2004.
35. Moskowitz HR. Product Testing and Sensory Evaluation of Foods: Marketing
and R&D Approaches. Westport, Mo, USA: Food and Nutrition Press; 1983.
36. Meilgaard M, Civille GV, Carr BT. Sensory Evaluation Techniques. Boca Raton,
Fla, USA: CRC Press; 2007. Descriptive analysis techniques; pp. 173–186.
37. Damasio MH, Costell E. Análisis sensorial descriptivo: generación de
descriptores y selección de catadores. Revista de Agroquimica y Technologia de
Alimentos. 1991;31:65–78.
38. Sheen LY, Ou YHT, Tsai SJ. Flavor characteristic compounds found in the
essential oil of Ocimum basilicum L. with sensory evaluation and statistical analysis.
Journal of Agricultural and Food Chemistry. 1991;39:939–943.
39. Borém A, Miranda GV. Melhoramento de Plantas. Viçosa, Brazil: UFV; 2009.

XXXIII. Sensorial pedagogies, hungry fat cells and the

limits of nutritional health education
Biosocieties. 2015 Jun; 10(2): 125–142.
Emilia Sanabriaa

Abstract

This article examines the way the category of ‘the sensorial' is mobilised across
obesity research and care practices for overweight persons in France. The ‘natural'
body is understood to have developed mechanisms that motivate eaters to seek out
energy-dense foods, a hardwiring that is maladaptive in today's plethoric food
environment. The article analyses the feedback models mobilised in scientific
literature on the neuroendocrine processes regulating appetite. The analysis of how
‘the sensorial' is studied and used to treat patients provides a vantage point onto the
ways foods and bodies transform each other. Recent findings show that fat cells
influence metabolism by secreting hormones, revealing that eaters are affected by
the materiality of the foods they ingest. ‘The sensorial' functions as a regulator in the
feedback mechanisms where social norms regulating foodscapes become enfolded
in the molecular processes that control appetite regulation. The article traces the
work that the category of ‘the sensorial' does as it flows through the loops and
feedbacks between scientific evidence, policy and care. It examines the way
pleasure and the sensations of eaters are increasingly foregrounded in French
nutritional health promotion strategies in a context where informing eaters is
increasingly deemed ineffective.

Keywords: sensorial, obesity, feedback, endocrinology, health education, France

Introduction

Valérie, 48, has gone on and off diets for over 30 years and has what she describes
as a terrible relationship to her body. She recently calculated her body mass index
(BMI) using an online calculator (www.imc.fr). At 79 kg for 1.77 m she has a BMI of
28 and is classified as overweight. When she entered her measurements, a flashing
red dialogue box appeared on her screen, warning her, “ATTENTION: you are
OVERWEIGHT. Consult your doctor or a nutritionist immediately to determine the
causes. You must lose weight”. Valérie has consulted innumerable doctors about
her weight and tried most of the mainstream weight loss diets. Whatever weight she
has lost, she has always gained back, sometimes in excess. As summer
approaches, she makes an appointment at her local hospital's endocrinology service
and a few tests later meets with an endocrinologist who enters her into an ‘eating
rehab' programme. “You have a slow metabolism”, he tells her. “You're unlucky, you
gain weight just by looking at food. It's not fair, but it's your reality. Your ancestors
were famine survivors and have passed on these genes to you. It's not your fault.
But you'll need to change the way you eat if you want to lose weight durably”.

This article is based on ethnographic work with nutritionists who believe that long-
term effective weight loss is only possible if you eat ‘to feed pleasure'. This group of
practitioners does not forbid any kind of food – not even greasy chips or chocolate
mousse – and even teaches patients the art of food dégustation. Patients who have
endured decades of restrictive diets are invited to taste and savour with their dietician
the very foods that have been proscribed and banished but which they feel they
irremediably succumb to. The GROS (literally fat/big), or Research Group on
Overweight and Obesity, is an association of health professionals comprising
nutritionists, dieticians, psychiatrists and psychologists working with overweight and
obese patients in reference centres or private practices across France. The network
counts nearly 200 specialists across France trained in the GROS method and
organises a yearly conference and training programme (which I have partially
undertaken). The GROS method is founded on a programme of therapeutic
education that is centred on the eater's bodily sensations. Pleasure is a leitmotif. 1
The GROS approach takes as its starting point the idea that food intake is regulated
by complex sensorial mechanisms that signal to the eater when to initiate or end a
meal. GROS practitioners consider that all dietary recommendations, including those
promoted by the French National Program for Nutrition and Heath (PNNS),
encourage eaters to “eat with their head rather than with their sensations,” as GROS
co-founder Gerard Apfeldorfer summarised in an interview. The GROS' explicit goal
is to reunite eaters with the pleasure and the sensations of eating. Rather than
‘cognitive messages' or dietary advice, the GROS considers that ‘sensations' should
guide eating. The GROS website's “regulating alimentary behaviour” information
pages asks, “Why tell the French population how they should eat, when every human
being is naturally equipped with a system of sensorial regulation that enables him or
her to eat adequately?”

As in most European nations, vast public health initiatives were launched in France
to tackle non-communicable, chronic disease associated with high BMIs. While the
causes, mechanisms and means of calculating the rising rates of obesity are
contested and uncertain, emphasis is recurrently placed on the shift between energy
intake and expenditure in societies where energy-dense foods have become widely
available and aggressively advertised. Given this context of uncertainty concerning
how to qualify and circumscribe the ‘problem', let alone define its aetiology,
developing effective public health responses presents considerable challenges. This
article traces some of the shifts in the epistemological understanding of appetite
regulation and examines how these are mobilised to different care or policy ends. It
aims to contrast two positions and map an emergent transformation in
understandings of obesity and its prevention. The first centres on informing eaters
about nutrients and on proscribing certain calorie-rich foods. The second, more
emergent, is the position developed by practitioners such as those trained in the
GROS method and builds on scientific understandings of the role of sensorial
processes in appetite regulation. It constructs its rationale on an explicit rejection of
dieting practices, noting that diets ultimately fail to bring about weight reduction and
lead to overall weight gain or disordered eating in the long term. 2

I examine the uncertain materialities these sensorial pedagogies bring to life and
trace the boundaries where foods and eaters, and matter and subjects, merge and
transform one another. By considering how the appetite control systems of eaters
are conceptualised and presented as undermined by a contemporary environment
of irresistible highly palatable foods, my aim is to examine how the promotion of
conscious and intimate sensory engagements with foods is understood by
practitioners such as those working with the GROS method to provide an alternative
response. In so doing, I examine the ways in which scientific evidence concerning
appetite regulation, taste, pleasure and sensoriality is mobilised to explain the
purportedly epidemic rise in obesity. What emerges is the idea of a misfit or
maladaptation between the plethoric environment in which ‘we' live today and the
body ‘we' are said to have inherited from a distant Palaeolithic past whose regulation
system is geared towards defending its fat stores.

The analysis of how sensorial mechanisms are studied (in scientific practice) and
used to treat patients or to promote health (in clinical practice or health education)
provides an interesting vantage point into the ways the materialities of foods and
bodies mutually transform each other. The sensorial systems that regulate food
intake are triggered, at one end, by materials (foods), information about which is
gathered through olfaction and sight and through taste and somesthesia, 3 once the
food has entered the eater's mouth (although olfaction continues and somesthesia
may begin earlier).4 Bodies and foods become so profoundly enfolded into each
other that it is impossible to determine where one begins and the other ends. The
cascade of metabolic processes that takes place in bodies once food has triggered
taste and olfaction, and the complex hormonal feedback mechanisms that follow,
leads across a threshold of sorts. As endocrine processes spark off neurological
responses, the tangibility and materiality of the process becomes more uncertain.
The relation between the body's sensorial apparatus, the environment, and the
relative importance of cognitive and automatic processes in food intake regulation
forms an important area of research in ‘obesity science'. The findings developed in
this area feed a host of popular science books and media coverage on topics such
as food addiction, willpower, dieting and overeating. They are in turn picked up by
actors in the field of food health promotion such as those I have been working with.
The questions that are being formulated by clinical and biomedical research around
the topic of the sensorial concern the biological mechanisms that are driving people
worldwide to eat more, and more of certain kinds of foods.

My aim is to attend to the way the category of the sensorial is used in the practices
put in place by the GROS (and similar groups outside France). I start by
contextualising their approach in a public health landscape dominated by health
information and education. I then analyse two key areas of scientific research that
the GROS draws on to found its rationale: the analysis of the active endocrine role
of fat tissue in the regulation of appetite, and debates surrounding the feedback
models that regulate energy in the body. Here, I draw on the distinction between
first- and second-order cybernetics to examine the way in which homoeostatic
regulation is repeatedly set against hedonism in discussions of obesity. Are the
hedonic processes that are said to be disrupting homoeostatic regulation of appetite
merely a dis-regulation in a closed loop system or a different kind of system
altogether? How might these conceptual differences map onto the debates taking
place around regulation in obesity science? The complex models of homoeostasis
and analyses of the mechanisms of its contemporary dis-regulation that circulate in
this field open up important questions for public intervention and the individualisation
of responsibility for health, and invite us to reconsider the ways in which nutritional
recommendations cast specific relationships between eaters and the foods they
consume.

Beyond Informing: Forging the Domain of the Sensorial

Mol (2012, p. 379) notes that “the overriding message of most dieting advice is that
a person who wants to lose weight needs to overrule the desires of her craving body”.
The fact that people are taught to relate to food in a calculative way suggests, she
argues, a particular enactment of a mind-body distinction according to which without
information, bodies would overeat, indulging in pleasure. The “ontonorms” at play in
the Dutch dieting advice she analyses construct the natural homoeostatic capacities
of bodies as an ideal and “hedonistic enjoyment” as a danger (Mol, 2012, p. 383).
Information on calories, nutritional content of foods and the importance of a balanced
diet is presumed in the dietary advice provided through public health nutrition
programmes to enable “the rational mind to take control of the pleasure seeking
body,” Mol notes. The GROS explicitly posit that information and rational (or
“cognitive,” in their terms) control not only fail to bring about durable weight loss but
actually impair eaters' natural bodily capacities to regulate weight. Like the
practitioners on the margins of Dutch dieting that Mol describes, the GROS – who
have developed a more structured and centralised approach than their Dutch
colleagues, encourage people to taste, savour and enjoy their food.

Much of the debate in the obesity arena concerns the relative importance of
willpower and biological factors in the development of adiposity and its control. This
debate goes back to the early part of the twentieth century and to the shifting
influence that endocrinologists and psychiatrists had in defining the aetiology and
outlining treatment for obesity. Rasmussen (2012) traces the shift in the United
States from an endocrinologically oriented approach that centred on glandular
disorders to a psychiatrically oriented one that focused on the neural determinants
of excess appetite. Obesity was framed as an addiction and constructed as a
disorder of willpower (Rasmussen, 2012, p. 890). Interestingly, a new current in
obesity research is attempting to re-qualify overeating as a form of addiction.
However, this can be seen as an attempt to map public policy responses to BigFood
on those that have been carried out against BigTobacco (see in particular Brownell
and Gold, 2012). So while food addiction is gaining renewed attention, the rationale
behind this is to shift the blame from individuals to industry-related activities that
deliberately harm health through what the Lancet Non-Communicable Disease
Action Group refer to as “corporate disease vectors” (Moodie et al, 2013). It is
beyond the scope of the current article to examine why educating eaters and
conveying information about nutritional and calorific content of foods emerged as a
key domain of intervention in nutrition. Suffice it to say that nutritional health
education is increasingly deemed to be at best a weak strategy in the face of the
strong biological drive inscribed in our hardwiring to seek out calorie-dense foods
and at worst a harmful or counter-productive one, shifting blame onto individuals
(James, 2007; Herrick, 2009; Guthman, 2011). Part of the problem lies in the fact
that the epidemiological patterns of obesity are contested (Gard, 2011), as are the
causes and mechanisms that lead to the widely denounced gain in girth. In the
absence of effective, population-wide interventions (such as pharmaceutical magic
bullets or surgical solutions), information-education-communication models of
behaviour change have thus been favoured.
The PNNS, French National Program for Nutrition and Heath (Ministère de la santé,
2011, p. 10), adopts a broad definition of ‘nutrition' that includes physical activity and
recognition that nutrition includes “nutrients, foods, social determinants, cultural,
economic, sensorial and cognitive” aspects of food behaviours. Nevertheless, the
PNNS identifies information, communication and education to shape food
behaviours and physical activity as its main ‘strategic lever'. The idea that obesity
would be solved if people ate balanced meals and were less sedentary remains a
tacitly shared one – largely reinforced by the tenor of health promotion messages.
One implication of such a view is that if people are informed and fail to change their
behaviour they are seen to be individually responsible for any ill health that may
ensue. The relationship between access to information and ‘positive' health
behaviours has been problematized in a number of ways in public health, notably
through greater attention to the social determinants of health (CSDH, 2008) or to the
complex interplay of environmental, urban and metabolic factors in obesity. The
Foresight Report on obesity put forward the notion of “passive obesity” to suggest
that default choices posed by the social and built environment mean that combatting
weight gain requires active strategies.

People in the UK today don't have less willpower and are not more gluttonous than
previous generations. Nor is their biology significantly different to that of their
forefathers. Society, however, has radically altered over the past five decades, with
major changes in work patterns, transport, food production and food sales. These
changes have exposed an underlying biological tendency, possessed by many
people, to both put on weight and retain it.
(Butland et al, 2007, p. 5)

At a recent conference on the social determinants of health held in France, one

economist even suggested that health education had deepened health inequalities
as more educated segments of the population had benefited disproportionally,
deepening the gap in health outcomes across the social economic gradient. The
relationship between providing information through public health campaigns and
changed behaviour has also been problematized by behavioural economists who
are paying increasing attention to the ‘non-rational' processes that shape health
decision making. The concept of nudging (Thaler and Sunstein, 2008), itself closely
tied to social marketing, draws on a distinction between reflective and impulsive
decision-making processes. Nudging opposes rational, reflective or cognitive
processes to ‘automatic, affective' reactions and ‘feelings' that are triggered by
environmental cues and driven by the lure of pleasure. In the context of a generalised
moral panic surrounding rising rates of obesity, disparate approaches such as those
appropriated by some eating disorder specialists adopting cognitive behavioural or
mindfulness therapies (for example, Kristeller and Wolever, 2011) or organisations
such as SlowFood that have taken up taste education (Barzanò and Fossi, 2010)
have coalesced around the questioning of nutritional interventions that address
overeating through abstract, disembodied information.5 The rational, autonomous
individual who has been the target of public health initiatives is attacked, so to speak,
on two fronts by the emergence of a critique of health information and
communication (Gastaldo, 1997; Evans et al, 2008; Whithead and Irvine, 2011; Ayo,
2012; Fitzpatrick and Tinning, 2014). On the first, she is seen to be fighting against
an obesogenic environment and structural forces that determine she will have
disproportionally worse health outcomes if she is female and of low social economic
status (in France, and where obesity is concerned). On the second, she is seen as
driven by unconscious, affective and automatic – largely biologised – forces that
determine her behaviour and actions against her will or rational intentions (Thaler
and Sunstein, 2008; Rice, 2013).

The GROS, for its part, draws a fairly radical distinction between what they refer to
as “cognitive” or informational approaches and those, such as theirs, that centre on
les sensations alimentaires (alimentary sensations). The ‘sensorial' emerges as a
central category in their practices. While there is an important literature on the
anthropology of the senses (Serres, 1985; Classen, 1993; Seremetakis, 1994;
Korsemeyer, 2002; Howes, 2003, 2005) and on the importance of attending to the
senses in the study of health and illness or eating (Hinton et al, 2008; Nichter, 2008;
Sutton, 2010), less has been written on how the sensorial may itself be rendered a
therapeutic tool. The GROS practices I examine participate in what Seremetakis
(1994, p. 123)6 has referred to as a modern concern with “sensory loss”. These
practices aim to “re-sensitise” eaters who have become “divorced” from their
sensations. They aim to bring heightened attention to the senses and the differences
in sensory input (differentiating textures, olfactory, gustative or visual stimuli) and the
importance of sensorial blending (sensations that are produced by the interaction of
multiple sensory inputs). Nichter (2008) notes that meaning and experience of
sensation are shifting and deeply biosocial (p. 186). He draws links between the
increasing medicalisation and pharmaceuticalization of bodily sensations and the
decreasing tolerance to discomfort and sensation. Sensations are so numerous in
our daily lives that only some of them are cognitively processed and given meaning.
They are typical hybrids and as such provide a zone through which biologies are
encultured. As certain sensations become more salient through being stimulated or
made culturally meaningful, certain neural pathways are reinforced, thus increasing
the “inducibility” of the sensation, a process that Hinton et al (2008) refer to as the
kindling of sensation.

Kindling aptly describes the industrial processing of “hyperpalatable foods”, or foods

that GROS practitioners refer to as “highly sensorial.” Palatability is the fine-tuning
of foods to enthral the senses, stimulate appetites, drive wanting, override satiety
and motivate eaters to pursue more of that taste. In The End of Overeating: Taking
Control of Our Insatiable Appetite, ex-FDA commissioner, paediatrician and Dean of
Yale medical school Kessler (2009) argues that the multisensory dimensions of
hyperpalatable foods reinforce reward (p. 49) and are drivers in overeating (p. 98).
“Mixture is where the magic happens,” a food industry consultant told Kessler (2009,
p. 100). Industry, Kessler tells us, is shifting towards greater complexity, combining
optimal amounts of sugar and fat to attain a “bliss point” and developing a versatile
range of “inclusion products” to add crunch, blasts of flavour, and dynamic contrasts
in flavour or colour, and to enhance the sensory properties of foods that drive
desires. Processing makes chewing increasingly unnecessary, facilitating
swallowing and reducing the time food spends in the mouth, thereby undermining
satiety mechanisms that arise when food is savoured and not just gobbled down. In
what follows I attend to the way the deeply encultured category of ‘the sensorial' is
increasingly being mobilised across a range of care and prevention practices in the
field of nutrition. I present some of the scientific evidence that is drawn upon in this
context with specific reference to the practices deployed by the GROS, and the
relations that they draw between sensorial overriding and overeating.

Returning to Alimentary Sensations

Cognitive restriction is a central concept for GROS practitioners. Initially developed

in 1975, the term is reworked by Polivy and Herman (1991) to refer to eating that
ceases to be controlled by internal factors and alimentary sensations and that
becomes controlled and planned according to cognitive factors such as dietary
prescriptions. Many of the obese patients that GROS practitioners treat in their
private practices have been through years of diets that GROS practitioners feel have
led them to lose touch with their bodily sensations. The aim of treatment is to shift
from a mode of eating that is cognitively determined (that is, by external injunctions
and beliefs) to one that is “intuitive and sensorial, governed by an attention to and a
respect of alimentary sensations and emotions” (Zermati and Apfeldorfer, 2010, p.
154, my translation).

During a GROS training session I attended in Spring 2012 along with two dozen
psychiatrists and dieticians from across France, GROS co-founder Zermati spoke of
a diet as any form of “deliberate control of food intake on the basis of external,
cognitive factors that take precedence over internal signals such as those that
ensure our energetic homeostasis”. He then provided a detailed explanation of why,
on the basis of the scientific evidence we have on adipose tissue hyperplasia and
negative feedback, people cannot lose weight efficiently or durably, let alone happily,
when they use cognitive restraint “against” their bodies. Apfeldorfer, Zermati's
colleague and co-founder of the GROS, explained the GROS method to me in the
following terms:

Our neurophysiological needs impose themselves upon us through a series of

sensorial mechanisms. Breathing is vital and independent of our control. It is an
automatically regulated processes. Eating like sleeping is a semi-automatically
regulated behaviour. If we are tired we can resist sleep, but the more we stray from
our sensations, the stronger the backlash. Because food was not always available
in our past, we have developed mechanisms to tell us when to seek out and end
meals as well as to stock calories. Sometimes we have to make do with the sensation
of hunger, and at others to eat without hunger. Pleasure was once the regulator, but
because our pleasure centres are overstimulated by hypersensorial foods, we have
lost track of our internal messaging system. The GROS defends an approach based
on the attention to and the respect of bodily sensations through work on the events
that impede us from respecting these sensations and which leads to addiction-type
processes.
The GROS method is put to work in a range of therapeutic settings, from private
dietetic practices to clinics specialising in the treatment of obese patients. Founded
in 1997, the GROS counts nearly 200 practicing members across France who have
received certification. The typical GROS patient is female, has been on restrictive
diets all her life and consults in a private practice through the liberal system of ‘town'
(as opposed to hospital) medicine. However, a growing number of specialised
services across public hospitals and private clinics are training their staff in the
GROS method and integrating aspects of the sensorial approach into the care
practices they offer their patients. The method itself centres on a series of practical
exercises and techniques that include experiments with taste. Many practitioners
begin with a dégustation of a ‘problem food'. Patients are invited to bring in a food
that they cannot resist and that they readily overeat. This invitation to explore and
consume a taboo food with a dietician is also a means of building trust with the
patient and is used by GROS practitioners to demarcate their approach from all the
‘restrictive' injunctions patients have often endured previously. Drawing on the model
of a wine tasting, patients are invited to consciously examine the food item they have
chosen, its shape, texture or smell, before beginning the dégustation. Each mouthful
is taken ‘in full consciousness' of the cascade of sensations that unfold as the
dietician explains the molecular and physiological processes that produce a taste
and through which the organoleptic properties of the food being consumed are
relayed into a mental image and a sensation of pleasure. As the dégustation
proceeds, the patient is invited to note changes in texture and sensation. Is the
pleasure experienced in the same way? Does it peak or drop? Does the food item
taste the same after the nth bite? Such an experience is expected to bring
awareness to the natural internal signalling mechanisms that the ‘disregulated'
eaters ‘we' have become fail to attend to. The objective is to find the point of
rassasiement7 at which the food no longer has the same taste and at which both the
wanting and liking of it decrease. Although the awareness that is sought through the
body is set against that derived from rational, mind-driven control, it is interesting to
note the way terms such as ‘mental images' or ‘full consciousness' creep back in.

The GROS training introduces practitioners to an explicitly cybernetic understanding

of appetite regulation. As Zermati explained during a training session, when the
homoeostatic system is internally unbalanced, a biological value (such as a drop in
blood sugar) is transformed into a specific sensorial message. When this message
is transformed into a sensation, it activates a behaviour such as the initiation of a
meal or a desire for a specific food. When the need has been satisfied, a sensorial
message in the form of the feeling of satiety is returned to end the meal. Negative
feedback characterises a balanced homeostatically regulated system, where the
need diminishes with the behaviour. In a ‘dis-regulated' system, the need (hunger)
augments with the activity: there is positive feedback. Satiety cannot be experienced
if eating takes place too fast, or while focusing attention on another activity
simultaneously. When foods are enjoyed and not just gulped down, chemically
induced gustative modifications occur, triggering sensory-specific satiety. Zermati
and Apfeldorfer (2010) ask how one can understand one's hunger and sensations if
one is plugged into the outside world, eating unconsciously and without truly
stopping to feel one's internal sensations.
GROS-led experiences also include work on the emotional aspects of hunger.
During their training, dieticians and health professionals seeking GROS certification
are themselves invited to experiment with the sensations and emotions raised by
hunger through exercises that include skipping meals and recording the physical and
emotional sensations this gives rise to. Ensuing discussions during training provide
a platform from which to introduce the scientific literature on which the GROS founds
its rationale. Hunger is graduated and the emotions it gives rise to (such as anxiety,
fear, anger, sadness) identified. The experiment aims to rid the sensation of hunger
of any anxiogenic emotions it may carry and return it to its primary physiological
function of indicating a need to feed. Patients may also be invited to keep a carnet
des sensations (sensations notebook) where all foods consumed in a week are
noted to bring attention to the temporal and emotional contexts of eating. This
teaches patients to differentiate eating that is triggered by hunger from eating that is
triggered by emotional states. It leads them to identify the emotional states that lead
to a desire to eat. As Zermati and Apfeldorfer note (2010, p. 138), contrary to
common understandings, it is not hunger that makes us eat but the motivation to
seek out food that results from hunger. This may also result from other factors such
as emotional states. One of the fundamental issues in GROS therapeutic practice is
learning to distinguish between desire to eat that is triggered by hunger and one that
is triggered by an emotion. Blame and guilt that result from a failure to follow dietary
recommendations are seen by GROS practitioners – and by others approaching
eating in a similar vein – as having an embodied effect and increasing the feelings
of hunger, decreasing the feeling of satiety and satisfaction, and putting the body in
‘starvation' mode, hence increasing the likelihood of stocking the calories ingested.
The sensorial thus becomes an interface between processes that are classified on
either side of body/mind or biological/social distinctions. Satiety, in this model, and
the stabilising of the energy balance are dependent on guilt-free pleasure.

Zermati and Apfeldorfer (2010, p. 137) note that obesity is due to two consecutive
dis-regulations. The first is a dis-regulation of the system of homoeostatic regulation
that leads eaters to consume more than their calorific needs, due to the fact that they
are no longer sensitive to the natural sensorial mechanisms that signal satiety. The
second is linked to what they refer to as the “set point”, which is a function of adipose
tissue signalling and which is determined by the quantity of adipocytes in the body.
As these multiply they raise the set point, making weight loss beyond a certain
threshold practically impossible. For these reasons, dieting cannot lead to durable
weight loss because the augmentation in adipose tissue is sometimes not reversible,
and mental control over alimentary behaviour cannot be maintained in the long run.
In the remainder of the article, I thus draw on two key issues that arise from the
scientific debate that the scientific committee of GROS draws on: the shift from
homoeostasis to hedonism and the role of adipose tissue in the regulation of
appetite.

Go to:

The Agency of Fat: Adiposity and Negative Feedback in Plethoric

Environments
This section focuses on some of the scientific work being done in the area of food
intake regulation and examines the disquieting effects foods are presented as having
on eaters in some of this literature. This raises specific questions for the issues I
opened with concerning preventative or educational strategies. If foods exert their
influence on eaters in ways that bypass or circumvent eaters' wills, what is the effect
of educating eaters about food groups, or the importance of eating five fruit and
vegetables? The regulation of human appetite is controlled by complex feedback
mechanisms involving a range of hormones that connect the different sensory
organs such as the tongue, taste buds and gut to nodes in the central nervous
system such as the hypothalamus. These elaborate pathways are still being
explored as new hormones and receptors are discovered and their functioning
explained. In 1994, the hormone leptin was discovered and shown to have a central
role in appetite regulation. Leptin is produced in the adipose tissue and plays a key
role in the regulation of hunger. Two key mechanisms are at play in appetite control
and food intake regulation: a homoeostatic mechanism and a hedonic one. Much of
the current work being carried out in this area points to the increased role of the latter
in the contemporary food-rich environment to which we are increasingly exposed.
The argument is that the marketing by BigFood of energy-dense hyperpalatable
foods stimulates the hedonic system, driving wanting and exciting pleasure, thus
overriding homoeostatic control mechanisms (Kessler, 2009; Gearhardt et al, 2011;
Brownell and Gold, 2012).

But first let us briefly consider how the homoeostatic system is said to function.
Energy homoeostasis works to maintain the stability of body fat stores by regulating
both food intake (the motivation to initiate and end meals) and energy expenditure
(the rate at which calories are burnt). In their conclusion to a review on central
nervous system food intake control published in Nature, Morton et al, 2006, note that
there is no consensus on the fundamental aspects of obesity parthogenesis. The
review examines evidence concerning the molecular and behavioural mechanisms
that link modulations in body fat to the regulation of food intake through mechanisms
such as satiation and motivation to initiate meals. It shows that central nervous
system-controlled energy homoeostasis responds to changes in body fat stores. As
these stores diminish (as weight is lost) “the motivation to find food and the size of
individual meals tend to increase until energy stores are replenished” (Morton et al,
2006, p. 289). Adipose negativity feedback signalling posits that signals inform the
brain of changes in body fat stores, leading to adaptive adjustments that aim to
stabilise fat stores by acting on the regulation of satiety perception and brain reward
circuitry. The authors thus make a claim that goes against commonly held
assumptions about weight gain, proposing that “obesity involves the defence of an
elevated body weight, rather than the absence of regulation”, and that the “global
obesity pandemic” (sic.) is the product of “deleterious interactions between obesity-
promoting environmental factors and homeostatic control systems”. This to say that
the homoeostatic regulatory mechanism is driven to defend against fat loss rather
than to prevent weight gain. This is often explained by reference to our
palaeontological past, in which periods of feast and famine were common and where
the capacity to maintain fat stores would have been an evolutionary advantage. In
this sense, the ‘natural' body ‘we' are said to have inherited from this collective past
has mechanisms that motivate eaters to seek out energy-dense foods, a hardwiring
that is maladaptive in today's plethoric environment. The adaptations that are
imagined to have favoured the survival of our ancestors, such as mechanisms to
maintain fat stores in prevision of periods of famine, are understood to be our modern
plagues, leading to the increase of cardiovascular disease and type-2 diabetes.

Kessler (2009) argues that the homoeostatic regulation system described above is
often overridden by another mechanism with different brain circuitry. “This is known
as the reward system. And […] in the fight between energy balance and reward, the
reward system is winning”. The reward system reinforces motivation to seek out
certain desirable foods. It is based on a feedback system that teaches eaters to
associate specific calorie intake with particular tastes and foods. This is an acquired
process, built up over time and one that is reinforced by the increasing presence of
hyperpalatable foods in global foodscapes. In a sense it can be argued that the turn
to the sensorial dimensions of appetite regulation in obesity prevention circles arises
subsequent to the close attention given by the food industry to tailored modifications
in the sensorial qualities of processed foods. Guyenet and Schwartz (2012) define
reward as “the process whereby certain behaviours are reinforced in response to
specific environmental stimuli.” Palatability, they argue, has been consistently shown
to influence meal size in humans. As Guyenet (2012) notes in a blog post, alluding
once more to an evolutionary explanation:

Our brains are highly attuned to these qualities because they're all elements of
nutritious, calorie-dense foods that would have sustained our ancestors in a natural
environment, but today, the exaggerated combinations of these qualities used by
processed food manufacturers […] overstimulate our natural reward pathways.
Commercial foods are professionally designed to maximize reward, because reward
is precisely what keeps you coming back for more.

These developments, and the shifts in language that are required to communicate
them, are interesting in that they give adipocytes substantial agency. These cells are
said to actively defend the fat stores in bodies through a range of metabolic,
neurologic and endocrine processes. They are, as it were, hungry, and in turn make
eaters hungry. Eaters' capacities to make rational choices are subverted by these
hungry fat cells. They decide for them, or interfere with their decision-making
capabilities. What does it imply to say that it is the fat that is hungry, rather than the
eater?

In its training programme the GROS gives considerable attention to recent findings
on the mechanisms of adipose tissue growth. They regularly refer to the fact that
there are two mechanisms of adipose tissue development: adipocyte hypertrophy
and adipocyte hyperplasia. In the first case, fat cells increase in size as they stock
excess energy. This mechanism of weight gain is easier to fight as weight is lost
through the ‘emptying out' of adipocytes. However, recent attention has turned to a
second process of fat accumulation that consists in the multiplication of fat cells (Jo
et al, 2009). There is debate concerning whether or not adipocyte hyperplasia occurs
after the age of 20 (Tchoukalova et al, 2010). What is significant for the present
purposes is that the adipose tissue has come to be portrayed in such literature as
having an active endocrine function. It is now conceived of as “an organ in and of
itself, like the pancreas or liver,” as one nutritionist put it to me during the 2012 GROS
congress. Attention is being given to the role of fat cell progenitors from different
parts of the body in generating adipose tissue with different inflammatory or ‘dis-
regulatory' impacts. While hypertrophied adipocytes can be expunged, their
multiplication at a young age implies that the future adult will have a much greater
susceptibility to weight gain due to the fact that the set point defended by negative
feedback is higher.

The relation between fat and appetite as it is described in scientific investigations

into the endocrine functions of adipose tissues and the active role fat tissue plays in
the regulation of appetites is a classic case of what Leslie Aiello, President of the
Wenner-Gren Foundation for Anthropological Research, has termed a
“paleofantasy”. Paleofantasies are stories about human behaviour that are
constructed upon evolutionary narratives based on limited fossil evidence. In an
essay citing Aiello, evolutionary biologist Zuk (2009) notes that “The notion that there
was a time of perfect adaptation, from which we've now deviated, is a caricature of
the way evolution works.” The idea of a mismatch between the body we have
inherited and the environment we now live in is so recurrent it appears self-evident.
The recursive nature of the argument gives weight to the idea that the complex
molecular pathways that regulate our appetite have developed in response to a poor,
famine-ridden environment. It is as if the environment we evolved in during the
Pleistone were now irremediably enfolded in our molecular make-up. This “primordial
gluttony”, as Kluger (2007) puts it in his review The Science of Appetite, arises from
a prehistoric time when we were “programmed” to overeat. In this imagined race for
survival, those who did not lose their appetite immediately put on more weight and
survived the next famine, passing on the genes.8 Yet, along with many public heath
nutritionists, I would like to suggest that this mismatch may not be so accidental, as
it were. The shape urban environments are taking worldwide, interspersed as they
increasingly are with a dense network of outlets for highly processed and highly
palatable foods available at all hours of the day and night at prices defying
competition, are the product of specific forces that have tailored their marketing
strategies to certain features of ‘our' inherited hardwiring (Hawkes and Buse, 2011;
Monteiro et al, 2011; Kleiman et al, 2012; Stuckler and Nestle, 2012; Stuckler et al,
2012).

Hedonic and homoeostatic mechanisms are deeply intermingled. As excessively

rewarding foods are consumed, body fat stores increase, and with them, the level of
body fat that is ‘defended' by homoeostatic feedback mechanisms. This raises
important questions concerning the burden of responsibility that is placed upon
eaters, particularly overweight or obese eaters. What are the implications of
recognising that adipose tissue stimulates appetites? Can this move us beyond the
responsibility placed on individuals to care for their own health despite living in
environments widely recognised to be deleterious to health? Or does this on the
contrary further render overweight people targets of specific risk reduction action?
As far as the GROS approach is concerned, the objective is explicitly to
déculpabiliser (de-blame) overweight people and to demystify commonly held ideas
concerning the fact that they eat more (which may not be true) by demonstrating that
their bodies are defending a larger body fat store (which does not necessarily imply
higher food intake). The objective of training practitioners on topics such as adipose
tissue negative feedback or adipose hyperplasia is not so much to biologise
behaviour as to reveal that the standard treatment – cognitive dietary restraint – is
inefficient and profoundly damaging for many patients. If we consider, as the GROS
suggests, that the body naturally strives to get back to its set point, as defined in part
by levels of adiposity, and that this set point cannot be modified (without surgical
intervention, but even then there is controversy), then dietary restraint is a very
limited response.

Feedback Models and Complex Systems in Changing Environments

Set point, negative feedback and homoeostasis are all terms that have been
borrowed from cybernetics and the science of complex systems regulation. As we
have seen, there is an extensive scientific literature that examines the way hunger
and satiety are regulated. Such processes are described as being regulated by
complex neuroendocrine processes that involve a number of brain centres (some of
which are even being located in the digestive system), sensory receptors throughout
the body and a list of hormones that continues to expand. Body weight is modulated
by interrelated systems that function on different temporalities, a shorter one
modulating blood sugar levels and a slower one modulating adiposity. Feedback
mechanisms maintain stability in changing environments, and are thus seen as
playing a vital evolutionary role as we have seen here. However, as we have seen
here, a growing number of obesity researchers are calling attention to the fact that
environmental conditions have shifted so drastically with the widespread availability
of calorie-dense foods that negative feedback mechanisms are failing, generating
pathology.

The sensorial approach defended by GROS raises important questions concerning

the complex feedback relations between the commensal, social and environmental
dimensions of eating that shape the sensorial mechanisms through which people
gauge hunger and satiety.9 Social norms concerning appropriate meal frequency, or
portion size and environmental factors concerning availability of foods in the various
environments people circulate through, become enfolded in the molecular processes
that control appetites. For example, hunger signals are known to occur at regular
intervals in accordance with habitual meal schedules. Nichter (2008, p. 173) notes
that the rural agriculturalists he worked with in India literally told the time of day by
their rumbling bellies. The sensorial domain is thus socially modulated in complex
ways. Although advocates of sensorial education make a strong distinction between
‘external' eating cues and ‘internal' sensations, the demarcation is not so clear if we
consider the extent to which internal sensations respond according to these
elaborate feedback mechanisms to external, societal, conditions. How, then, can we
effectively return to our bodily signals in a context where these are always already
shaped by the social context that is understood to disrupt our natural capacity to self-
regulate? There is a reflexive loop here that poses considerable difficulties to
practitioners and patients alike, as attested by the centrality of the task of learning
how to differentiate between ‘real' physiological hunger and ‘emotional' or ‘triggered'
hunger. Such issues, for example, took on a particular difficulty in the context of
discussions in the GROS training sessions I attended and in debates following
GROS method presentations at conferences, concerning child feeding. While
children are often taken as a prime example of our innate capacity to eat in
accordance with our energetic needs (and not with emotions or as driven by hedonic
processes), the importance of socialising children's senses is emphasised as,
without guidance, it is suggested they would eat only sweet and calorie-dense foods.
These contradictions point to the subtle and fragile nature of the feedback
mechanisms that regulate appetite. I would like to suggest that the conceptual
difficulties raised through such discussions point to the ways in which bodies and
their environments are tacitly classified on different ends of a polarity. The sensorial
mechanisms these different practices seek to engage are interesting precisely in that
they straddle these conceptual divides between the inside and the outside of bodies,
between matter and subjects, between conditioned and inherited processes.

Feedback models are central to the way endocrinologists are trained to think about
the regulation of the energy balance. Complex systems-thinking is increasingly
important to analyses of obesity policy (Ulijaszek, forthcoming). However, systems
thinking is itself a complex and evolving field. I briefly turn to some important
distinctions between models of complexity and consider their implications for the
ways in which the complexity of obesity is conceptualised on the one hand, and for
the practices promoted by the GROS on the other. Different models to analyse and
predict the behaviour of complex systems are more or less able to accommodate
internal complexity and the regulation of variation in input from outside the regulated
system. This lead to a break between what has been referred to as first- and second-
order cybernetics. First-order cybernetics, of the kind associated with Bateson
(1972), sought to describe the total patterns that connect the elements of a system
through loops of negative and positive feedback. It adopted a putative distinction
between the inside and the outside of the system and aimed to provide an
overarching, integrated perspective that assumed the totality or wholeness of the
system. By contrast, theorists of second-order cybernetics, most notably Maturana
and Varela (1992), aimed to bypass the “geography of inner versus outer” (p. 172).
For Maturana and Varela, the looping that occurs in complex systems is always
contingent on the observer. While first-order cybernetic systems are conceptualised
as closed, total systems, second-order cybernetics posits a circular looping system
more aptly represented by the image of the Mobius strip. Maturana and Varela
(1992) propose to think the distinction between first- and second-order cybernetics
in terms of a contrast between an integrated totality regulating the smaller parts and
“an unruly conversational interaction: the very presence of this unruliness allows a
cognitive moment to come into being according to the system's constitution and
history” (p. 336). In such systems, referred to as ‘autopoeitic', the relationship
between an organisation (or system) and its elements (or structure) is open-ended
but not random. The elements in the system are neither just analytical constructs nor
ontological: their existence is a factor of the reference point from which they are
described. So while first-order cybernetics placed an emphasis on systemic
homoeostasis, second-order cybernetics is based on the idea that it is not possible
to see the totality from any particular point of view.

To which order of cybernetics do the feedback mechanisms mobilised in the obesity

literature belong? Clearly, there are important differences within the scientific fields
engaged in these debates. Most of the literature reviewed for this article provides a
fairly closed and integrated model of regulation in which the biological system
analysed (in rats or humans) and its – for the most part lab – environment are
presented as ontologically distinct. Nevertheless, in their analysis of the cybernetics
of body weight regulation, Fricke et al (2006) examine the role of systems thinking
in analysing the molecular aspects of homoeostatic control of food intake. Their
conclusion is worth citing at length for it marks a clear break with the tenor of much
of the literature on homoeostatic regulation:

The attribution of control elements to biological structures always contains the

problem of semantic simplification, because the biological structure is reduced to a
singular meaning. The same biological structure can be a controlled variable in a
certain feedback loop and a manipulated variable in a different loop. Therefore,
feedback loops might be organized in chains, where elements are parts of different
loops in different functions.
(Fricke et al, 2006, p. 171)

An important question for future work in this area – both for scientific research and
the examination of its policy implications – concerns the kinds of models that are put
to work to conceptualise shifts in the way environments and bodies interact through
‘the sensorial.' Does the rise of ‘hedonism' and renewed attention to ‘addiction' in
obesity science simply signal a historic shift in the food system, or does it also reflect
a shift in epistemological understandings of the processes of appetite regulation? To
what extent do first- or second-order cybernetic models actually influence the way
research is carried out in this field? What are the genealogies of the models used in
the different domains of obesity science and how do they come to impose
themselves in the academic practices of different scientific communities? What
political and conceptual implications do these different models have for the way we
think about eating, treat its disorders and regulate the food system? In practice, the
GROS method is messy. The interactions between sensations, food cues, desires,
fears, habits and so on that it re-sensitizes eaters to are not neatly classifiable
according to a topology of inside or outside, input or output. The method that has
been drawn up experimentally from work with patients aims to return eaters to liking,
rather than being driven by wanting (see Finlayson and Dalton, 2012).

Conclusion

If obesity involves the biological defence of an elevated level of body fat, as current
evidence suggests, advice to simply “eat less, move more” cannot be expected to
remedy the problem. This is because interventions that reduce body fat stores
without a corresponding decrease in the defended level of fat mass elicit
compensatory responses that promote the recovery of lost fat and are difficult to
consciously override.
(Guyenet and Schwartz, 2012)

Public health initiatives aiming to curb the rising rates of obesity are often still
implicitly based on the idea that controlling weight is a matter of willpower and choice.
In this model, people are given information and are expected to understand it and
change their behaviour in accord. Little recognition is given to the fact that there are
a series of factors, ranging from the structural to the neurophysiological (themselves
intrinsically tied up with global economic factors), that intercede in the neat
progression that would lead from the provision of health information to changed
eating behaviour.

In this article, I have mobilised scientific literature on the question of appetite

regulation in order to reveal the models that are used to think about the complex
interplay between foods, bodies and their environments. What has captured my
attention is the way in which the different findings emerging from the fields of
adiposity research or the science of appetite reconfigure the relative balance of
agency between foods and eaters. Here foods are seen to subvert eaters' will by
exerting certain influences on them that are beyond the realm of rational control. The
intimate encounter between foods and eaters is not accidental or incidental but the
product of fine-tuning between food production and homoeostatic and hedonic
pathways of food intake regulation. This opens the question of the implications, for
public health, of recognising the ways in which food may have agency over eaters.
While industry has long understood and capitalised upon this, public health has been
slower to do so.

I have grappled with two rather different sets of material effects in this article. The
first concerns the material effects of foods (or fat, as stocked in adipose tissue) on
eaters. The second concerns a political economy notion of materialism as it relates
to the role of the food industry in both driving the first set of material effects and
subsidising efforts to seek out solutions to the problems generated. Examining the
way the science of sensoriality, food intake regulation and appetite control is
mobilised in public health strategies targeting le comportement alimentaire
(alimentary health behaviours) provides a vantage point onto some of the ways
politics and metabolism collide and into the complex effects of scale that eating
engages. What is striking is the recurrent play on a balance between reflective,
information-based approaches understood to relate directly to people's will and
agency and a growing consensus around the idea that food choices are in fact
largely guided by affective, physiological, endocrine or otherwise unconscious
mechanisms. This raises questions as to the kinds of initiatives that can be
effectively mobilised to target what – notwithstanding the controversy surrounding
the epidemiological claims regarding ‘overnutrition' – continues to be presented as
a major public health priority.

About the Author

Emilia Sanabria (PhD. Anthropology, University of Cambridge) is a Lecturer in
Anthropology at ENS-Lyon (France). She carried out research in Brazil on
experiences of the body, sex hormones and pharmaceuticals. Her current research
examines the multiple relational properties of the category “chemical” through a
range of localities, from the sensorial dimensions of eating to the making of new
efficacies for different pharmaceutical and herbal substances.

Footnotes

The online version of this article is available Open Access

1There is much to be said about the way industry has seized this notion of ‘pleasure'
and the anti-dieting sentiment (notably through the industry support groups such as
GROS receive). To give but one example, France's Weight Watchers launched a
shock campaign provocatively titled “Stop Dieting! Relearn how to eat!” In January
2011 onwards, huge billboards appeared across France with provocative and highly
sexualised images of voracious women's mouths bursting with all sorts of foods,
from fries to sweets to broccoli.
2This was corroborated by a report by the French National Agency of Food Security,
Environment and Work (ANSES, 2010) on the physical and psychological risks of
weight-loss dietary practices. The 10 most practised diets were evaluated for their
risks for vital organ functioning, nutrient deficiencies and impact on food intake
regulation mechanisms. The most damning aspect of the report concerns weight
regain, estimated at 80 per cent of subjects at 1 year and higher beyond. This idea
is not new to GROS, who has built its action around an anti-diet stance for over 10
years. However, the publication of the report lent support to a more diffuse anti-
dieting sentiment arising from uncertainty about both the efficiency of diets and their
long-term effects on health.
3This faculty of bodily perception includes skin senses and proprioception of the
internal organs.
4On this topic, see Mann et al (2011).

5This distinction between rational, will-governed behaviour and emotional, irrational

or pleasure-driven behaviour finds parallels in public health approaches to the
consumption of harmful substances that I do not address here but that would merit
greater attention (see, for example, Greco, 1993, or Valverde, 1997).
6Although she is referring to a discursive production concerning the loss of senses
for ‘us' and a turn to the sensoriality of the ‘Other' in anthropology.
7French distinguishes between ‘rassasiement' and ‘satiété'. The first concerns the
decrease in hunger signals during a meal; it takes place roughly 15–20 min after
meal initiation. The second occurs just after the first and is the period during which
hunger is no longer experienced.
8While not denying the complex processes of evolution that link the metabolic
process at work in human bodies today to those of their prehistoric predecessors,
my point is simply that such self-evident conclusions appear repeatedly in
neurobiology or endocrinology articles without any reference to published research
on these questions. Past environments are presented as homogenously poor in
nutriments without any recognition of the vast diversity of environments in which
humans have lived, thrived and evolved.

(2011) proposes that the French have been “protected” against obesity
9Fischler

because they spend more time eating, and eating together, than the British or
Americans, a fact that increases the chances of pleasure and satiety.

References

1. ANSES 2010Évaluation des risques liés aux pratiques alimentaires

d'amaigrissement. Rapport d'expertise collective. Agence nationale de
sécurité de l'alimentation, de l'environnement et du travail.
2. Ayo N. Understanding health promotion in a neoliberal climate and the making
of health conscious citizens. Critical Public Health. 2012;22 (1:99–105.
3. Barzanò C., Fossi M. In What Sense? A Short Guide to Sensory Education.
Bra, Italy: Slow Food®; 2010.
4. Bateson G. Steps to an Ecology of Mind: Collected Essays in Anthropology,
Psychiatry, Evolution, and Epistemology. Chicago, IL: University of Chicago
Press; 1972.
5. Brownell K., Gold M. Food and Addiction: A Comprehensive Handbook.
Oxford: Oxford University Press; 2012.
6. Butland B., et al. 2007Foresight: Tackling Obesities: Future Choices Project
Report. London: Government Office for Science. [PubMed]
7. Classen C. Worlds of Sense: Exploring the Senses in History and Across
Cultures. London and New York: Routledge; 1993.
8. Commission on Social Determinants of Health (CSDH) 2008. Closing the gap
in a generation: health equity through action on the social determinants of
health. Final Report of the Commission on Social Determinants of Health.
Geneva, World Health Organization. [PubMed]
9. Evans J., Rich E., Davies B., Allwood R. Education, Disordered Eating and
Obesity Discourse: Fat Fabrications. Abingdon, UK: Routledge; 2008.
10. Finlayson G., Dalton M. 2012Current progress in the assessment of ‘liking'
vs. ‘wanting' food in human appetiteCommentAppetite 58(1373–378.
[PubMed]
11. Fischler C. Commensality, society, and culture. Social Science Information.
2011;50 (3–4:528–548.
12. Fitzpatrick K., Tinning R. Health Education: Critical Perspectives. London:
Routledge; 2014.
13. Fricke O., Lehmkuhl G., Pfaff D.W. Cybernetic principles in the systematic
concept of hypothalamic feeding control. European Journal of Endocrinology.
2006;154:167–173. [PubMed]
14. Gard M. The End of the Obesity Epidemic. London: Routledge; 2011.
15. Gastaldo D. 1997Is health education good for you? Rethinking the concept of
health education through the concept of biopowerIn: R. Bunton and A.
Peterson (eds.)Foucault, Health and Medicine London: Routledge; 113–133.
16. Gearhardt A.N., Yokum S., Orr P.T., Stice E., Corbin W.R., Brownell K.D.
Neural correlates of food addiction. Archives of General Psychiatry. 2011;68
(8:808–816. [PMC free article] [PubMed]
17. Greco M. Psychosomatic subjects and the ‘duty to be well': Personal agency
within. Economy and Society. 1993;22 (3:357–372.
18. Guthman J. Weighing In: Obesity, Food Justice and the Limits of Capitalism.
Berkeley, Los Angeles, London: University of California Press; 2011.
19. Guyenet S.J. 2012. Seduced by food: Obesity and the human brain, blog post,
9 March, , http://boingboing.net/2012/03/09/seduced-by-food-obesity-and-
t.html#disqus_thread , accessed 29 June 2012.
20. Guyenet S.J., Schwartz M.W. Regulation of food intake, energy balance, and
body fat mass: Implications for the pathogenesis and treatment of obesity.
Journal of Clinical Endocrinology and Metabolism. 2012;97 (3:745–756.
[PMC free article] [PubMed]
21. Hawkes C., Buse K. Public health sector and food industry interaction: It's
time to clarify the term “partnership” and be honest about underlying interests.
The European Journal of Public Health. 2011;21 (4:400–401. [PubMed]
22. Herrick C. Shifting blame/selling health: Corporate social responsibility in the
age of obesity. Sociology of Health & Illness. 2009;31 (1:51–65. [PubMed]
23. Hinton D., Howes D., Kirmayer L. Toward a medical anthropology of
sensations: Definitions and research agenda. Transcultural Psychiatry.
2008;45 (2:142–162. [PubMed]
24. Howes D. Sensual Relations: Engaging the Senses in Culture and Social
Theory. Ann Arbor, MI: University of Michigan Press; 2003.
25. Howes D.(ed.) (2005Empire of the Senses: The Sensual Culture Reader
Oxford and New York: Berg
26. James P. Developing national obesity prevention policies: An international
perspective. Obes. 2007;2 (1:122–132.
27. Jo J., et al. 2009. Hypertrophy and/or hyperplasia: Dynamics of adipose tissue
growth. in PLoS Computational Biology 5(3): e1000324. [PMC free article]
[PubMed]
28. Kessler D. The End of Overeating: Taking Control of our Insatiable Appetite.
London: Penguin Books; 2009.
29. Kleiman S., Ng S.W., Popkin P. Drinking to our health: Can beverage
companies cut calories while maintaining profits. Obesity Reviews. 2012;13
(3:258–274. [PMC free article] [PubMed]
30. Kluger J. 2007. The science of appetite. Time 31 May.
31. Korsmeyer C. Making Sense of Taste: Taste, Food, and Philosophy. Ithaca:
Cornell University Press; 2002.
32. Kristeller J.L., Wolever R.Q. Mindfulness-based eating awareness training for
treating binge eating disorder: The conceptual foundation. Eating Disorders.
2011;19 (1:49–61. [PubMed]
33. Mann A., et al. Mixing methods, tasting fingers Notes on an ethnographic
experiment. HAU: Journal of Ethnographic Theory. 2011;1 (1:221–243.
34. Maturana H., Varela F. 1992The Tree of Knowledge: The Biological Roots of
Human UnderstandingRev. ed.Boston, MA: Shambhala
35. Ministère du Travail, de l'Emploi et de la Santé 2011Programme national
nutrition santé 2011-2015, http://www.sante.gouv.fr , acessed 18 March
2014.
36. Mol A. Mind your plate! The ontonorms of Dutch dieting. Social Studies of
Science. 2012;43 (3:379–396.
37. Monteiro C.A., Levy R.B., Claro R.M., Castro I.R.R., Cannon G. Increasing
consumption of ultra-processed foods and likely impact on human health:
Evidence from Brazil. Public Health Nutrition. 2011;14 (1:5–13. [PubMed]
38. Moodie R., et al. Profits and pandemics: Prevention of harmful effects of
tobacco, alcohol, and ultra-processed food and drink industries. The Lancet.
2013;381 (9867:670–9. [PubMed]
39. Morton G.J., Cummings D.E., Baskin D.G., Barsh G.S., Schwartz M.W.
Central nervous system of control of food intake and body weight. Nature.
2006;443 (7109:289–295. [PubMed]
40. Nichter M. Coming to our senses: Appreciating the sensorial in medical
anthropology. Transcultural Pyschiatry. 2008;45 (2:163–197. [PubMed]
41. Polivy J., Herman H. Good and bad dieters. Self-perception and reaction to a
dietary challenge. International Journal of Eating Disorders. 1991;10 (1:91–
99.
42. Rasmussen N. Weight stigma, addiction, science, and the medication of
fatness in mid-twentieth century America. Sociology of Health & Illness.
2012;34 (6:880–895. [PubMed]
43. Rice T. The behavioral economics of health and health care. Annual Review
of Public Health. 2013;34:431–47. [PubMed]
44. Seremetakis N.(ed.) (1994The Senses Still Chicago, IL: University of Chicago
Press
45. Serres M. Les cinq sens. Philosophie des corps mêlés I. Paris, France:
Grasset; 1985.
46. Stuckler D., McKee M., Ebrahim S., Basu S. Manufacturing epidemics: The
role of global producers in increased consumption of unhealthy commodities
including processed foods, alcohol, and tobacco. PLoS Medicine. 2012;9
(6:e1001235. [PMC free article] [PubMed]
47. Stuckler D., Nestle M. Big food, food systems, and global health. PLoS
Medicine. 2012;9 (6:e1001242. [PMC free article] [PubMed]
48. Sutton D.E. Food and the senses. Annual Review of Anthropolology.
2010;39:209–23.
49. Tchoukalova Y., et al. Regional differences in cellular mechanisms of adipose
tissue gain with overfeeding. Proceedings of the National Academy of
Sciences. 2010;107 (42:18226–18231. [PMC free article] [PubMed]
 50. Thaler R., Sunstein C. Nudge: Improving Decisions about Health, Wealth and
Happiness. New Haven, CT: Yale University Press; 2008.
51. Ulijaszek S. Biosocieties. in press; forthcoming. With the benefit of Foresight:
obesity, complexity and joined-up government.
52. Valverde M. Slavery from within: The invention of alcoholism and the question
of free will. Social History. 1997;22 (3:251–268.
53. Whithead D., Irvine F. Ottawa 25+ ‘all aboard the dazzling bandwagon' –
developing personal skills: What remains for the future. Health Promotion
International. 2011;26 (S2:ii245–ii252. [PubMed]
54. Zermati J.P., Apfeldorfer G. 2010Vers une nouvelle théorie de l'obésitéIn: J.P.
Zermati, G. Apfeldorfer and B. Waysfeld (eds.), Traiter l'obésité et le surpoids
Paris: Odile Jacob
55. Zuk M. 2009. The evolutionary search for our perfect past. The New York
Times 20 January.

XXXIII. Effects of Partial Substitutions of NaCl with KCl,

CaSO4 and MgSO4 on the Quality and Sensorial Properties
of Pork Patties
Korean J Food Sci Anim Resour. 2014; 34(4): 500–506.
Munkhtugs Davaatseren, Ji-Yeon Chun,1 Hyung-Yong Cho,2 Sang-Gi Min,1 and Mi-
Jung Choi*
Author information ► Article notes ► Copyright and License information ►

Abstract

This study investigated the effects of NaCl replacers (KCl, CaSO4, and MgSO4) on
the quality and sensorial properties of pork patty. In the characteristics of spray-dried
salt particles, KCl showed the largest particle size with low viscosity in solution.
Meanwhile CaSO4 treatment resulted in the smallest particle size and the highest
viscosity (p<0.05). In comparison of the qualities of pork patties manufactured by
varying level of Na replacers, MgSO4 treatment exhibited low cooking loss
comparing to control (p<0.05). Textural properties of KCl and MgSO4 treatments
showed similar pattern, i.e., low level of the replacers caused harder and less
adhesive texture than those of control (p<0.05), whereas the hardness of these
products was not different with control when the replacers were added more than
1.0%. The addition of CaSO4 also manifested harder and less adhesive than control
(p<0.05), but the textural properties of CaSO4 treatment was not affected by level of
Ca-salt. Eventually, sensorial properties indicated that KCl and CaSO 4 influenced
negative effects on pork patties. In contrast, MgSO4 showed better sensorial
properties in juiciness intensity, tenderness intensity as well as overall acceptability
than control, reflecting that MgSO4 was an effective Na-replacer in meat product
formulation.

Keywords: sodium-replacements, potassium, calcium, magnesium, porcine-patty

Introduction

Sodium is an essential mineral to regulate physiological reaction in living bodies,

thereby maintaining life (Durack et al., 2008; Kuwabara, 2010; SACN, 2003). Major
dietary sources of sodium are from salt as a condiment or processed food containing
salt as an additive (Alino et al., 2009; Engstrom et al., 1997; Stamler, 1997). Sodium
is a viral constituent of the body and thus is an essential nutrient. However, high
levels of salt intake have been associated with hypertension, also known as high
blood pressure (HBP) which leads to cardiovascular disease (CVD) (Alino et al.,
2010c; He and MacGregor, 2008). In meat processing, NaCl plays a key role in
extending shelf-life and improving processing qualities such as extraction of salt-
soluble proteins, thermal protein gelation and waterbinding. In addition, salty flavor
manifested by NaCl is an important sensorial property of meat products (Ruusunen
and puolanne, 2005).

From the human health point of view, Na content has to be reduced in the
formulation. Two kinds of approaches are possible to achieve low Na product
manufacturing, i.e., the usage of Na replacer and flavor enhancer (Armenteros et al.,
2012; Matthew and Strong, 2005). KCl is representatively used to provide salty taste
for the purpose of the former. Meanwhile, the letter includes hydrolyzed proteins or
yeast extract (Canto et al., 2014). Regardless of the mentioned approaches, both
techniques are effective in reducing Na content in food formulation. On the other
hand, reducing NaCl directly triggers deterioration of quality characteristics such as
poor water-binding and texturization in the case of meat products (Desmond, 2006;
Doyle and Glass, 2010; Ruusunen et al., 2005). For that reason, the usage of Na
replacer would be appropriate technique to apply meat processing rather than flavor
enhancer.

Theoretically, replaced salts, particularly KCl would maintain the quality and
sensorial characteristics of meat products like the products manufactured with NaCl.
Armenteros et al. (2009a) and Ripolles et al. (2011) reported that sensorial
properties of meat products with substitution of NaCl up to 50% of KCl were not
significantly different. One of disadvantage in the usage of KCl was that it affected
sensorial properties of meat products, negatively (Gou et al., 1996). There were
several studies reported that replacing 40% NaCl with KCl generated bitter flavor
which attribute negative sensorial properties (Gelabert et al., 2003; Gou et al., 1996).

Sensory evaluation of dry-cured loin formulated with 15% NaCl substitution to CaCl2
or MgCl2 did not show significant difference (Armenteros et al., 2009b). According to
Ruusunen et al. (2005), phosphate (P2O5) effectively reduced sodium content in
ground meat patties. Toldra and Barat (2009) noted that salt mixture with low sodium
chloride and high organic salts including magnesium and calcium orthophosphate
were preferable for reduction of sodium content in food formulation. Ripolles et al.
(2011) and Armenteros et al. (2012) indicated that replacement of NaCl content with
other chloride salts could effectively control the final flavor of the dry-cured ham. In
other case, it was still unclear if there was an optimum substitution level (Na-
equivalent) for those Nareplacers (Alino et al., 2010a; Horita et al., 2011). Alino et
al. (2010b) reported that 25% of potassium, 15% of calcium, and 5% of magnesium
substitutes had no significant effect on the physicochemical characteristics and
microbial counts of dry-cured loins. Despite of numerous in- vestigations, the results
regarding the type and replacing level of non-Na salts are still contradictory. For that
reason, this study investigated the effects of NaCl replacer (KCl, CaSO 4, and
MgSO4) on the quality and sensorial characteristics of pork patties.

Materials and Methods

Materials

All salt replacers (KCl, CaSO4 and MgSO4) were purchased from Sigma-Aldrich Co.
(USA). Food grade NaCl and maltodextrin (MD) was obtained from a local market.
Pork leg and backfat were purchased at 48 h post-mortem from three carcasses
randomly. Meat and fat were separately ground through 3 mm plate and vacuum-
packed with polyethylene pouch. Prior to pork patty preparation, the ground meat
and fat were kept at 4℃ within 3 h. For repeated experiments, the meat was
purchased at the same market on three different days (n=3).

Preparation of low-Na salt particles

In order to reduce the sodium ion (Na+) levels, K-salt (KCl), Ca-salt (CaSO4), and
Mg-salt (MgSO4) were substituted with partial amount of Na-salt (NaCl). Control
group was prepared by mixing 10% (w/v) of MD with 20% (w/v) of NaCl in water.
Low-Na salts were also prepared by mixing 13% (w/v) NaCl and 10% (w/v) MD in
water with 7% (w/v) KCl, CaSO4, and MgSO4, respectively. Each mixture was
completely dissolved using a magnetic stirrer at 500 rpm for 30 min. After fully
dissolution of all mixtures, solutions were spray-dried by the condition of 0.6 m3/min
blowing power, 500 mL/h flow rate, 150℃ inlet temperature, and 180 kPa atomizing
pressure.

Microstructure of low-Na salt particles

All powdered salts were kept in desiccator prior to analyses. For morphology
observation, sample powders were gold-coated with platinum for 40 s using an ion
sputter (E-1010, Hithachi, Japan) and microstructures were taken using a scanning
electron microscope (FE-SEM, S-4700, Hitachi, Japan) at 15 mA current. The
images contained above 100 particles were chosen for particle size determination.
Particle diameters were automatically measured and analyzed with image analysis
software (UTHSCSA, USA), whereby the average surface dimension and cumulative
frequency graph were obtained.

Rheological property of low Na-salts

Salt powders (1%, w/w) were dissolved in water and aliquot 1 mL of the solution was
loaded into sample cup. Rheology of the solution was determined using a Rheometer
(MCR-302, Anton-Paar, USA) equipped with a concentric cylinder measuring
geometry 303316 standard (DG26.7/T200/AL). The solution was sheared by linearly
increasing shear rate (γ) from 1/s to 100/s for 500 s under the constant temperature
of 25℃. Shear force (τ) was measured at 5 s interval. Samples viscosity (μ) was
calculated by μ=τ·γ relationship.

Preparation of pork patty

Pork patties were formulated as followings; ground pork (78.8%, w/w), fat (19.7%,
w/w), and various salts (total 1.5%, w/w) were mixed using a food mixer (5K5 SS,
kitchen Aid, USA), and shaped with patty thickness of 3 cm. Pork patties were
separately vacuum-packed in polyethylene pouch and cooked in 95℃ water bath for
30 min. Cooked patties were cooled down at ambient prior to analyses.

Qualities of pork patties

Quality analysis of the pork patty was analyzed by cooking loss and compression
test. The pork patties were weighed before and after cooking for the analysis of
exudation value, and cooking loss were expressed as percentage of initial weight.
For evaluation of textural properties, pork patties were cut into 2 cm 3 cubes and
compressed using a texture analyzer (CT3 Texture Analyzer, BROOKFIFLD, USA)
equipped with TA43 sphere 25.4 mm D probe (BROOKFIFLD, USA) under 10 mm
target value, 20 g trigger load, 0.50 mm/s test speed. All samples were determined
6 times to eliminate technical errors.

Sensory test

For the sensory evaluation, porcine patties were cut into 2 cm cubes, and saltiness,
juiciness, tenderness, and overall acceptability were evaluated. Sensory evaluation
was carried out by using ranking method. For data analysis, samples were scored
from 1 (least value) to 4 (highest value). The ratings were evaluated by relative ratio
of total score of each group against control group. For salt release rate analysis, 2
cm cubes of pork patties were wrapped in cotton and submerged into 100 mL of
water. Release rate graphs were obtained by determining salinity of each sample for
180 min using a SevenCompact S230 conductivity meter (Mettler Toledo AG, CH-
8603, Switzerland).

Statistical analysis

The evaluation of the physicochemical and sensory properties of pork patty was
done by the one-way analysis of variance (ANOVA) using a SAS statistical analysis
program (ver. 9.1) and the means were compared by Duncan’s multiple range test
when the main effect was significant (p<0.05).

Results and Discussion

Characteristics of low-Na salts

The morphology of salt crystals was different depending on type of salts (Fig. 1). For
KCl particles, its boundaries were not clear. In particular, cracks or hollows on the
particles were observed at ×10,000 magnification. The surface of KCl was not
smooth, uneven rugged appearance and also empty micro-particles were observed.
Both CaSO4 and MgSO4 salts were also showed a similar tendency but smoother
surface appearance comparing to that of KCl. For particle size distribution (Fig. 2),
KCl exhibited the largest particle size followed by MgSO 4 (p<0.05). CaSO4 showed
the smallest particle size (p<0.05). On the basis of 60% cumulative frequency,
particle sizes of CaSO4, MgSO4, and KCl were 50 m2, 80 m2, and 110 m2,
respectively. No information regarding how salts attributed the morphology and
particle size is available. Because the particle sizes of spray-dried samples were
depending on the mass flow ratio (Costantino et al., 2000), it would also be closely
related with rheological properties of the salt solution during spray-drying.
Fig. 1. SEM image of spray-dried salt micro-particles. Red arrows indicate
cracks and holes on particle surface.
Fig. 2. Cumulative distributions of surface dimensions of spray-dried salt
micro-particles.

Flow behaviors of each salt solution are given in Fig. 3. All salts were highly soluble
in water and the solution contained only 1% of salts, and the tested samples
displayed a Newtonian fluid-like flowing behavior (Fig. 3A). Meanwhile, shear stress
of the samples indicated different responses with type of salts, reflecting Na-
replacers had a different viscosity (Fig. 3B). KCl treatment (7% KCl + 13% NaCl)
tended to decrease the viscosity of the solution comparing to control (20% NaCl),
although the difference was not significant. Nevertheless, low viscosity of KCl would
account for the largest particle size as mentioned above. Owing to low viscosity, the
KCl solution would have high mass flow in the nozzle of spray-dryer, reflecting the
large particle formation (Soottitantawat et al., 2003). The relationship between
particle size and viscosity was also identified in CaSO4 and MgSO4 treatments. The
former did not show a difference in viscosity comparing to control, while that of the
latter was significantly higher than control (p<0.05). The higher the viscosity, the
smaller the particle size of the salts. In addition, it was hypothesized that rheological
properties of salts would influence on the saltiness. Depending on the flow behavior,
sodium release rate will be different, that plays important role for recognition in the
mouth.
Fig. 3. Changes in (A) shear stress-strain relationship and (B) viscosity of salt
solutions.

Characteristics of pork patties

Since Na replacers exhibited different particle size and rheological properties, Na

release rate would vary and it possibly resulted in different organoleptic properties.
The release rate of Na as a function of time is presented in Fig. 4. Irrespective of
type of salts, increasing level of Na replacers delayed the Na release from pork
patties. In comparison of Na replacers, release rate was fast in the order of KCl,
MgSO4 and CaSO4. Probably, CaSO4 would be adsorbed on the surface of NaCl
more compact, causing small particle size and slow release of Na. In contrast, KCl
formed a porous or cracked layer on the surface of NaCl, hence accelerating Na
release. Eventually, the results reflected that rheological properties attributed the
size of salt particles and rheology of the surface layer by which Na-release was
affected.
Fig. 4. Changes in salinity release rates of pork patties manufactured with NaCl
substitution with (A) KCl, (B) CaSO4 and (C) MgSO4.

For quality characteristics, Na replacers provided different characteristics comparing

to control. Cooking loss of KCl or CaSO4 treatments was significantly higher than
control regardless of the replacer levels (p<0.05) with the exception of 1.0% KCl
treatment (K-1.0%) (Table 1). Meanwhile, 0.5-1.0% MgSO4 treatments showed
significantly lower cooking loss than control (p<0.05). Although there was no report
about the usage of MgSO4 in meat product, it was thought that Mg2+ formed less
rigid myofibrillar gel network comparing to those of K+ or Ca2+. The result was not in
agreement with Panyathipong and Puechkamut (2008) who postulated that Mg-salts
were more effective protein coagulants than Ca-salts in Tofu preparation. Type of
protein sources might result in opposite results. For textural properties, all Na
replacers modified the texture of pork patty including harder and less adhesive
texture (Fig. 5). However, the textural modification was depending on the levels of
Na replacers. For KCl and MgSO4, 0.5% replacements produced harder and less
adhesive than control (p<0.05), whereas their effect on hardness was disappeared
when the level of replacer increased to > 1.0%. Texture of patties manufactured with
CaSO4 also attributed similar pattern to those of KCl and MgSO4, while the patterns
were still maintained with increasing CaSO4 level.

Table 1. Effect of salt substitutes on cooking loss of porcine patties

Replacer concentration (%)

Replacer
0 0.5 1.0 1.5
Cooking loss (%)
KCl 27.7±0.94a 32.0±2.77a 25.5±2.13b 30.9±1.36a
CaSO4 27.7±0.94a 31.6±0.90a 31.2±2.71a 31.1±0.48a
MgSO4 27.7±0.94a 13.3±1.22c 19.4±0.44b 26.9±0.83a

Mean±S.D.
a-cMeans with different superscript in each row are significantly different (p<0.05).
Fig. 5. Changes in hardness and adhesiveness of pork patties manufactured
with NaCl substitution with (A) KCl, (B) CaSO4 and (C) MgSO4. Vertical bars
indicate standard deviations. Means with different letters are significantly
different (p<0.05).

Sensorial properties of pork patties were revealed that the addition of Na-replacers
decreased the saltiness intensity significantly (p<0.05), and the saltiness was lower
with increasing level of replacers (Fig. 6). The results were also identified in Na-
release rate, while the decrease in saltiness was relatively low when MgSO 4 was
used as a Na replacer. In addition, MgSO4 treatment resulted in better juiciness,
tenderness as well as overall acceptability than control. In contrast, KCl and CaSO4
treatments showed negative sensorial attribute with increasing the level of them, as
also supported by Gelabert et al. (2003) and Gou et al. (1996). Our data suggested
that Mg-salt could possibly be the one of the potent candidate for beneficial salt
replacement for reduction of Na in the formulation.
Fig. 6. Sensorial properties of pork patties manufactured with NaCl
substitution with (A) KCl, (B) CaSO4 and (C) MgSO4.

Conclusion

Based on the results of this study, MgSO4 showed a potential application as NaCl
replacer. Still, main mechanisms involved in rheological properties and Na-releasing
rate in actual food matrix were not fully understood, MgSO4 improved water-binding
property and textural modification was also involved in better consumers’
acceptability. Therefore, this study indicated that MgSO4 could be applied as NaCl
replacer with simultaneously improving quality characteristics of meat products.

Acknowledgments

This study was supported from Korea Institute of Planning & Evaluation for
Technology in Food, Agriculture Forestry & Fisheries (IPET) [2013-A008-0013], and
by the 2014 KU Brain Pool of Konkuk University, Seoul, Korea.p

References

1. Alino M., Grau R., Baigts D., Barat J. M. Influence of sodium replacement on pork
loin salting kinetic. J. Food Eng. (2009);95:551–557. doi:
10.1016/j.jfoodeng.2009.06.016. [Cross Ref]
2. Alino M., Grau R., Toldra F., Barat J. M. Physicochemical changes in dry-cured
hams salted with potassium, calcium and magnesium chloride as a partial
replacement for sodium chloride. Meat Sci. (2010a);86:331–336. doi:
10.1016/j.meatsci.2010.05.003. [PubMed] [Cross Ref]
3. Alino M., Grau R., Fuentes A., Barat J. M. Influence of low-sodium mixtures of
salts on the post-salting of dry-cured ham process. J. Food Eng. (2010b);99:198–
205. doi: 10.1016/j.jfoodeng.2010.02.020. [Cross Ref]
4. Alino M., Grau R., Toldra F., Blesa E., Pagan M. J., Barat J. M. Physicochemical
properties and microbiology of dry-cured loins obtained by partial sodium
replacement with potassium, calcium and magnesium. Meat Sci. (2010c);85:580–
588. doi: 10.1016/j.meatsci.2010.03.009. [PubMed] [Cross Ref]
5. Armenteros M., Aristoy M. C., Barat J. M., Toldra F. Biochemical and sensory
properties of dry-cured loin as affected by partial replacement of sodium by
potassium, calcium, and magnesium. J. Agr. Food. Chem. (2009a);57:9699–9705.
doi: 10.1021/jf901768z. [PubMed] [Cross Ref]
6. Armenteros M., Aristoy M. C., Barat J. M., Toldra F. Biochemical changes in dry-
cured loins salted with partial replacements of NaCl by KCl. Food Chem.
(2009b);117:627–633. doi: 10.1016/j.foodchem.2009.04.056. [Cross Ref]
7. Armentos M., Aristoy M. C., Barat J. M., Toldra F. Biochemical and sensory
changes in dry-cured ham salted with partial replacements of NaCl by other chloride
salts. Meat Sci. (2012);90:361–367. doi: 10.1016/j.meatsci.2011.07.023. [PubMed]
[Cross Ref]
8. Canto A. C. V. C. S., Lima B. R. C. C., Suma S. P., Lazaro C. A., Monterio M. L.
G., Conte-Junior C. A., Freitas M. Q., Cruz A. G., Santos E. B., Silva T. J. P.
Physicochemical and sensory attributes of low-sodium restructured caiman steaks
containing microbial transglutaminase and salt replacers. Meat Sci. (2014);96:623–
632. doi: 10.1016/j.meatsci.2013.08.003. [PubMed] [Cross Ref]
9. Costantino H. R., Firouzabadian L., Hogeland K., Wu C., Beganski C.,
Carrasquillo K. G., Córdova M., Griebenow K., Zale S. E., Tracy M. A. Protein spray-
freeze drying. Effect of atomization conditions on particle size and stability. Pharm.
Res. (2000);17:1374–1382. doi: 10.1023/A:1007570030368. [PubMed] [Cross Ref]
10. Desmond E. Reducing salt: a challenge for the meat industry. Meat Sci.
(2006);74:188–196. doi: 10.1016/j.meatsci.2006.04.014. [PubMed] [Cross Ref]
11. Doyle M. E., Glass K. A. Sodium reduction and its effect on food safety, food
quality, and human health comprehensive. Compr. Rev. Food Sci. F. (2010);9:44–
56. doi: 10.1111/j.1541-4337.2009.00096.x. [Cross Ref]
12. Durack E., Gonzalez M. A., Wilkinson M. G. Salt: A review of its role in food
science and public health. Curr. Nutr. Food Sci. (2008);4:290–297. doi:
10.2174/157340108786263702. [Cross Ref]
13. Engstrom A., Tobelmann R. C., Albertson A. M. Sodium intake trends and food
choices. Am. J. Clin. Nutr. (1997);65:704S–707S. [PubMed]
14. Gelabert J., Gou P., Guerrero L., Arnau J. Effect of sodium chloride replacement
on some characteristics of fermented sausages. Meat Sci. (2003);65:833–839. doi:
10.1016/S0309-1740(02)00288-7. [PubMed] [Cross Ref]
15. Gou P., Guerrero L., Gelabert J., Arnau J. Potassium chloride, potassium lactate
and glycine as sodium substitutes in fermented sausages and in dry-cured pork
loins. Meat Sci. (1996);42:37–48. doi: 10.1016/0309-1740(95)00017-8. [PubMed]
[Cross Ref]
16. He F. J., MacGregor G. A. A comprehensive review on salt and health and
current experience of worldwide salt reduction programmes salt and health. J. Hum.
Hypertens. (2008);23:363–384. [PubMed]
17. Horita C. N., Morgano M. A., Celeghini R. M. S., Pollonop M. A. R. Physico-
chemical and Sensory properties of reduced-fat mortadella prepared with blends of
calcium, magnesium and potassium chloride as partial substitutes for sodium
chloride. Meat Sci. (2011);89:426–433. doi: 10.1016/j.meatsci.2011.05.010.
[PubMed] [Cross Ref]
18. Kuwabara A. Nutrition and bone health. Dietary sodium intake and bone health.
Clin. Calcium. (2010);20:590–595. [PubMed]
19. Matthew K., Strong M. Salt - its role in meat products and the industry’s action
plan to reduce it. Nutr. Bull. (2005);30:55–61. doi: 10.1111/j.1467-
3010.2005.00469.x. [Cross Ref]
20. Panyathitipong W., Puechkamut Y. Qualities of tofu powder as affected by
soybean coagulant and drying method. Kasetsart J. (2008);42:156–162.
21. Ripolles S., Campagnol P. C. B., Armenteros M., Aristoy M., Toldra F. Influence
of partial replacement of NaCl with KCl, CaCl2 and MgCl2 on lypolysis and lipid
oxidation in dry-cured ham. Meat Sci. (2011);89:58–64. doi:
10.1016/j.meatsci.2011.03.021. [PubMed] [Cross Ref]
22. Ruusunen M., Puolanne E. Reducing the sodium intake from meat products.
Meat Sci. (2005);70:531–541. doi: 10.1016/j.meatsci.2004.07.016. [PubMed] [Cross
Ref]
23. Ruusunen M., Vainionpaa J., Lyli M., Lahteenmaki L., Niemisto M., Ahvenainen
R., Puolanne E. Reducing the sodium content in meat products: The effect of the
formulation in low-sodium ground meat patties. Meat Sci. (2005);69:53–60. doi:
10.1016/j.meatsci.2004.06.005. [PubMed] [Cross Ref]
24. Stamler J. The INTERSALT Study: Background, methods, findings, and
implications. Am. J. Clin. Nutr. (1997);65:626S–642S. [PubMed]
25. Scientific Advisory Committee on Nutrition (SACN) Salt and Health. the
stationary office; London: (2003).
26. Soottitantawat A., Yoshii H., Furuta T., Ohkawara M., Linko P.
Microencapsulation by spray drying: influence of emulsion size on the retention of
volatile compounds. J. Food Sci. (2003);68:2256–2262. doi: 10.1111/j.1365-
2621.2003.tb05756.x. [Cross Ref]
27. Toldra F., Barat J. M. Recent patents for sodium reduction in foods. Recent Pat.
Food Nutr. Agric. (2009);1:80–86. doi: 10.2174/2212798410901010080. [PubMed]
[Cross Ref]

XXXIV. Polyphenols from halophytes and modified

atmosphere packaging improve sensorial and
biochemical markers of quality of common
dolphinfish (Coryphaena hippurus) fillets
2016 Sep; 4(5): 723–732.
Concetta Maria Messina, 1 Gioacchino Bono, 2 Rosaria Arena, 1 Mariano Randazzo,
3 Simona Manuguerra, 3 and Andrea Santulli 1 , 3

Abstract

Quality and shelf‐life of whole and filleted Coryphaena hippurus, stored with modified
atmosphere packaging (MAP) and natural antioxidants from halophytes (HAL), were
investigated. Fillets were divided into control, simply sealed in trays; MAP, preserved
by modified atmosphere (45% CO2, 50% N2, 5% O2); and MAP‐HAL, pretreated with
antioxidants and preserved by MAP. Whole and filleted fish were stored at
−1 ± 0.5°C for 18 days. The quality of the samples was analyzed at the time of
packaging and after 3, 6, 9, 12, 15, and 18 days. The MAP and MAP‐HAL groups
maintained the best sensorial profile, pH, and drip loss with respect to the untreated
fillets. Higher levels of total volatile basic nitrogen (TVB‐N) and oxidized proteins
were observed in untreated samples with respect to the MAP and MAP‐HAL groups.
The principle component analysis revealed a different quality profile for untreated
and MAP‐treated fish.

Keywords: Fish, modified atmosphere packaging, polyphenols, shelf‐life

Introduction

Recently, consumers began associating the consumption of seafood with beneficial

effects on human health, which has led to an increasing demand for seafood. In
order to meet these demands of both domestic and public consumption, new fish
products and new technologies, able to extend the shelf‐life of seafood, are of
interest. Currently, many new fish and fishery products are imported from
aquaculture in Pacific areas to the European market as deep‐frozen fillets, not
always of good quality (Karl et al. 2014).

The growing demand for seafood can be met by introducing new fish products, by
promoting the consumption of local resources and low commercial value fish
species, and by adopting innovative methods of processing, which are able to extend
the shelf‐life of fish. Modified atmosphere packaging (MAP) has received increasing
attention, also in the seafood industry. MAP is defined as the enclosure of food
products in gas‐barrier materials; the gaseous environment inside the package is
changed in order to reduce spoilage and extend the shelf‐life of the food product
(Sivertsvik et al. 2002; Masniyom 2011). Additionally, many recent studies have focused
on the use of natural polyphenols, instead of synthetic products, for extending the
shelf‐life of products in the seafood industry (Mastromatteo et al. 2010).

We designed the present study to investigate the effect of coupling MAP with natural
antioxidant treatment, on biochemical quality traits of dolphinfish (Coryphaena
hippurus), a Mediterranean fisheries resource. Results from another recent study
demonstrated that MAP, coupled with natural antioxidants, prevented the oxidation
of polyunsaturated fatty acids and the production of molecules that lead to
peroxidation, preserving the quality of dolphinfish (Messina et al. 2015). In this study,
we aimed to verify the effects of the same packaging process on others aspect of
quality, related to the evolution of the protein composition of dolphinfish fillets. In fact,
during fish storage, chemical and enzymatic changes affect proteins and ultimately
give rise to off‐flavors, odors, and amino acid breakdown products (Antoine et al.
2002a). Shelf‐life evolution of fishes is known to be species‐specific and related to the

chemical composition of the fish (Antoine et al. 2002a). Therefore, the observations of
quality by an integrated sensorial and biochemical approach can help to describe
the susceptibility of a species to spoilage and to identify the most effective method
of processing. As source of antioxidants, in this study, we used a species of
halophyte, Halocnemum strobilaceum, which is endemic in the region of Sicily,
where the dolphinfish were caught. This halophyte, as previously reported, is rich in
polyphenols, tested for antioxidant power and total polyphenol content (Messina
et al. 2015). In addition, in this study, we compared data of filleted fish to whole fish,
stored at the same temperature of fillets.

Materials and Methods

Fish sampling and processing

Fish were caught from the west side of the Strait of Sicily in the central Mediterranean
Sea in October 2014. Approximately 60 fish (average total length, 50.2 ± 2.78 cm;
average total weight, 980.2 ± 123.45 g) were caught in the same haul. The fish were
washed in flowing seawater, put on ice and delivered to the laboratory in less than
1 h. Upon arrival at the laboratory, 12 fish were maintained to be preserved on the
whole (WH) and 40 fish were washed under running tap water, headed, gutted,
cleaned, and rewashed prior to be filleted.

From the 80 obtained fillets, eight were utilized for the analyses at T zero and the
others (72) were destined to the shelf‐life experiments. The 72 fillets were randomly
divided into three lots that were preserved in three different ways: 24 fillets of the
control group (CO) were simply sealed in 12 boxes (two fillets/box); 24 fillets of the
MAP group were preserved in 12 boxes (two fillets/box) under modified atmosphere
(45% CO2, 50% N2, 5% O2); and 24 fillets of the MAP‐HAL group were immersed in
a solution of natural antioxidants from halophytes and afterward preserved in 12
boxes (2 fillets/box), with the same modified atmosphere conditions adopted for the
MAP group. The treatment with the antioxidant solution was done according to
Messina et al. (2015) by immersing the fillets in 1 L of antioxidant solution (10 g L−1 of
distilled water) for 2 min.

Packaging materials and gas composition

The gas blends were prepared using a gas mixer (MAP Mix 9000, PBI‐Dansensor
A/S, Ringsted, Denmark). All samples (2 fillets/package) were packaged in fully
transparent APET/EVOH/PE barrier boxes (bag volume: 1800 cc; laminate density:
1.39 g/cm3; thickness: 500 μm) with an oxygen permeability (at 23°C and 0% relative
humidity [RH]) of 1.8 cm3 m−2 day−1 atm−1 and water vapor permeability of
4 g m−2 day −1. To prevent the undesirable accumulation of tissue fluids, an
absorbent food pad (LLP34, Cryovac‐Sealed Air, NJ) was inserted between the fish
and the bottom of the packaging bag. The bags were heat sealed with a barrier
shrink film that was 25‐μm thick (LID 2050, Cryovac‐Sealed Air) with O2 and CO2
permeabilities (at 23°C and 0% RH) of 24 and 95 cm3 m−2 day−1 atm−1, respectively.
The heat‐sealing operation was completed using a semiautomatic packaging
machine (Mondini, Brescia, Italy). A gas‐to‐product ratio of 2.5:1 was used for
samples packed under MAP.

Storage and sampling

Both fillets and WH fish were stored in a refrigerator at −1 ± 0.5°C for 18 days. We
analyzed the quality of the samples by sensorial, chemical, and biochemical
methods at the time of packaging and after 3, 6, 9, 12, 15, and 18 days of storage.
We analyzed two WH fish and two boxes, each containing two fillets, from each
treatment group on each day of sampling.

Sensorial analyses

Seven members of the laboratory, trained in fish quality evaluation, completed the
sensory evaluations of the WH fish and fillets preserved under the experimental
conditions (CO, MAP, and MAP‐HAL), according to the scheme adopted from
Antoine et al. (2002b) for dolphinfish, independently, under the same environmental
condition and after the box were opened. The panelists assessed the attributes of
odor (very fresh, slightly fishy, moderately fishy, and ammoniacal), color (very bright,
bright, dull, and brown), gaping (none, slight, moderate, and excessive), and texture
(very firm, firm, soft, and very soft) (Antoine et al. 2002a). A categorical scale of 1–10
was used for each attribute; a score of 1–2 indicated very poor quality and a score
of 9–10 indicated excellent quality. Sensory evaluations were completed at the same
time on each day of evaluation. The results are expressed as the mean score
assigned by the panelists.

Chemical and physical analyses: pH and drip loss

Determinations of pH and drip loss were completed according to the method of

Benjakul et al. (1997). The samples were homogenized by ultraturrax T25 (KA®‐
Werke GmbH & Co. KG, Staufen, Germany) with 10 volumes of deionized water
(w/v) at a speed of 11,000 rpm for 1 min. The pH of the homogenate was measured
using a microprocessor controlled pH meter (CRISON MicropH 2001, Barcelona,
Spain). The drip loss was measured gravimetrically only in filleted fish, after 48 h of
packaging (day 3 of storage). First, the entire package (sample and film) was
weighed. Then, the sample and any purge were removed from the package, and the
fish and the entire package surface were wiped clean with a paper towel. Finally, the
fish sample was placed back into its package and reweighed. The drip's mass (g)
was divided by the initial mass of the product (g) and reported as a percentage (%).

Biochemical analyses

Fillets from treatments and from WH fish were homogenized and used for analyses,
in triplicate, to assess the quality and shelf‐life of the fish. Aliquots of the homogenate
were processed for moisture (method 950.46), ash (method 938.08), and crude
protein (method 981.10), according to the AOAC (1990) procedures, and total lipid
(Folch et al. 1957) contents.

The total volatile basic nitrogen (TVB‐N) content was determined by direct distillation
of the homogenized samples, according to the EU Commission Decision 95/149/EC
(EEC, 1995). Protein deterioration by oxidation was determined according to the
method described by Lin and Lin (2005). Each 0.4 mL aliquot of the homogenate was
mixed thoroughly with 4 mL of distilled water and 5 mL chloroform–methanol (2:1,
v/v); the total sample was centrifuged at 400 rpm for 20 min. The water layer was
collected and used to determine the fluorescence intensity (excitation wavelength of
340 nm and emission wavelength of 445 nm) using a spectrofluorometer (Cary
Eclipse Fluorescence Spectrophotometer, Agilent Technologies, CA).

The molecular weights of the proteins in the fish samples were determined with
sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS‐PAGE). Total
proteins (TP) were extracted from the fillet homogenate (1:2 w/v) by a solution of
HEPES 10 mmol/L (pH 7.4), NaCl 4 mol/L, EDTA 2 mmol/L, EGTA 2 mmol/L, and a
cocktail of protease inhibitors. TP were diluted to a final protein concentration of
6 mg/mL in a sample buffer (2 mL of 2‐mercaptoethanol, 20 mL 10% SDS, and 5 mL
0.05% bromophenol blue) in 10 mmol/L Tris/HCl and 1 mmol/L EDTA at pH 8.0. The
final volume was adjusted to 100 μL. The samples were heated for 10 min in a boiling
water bath and centrifuged at 16,000g for 1 min. The electrophoresis was performed
on a Bio‐Rad system using a 4–15% Mini‐PROTEAN polyacrylamide gel gradient
(Bio‐Rad Laboratories, Inc., CA). The gel was run at 120 V and 20 mA at 15°C. The
gel was stained with Coomassie blue (0.1% PhastGel Blue R solution in 30%
methanol and 10% acetic acid), destained with 30% methanol and 10% acetic acid,
and preserved in 10% acetic acid and 5% glycerol. The approximate molecular
weights of the proteins were determined using wide range molecular weight
standards (PrecisionPlus standards, Bio‐Rad).

Statistical analyses

The statistical differences were evaluated for each parameter by the analysis of
variance (ANOVA). The differences among the mean values were subjected to the
Student–Newman–Keuls test (SNK). Before analysis, the degree of heterogeneity
was assessed by the Cochran's test (Underwood 1997). Data were processed for one‐
way ANOVA and principal component analysis (PCA) by Statistica (version 8.0,
Statsoft, Inc., Oklahoma).
Results and Discussion

Fish packed as ready‐to‐use products could respond to the growing demand for
minimally processed food and combine high quality with high convenience (Torrieri
et al. 2011). MAP, combined with molecules of natural origin, environmentally safe,
inexpensive, and not limiting is one of the most successful preservation techniques
available to extend the shelf‐life of seafood (Mastromatteo et al. 2010). This technique
is especially desirable for underutilized fisheries species in order to increase the
market of less‐popular species and diversify the offerings of commercial seafood.
Several reports are available on dolphinfish fillet quality, including studies of fillet
refrigeration (Du et al. 2000; Antoine et al. 2002a,b) and fillets processed by high
pressure and cold‐smoking techniques (Yagiz et al. 2007; Gómez‐Guillén et al. 2009;
Gómez‐Estaca et al. 2013). There is increasing interest in expanding the use of this
resource by applying alternative methods of processing for preservation and
storage.

Sensorial analyses

Consumers use postmortem biological changes, including smell and other sensory
attributes, to assess fish quality. As such, they assess some characteristics not
usually considered by experts. Fish tissues contain large quantities of nitrogenous
compounds that generate volatile compounds responsible for the variation in some
sensorial attributes, such as odor, which are generally perceived to indicate
putrefaction (Antoine et al. 2002a). Of these malodorous compounds, volatile bases
are considered the most characteristic feature of fish spoilage (Shewan et al. 1971).

The results of the sensory evaluations of WH and filleted fish are presented in
Figure 1.
Figure 1. Sensorial evaluation scores attributed to Coryphaena hippurus from 3 to 18 days
of storage at −1°C according to the method of preservation: untreated control group (CO)
(▲), modified atmosphere packaging MAP (□), modified atmosphere packaging combined
with natural antioxidant treatment MAP‐HAL (halophytes) (♦), whole fish (WH) (○). Error bar
are smaller than symbols. Each point represents the mean of seven independent
determinations with a standard deviation smaller than 0.4.

At the time of arrival in the laboratory, fish were in the extra freshness status,
represented by a score of 10, referred from all panelists. The sensorial evaluation of
WH and filleted fish, from the third day onward, is represented in Figure 1. WH fish,
at day 6 of storage, were considered not acceptable for consumption. In general,
samples stored under MAP and MAP‐HAL showed better appearances than
samples stored under CO conditions, and the MAP and MAP‐HAL groups
maintained a level of acceptability until day 9. After day 9, fillets preserved with MAP‐
HAL maintained the best general sensorial characteristics, followed by the MAP
group (P < 0.05; Fig. 1). Results of the sensorial evaluations for WH fish agree with
data reported by Antoine et al. (2002b) on fillets of the same species, which were
preserved at 7°C, in which significant variations in sensorial traits start on day 3 of
storage.

The combination of MAP with antioxidants (MAP‐HAL), resulted in a better

maintenance of fish sensorial quality, which is similar to findings obtained with other
fish species processed in a similar manner, such as cod (Wang et al. 2008),
seabream, and salmon fillets (Giménez et al. 2004, 2005). Our results confirm that MAP,
in combination with refrigeration at low temperatures and treatment with natural
antioxidants, can maintain some sensorial aspects of fish quality, which are the first
parameters evaluated by consumers in food grading.

Chemical and physical analyses: pH and drip loss

The increase in pH during fish storage is commonly associated with spoilage and is
considered a general consequence of the accumulation of basic amines in the
muscle (Torrieri et al. 2006).The pH values registered during the storage of WH fish
and fillets are shown in Figure 2.
Figure 2. pH variations in Coryphaena hippurus during 18 days of storage at −1°C
according to the method of preservation: untreated control group (CO) (▲), modified
atmosphere packaging MAP (□), modified atmosphere packaging combined with natural
antioxidant treatment MAP‐HAL (halophytes) (♦), whole fish (WH) (○). Each point represents
the mean of four independent determinations with a standard deviation smaller than 0.5.

Modified atmosphere packaging (MAP)‐HAL and MAP fillets tended to have lower
pH values than WH and CO samples during the entire storage period (P < 0.05). The
pH increase in untreated samples can be explained by the production of basic
amines, which are responsible for rapid spoilage (Ordóñez et al. 2000; Torrieri et al.
2006; Kostaki et al. 2009).

The lower pH for samples stored under MAP can be attributed to the effect of CO2
in inhibiting protein breakdown and, thus, amine production. Similar trends were
observed in other studies of fish processed by MAP (Ordóñez et al. 2000; Cakli et al.
2006; Torrieri et al. 2006; Goulas and Kontominas 2007; Kostaki et al. 2009).

The pH variation modifies protein quality and properties and influences water‐
holding capacity, which is responsible for the drip loss that is negatively associated
with fish quality evolution. In the present study, drip loss increased slightly with
storage time in all fillet samples (Fig. 3). No obvious differences in drip loss were
observed among the treatment groups during the first 9 days of storage, which
confirms the favorable response of this species to filleting and water‐holding
capability. Drip loss was higher in the CO group during the latter part of the storage
period (Fig. 3).

Figure 3
Drip loss variations (%w/w) in fillets of Coryphaena hippurus during 18 days of
storage at −1°C according to the method of preservation: untreated control group
(CO) (▲), modified atmosphere packaging MAP (□), modified atmosphere
packaging combined with natural antioxidant treatment MAP‐HAL (halophytes) (♦).

The drip loss reached approximately 2% in the MAP group, which was lower than
the values (4–8%) reported for MAP‐preserved cod (Dalgaard et al., ), salmon
(Sivertsvik et al. 2002), sea bream (Goulas and Kontominas 2007), and sea bass
(Torrieri et al. 2006); other studies reported that super‐chilled MAP storage did not
lead to excessive drip formation and that water losses lower than 3–5% do not
significantly affect the juiciness of fish flesh (Torrieri et al. 2006). The inclusion of
absorbing pads in the packages, as in our study, can mitigate the negative impact of
drip loss on both quality and consumer acceptability (Torrieri et al. 2006).

Biochemical analyses

The proximate compositions of WH dolphinfish and fillets, preserved in three

different manners, are shown in Figure 4. In general, the proximate composition,
determined on fresh fish at t zero, agree with data reported by Yagiz et al. (2005, 2007).
The total proteins account for 18.97 ± 0.15%, total lipids 4.6 ± 0.62%, total water
72.68 ± 0.29%, and ash 1.47 ± 0.3%. The evolution of proximate composition of
whole fish and fillets, after 3 days, is represented in Figure 4.
Figure 4. Proximate composition (%) of fillets of Coryphaena hippurus during 18 days of
storage at −1°C according to the method of preservation: untreated control group (CO) (▲),
modified atmosphere packaging MAP (□), modified atmosphere packaging combined with
natural antioxidant treatment MAP‐HAL (halophytes) (♦), whole fish (WH) (○). Each point
represents the mean of four independent determinations with a standard deviation smaller
than 0.4.

Modified atmosphere packaging and MAP‐HAL fillets did not show significant
variations in composition during the storage time as did WH and CO samples
(Fig. 4). This observation confirms that MAP, with or without natural extracts, can
contribute to maintain the global quality of nutritional components in this species and
that could be an alternative to synthetic preservatives used in the food industry.

Only a small reduction in water content (approximately 3–5%) was observed in fillets
from the MAP and MAP‐HAL groups during the entire storage period and it was
responsible of the slight protein content variation during the preservation (Fig. 4). A
similar trend was observed in sea bass preserved under MAP and can be considered
a result of species‐specific characteristics and CO2 concentration (Torrieri et al. 2006).
In fact, the increase in CO2 concentration enhances exudation by acidifying fish
muscle and reduces the water‐holding capacity of fish proteins (Pastoriza et al. 1998).

Many indicators are available as biomarkers of fish spoilage, especially nitrogen

degradation, including TVB‐N, biogenic amines, ammonia, however none of these
markers are univocally applicable. TVB‐N is universally considered an easy and
reliable indicator of fish shelf‐life if coupled with others indicators, and its correlations
with other biomarkers of shelf‐life have been demonstrated for dolphinfish (Antoine
et al. 2002b). The TVB‐N levels recorded in the WH fish and fillets in our study are
presented in Figure 5. Until day 12, neither unprocessed nor processed samples
showed a significant increase in TVB‐N levels, although fillets in the MAP and MAP‐
HAL groups had lower TVB‐N values than fillets in the CO group (Fig. 5). After day
12, only WH fish reached the threshold of spoilage (30 mg of TVB‐N/100 g), followed
by samples of CO group. Fillets in the MAP‐HAL and MAP groups showed TVBN
values below the threshold for spoilage, even after 15 days of storage (Fig. 5).

Figure 5. TVB‐N variations (mg/100 g tissue) in fillets of Coryphaena hippurus during

18 days of storage at −1°C according to the method of preservation: untreated control group
(CO) (▲), modified atmosphere packaging MAP (□), modified atmosphere packaging
combined with natural antioxidant treatment MAP‐HAL (halophytes) (♦), whole fish (WH) (○).
Different superscript letters, within the same storage time, indicate significant differences
among treatments (a, b, c: P < 0.05).

TVB‐N variations (mg/100 g tissue) in fillets of Coryphaena hippurus during 18 days

of storage at −1°C according to the method of preservation: untreated control group
(CO) (▲), modified atmosphere packaging ...

The levels of TVB‐N did not differ between fillets of the MAP and MAP‐HAL group,
until day 12; after day 12, MAP samples exhibited an increase in TVB‐N production,
with respect to MAP‐HAL (P < 0.05). The inhibitory effect of CO2 on the production
of TVB‐N was more pronounced in samples from the MAP‐HAL group, suggesting a
synergistic role of MAP and antioxidants (Fig. 5).

For all samples in our study, the levels of TVB‐N were lower than those recorded in
the same species and stored under refrigeration at 7°C (Antoine et al. 2002b); our
results were similar to those recorded in cold‐smoked dolphinfish, which supports
the fact that some processing techniques, coupled with natural antioxidants, can
significantly contribute to the preservation of quality and the extension of shelf‐life of
seafood by limiting the production of TVB‐N (Gómez‐Guillén et al. 2009). The validity
of this indicator in evaluating the shelf‐life of dolphinfish was established by Antoine
et al. (2002b), who also reported a positive correlation between TVB‐N levels and
sensorial evaluations.

Other fish species processed by coupling MAP with antioxidants exhibited a

decrease in TVB‐N levels, which confirms the effectiveness of this processing
method for preserving the quality of fish (Kostaki et al. 2009; Mastromatteo et al. 2010).
The European Union (Commission Decision 1995–95/149/EEC) set an upper limit
of 25–35 mg TVB‐N per 100 g of fish for consumption, but additional data are
needed to provide new limits for processed fish, as suggested by Gómez‐Guillén
et al. (2009).

Others aspects of protein deterioration can be considered in the evaluation of the

shelf‐life of fish, such as the loss of quality, due to protein oxidation. Protein oxidation
is one of the factors influencing the deterioration of fish meat, although most enzyme
activity is slowed or inhibited during frozen storage. Srinivasan and Hultin ( 1994)
reported that protein denaturation was mainly affected by lipid oxidation, suggesting
that protein oxidation is likely coupled with lipid oxidation.

Our previous results showed that storage with MAP and natural antioxidants can
reduce lipid peroxidation in dolphinfish (Messina et al. 2015). In this study, the analysis
of protein degradation by oxidation revealed less oxidation than that reported by Lin
and Lin (2005) in both untreated and treated samples of bonito fillets (Fig. 6).
Furthermore, higher levels of oxidized proteins were observed in WH fish and CO
fillets than in MAP and MAP‐HAL fillets. These findings confirm that the fillets
resulted protected against oxidation by treatment with MAP and antioxidants (Fig. 6).

Figure 6. Protein degradation (μmol/g) in fillets of Coryphaena hippurus during 18 days of

storage at −1°C according to the method of preservation: untreated control group (CO) (
), modified atmosphere packaging MAP ( ), modified atmosphere packaging combined
with natural antioxidant treatment MAP‐HAL (halophytes) ( ), whole fish WH ( ).

The reduction of protein oxidation levels in MAP‐HAL fillets confirmed that HAL
extract induced a significant antioxidant activity when coupled with MAP, also on
protein fraction of the fillets and confirms our previous work, which demonstrated
that MAP and antioxidants inhibited lipid oxidation in dolphinfish (Messina et al. 2015).
Overall, the low levels of protein oxidation in both treated and untreated dolphinfish
samples did not compromise the pattern and integrity of TP, as confirmed by SDS
electrophoresis. No significant variations were observed between the electrophoretic
patterns of the TP extracted from the samples, at the beginning and at the end of
our study (Fig. 7).
Figure 7. SDS‐PAGE profiles of total proteins extracted from Coryphaena hippurus,
according to the method of preservation. Lane Std: wide range molecular weight markers;
lane 1: whole fish (WH) at T zero; lane 2: WH at T3; lane 3: control group CO at T3; lane 4:
modified atmosphere packaging MAP at T3; lane 5: MAP combined with natural antioxidant
treatment MAP‐HAL (halophytes) at T3; lane 6: CO at T18; lane 7: MAP at T18; lane 8:
MAP‐HAL (halophytes) at T18; lane 9: WH at T18.

Correlations among sensorial, chemical, and biochemical parameters of

quality

Sensorial aspects of quality in fish, including dolphinfish, are closely related to the
chemical and biochemical composition of the fish and to their variations during
storage (Du et al. 2000; Antoine et al. 2002b; Gómez‐Guillén et al. 2009). The
temperature of preservation and the methods of processing contribute to preventing
spoilage, preserving quality, and extending the shelf‐life of fish, as well as improving
the overall quality (Mastromatteo et al. 2010; Masniyom 2011). In our study, the
methods of processing effectively preserved the quality and shelf‐life of dolphinfish
and improved some parameters related to sensorial, chemical, and biochemical
quality in the MAP and MAP‐HAL groups compared to the untreated samples.

PCA was applied to sensorial, chemical, and biochemical data obtained in this study
(Fig. 8). Five variables (sensorial analysis, pH, drip loss, TVB‐N, and oxidized
proteins) were considered for the analysis; PCA examined 72 cases that represented
the means of data collected from WH fish and CO, MAP, and MAP‐HAL fillets, from
day 3 to day 18. This analysis explained 87% of the variability in the original data
(Fig. 8). The first principal component (PC1) explained 55.6% of the combined
variance and the second component (PC2) explained 33.3% (Fig. 8). This
representation indicates that the quality of untreated fish (WH fish and CO fillets)
was different than the quality of MAP‐treated dolphinfish fillets (Fig. 8). By this
representation, it is possible to appreciate that the points representing the quality of
untreated fish, during the storage period, are dispersed within the graphic; on the
contrary, the points representing the quality of MAP‐ and MAP‐HAL‐treated fillets
are concentrated within the same area, independently from the time of preservation,
indicating a more stable profile of quality in processed fillets than in untreated ones
(Fig. 8). Finally, sensorial quality resulted strongly correlated to drip loss, TVB‐N
level, and oxidized proteins, which confirms the effect of nitrogen and protein
compounds on shelf‐life and perceived quality.

Figure 8. Principal component analysis, obtained from correlation of five components

(sensorial analysis, pH, drip loss, TVBN, and oxidized proteins) and matrix of correlations
among parameters. The 72 represented cases are mean of multiple determinations,
collected on Coryphaena hippurus during the experiment according to the method of
preservation: untreated control group CO and whole fish (black label), modified atmosphere
packaging MAP (red label), MAP with natural antioxidant treatment MAP‐HAL (halophytes)
(blue label).

Conclusions

Our results demonstrated that processing by filleting can ameliorate the quality
evolution of dolphinfish during storage at −1°C. Overall, MAP, coupled with natural
polyphenols from HAL, contributed to improved dolphinfish quality from sensorial,
chemical, and biochemical points of view.

The biomarkers that are related to the quality of the proteins, such as TVB‐N and
oxidized proteins, were lowest in samples processed by MAP and MAP‐HAL. The
processing method that coupled MAP with natural antioxidants, resulted effective in
extending the shelf‐life of dolphinfish fillets by almost 6 days, compared to WH fish.
These findings support the application of this method of processing to increase both
the shelf‐life and marketability of fishery species.

Conflict of Interest

None declared.

Acknowledgments

This study was supported by grants from the Italian Ministry of Education, University
and Research, University of Palermo, FFR 2012/2013 to Concetta Maria Messina,
IEVP 2007–2013 (project BioVecQ Cod PS1.3_08), and PON R&C 2007/2013
(project PESCATEC, code 004513362121).

Notes

This paper was supported by the following grant(s):

Italian Ministry of Education, University and Research.

Notes

This paper was supported by the following grant(s):

University of Palermo.

Notes

This paper was supported by the following grant(s):

FFR.

Notes

This paper was supported by the following grant(s):

IEVP.

Notes
This paper was supported by the following grant(s):

PON R&C.

References

1. Antoine F. R., Wei C. I., Otwell W. S., Sims C. A., Littell R. C., Hogle A. D.,
et al. 2002a. TVB‐N correlation with odor evaluation and aerobic plate count
in Mahi Mahi (Coryphaena hippurus). J. Food Sci. 67:3210–3214.
doi:10.1111/j.1365‐2621.2002.tb09567.x.
2. Antoine F. R., Wei C. I., Otwell W. S., Sims C. A., Litterel R. C., Hogle A. D.,
et al. 2002b. Analysis of biogenic amines and their precursor free amino acids
in Mahi‐Mahi (Coyphaena hippurus). J. Food Biochem. 26:131–152.
doi:10.1111/j.1745‐4514.2002.tb00870.x.
3. AOAC . 1990. AOAC official methods of analysis. 15thed. Association of
Official Analytical Chemists, Arlington, VA.
4. Benjakul S., Seymour T. A., Morrissey M. T., and An H.. 1997.
Physicochemical Changes in Pacific Whiting Muscle Proteins during Iced
Storage. J. Food Sci. 62:729–733. doi:10.1111/j.1365‐2621.1997.tb15445.x.
5. Cakli S., Kilinc B., Dincer T., and Tolasa S.. 2006. Comparison of the shelf
lifes of map and vacuum packaged hot smoked rainbow trout (Onchoryncus
mykiss). Eur. Food Res. Technol. 224:19–26. doi:10.1007/s00217‐006‐0283‐
3.
6. Dalgaard P., Gram L., and Huss H.H.. 1993. Spoilage and shelf‐life of cod
fillets packed in vacuum or modified atmospheres. Int. J. Food Microbial.
19:283–294. http://dx.doi.org/10.1016/0168-1605(93)90020-H
doi:10.1016/0168‐1605(93)90020‐H [PubMed]
7. Du W.‐X., Huang T., Kim J., Marshall M. R., and Wei C.. 2000. Chemical,
Microbiological, and Aroma Scan Evaluation of Mahi‐mahi Fillets under
Various Storage Conditions. J. Agric. Food Chem. 49:527–534.
doi:10.1021/jf0011135. [PubMed]
8. EEC . 1995. Total volatile basic nitrogen (TVB‐N) limit values for certain
categories of fishery products and specifying the analysis methods to be
used. Commission Decision 95/149/EEC of 8 March 1995. Official J. Euro
Communities 97:84–87.
9. Folch J., Lees M., and Stanley G. H. S.. 1957. A simple method for the
isolation and purification of total lipids from animal tissues. J. Biol. Chem.
226:497–509. [PubMed]
10. Giménez B., Roncalés P., and Beltrán J. A.. 2004. The effects of natural
antioxidants and lighting conditions on the quality characteristics of gilt‐head
sea bream fillets(Sparus aurata) packaged in a modified atmosphere. J. Sci.
Food Agric. 84:1053–1060. doi:10.1002/jsfa.1785.
11. Giménez B., Roncalés P., and Beltrán J. A.. 2005. The effects of natural
antioxidants and lighting conditions on the quality of salmon (Salmo salar)
fillets packaged in modified atmosphere. J. Sci. Food Agric. 85:1033–1040.
doi:10.1002/jsfa.2055.
12. Gómez‐Estaca J., Gómez‐Guillén M. C., and Montero P.. 2013. The effect of
combined traditional and novel treatments on oxidative status of dolphinfish
(Coryphaena hippurus) and sardine (Sardinapilchardus) muscle lipids. Food
Sci. Technol. Int. 1–10. [PubMed]
13. Gómez‐Guillén M. C., Gómez‐Estaca J., Giménez B., and Montero P.. 2009.
Alternative fish species for cold‐smoking process. Int. J. Food Sci. Technol.
44:1525–1535. doi:10.1111/j.1365‐2621.2008.01762.x.
14. Goulas A. E., and Kontominas M. G.. 2007. Combined effect of light salting,
modified atmosphere packaging and oregano essential oil on the shelf‐life of
sea bream (Sparus aurata): biochemical and sensory attributes. Food Chem.
100:287–296. doi:10.1016/j.foodchem.2005.09.045.
15. Karl H., Lehmann I., Manthey‐karl M., Meyer C., and Ostermeyer U.. 2014.
Original article Comparison of nutritional value and microbiological status of
new imported fish species on the German market. Int. J. Food Sci. Technol.
49:2481–2490. doi:10.1111/ijfs.12543.
16. Kostaki M., Giatrakou V., Savvaidis I. N., and Kontominas M. G.. 2009.
Combined effect of MAP and thyme essential oil on the microbiological,
chemical and sensory attributes of organically aquacultured sea bass
(Dicentrarchus labrax) fillets. Food Microbiol. 26:475–482.
doi:10.1016/j.fm.2009.02.008. [PubMed]
17. Lin C. C., and Lin C. S.. 2005. Enhancement of the storage quality of frozen
bonito fillets by glazing with tea extracts. Food Control 16:169–175.
doi:10.1016/j.foodcont.2004.01.007.
18. Masniyom P. 2011. Deterioration and shelf‐life extension of fish and fishery
products by modified atmosphere packaging. Songklanakarin J. Sci. Technol,
33:181–192.
19. Mastromatteo M., Conte A., and Del Nobile M. A.. 2010. Combined Use of
Modified Atmosphere Packaging and Natural Compounds for Food
Preservation. Food Eng. Rev. 2:28–38. doi:10.1007/s12393‐010‐9013‐5.
20. Messina C. M., Bono G., Renda G., La Barbera L., and Santulli A.. 2015.
Effect of natural antioxidants and modified atmosphere packaging in
preventing lipid oxidation and increasing the shelf‐life of common dolphinfish
(Coryphaena hippurus) fillets. LWT ‐ Food Sci. Technol. 62:271–277.
doi:10.1016/j.lwt.2015.01.029.
21. Ordóñez J. A., López‐Gálvez D. E., Fernández M., Hierro H., and Hoz L..
2000. Microbial and physicochemical modifications of hake (Merluccius
merluccius) steaks stored under carbon dioxide enriched atmospheres. J. Sci.
Food Agric. 80:1831–1840. doi:info:doi/10.1002/1097-
0010(200010)80:13%3c1831:AID-JSFA707%3e3.0.CO;2-Z">10.1002/1097‐
0010(200010)80:13<1831:AID‐JSFA707>3.0.CO;2‐Z.
22. Pastoriza L., Sampedro G., Herrera J. J., and Cabo M. L.. 1998. Influence of
sodium chloride and modified atmosphere packaging on microbiological,
chemical and sensorial properties in ice storage of slices of hake (Merluccius
merluccius). Food Chem. 61:23–28. doi:10.1016/S0308‐8146(97)00130‐1.
23. Shewan J. M., Gibson D. M., and Murray C. K.. 1971. The estimation of
trimethylamine in fish muscle Pp. 183–186 in Kreuzer R., editor. , ed. Fish
inspection and quality control. Fishing News Books Ltd., London, England.
24. Sivertsvik M., Jeksrud W. K., and Rosnes J. T.. 2002. A review of modified
atmosphere packaging of fish and fishery products – significance of microbial
growth, activities and safety. Int. J. Food Sci. Technol. 37:107–127.
doi:10.1046/j.1365-2621.2002.00548.x.
25. Srinivasan S., and Hultin H. O.. 1994. Hydroxyl radical modification of fish
muscle proteins. J. Food Biochem. 18:405–425. doi:10.1111/j.1745‐
4514.1994.tb00513.x.
26. Torrieri E., Cavella S., Villani F., and Masi P.. 2006. Influence of modified
atmosphere packaging on the chilled shelf life of gutted farmed bass
(Dicentrarchus labrax). J. Food Eng. 77:1078–1086.
doi:10.1016/j.jfoodeng.2005.08.038.
27. Torrieri E., Carlino P. A., Cavella S., Fogliano V., Attianese I., Buonocore G.
G., et al. 2011. Effect of modified atmosphere and active packaging on the
shelf‐life of fresh bluefin tuna fillets. J. Food Eng. 105:429–435.
doi:10.1016/j.jfoodeng.2011.02.038.
28. Underwood A. J. 1997. Experiments in ecology: their logical design and
interpretation using analysis of variance. Cambridge University Press,
Cambridge.
29. Wang T., Sveinsdóttir K., Magnússon H., and Martinsdóttir E.. 2008.
Combined application of modified atmosphere packaging and superchilled
storage to extend the shelf life of fresh cod (Gadus morhua) loins. J. Food
Sci. 73:11–19. doi:10.1111/j.1750‐3841.2007.00590.x. [PubMed]
30. Yagiz Y., Kristinsson H. G., Balaban M. O., and Marshall M. R.. 2005. Effect
of high pressure treatment on Atlantic salmon (Salmo salar) muscle In IFT
annual meeting and food expo, aquatic food products, Pp. 89, July, New
Orleans, LA.
31. Yagiz Y., Kristinsson H. G., Balaban M. O., and Marshall M. R.. 2007. Effect
of high pressure treatment on the quality of rainbow trout (Oncorhynchus
mykiss) and mahimahi (Coryphaena hippurus). J. Food Sci. 72:509–515.
doi:10.1111/j.1750‐3841.2007.00560.x. [PubMed]
XXXV. Increasing Strawberry Fruit Sensorial and
Nutritional Quality Using Wild and Cultivated
Germplasm
PLoS One. 2012; 7(10): e46470.
Jacopo Diamanti,1 Franco Capocasa,1 Francesca Balducci,1 Maurizio Battino,2 Jim
Hancock,3 and Bruno Mezzetti1,*
Cameron Peace, Editor

Abstract

Background

Increasing antioxidant levels in fruit through breeding is an important option to

support higher antioxidant intake particularly when fruit consumption is low. Indeed,
if nutritional components are also combined with a high standard of sensorial fruit
quality, the perspective for consumer health can be further improved by encouraging
more fruit consumption. Wild species are valued by strawberry breeders as sources
of novel traits, especially for pest resistance and abiotic stress tolerance.
Furthermore, previous investigations have shown improvements in fruit nutritional
quality in breeding material that originated from Fragaria virginiana ssp. glauca
(FVG) inter-species crosses. Recently, commercial varieties of strawberries have
also shown interesting variability in fruit nutritional quality.

Results

Strawberry fruit sensorial and nutritional qualities generated by Fragaria inter-

species and intra-species crosses were evaluated on 78 offspring derived from 8
families: two that originated from F. × ananassa intra-species crossing; three from
back-crossing of F1– FVG × F. × ananassa; and three from back-crossing of BC1–
FVG × F. × ananassa. The genetic variability from the three types of cross
combinations was analyzed by calculation of the correlations among the fruit
sensorial and nutritional parameters. The results obtained show that two subsequent
back-crossing generations from an inter-species crossing combination with F.
virginiana ssp. glauca provides useful improvement of the fruit nutritional and
sensorial qualities that is combined with agronomic standards that are close to those
requested at the commercial level. Improvements of these traits can also be
achieved by programming F. × ananassa intra-species crosses and producing
progeny with productivity traits more similar to those of the commercial cultivars.

Conclusions

The two types of combination programs (inter-species back-crosses, and intra-

species crosses) can be used to improve strawberry nutritional quality.

Introduction

Greater consumption of fruit and vegetables is considered as one way of increasing

the intake of antioxidants, and like other berries, strawberry represents one of the
most important sources of bioactive compounds with high antioxidant capacity [1]–
[6]. Accordingly, an increase in the consumption of berries richer in ‘healthy
compounds’ is seen as an appropriate strategy for improving human health.

Increasing the antioxidant levels in fruit through breeding and/or biotechnology is an

important option to maintain a higher antioxidant intake, particularly when fruit
consumption is low. Indeed, if nutritional bioactive components are combined with a
high standard of sensorial fruit quality, consumer fruit consumption can be
encouraged, with much greater positive health effects [3], as would be the case for
strawberry fruit [7].

The breeding of more nutritious, better-tasting cultivars can be successful if the

variability and heritability of the bioactive compounds, which is defined as the total
antioxidant capacity (TAC), indicate the possibility of achieving breeding progress. It
is well know that the availability of genetic diversity within compatible species of any
given crop will enhance the extent of any improvement [8]. The biotechnological
approach is also an option to supplement this improvement, through modifications
of specific biosynthetic pathways [9]. However, the success of both breeding and
biotech approaches is dependent on deep knowledge of the sources of the most
useful wild and cultivated genetic diversity to be used in genetic and genomic
studies.

The effect of genotype on the nutritional quality of strawberry is well known [10]–[13].
The levels of antioxidants and the TAC in strawberry extracts from whole fruits vary
considerably among genotypes [4], [12], [14], although few genotypes have been
well characterized for these important features. Furthermore, limited knowledge is
available on the possibility of improving strawberry nutritional traits by breeding [15]–
[16]. Some results have been obtained for blueberries and raspberries, with a
moderate heritability for TAC, total phenols content (TPH) and anthocyanins content
(ACY) demonstrated [17]–[21].

Accessions of the progenitor wild species are valued by strawberry breeders as

sources of novel traits, and especially for pest resistance and abiotic stress tolerance
[22]. Furthermore, previous investigations [4], [5], [15], [16] have shown
improvements in fruit nutritional quality in breeding material that originated from
Fragaria virginiana ssp. glauca (FVG) in inter-species crosses. This result is
supported by a study carried out by Wang et al. [14] that showed that the fruit of F.
virginiana accessions have significantly higher TAC, TPH, and total ACY than the
fruit from different lines of F. chiloensis and F. × ananassa.

Since 2006, a breeding program based on the evaluation and comparison of different
populations of F. × ananassa crosses and FVG inter-species back-crosses has been
conducted by the Università Politecnica delle Marche, with the objective being to
generate new selections with increased fruit nutritional quality.

With the aim of studying strawberry fruit sensory and nutritional quality generated by
Fragaria inter-species and intra-species crosses, 78 offspring derived from 8 families
were evaluated, two originating from F. × ananassa intra-species crosses; three from
back-crossing of F1– FVG × F. × ananassa; and three from back-crossing of BC1–
FVG × F. × ananassa (Figure 1). The selection AN94,414,52 (F1 FVG) used in the
breeding program was originated by the inter-species cross of F. virginiana ssp.
glauca “FVG22” × F. × ananassa selection 91,143,5. FVG22 is a clone of F.
virginiana ssp. glauca that is maintained in the National Clonal Germplasm
Repository in Corvallis (OR, USA), while the F. × ananassa selection 91,143,5
originated from the breeding program of the Italian Experimental Agriculture
Research Council, Forli Fruit Research unit (CRA-FRF, Italy).
Figure 1. Crossing combination scheme.

Scheme of the crossing combinations, including F. × ananassa and F. virginiana

glauca back-crossing, and F. × ananassa intra-species crossing.

Fruit sensorial and nutritional parameters of all of the offspring were evaluated in
2009 from single seedling plants, and in 2010 on vegetatively propagated selections.
For both years, all of the offspring and the corresponding parents were evaluated for
their fruit nutritional qualities (fruit TAC, TPH and ACY), while in the second year,
they were evaluated for their production (fruit weight and total plant production,
commercial, misshapen, undersized and rotten fruit production, per plant) and
sensorial (fruit firmness, color, solid soluble [SS] content and total acidity [TA])
parameters.

The genetic variability that originated from the three types of cross-combinations
(BC1 back-cross, BC2 back-cross, and intra-species cross) was analyzed by
correlating fruit sensorial and nutritional parameters. With such an analysis, we were
able to evaluate the relative effects of intra-species and inter-species back-cross
populations.

Results
Fruit nutritional and sensorial parameters of the different breeding populations were
evaluated by two-way nested ANOVA, which revealed significant differences
between the inter-species back-crosses and the intra-species crosses (Table 1).

Table 1
Two-way nested ANOVA for the type of cross combination and the individual
BC1, BC2 and intra-species populations for the fruit sensorial and nutritional
parameters.

SS content, TA, TAC, TPH and ACY were significantly higher in the BC1 and BC2
back-crossed populations that contained genes from Fragaria virginiana ssp glauca
(Table 2) than in the F. × ananassa intra-species crosses, while the intra-species
crosses were firmer and more darkly colored (Table 2). For all of the parameters
examined, there were no significant differences between the BC1 and BC2 back-
crosses, although the individual populations were significantly different.

Table 2
Fruit sensorial and nutritional parameters of the clonal populations of selected
seedlings for each of the individual cross combinations within the BC1 and
BC2 back-crosses and the intra-species cross populations.

Selections of the BC1 Populations

The two-way nested ANOVA evaluated for the BC1 back-cross populations showed
significant differences for fruit TA, TAC, TPH and ACY, but not for SS content,
firmness, Chroma and L* (Table 1).
The highest mean values of the fruit TA were seen for AN07,004 and AN07,005,
while the AN07,003 cross-combination showed a significantly lower TA than for
AN07,004, although not a significantly different TA than for AN07,005 (Table 2). The
highest fruit TAC was seen for the fruit of AN07,004, while the lowest was for the
fruit of AN07,005; these TACs were significantly different. Population AN07,004 also
showed the highest fruit TPH, which was significantly different from the AN07,003
back-cross population with the lowest TPH (Table 2).

In contrast to these results, population AN07,003 showed the highest fruit ACY,
which was significantly higher than for AN07,004 (Table 2).

The BC1 populations also differed significantly in their plant production parameters.
Indeed, AN07,004 had the highest mean for fruit weight and commercial production,
and the lowest for misshapen fruit per plant (Table 3). The lowest mean fruit weight
was seen for AN07,003, while the lowest commercial production was for AN07,005.
The lowest amount of rotted fruit was seen for BC1 AN07,004 (Table 3).

Table 3
Plant production parameters of the selections from the BC1 and BC2 back-
cross and the intra-species cross populations.

Fruit sensorial and nutritional parameters of the BC1 populations had different levels
of correlation (Table 4). Fruit TA was negatively correlated with fruit SS content
(−0.24), firmness (−0.50) and ACY (−0.30). A high negative correlation was also
seen between fruit firmness and TAC (−0.34). The Chroma index had a high negative
correlation with TAC (−0.29) and ACY (−0.46). Finally, L* grade showed a high
negative correlation with fruit ACY (−0.47) (Table 4).

Table 4
Pearson’s correlation matrix of the fruit sensorial and nutritional parameters
for type of crossing evaluated on clonal populations of selected seedling.

Positive correlations were seen between SS content and fruit ACY (0.26), Chroma
index and L* grade (0.83), and fruit TAC and TPH (0.46) and ACY (0.33) (Table 4).

These data show the positive effects of using F. virginiana glauca in strawberry
breeding, in terms of improving the fruit nutritional parameters, although there was
a negative effect on fruit sensorial parameters, as shown by the correlation matrix
for the BC1 FVG back-crosses. Indeed, the increase in fruit nutritional parameters
was associated with a large increase in fruit TA, which was linked with a decrease
in fruit SS content, firmness and ACY. In particular, fruit TAC was negatively
correlated with fruit firmness and Chroma index. This indicates that in BC1
progenies, the higher content of bioactive compounds that determine the fruit TAC
is associated with negative effects on fruit firmness, color (darker fruit) and ACY. The
correlation matrix of the same BC1 progenies reveals negative correlation between
ACY and both colorimetric parameters, while in the BC1 population, ACY was
positively correlated with an increase in fruit SS content. Among the nutritional
parameters, the positive correlations between fruit TAC and TPH, and fruit TAC and
ACY were consistent with what was reported by Scalzo et al. (2005).

Principal component analysis (PCA) bi-plots of the BC1 families showed strong
associations between the TAC and TPH nutritional parameters for AN07,004, and in
particular for AN07,004,51 and AN07,004,54 (Figure 2b), confirming the data for the
nutritional parameters in Table 4. AN07,003 showed the highest fruit ACY (Table 2)
and PCA analysis confirmed this trend (Figure 2b). The PCA bi-plots combining the
data of the fruit TAC, TPH, ACY, and weight, and the commercial production of the
selections and corresponding parents of the BC1 populations (Figure 2a, vector
distribution plot; Figure 2b, case distribution plot), highlight the selections
AN07,004,51, AN07,004,52, AN04,007,54 and AN07,003,60 for their positions in the
upper right-hand section of the plot, corresponding to high values of the fruit ACY,
TPH and TAC variables. Thus, these selections have the highest fruit nutritional
value, although they still have some deficiencies in fruit fresh weight and the levels
of commercial production. Conversely, selections AN07,004,57 and AN07,005,58
are located in the upper left-hand quadrant of the plot, located towards the vectors
of fruit weight and commercial production. All of the other offspring are located
between the positions of the parents, which are in the central area of the plot. The
male parents (cv. Alba and Clery, selection Romina) are situated on the left-hand
side of the plot, again, corresponding to the vectors of fruit weight and commercial
production, while the common female parent, the F1 selection AN94,414,52, is
located on the right-hand side of the plot, corresponding to the vectors of fruit TAC,
TPH and ACY. This confirms the offspring improvement in fruit size and plant
production with regard to the FVG mid-parent [22].
Figure 2
Principal component analysis.

The PCA bi-plots of the BC1 families for fruit sensorial parameters (SS content, TA)
combined with fruit weight and commercial production variables (Figure 3a) locates
most of the selections towards the SS content vector instead of the fruit weight and
commercial production vectors, revealing an improvement in the sensorial quality of
the progenies, with respect to their parents (Figure 3b). With all three of the families
more towards the SS content vector, AN07,003,60 is the selection that is most
oriented towards the fruit SS content vector.

Figure 3
Principal component analysis.

In the PCA bi-plots of the BC1 families for Chroma index, fruit ACY and TPH (Figure
4a, vector distributions), most of the offspring and parents are distributed in the left-
hand central area of the plot, where the Chroma index vector is located (Figure 4b,
case distributions). The only exceptions are selections AN07,003,60 and
AN07,004,54: AN07,003,60 is located on the right-hand side of the plot, towards the
vector of the fruit ACY variable, and AN07,004,54 is located in the upper right-hand
quadrant, in the direction of the fruit TPH.

Figure 4
Principal component analysis.

These data on the fruit sensorial and nutritional parameters demonstrate the
importance of using the F1 FVG parent in the breeding, as it positively influences
the fruit nutritional characteristics of the progeny seedlings.
Selections of the BC2 Populations

The ANOVA analysis of the BC2 population in 2010 showed significant differences
for fruit TA, Chroma index, L* grade, TAC and TPH, while there were no significant
differences for fruit SS content, firmness and ACY (Table 1).

The significant highest mean values for fruit TA, Chroma index, TAC and TPH were
seen for AN07,215, while the significant highest values for SS content and L* were
in AN07,006 (Table 2). Fruit firmness and ACY did not show significant differences
among the BC2 populations; however, the lowest fruit TA, Chroma, L* and TAC were
seen in AN07,216 (Table 2).

The highest means for fruit weight and commercial production were seen in the
AN07,216 and AN07,006 populations, although these also had the highest levels of
misshapen and undersized fruit per plant (Table 3). The lowest mean commercial
production was seen for AN07,215, together with the lowest level of misshapen and
undersized fruit production per plant, although also with the highest value for rotten
fruit production (Table 3). Few misshapen fruit were seen in the BC2 populations, in
comparison with the intra-species cross population (Table 3).

The Pearson matrix for the fruit of the BC2 populations showed a number of negative
correlations among the various sensorial and nutritional parameters (Table 4).
Indeed, significant negative correlations were seen between fruit firmness and fruit
SS content (−0.22), TA (−0.24), L* grade (−0.26) and fruit antioxidant capacity
(−0.21). Most significantly, negative correlations were seen between fruit ACY and
Chroma index (−0.62) and L* grade (−0.65) (Table 4).

Significant positive correlations were seen between TA and fruit TAC (0.45), TA and
TPH (0.44), fruit TAC and TPH (0.47), and Chroma index and L* grade (0.82); also,
stronger significant positive correlations were seen between SS content and TAC
(0.29), and between SS content and TPH (0.29). Finally, firmness and ACY were
positively correlated (0.24) (Table 4).

The results from the correlation matrix of the BC2 families show that in moving ahead
by one back-cross generation there is a substantial improvement, seen as favorable
correlations between sensorial and nutritional parameters. Indeed, in the BC2
populations, high fruit TAC and TPH content correspond to an increase in the fruit
SS content and TA, and an increase in TAC corresponds to an increase in fruit
brightness (L*) (Tables 2, ,4).4). There are still some negative correlations, mainly
for fruit firmness, which is negatively correlated with SS content, TA, L* grade and
fruit TAC. So, highlighting these effects, a second back-cross generation can have
positive effects for the improvement of fruit sensorial (SS content, TA, and L*) and
nutritional (TAC) quality, but not of fruit firmness (Table 4). Negative correlations also
remain between the fruit color parameters and ACY; indeed, in the BC2 families,
high fruit ACY corresponds to a darker and more opaque fruit (Tables 2, ,44).
The PCA bi-plots of fruit TAC, TPH, ACY, fruit weight and plant commercial
production for the offspring and parents of these BC2 back-cross combinations (male
parents: Clery, Romina and AN03,338,56; female parent: AN00,239,55) (Figure 5a)
show a unique distribution for each cross combination; indeed, the seedlings of
AN07,215 were almost all located in the lower right-hand quadrant of the plot, where
the vectors for the fruit TAC and TPH variables are seen, in particular for the
selections AN07,215,55 and AN07,215,57 (Figure 5b). The other two cross
combinations are located mainly on the mid left-hand side of the plot (Figure 5b),
where the vector for plant commercial production is located (Figure 5a), showing
their lower affinity for the nutritional parameters. The selection AN07,216,60 here
deserves special attention, as it is located in the upper right-hand quadrant where
the vector for fruit ACY is positioned (Figure 4a), together with the female parent of
the BC1 selection AN00,239,55.

Figure 5
Principal component analysis.

The PCA plots for fruit sensorial parameters (SS content, TA) and productive
parameters (fruit weight, commercial production) (Figure 6a) show wider
distributions of the offspring for the cases plot (Figure 6b), in the offspring that are
characterized by higher sensorial parameters and the offspring that are
characterized by higher productive parameters.

Figure 6
Principal component analysis.

The selections AN07,215,55 and AN07,215,53 are in the lower left-hand quadrant
of the plot (Figure 6b), where the SS content and TA vectors are located (Figure 6a).
The selections AN07,006,62 and AN07,216,55 and the male parents (Clery and
AN03,338,56) (Figure 6b) are located in the lower right-hand quadrant of the plot,
showing more affinity with the fruit weight and commercial production vectors (Figure
6a).

The other selections of BC2 are positioned in the central area of the plot, or in the
opposite quadrants from where the vectors are located. From this distribution, it can
be noted that most of the BC2 offspring are more oriented towards the commercial
production and fruit weight vectors, in comparison with the BC1 parent AN00,239,55
(Figure 6b). This confirms the greater improvement of this trait in the BC2 genotypes
than in their FVG parent [22].

The PCA plot for the fruit Chroma index, TPH and ACY variables (Figure 7a) shows
the distribution of the offspring (Figure 7b) around the central area of the plot, with
only the selection AN07,215,55 differing from the others; indeed, it is located in the
lower left-hand quadrant of the plot (Figure 7b), where the TPH vector is located.
The location of the female parent AN00,239,55 is in the lower right-hand quadrant
and the location of the male parent AN03,338,56 is in mid-left-hand area of the plot
(Figure 7b), following the orientations of the ACY vector and the Chroma index
vector, respectively (Figure 7b).

Figure 7
Principal component analysis.

The offspring AN07,215,55 showed high affinity for fruit nutritional quality (Figure
5b), and the PCA bi-plot also revealed its high fruit sensorial quality (Figure 6b),
which indicates a good equilibrium between fruit SS content and TA. Also, the other
offspring of the AN07,215 back-cross population showed a good combination of fruit
nutritional (Figure 5b) and sensorial (Figure 6b) qualities, with respect to the other
BC2 populations. These selections can be considered as an important improvement
from the original FVG inter-species cross combination, and to be very close to the
characteristics now requested for a commercial variety.

Selections from the Intra-species Crosses

The selections from the intra-species cross populations evaluated in 2010 differed
significantly for fruit TA, Chroma index, brightness (L*), TPH and ACY, while the SS
content, fruit firmness and TAC remained more stable within the populations (Table
1) and showed no significant differences.

The intra-species cross population AN07,007 (Romina × AN03,338,56) showed the

most significant values for fruit TA, Chroma index, brightness (L*) and fruit TPH,
while the highest fruit firmness and ACY were seen for the AN07,009 intra-species
cross combination (Romina × Clery) (Table 2).

For the production parameters, the highest fruit weight was seen for AN07,009,
which also had the lowest values for undersized and misshapen fruit per plant (Table
3), while AN07,007 showed the highest values of commercial production and total
production (Table 3). The rotten fruit per plant were remarkable for both of these
cross combinations, as this was almost half of that seen for the commercial
production (Table 3).

For the Pearson’s correlations, the Chroma index showed a high level of negative
correlation with firmness (−0.48) and fruit ACY (−0.64), and even the L* grade
showed high negative correlation with fruit firmness (−0.44) and ACY (−0.53).
Finally, negative correlation was also seen between fruit TA and ACY (−0.22) (Table
4).

Positive correlations were seen for fruit TA and Chroma index (0.28), TAC (0.27)
and TPH (0.34). Fruit TPH also showed positive correlation with Chroma index (0.23)
and L* grade (0.30), and high positive correlation with TAC (0.83). Fruit ACY also
shown positive correlation with firmness (0.28) and high correlation with fruit TAC
(0.37). Finally, fruit SS content showed positive correlation with L* grade (0.26)
(Table 4).

For the intra-species cross populations, positive correlations were seen between the
fruit sensorial TA and Chroma index parameters and the TAC and TPH nutritional
parameters (Table 4). AN07,007 had the highest TA, Chroma index, TPH and TAC,
even if the TAC was not significantly different from that detected for fruit of the
AN07,009 family (Table 2). In addition, positive correlation was seen for TPH and
both of the colorimetric parameters, for both families, even if the fruits of the
AN07,007 cross combination showed significantly higher TPH content, Chroma
index and L* grade, in comparison with the fruit of AN07,009 (Table 2). In both of
these cross combinations, Romina was used as the common male parent (Table 5).
Romina is a new variety that was released by the Università Politecnica delle
Marche, and it is characterized by improved fruit sensorial and nutritional quality. It
was crossed with the selection AN03,338,56 from the University breeding program,
which was chosen because of its interesting sensorial and nutritional parameters,
and with Clery, a well-know commercial variety on the European Union market. The
differences detected for these two intra-species families reveal the influence of the
Romina genetic background in the production of a high combination of fruit sensorial
and nutritional parameters, and indicate that these values can be improved even
more if further genetic material identified for additional improved fruit sensorial and
nutritional values can be used.

Table 5
The families with the corresponding cross and back-cross (BC) combination,
as used in the present study.

These two intra-species cross populations also showed negative correlations

between both of the color parameters, Chroma index and L* grade, and the fruit
firmness and ACY, which means that the high color parameters correspond to a
softer fruit with lower ACY (Table 4). Indeed, particularly for the fruits of the
AN07,007 family, the significantly higher values of both of the color parameters
correspond to significantly lower fruit firmness and ACY.

The PCA bi-plot of fruit nutritional parameters showed a clear distribution of the
populations on the plot that highlights the selection AN07,007,55 as having the high
values of the fruit nutritional parameters (TAC and TPH).

The PCA bi-plot of fruit weight, plant commercial production, and fruit nutritional
parameters (TAC, TPH and ACY) for the offspring and parents of the intra-species
cross combinations (Figure 8a) showed distinct distributions for the two populations
(Figure 8b). Indeed, the offspring of AN07,007 were mainly positioned in the central-
right-hand area of the plot, where the TPH and TAC vectors are located, while only
AN07,007,57 is located in the lower right-hand quadrant. The offspring of AN07,009
are mainly located in the lower quadrant (Figure 8b), where the vectors of the fruit
weight and commercial production variables are located (Figure 8a). The female
parents, Clery and AN03,338,56, are located on the right-hand side of the plot
(Figure 8b), where the TPH and TAC vectors are located (Figure 8a), while the
common male parent, Romina, is located on the lower part of the plot, where the
vectors for fruit weight, commercial production and ACY are located.

Figure 8
Principal component analysis.

The PCA bi-plot for fruit weight and commercial production combined with the
sensorial parameters of SS content and TA (Figure 9a) distributes the offspring and
parents widely in all four of the quadrants of the plot (Figure 9b). This distribution
makes the different affinities of the two cross combinations towards the TA sensorial
parameter clear. Indeed, the fruit of AN07,007 are located mainly towards the
sensorial parameters vector TA (Figure 9b), while the fruit of the AN07,009 cross
combination show a distribution that is mainly in the lower quadrants, on the opposite
side of the TA vectors. From the distribution of the offspring, it can be noted that
AN07,009,54, AN07,009,56 and AN07,009,60 are located in the lower left-hand
quadrant, just in the opposite direction of the vectors that identify the TA variable,
and also AN07,009,53, AN07,009,52 and AN07,009,51 are located on the lower part
of the plot (Figure 9b), more oriented towards the SS content vector (Figure 9a). Half
of the offspring of AN07,007, together with the female parents, AN03,338,56 and
Clery, are located in the upper part of the plot (Figure 9b), between the vectors for
the production variables and the fruit TA (Figure 9a). The offspring AN07,007,57 and
AN07,009,57, together with the male parent (Romina), are located on the left-hand
side of the plot (Figure 9b), where the production variables are located (Figure 9a).
The selections AN07,007,53 and AN07,009,51, are located towards the SS content
vector.

Figure 9
Principal component analysis.

The PCA bi-plot for the fruit Chroma index, TPH and ACY parameters (Figure 10a)
shows clear-cut distributions of the parents and offspring (Figure 10b). Indeed, the
female parents together with most of the AN07,007 offspring are located on the left-
hand side of the plot (Figure 10b), where the Chroma vector is located (Figure 10a),
while the male parents and most of the AN07,009 offspring are on the right-hand
side of the plot, which shows the separation between these two populations. The
AN07,007 family is mainly located in the mid-upper part of the plot (Figure 10b), just
on the opposite side of the TPH vector (Figure 10a).

Figure 10
Principal component analysis.

The selection AN07,007,55 is located between the ACY and TPH vectors, while the
selection AN07,009,56 is located towards the ACY vector, and the other offspring
are not influenced by these variables (Figure 10b).

Go to:

Discussion

To date, strawberry breeding programs have been mainly focused on the

improvement of agronomic and commercial traits [23], although more recently the
major breeding targets have shifted more towards the sensorial and nutritional
qualities of the strawberry fruit [24]–[25]. Very recent studies that have focused on
the sensorial [26]–[27] and nutritional [4]–[7], [13]–[15], [24]–[25], [27] parameters in
strawberry fruit have given more information and have considered further the
concepts relating to their origins in plants and their importance for human health.

The major aim of the present study was to compare different breeding approaches
to increase the narrow genetic base of the cultivated strawberry, and in particular, to
increase the strawberry fruit sensorial and nutritional values. To achieve this, the
contributions of Fragaria virginiana ssp. glauca, a wild octoploid strawberry, were
tested in inter-species crosses and through subsequent back-cross cycles. The use
of wild strawberry in this breeding program, which is aimed at producing new
commercial varieties, is motivated by the need to increase the genetic variability of
Fragaria × ananassa [28]–[29], because of the breeding selection process that has
developed since the domestication of the first strawberry hybrids. The usefulness of
FVG as an original source of fruit nutritional quality has already been reported [14];
however, long-term back-cross generations are necessary to provide new genotypes
with high commercial value [5].

From the study of the inter-species back-cross populations, the offspring of BC1 and
BC2 from FVG showed overall high fruit sensorial and nutritional qualities, as seen
in the original F1 inter-species crosses [5]. These were combined with improvements
in the other important commercial traits (mostly for fruit weight and commercial
production), which were lost in the F1 lines. Indeed, the lack of association between
the productivity and quality traits is well represented by the different PCA plots
(Figures 2, ,3,3, ,55 and and6).6). These results highlight the independent genetic
control of these important plant and fruit traits.

The use of FVG was also particularly efficient for the creation of greater variability,
as seen with the production of a relatively high number of transgressive segregants.
This result is considered to be particularly relevant in comparison with what was
achieved from the F. × ananassa intra-species crosses included in the present study
and in other studies [24]. Indeed, the FVG-derived populations had a higher number
of new transgressive segregants, and also higher values for each of the sensorial
and nutritional traits. This achievement is probably the result of an increased number
of alleles that can interact epistatically [30], and also dominance/overdominance
effects might have contributed to the presence of transgressive segregants for fruit
quality [31].

The increase in fruit ACY is reflected phenotypically by the darker red coloration of
the berry [24]. Fruit color is controlled at multiple levels that involve transcription
factors and structural genes [32]–[33]. Quantitative trait loci for anthocyanin and fruit
color have been evaluated for the flavanone 3-hydroxylase, chalcone isomerase,
flavonol synthetase and dihydroflavonol reductase genes, which are involved in
anthocyanin biosynthesis, but no gene co-location with fruit color or ACY was seen
[25]. Improvements in the nutritional values in the BC1 and BC2 FVG populations,
which were due in particular to increased ACY (Table 6), were associated with a
decrease in the Chroma index. This was supported by the correlation matrix (Table
4), where these parameters were highly negatively correlated, both in the BC and
intra-species offspring populations (Table 4) [24]. While for the intra-species
populations, these were the only ones where a high negative correlation between
colorimetric parameters and phenol content was confirmed [5], [15]. The PCA plots
for the BC1 populations (Figure 4) show a high number of transgressive segregants
for fruit color and ACY, while the BC2 populations do not show segregants. Hence,
these data confirm the concept that fruit color is controlled by epistatic and
dominance/overdominance effects [30]–[31], as seen for the nutritional compounds.

Table 6
Fruit sensorial and nutritional parameters of the BC1 and BC2 back-cross
combinations and the intra-species cross combinations (evaluated in 2010 on
clonal population of selected seedlings).

The importance of F. virginiana, and in particular ssp. glauca, for increases in

pathogen resistance of the hybrid progenies was evaluated in a breeding program
that was designed to widen the genetic base of the strawberry germplasm [22]. The
BC1 and BC2 populations showed marked reductions in rotten fruit production in
comparison to the F. × ananassa populations. This lower rotten fruit production was
accompanied by a higher content of bioactive compounds, and specifically of
anthocyanins and phenols, which have active roles in the protection of plants and
fruit against pathogens [34]–[36]. Hence, the increases in the bioactive compounds
in the BC1 and BC2 populations, and the data showing lower levels of rotten fruit
production, can be considered as confirmation that the increased content of these
bioactive compounds can have effects on increased tolerance to the major
strawberry diseases [22]. Therefore, these achievements underline the importance
of the wild genetic resources as sources of different important genes and the need
for continued evaluation of back-cross generations for the creation of new genotypes
with both improved nutritional compound content and disease resistance. Moreover,
studies performed by Tulipani et al. [37] also showed that wild species can contribute
to increases in not only interesting nutritional features in cultivated strawberry, but
also in non-nutritional compounds, which might have implications for human
consumption. From these findings, it is clear that intra-species crosses induce low
increases in allergens contents, while the fruit from the F1 and BC1 selections had
much higher contents of allergens.
In general, the performances of the BC1 and BC2 FVG populations can be further
improved by additional back-cross generations, so as to be closer to the overall
commercial quality that is required for any new commercial variety. By the third back-
cross generation, the plant and fruit commercial standards can be very near to those
requested for commercial cultivars [8].

The results for the F. × ananassa intra-species crosses show that there is a genetic
background in the cultivated species that can induce an increase in the fruit sensorial
and nutritional qualities. Even if these are of lower value in comparison with the FVG
back-cross program, they are already combined with a high standard of commercial
performance. However, this positive increase is strictly related to the genetic
background of the selected parents. Indeed, the parents used in the cross
combinations were chosen because they had already been identified as having high
fruit nutritional quality in comparison with other cultivars [5], [12], [15].

Conclusions

The present study has demonstrated that these two types of combination programs
(inter-species back-crosses and intra-species crosses) can be used for the
improvement of strawberry nutritional quality, although the success of the program
is strictly related to the combining attitude of the different parents. For this reason,
this type of experiment can be considered important for the production of new
genotypes with improved sensorial and nutritional qualities, and for the identification
of cultivars and selections that can perform in the same way as the most useful
parents. These can be used in strawberry breeding programs that are aimed at the
production of new improved varieties, for the better combination of the agronomic
and overall fruit quality traits.

These evaluations carried out on the genetic pools that originated from the F.
virginiana ssp. glauca inter-species back-cross and the F. × ananassa intra-species
cross confirm the importance of using FVG, or genotypes with an FVG genetic
background, in breeding programs aimed at increasing the fruit nutritional (TAC, TPH
and ACY) and sensorial (SS content, TA, Chroma, L*, firmness) qualities. Indeed,
both the BC1 and BC2 populations showed these characteristics, thus providing
useful improvements in the fruit nutritional and sensorial qualities combined with
agronomic standards that are closer to the those requested at the commercial level
with respect to their respective FVG hybrid parents.

The results obtained in the present study show that two sequential back-crossing
generations (BC1 and BC2) from an inter-species hybrid with a wild genotype of F.
virginiana ssp. glauca can improve sensorial and nutritional traits. These data
confirm that at least two back-cross generations are needed to be close to the
agronomic values requested at the commercial level [8].

This study has created new genotypes with the FVG genetic background introduced
by inter-species crosses, thus increasing the genetic variability in cultivated
strawberry, particularly for their specific nutritional traits [5]. This helps to counteract
the narrow germplasm base that is used for breeding, which has caused deleterious
effects of inbreeding and genetic vulnerability [27]. From our data, it is quite clear
which parameters have to be increased in a back-crossing program. Fruit firmness,
size, acidity and color are the main traits that can be recovered in the back-crossing
generations. Clearly, the high positive values detected for some of the parameters
in the F1 genotypes (TAC, polyphenols and folate) are reduced in the back-crossing
generations, although they always remain higher than those in the commercial F. ×
ananassa genotypes. These findings suggest that controlled crosses of wild species
with cultivated strawberry will contribute to introgression of interesting nutritional
features; however, this can also lead to an increase in unfavorable compounds that
might have health implications for humans, and that thus need specific evaluation.

Successive breeding studies using wild Fragaria octoploid species can introduce
new genetic variability for nutritional, flavor and disease-resistance traits, and can
contribute to our understanding of their heritability ratios and the hypothetical
correlations with nutritional parameters [27]. Indeed, further studies on wild
genotypes have evaluated the importance of FVG [22], and also of other wild
genotypes with different ploidy levels [38]–[40], in terms of their flavor and resistance
parameters, using gene expressions studies [41].

Improvements in sensorial and nutritional traits can also be achieved by the

programming of F. × ananassa intra-species crosses and the production of progeny
with productivity traits that are closer to those of the commercial cultivars, even if
these improvements are fewer in number and quality with respect to the FVG back-
cross populations.

Go to:

Materials and Methods

Cross Combinations and Seedling Evaluations

A breeding program was carried out to obtain eight half-sib cross combinations
originated using the following six different strawberry inter-species back-cross
combinations and two intra-species cross combinations (Table 5, Figure 1):

a. Three half-sib cross combinations originated as back-cross populations (BC1)

with the F1 selection AN94,414,52, and an inter-species cross of F. ×
ananassa (Don) × F. virginiana glauca (FVG22) as the common parent;
b. Three half-sib cross combinations originated as back-cross populations (BC2)
with AN00,239,55, a BC1 selection derived from the back-crossing of
selection F1 AN94,414,52 with 91,143,5 (F. × ananassa advanced selection)
as the common parent;
c. Two half-sib cross combinations originated by an intra-species cross with the
selection Romina as the common parent.
Seedlings from all of these cross combinations were planted in 2008 in non-
fumigated soil at the ‘P. Rosati’ University experimental farm, Agugliano (Ancona,
Italy), and grown under open-field conditions using the plastic hill culture production
system. For each cross combination of about 50 seedlings, seven to 11 elite
seedlings were selected, based on a subjective phenotypic evaluation of the major
plant vegetative and fruit sensorial traits, which was performed at the stage of fruit
ripening. All of the seedlings of the different families were identified by code number,
including AN as the Ancona site of the crossing, followed by 07 as the year of the
crossing, with the code for the type of crossing, followed by progressive numbering
(starting from 51) for each selected seedling. From this subjective selection, a total
of 80 elite seedlings were identified and propagated by runners to create clonal
population of the selected seedling. After the 2009 evaluation, 78 elite seedlings (two
were lost) were planted as fresh plants in July 2009, in single plots of six plants each.
They were then harvested and evaluated in May 2010, as selections from clonal
population of selected seedlings.

This second evaluation was performed under the same open-field conditions in non-
fumigated soil at the ‘P. Rosati’ University experimental farm, Agugliano (Ancona,
Italy), again using the plastic hill culture production system. In the same field, under
the same conditions, the corresponding parents of each of the cross combinations
were also grown. Fruit samples of the fully red berries were harvested at the second,
third and fourth main pickings. The evaluations took into account plant productivity
(fruit weight and total plant production), and fruit nutritional (fruit TAC, phenols and
ACY) and sensorial (fruit firmness, color, SS content and TA) parameters.

Chemicals

The 99% methanol (ACS-ISO) for the analysis, and the bromothymol blue were
purchased from Carlo Erba Reagenti (Milan, Italy). Folin-Ciocalteu’s reagent
anhydrous, sodium carbonate, potassium chloride, sodium acetate, chloridric acid,
glacial acetic acid, dihydrogen potassium phosphate, dipotassium hydrogen
phosphate, 2,2′-azino-bis (3-ethylbenzothiazoline-6-sulphonic acid) diammonium
salt (ABTS•+), 6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic acid (Trolox),
potassium persulfate, 3,4,5-trihydroxybenzoic acid (gallic acid), and sodium
hydroxide, were purchased from Sigma-Aldrich (Milan, Italy).

Strawberry Production Parameters

In 2010, the strawberry plant production was evaluated in the single plot of all of the
selections at each harvest, by evaluating the following parameters:

a. Total fruit production (g/plant).

b. Commercial production: total weight (g/plant) of ripe fruit with a commercial
diameter (Ø≤0.22 mm) and without injuries.
c. Fruit weight: average weight (g) of 20 commercial fruits from each harvest.
d. Undersized fruit: total weight (g/plant) of ripe fruits with diameter Ø ≥0.22 mm,
without injuries.
 e. Misshapen fruit: total weight (g/plant) of misshapen fruits.
f. Rotten fruit: total weight (g/plant) of rotten fruits.

Strawberry Instrumental Sensorial Parameters

Fruit sensorial quality of the selections and the parents was analyzed on the
harvesting day, at each harvest (2nd, 3rd, 4th), on 10 commercial fruits, taking in
account the following parameters:

a. SS content: determined using a hand-held refractometer (ATAGO), with

results expressed as °Brix.
b. TA: determined from 10 mL juice diluted with distilled water (1/2, v/v) and
titrated with 0.1 N NaOH solution to pH 8.2, and expressed as mEQ of NaOH
per 100 g fruit (fresh weight; FW).
c. Fruit color: determined as the Chroma index using a Minolta Chromameter
CR 400, for two sides of 10 sound, ripe, undamaged and uniform fruits. The
instruments included three parameters L* (luminescence), a* (red tone), b*
(yellow tone). The Chroma index was evaluated from a and b [(a* 2+b*2)½],
with a higher Chroma index representing pale fruit and a low Chroma index
representing dark fruit.
d. Fruit firmness: measured using a penetrometer 327 (Effegì, Ravenna, Italy),
with the results expressed in g.

Strawberry Nutritional Parameters

Fruit extraction method

From a pool of fruit for each genotype harvested in 2010 and stored at −20°C, 10
ripe commercial fruits were sampled, and from each fruit two slices from opposite
sides were cut and milled into small pieces. Ten gram of this blend was weighed and
placed into a tube for extraction with methanol (1∶4, fruit: methanol, w/w), including
two subsequent steps. The first step consisted of homogenization with an Ultraturrax
T25 homogenizer, for a fruit blend in 20 mL methanol (Janke and Kunkel, IKA
Labortechnik, Staufen, Germany). The homogenized suspension was continuously
agitated for 30 min in the dark. The suspension was centrifuged at 4,500×g for 10
min (Centrifuge Rotofix32, Hettich Zentrifugen, Tuttlingen, Germany), and the
supernatant was collected and stored in three amber vials, of 1 mL each. For a more
complete and extensive extraction, the pellet of the fruit was extracted a second time
by adding another 20 mL methanol and repeating the procedure described above.
The second supernatant was collected and added to the first one, always using 1
mL for each of the 3 amber vials per sample, and then stored at −20°C.

TAC

The fruit extract from each genotype was used to analyze the fruit TAC using the
ABTS assay, according to a previously validated procedure [42]–[43]. ABTS is a
chromogen and a colorless substance; it changes into its colored monocationic
radical form (ABTS•+) with exposure to an oxidative agent. Addition of antioxidants
reduces ABTS•+ into its colorless form. The extent of decolorization as a percentage
of inhibition of ABTS•+ was determined as a function of the concentration and was
calculated relative to the reactivity of Trolox, a water-soluble vitamin E analog. The
antioxidant activity is expressed as mg Trolox equivalent per kg fresh pulp weight.
The calibration was calculated as the linear regression from the dose-response of
the Trolox standard.

TPH

The fruit TPH was evaluated on fruit extracts using the Folin-Ciocaltou’s reagent
method [44], with gallic acid as the standard for the calibration curve. Briefly, a glass
test-tube was filled with 7.0 mL water. Afterwards, 1 mL of the diluted sample (1∶20)
was added, followed by the addition of 500 µL Folin-Ciocalteu-Reagent and
vortexing. After 3 min, 1.5 mL sodium carbonate (0.53 mol/L) was added, and the
tube was mixed once more and then stored in the dark for 60 min. The absorbance
of the sample was measured at 760 nm after exactly 60 min. The data were
calculated and expressed as mg gallic acid per kg fresh fruit, using a standard curve
with the range of standard values of 10 to 50 mg gallic acid/L.

ACY

Total fruit ACY was measured on the fruit extracts using the pH differential shift
method [45]. This assay is based on the anthocyanins characteristic of a change in
intensity of hue depending on the pH shift. Briefly, the samples were diluted to a ratio
of 1∶10 with potassium chloride (pH 1.00) and with sodium acetate (pH 4.50), and
then the corresponding maximum absorbance for both of the solutions was
measured at λ = 500 nm and at λ = 700 nm, respectively. The data were expressed
as mg pelargonidin-3-glucoside (the anthocyanin more representative in strawberry)
per kg fresh weight (mg Pel-3-Glu/kg FW).

Experimental Trial and Statistical Analysis

In the 2010 harvest season, the fruit sensorial (SS content, TA, firmness, Chroma,
L*) and nutritional (TAC, TPH, ACY) parameters were analyzed in triplicate for each
clonal population of selected seedlings identified in 2009, and for their corresponding
parents.

All of these data were subjected to the two-way nested analysis of variance,
MANOVA test. The analysis of variance was performed for single cross types and
single cross combinations as random effects. The data on fruit sensorial and
nutritional parameters are reported as means±standard error (SE). Multiple
comparisons were calculated according to the SNK test and were considered
significant at p≤0.05.
Correlations among the sensorial (SS content, TA, firmness, Chroma, L*) and
nutritional (TAC, TPH and ACY) parameters were analyzed using Pearson’s
correlations. PCA was also used to evaluate the levels of association among the
various production, sensorial and nutritional parameters of the selections that
originated from the different cross combinations and their corresponding parents,
calculated on a genotype-mean basis. Based on the theoretical arguments of PCA
described by [46], the significant factor loading values ≥0.7 were used to identify the
most important variables and observations in each dimension (PC). The factor
loading values are the correlations of each variable with the PC. They are
represented as vectors (positions) in the space represented by the axes of the PCA
bi-plots (Figures 2a to 10a). In the graphs, the variables (Figures 2a to 10a) and
observations (Figures 2b to 10b) that are closest to each other in the same geometric
plane of the bi-plot are considered as interrelated, and as a result they are distant
from the variables and observations to which they are not related, or are negatively
related. The greater the distance of a vector from the origin of the axis, the higher
the correlation of the variable with the PC represented in that dimension (axis). All
of the analyses were performed using STATISTICA (Statsoft Inc., Tulsa, OK, USA).

Go to:

Acknowledgments

Our thanks are expressed to Dr. Enrico Muzzi for assistance in planning and
discussing statistical analyses.

Go to:

Funding Statement

The research work was carried out thanks to the project on ‘Valorization of Red Fruit
Genetic Resources’ supported by the Italian Ministry of Environment and for the last
analyses to the EU FP7 EUBerry project No. 265942. The funders had no role in
study design, data collection and analysis, decision to publish, or preparation of the
manuscript.

Go to:

References

1. Meyers KJ, Watkins CB, Pritts MP, Liu RH (2003) Antioxidant and antiproliferative
activities of strawberries. J Agric Food Chem 51: 6887–6892 [PubMed]
2. Prior RL, Cao G, Martin A, Sofic N, McEwen J, et al. (1998) Antioxidant capacity
as influenced by total phenolic and ACY, maturity and variety of Vaccinium species.
J Agric Food Chem 46: 2686–2693
3. Proteggente AR, Pannala AS, Pagana G, Van Buren L, Wagner E, et al. (2002)
The antioxidant activity of regular consumed fruit and vegetables reflects their
phenolic and vitamin C composition. Free Radic Res 36: 217–233 [PubMed]
4. Shiow Y, Wang L, Kim S (2007) Antioxidant capacity and flavonoid content in wild
strawberries. J Amer Soc of Hort Sci 132: 629–637
5. Tulipani S, Mezzetti B, Capocasa F, Bompadre S, Beekwilder J, et al. (2008)
Antioxidants, phenolic compounds, and nutritional quality of different strawberry
genotypes. J Agric Food Chem 2008 56: 696–704 [PubMed]
6. Battino M, Beekwilder J, Denoyes-Rothan B, Laimer M, McDougall JG, et al.
(2009) Bioactivities of berries relevant to human health. Nutr Rev 67: 145–150
7. Tulipani S, Romandini S, Busco F, Bompadre S, Mezzetti B, et al. (2009)
Ascorbate, not urate, modulates the plasma antioxidant capacity after strawberry
intake. Food Chem 117: 181–188
8. Bringhurst RS & Voth V (1978) Origin and evolutionary potentiality of the day-
neutral trait in octoploid. Fragaria Genetics 90: 510
9. Della Penna D (2001) Plant metabolic engineering. Plant Physiology 125: 160–
163 [PMC free article] [PubMed]
10. Azodanlou R, Darbellay C, Luisier J, Villettaz J, Amadò R (2003) Quality
assessment of strawberry (Fragaria species). J Agric Food Chem 51: 715–721
[PubMed]
11. Olsson ME, Ekvall J, Gustavsson K, Nilsson J, Pillai D (2004) Antioxidants, low
molecular weight carbohydrates, and TAC in strawberry (Fragaria x ananassa). J
Agric Food Chem 52: 2490–2498 [PubMed]
12. Scalzo J, Politi A, Mezzetti B, Battino M (2005) Plant genotype affects TAC and
phenolic contents in fruit. Nutrition 21: 207–213 [PubMed]
13. Wang SY, Zheng W, Galletta GJ (2002) Cultural system affects fruit quality and
antioxidant capacity in strawberries. J Agric Food Chem 50: 6534–6542 [PubMed]
14. Wang SY, Lewers KS (2007) Antioxidant capacity and flavonoid content in wild
strawberries. J Amer Soc Hort Sci 132: 629–637
15. Capocasa F, Diamanti J, Tulipani S, Battino M, Mezzetti B (2008) Breeding
strawberry (Fragaria X ananassa Duch) to increase fruit nutritional quality. Bio
Factors 34: 67–72 [PubMed]
16. Capocasa F, Bordi M, Mezzetti B (2009) Comparing frigo and fresh plant in not
fumigated and heavy soil: the response of ten strawberry genotypes. Proc VIth
International Strawberry Symposium Acta Hortic (ISHS) 842: 129–133
17. Connor AM, Luby JJ, Tong CBS, Finn CE, Hancock JF (2002) Variation and
heritability estimates for antioxidant activity, total phenolic content and ACY in
blueberry progenies. J Amer Soc Hort Sci 1: 82–88
18. Connor AM, Stephens MJ, Hall HK, Alspach PA (2005) Variation and
heritabilities of antioxidant activity and total phenolic content estimated from a red
raspberry factorial experiment. J Amer Soc Hort Sci 3: 130
19. Deighton N, Brennan R, Finn C, Davies HV (2000) Antioxidant properties of
domesticated and wild Rubus species. J. Sci. Food Agric. 80: 1307–1313
20. Wang SY, Lin HS (2000) Antioxidant activity in fruits and leaves of blackberry,
raspberry, and strawberry varies with cultivar and developmental stage. J Agric Food
Chem 48: 140–146 [PubMed]
21. Zhen W, Wang SY (2003) Oxygen radical absorbing capacity of phenolics in
blueberries, cranberries, chokeberries, and lingonberries. J Agric Food Chem 51:
502–509 [PubMed]
22. Hancock JF, Luby JJ, Dale A, Callow PW, Serce S, et al. (2002) Utilizing wild
Fragaria virginiana in strawberry cultivar development: inheritance of photoperiod
sensitivity, fruit size, gender, female fertility and disease resistance. Euphytica 126:
177–184
23. Darrow GM (1966) The strawberry: history, breeding and physiology. Holt,
Rinehart, and Winston, New York.
24. Lerceteau-Köhler E, Moing A, Guérin G, Renaud C, Petit A, et al. (in press)
Genetic dissection of fruit quality traits in the octoploid cultivated strawberry
highlights the role of homoeo-QTL in their control. Theor Appl Gen DOI:
10.1007/s00122-011-1769-3. [PMC free article] [PubMed]
25. Zorrilla-Fontanesi Y, Cabeza A, Domínguez P, Medina JJ, Valpuesta V, et al.
(2011) Quantitative trait loci and underlying candidate genes controlling agronomical
and fruit quality traits in octoploid strawberry (Fragaria × ananassa), Theor Appl Gen.
123: 755–778 [PubMed]
26. Kader AA (1991) Quality and its maintenance in relation to the post-harvest
physiology of strawberry. In: Luby JJ, Dale A (eds) Quality and its maintenance in
relation to the post-harvest physiology of strawberry. Timber press, Portland, 145–
152.
27. Aharoni A, Giri AP, Verstappen FWA, Bertea CM, Sevenier R, et al. (2004) Gain
and loss of fruit flavor compounds produced by wild and cultivated strawberry
species. Plant Cell 16: 3110–3131 [PMC free article] [PubMed]
28. Sjulin TM, Dale A (1987) Genetic diversity of North American strawberry
cultivars. J Am Soc Hortic Sci 112: 375–385
29. Hancock J, Luby J (1995) Adaptative zones and ancestry of the most important
North American strawberry cultivars. Fruit Varieties J 49: 85–90
30. Coelho CM, Wu S, Li YC, Hunter B, Dante RA, et al. (2007) Identification of
quantitative trait loci that affect endoreduplication in maize endosperm. Theor Appl
Genet 115: 1147–1162 [PubMed]
31. Flagel LE, Wendel JF (2009) Gene duplication and evolutionary novelty in plants.
New Phytol 183: 557–564 [PubMed]
32. Carbone F, Preuss A, De Vos RC, D’Amico E, Perrotta G, et al. (2009)
Developmental, genetic and environmental factors affect the expression of flavonoid
genes, enzymes and metabolites in strawberry fruits. Plant Cell Environ 32: 1117–
1131 [PubMed]
33. Allan AC, Hellens RP, Laing WA (2008) MYB transcription factors that colour our
fruit. Trends Plant Sci 13: 99–102 [PubMed]
34. Karjalainen R, Lehtinen A, Hietaniemi V, Pihlava JM, Jokinen K, et al. (2002)
Benzothiadiazole and glycine betaine treatments enhance phenolic compound
production in strawberry. Acta Hort. 567: 353–356
35. Hukkanen AT, Kokko HI, Buchala AJ, McDougall GJ, Stewart D, et al. (2007)
Benzothiadiazole induces the accumulation of phenolics and improves resistance to
powdery mildew in strawberries. J. Agric. Food Chem. 55: 1862–1870 [PubMed]
36. Hebert C, Charles MT, Gauthier L, Willemot C, Khanizadeh S, et al. (2002)
Strawberry proanthocyanidins: biochemical markers for Botrytis cinerea resistance
and shelf-life predictability. Acta Hort. 567: 659–661
37. Tulipani S, Marzban G, Herndl A, Laimer M, Mezzetti, et al (2011) Influence of
environmental and genetic factors on health-related compounds in strawberry. Food
Chem 124: 906–913
38. Amil-Ruiz F, Blanco-Portales R, Muñoz-Blanco J, Caballero JL (2011) The
strawberry plant defense mechanism: a molecular review. Plant Cell Physiol 52:
1873–1903 [PubMed]
39. Gooding HJ, Mcnicol RJ, Macintyre D (1981) Methods of screening strawberries
for resistance to Sphaerotheca macularis (wall ex Frier) and Phytophthora cactorum
(Leb. and Cohn). J. Hort. Sci. Biotechnol. 56: 239–245
40. Harland SC, King E (1957) Inheritance of mildew resistance in Fragaria with
special reference to cytoplasmatic effects. Heredity 11: 257
41. Maas J (1998) Compendium of Strawberry Diseases. American
Phytopathological Society, St Paul, MN.
42. Miller NJ, Rice-Evans C, Davis MJ (1993) A novel method for measuring
antioxidant capacity and its application to monitoring the antioxidant status in
premature neonates. Clinical Science 83: 407–412 [PubMed]
43. Re R, Pellegrini N, Proteggente A, Pannala A, Yang M (1999) Antioxidant activity
applying an improved ABTS radical cation decolorization assay. Free Radic Biol &
Med 26: 1231–1237 [PubMed]
44. Slinkard K, Singleton VL (1977) Total phenol analysis: automation and
comparision with manual methods. Amer J Enol and Vitic 28: 49–55
45. Giusti M, Wrolstad RE (2001) Characterization and Measurement of
Anthocyanins by UV-visible Spectroscopy. In Current Protocols in Food Analytical
Chemistry. Hoboken: John Wiley and Sons F1.2.1–F1.2.13.
46. Hair F, Anderson J, Tatham L, Black C (1998) Multivariate data analysis, 5th
edition. Prentice Hall.

Volume 80, Issue 5 May 2015 Pages R 901–R909

Examining Extrinsic Factors that Influence Product Acceptance: A Review

X.L Li, S.M. Jervis, M.A. Drake

Abstract

Drivers of liking (DOL) studies are useful for product development to formulate
acceptable products; however, DOL alone are insufficient for understanding why a
product is purchased and repurchased, which is ultimately the indication of a
successful product. Ultimately sensory attributes drive product success (that is,
repeat and continued purchase). However, ignoring the importance of extrinsic
factors may neglect the vital product attributes responsible for the initial purchase,
which may in turn, affect repeat purchase. The perception of sensory attributes
assessed by DOL is mitigated by external perceptions of quality. If the sensory
attributes do not deliver based upon the quality cues, the product will not be
acceptable. Four key extrinsic factors that affect DOL are the perceived satiety,
brand and labeling, price, and the emotional impact to decision making. In order to
more thoroughly understand what the DOL for a product is, these 4 product cues
should be considered in conjunction with sensory attribute perception to gain a
holistic understanding of product acceptance.

Practical Implication

Drivers of liking (DOL) studies gain insight into the sensory attributes responsible for
liking. Assessing only sensory attributes ignores any effect that product cues have
on enhancing attribute perception. Evaluation of extrinsic factors enhances
understanding of DOL and will aid food and beverage companies in positioning their
products in the best scenario for product success.

Introduction

Decision-making is made up of preferences that are dependent on the subjective

utilities of anticipated outcomes that are weighted by their probabilities (Angie and
others 2011). Individuals tend not to have clear organized preferences, but instead
assemble these preferences when put in a choice decision where the decision is
heavily influenced by the nature and context of the decision (Angie and others 2011).
When a consumer is choosing between bags of potato chips, their decision is
influenced by all of the information surrounding them at the moment of choice—in
the grocery store before tasting. These external influencing factors include the brand,
price, package size, label claims, fat content, perceived satiety, emotional response,
the decision making process, and so on. To only measure a product's liking by the
sensory attributes is to ignore all of the extrinsic attributes that influence the
purchase decision. Extrinsic attributes are attributes that are “related to the product
but do not form part of the physical product itself” (Ampuero and Vila 2006). To truly
understand why a product is liked, one needs to understand not only the perception
of the sensory attributes, but all of the related attributes that influence the decision-
making process. This review will discuss 4 extrinsic attributes that affect product
acceptance: perceived satiety and portion size, brand and labeling information, price
and willingness-to-pay (WTP), and finally the influence of emotions in the decision-
making process.
Drivers of Liking

A DOL study determines the sensory attributes of a food or beverage product

through descriptive analysis profiling, and models those attributes with consumer
overall liking scores to determine the degree of importance of each sensory attribute
on liking (also called external preference mapping). Multivariate analysis techniques
for DOL studies were recently reviewed by Bi (2012).

Drivers of liking (DOL) are defined as those attributes responsible for product
acceptance (Moskowitz 2001). Based on published literature, most DOL studies are
focused on sensory related attributes of taste, texture, and appearance of the
product, generally in the absence of packaging or price information. DOL studies can
be very useful for product developers for new product development before extrinsic
factors are a concern. Utilizing DOL techniques, researchers have identified key
ingredients, ingredient levels, and ideal sensory attributes to target in optimization of
various product categories (Table 1; Cardello and others 2003; Allgeyer and others
2010; Cruz and others 2011; Kreger and others 2012; Tomaschunas and others
2013).

Table 1. Summary of drivers of liking studies

Authors Subject studied Key DOL
Allgeyer and Yogurt drinks with pre Sweet and honey aromas, medium sweet
others (2010) and probiotics taste, and high viscosity.
Cheese odor, cheese flavor, consistency,
Bayarri and
Spreadable cheese creaminess, smoothness, cohesiveness,
others (2012)
and mouthcoating.
Females were more concerned about
Novel processing processing technologies than males.
technologies, Negative perceptions of foods after being
Cardello and (irradiation, pulsed informed about the processing technology
others (2003) electric fields, etc.)— used to make the pudding varied by
chocolate pudding as process. High levels of concern, genetic
the subject engineering, irradiation, pulsed X-ray and
bacteriocins.
Wet cheese sauces, beefy/broth, free fatty
Childs and others Wet and dry cheese acid, salty taste, oiliness in mastication,
(2009) sauces residual particle mouth coating. Salty
taste was the primary DOL for dry sauces.
UHT and HTST milk,
High cooked flavor. Preferred milks
Chung (2009) Korean female
available in Korea.
consumers
Cruz and others Taste and texture more important than
Probiotic yogurt
(2011) appearance.
Delgado and Nectarines and Sweetness was the main driver. Fruity
others (2013) peaches aromas and pit aromas were cluster
Authors Subject studied Key DOL
specific drivers. High ripe soluble solids
concentration and flesh firmness were
correlated to overall liking
Color, cooked/milky, whey, brothy flavors,
Drake and others and sour taste. Four consumer clusters
Mild Cheddar cheese
(2008) identified based upon preferences for
flavor and color.
Cluster preferences identified for brothy,
Drake and others Sharp Cheddar free fatty acid, and nutty flavors, salty and
(2009) cheese sour taste, young/mild flavors, and milk fat
flavor.
Herrera-Corredor
Corn tortillas, Mexican Chewiness, appearance, rollability, and
and others
consumers taste.
(2007)
Fermented soybean Sweet and MSG primary DOL. Salty taste,
Kim and others
paste (doenjang), meju (traditional Korean soy sauce), and
(2010)
Korean consumers fermented fish primary drivers of dislike.
Kim and others Orange juice, Korean
High intensity orange flavor.
(2013b) consumers
Krause and Butter flavors, salty taste, and spread
Butter
others (2007) ability.
Kreger and Extruded high protein
Protein type primary DOL.
others (2012) snack
Aroma and flavor attributes associated
with country of origin were the biggest
Imported and
Lee and others drivers. Californian olives were the most
domestic sliced black
(2012) liked which were the most familiar to the
ripe olives
consumers studied. Drivers of dislike were
alcohol and artificial fruity/floral.
Flavor and texture attributes more
Leksrisompong
important than appearance. Smooth
and others Sweet potatoes
texture, brown sugar, dried apricot flavor,
(2012)
and sweet taste.
Lemon lime
Leksrisompong
carbonated Mouthfeel for both regular and diet
and others
beverages, diet, and beverages.
(2013)
regular
Diacetyl (buttery), whey, milk fat, and
Liggett and
Swiss Cheese umami. Cabbage, cooked, and vinegar
others (2008)
were drivers of dislike.
Michon and Age effect reported with highest age
Jam filled cakes
others (2010) group having the highest liking. Gender
Authors Subject studied Key DOL
effect, males preferred cakes more so
than females.
Morais and Prebiotic gluten-free Texture and sweetness. Rubbery, hard,
others (2014) bread and chewy were drivers of dislike.
Firmness, herbaceous aroma, sour taste,
thick epicarp, and seeds were identified
Piombino and as primary drivers of liking. Drivers of
Tomatoes
others (2013) dislike were identified as mealiness,
diacetyl-like odor, fruity odor, and thin fruit
pulp.
Cooked and milk fat flavors.
Segmentation identified for dense, and
Shepard and
Sour cream highly cooked flavors. Across
others (2013)
segmentation high fat content and milk fat
attributes preferred.
Thompson and Chocolate milk, Milk fat flavor, cocoa flavor. Cooked and
others (2007) Hispanic consumers malty specific to cluster membership.
Tomaschunas
Vanilla custard Medium fat contents, fat related texture
and others
dessert attributes.
(2013)
Australian and Chinese consumers had
different DOL for red wine. Australian DOL
Red wine, Chinese
Williamson and primarily oak flavor and dark fruit. Chinese
and Australian
others (2012) consumers preferred brown color and had
consumers
stronger negative reactions to
astringency.

Understanding the target consumer population for a specific product is fundamental

for understanding pertinent DOL for purchase intent (PI). Intrinsic factors, which are
ones “that cannot be changed without altering the physical composition of the
product” (Ampuero and Vila 2006), and extrinsic cues influence food choice and
these cues are often influenced by culture (Kim and others 2013a). Several studies
have been conducted to determine the DOL of a product category with a specific
cultural group and have identified key DOL for products specific for the cultural group
studied (Table 1; Herrera-Corredor and others 2007; Thompson and others 2007;
Chung 2009; Kim and others 2010; Williamson and others 2012; Kim and others
2013a). Not only is the target purchasing population important to know when
evaluating DOL for a product category, but previous history of product exposure and
demographics (gender, age, etc.) will also influence DOL (Cardello and Sawyer
1992; Deliza and Macfie 1996; Solheim and Lawless 1996; Bayarri and others 2012).
Intrinsic and extrinsic cues can also be affected by age, gender, and consumption
habits. The eating habits of men and women, or the young versus the elderly can be
quite different. Several studies have reported different DOL for various product
categories by demographic breakdown and product familiarity (Table 1; Cardello and
others 2003; Drake and others 2008, 2009; Liggett and others 2008; Childs and
others 2009; Michon and others 2010; Bayarri and others 2012; Morais and others
2014).

There is no set of sensory attributes that will be important across all products. The
importance of flavor and texture attributes in DOL is product dependent (Table 1;
Herrera-Corredor and others 2007; Krause and others 2007; Liggett and others
2008; Kim and others 2010; Bayarri and others 2012; Lee and others 2012; Delgado
and others 2013; Leksrisompong and others 2013; Piombino and others 2013;
Morais and others 2014). This concept was explored in a seminal study conducted
by Moskowitz and Krieger (1995), which investigated the DOL of 6 different food
categories, bologna, hot dogs, carbonated fruit beverages, blueberry pie filling,
peanut butter, and salad dressing. The authors attempted to demonstrate how the
complexity of the food system may prevent researchers from obtaining the DOL for
a particular food system. On average across all food systems, the DOL categories
were in order of importance taste/flavor, texture, and appearance. The authors also
reported that no one pattern applied to all individuals however they did report that
the ranking between products within a product category was the same. This is
consistent with other DOL studies, which show distinct consumer segmentation
based upon DOL (Cardello and others 2003; Childs and others 2009; Drake and
others 2009, 2008; Kim and others 2010; Michon and others 2010; Kreger and others
2012; Lee and others 2012; Leksrisompong and others 2013; Shepard and others
2013).

The role of appearance is variable. Appearance was reported to be irrelevant to

some consumers for blueberry pie filling (Moskowitz and Krieger 1995). However,
appearance may be an important attribute based upon consumer expectations of
what the product should look like, but ultimately is not a primary driver of liking.
Leksrisompong and others (2012) investigated the DOL of different sweet potato
cultivars with varying flesh colors. In agreement with Moskowitz and Kreiger (1995),
the authors reported flavor and texture attributes to be the stronger DOL followed by
appearance. Leksrisompong and others (2012) confirmed the importance of
appearance by having consumers evaluate products in blinded and unblinded
conditions. Regardless of condition, smooth texture, brown sugar, dried apricot
flavor, and sweet taste were the major DOL for sweet potato with bitter, umami,
astringent mouthfeel, vanilla aroma, and residual fiber texture as negative attributes
of sweet potatoes. Appearance attributes of visual moistness and color homogeneity
helped differentiate consumer segments, but were not the primary DOL. Familiar
orange color received the highest liking scores compared to more unfamiliar colors
of yellow and purple (Leksrisompong and others 2012). The results of this study are
in agreement with the study conducted by Moskowitz and Kreiger (1995) in that
appearance may be important in specific products for acceptability, but ultimately
flavor/taste and texture drive liking. Shepard and others (2013) conducted an
exhaustive study of sour creams available across the U.S. market. Appearance,
flavor, and texture attributes were all compared in a DOL study with sour cream
consumers. Flavor attributes were the primary drivers (milk fat, diacetyl,
cooked/milky flavor attributes), followed by texture attributes, (denseness). Three
consumer clusters were identified, out of the 3 clusters, only one cluster had an
appearance attribute as a DOL (opacity). The importance of appearance as a DOL
will be product specific.

Expectation plays an important role in product purchase because it may improve or

degrade the perception of the product even before it is tasted (Deliza and MacFie
1996). Expectation is strongly related to consumer satisfaction (Anderson 2012).
Food acceptability research has defined 2 types of expectations: “1) sensory-based
expectations: belief that the stimulus/food product will possess certain sensory
attributes, each at certain intensities, or (2) hedonic-based expectations: belief that
the product will be liked/disliked to a certain degree” (Cardello and Sawyer 1992).
Expectation is strongly driven by extrinsic factors because products are rarely tasted
prior to purchase. A thorough understanding of consumer expectations of a product
based on extrinsic factors is important as extrinsic factors can be an indication of
product quality and perceived product quality can impact product performance and
therefore repeat purchase (Rao 2005; Stefani and others 2006). Consumers utilize
cues from cost, extrinsic properties, expected intrinsic quality cues of sensory,
health, and convenience, when assessing their perceived quality of a product before
it is purchased (Grunert 2005). The difference between expected intrinsic quality
cues and actual product performance is assessed in DOL studies. Understanding
the expectations of consumers from cues due to perceived satiety, brand and
labeling, price, and emotions, is paramount for understanding initial purchase.

Perceived Satiety and Portion Size

Perceived satiety of a food or beverage is the expectation of the degree of fullness

(satiety) after consuming that product. Perceived satiety and portion size go hand-
in-hand as the portion size is already dictated by the packaging of that
food/beverage. Before consumers actually taste the product, expected satiety from
just looking at food products with different portion sizes, will impact purchase intent.
Brunstrom and others (2008) define this concept as “expected satiety,” where
consumers consider if a particular food/beverage portion is large enough to stave off
hunger until the next meal. The importance of perceived satiety and portion size on
purchase intent was investigated by Brunstrom and Shakeshaft (2009), where
participants were asked to indicate the amount of money they would spend for
particular portion sizes of 8 different snack foods. The authors used pictures of the
different snack foods of specific food amounts that were correlated to specific fat,
carbohydrate, and protein contents. Participants then conducted a trade-off between
2 images as to which food would prevent them from feeling hungry for the longest
period of time, and indicated their liking of each food with size of food portion also
being decided. The authors reported that foods with high expected satiety were
regarded as more rewarding and were also chosen in smaller portions (Brunstrom
and Rogers 2009; Brunstrom and Shakeshaft 2009). The study concluded that
reasonably familiar and palatable foods are not selected in equal portion sizes. The
authors also reported that individuals with a higher body mass index (BMI) were
more likely to rely on expected satiety when making their portion size decisions.
Expected satiety was a better predictor of purchase intent over food liking
(Brunstrom and Shakeshaft 2009). No tasting was conducted in this experiment. It
is possible that sensory influences, cognitive factors, and post-ingestive
consequences all interact to affect the amount of food that is consumed (Kral 2006).
The availability of foods/beverages in specific packaged portion sizes may not be
the ideal portion size for each consumer and would affect purchase intent and would
therefore be a potential driver of purchase intent (DPI; Brunstrom and Rogers 2009).

Besides expected satiety from appearance, sensory-related satiety also plays an

important role on purchase intent, especially repeated purchase. Studies have
shown that repeated exposure, product ingredients information and form of food all
impact satiety, which eventually impact purchase intent. Sensory specific satiety
(SSS) is defined as the decrease in pleasantness of a product after eating the
product to satiety (Rolls 1986). Weenen and others (2005) reported that daily
exposure to cheese biscuits and pears resulted in decreased liking of both products
with a longer effect of food aversion from repeat exposure observed with cheese
biscuits. The authors speculated that the eating of cheese biscuits lead to a larger
degree of boredom than pears. The decrease in liking after consuming a
food/beverage product to satiety may affect the frequency of repeat purchase.
Havermans and others (2009) investigated the degree to which participants would
work for either chocolate milk or crisps to acquire more of either item to consume
when hungry. The “work” was determined by way of a computer game where
participants could click on a computer screen to earn points that would allow them
to earn more of either item. The amount of work that went into the task was then
equated to purchase intent. The authors reported participants were more motivated
to obtain points for crisps than for chocolate milk. Chocolate milk hedonic ratings
decreased more so than crisps with continued consumption. The authors concluded
that SSS was food/beverage product dependent and would affect PI differently
based upon that food product.

When consumers have knowledge of certain product ingredients, pre-existing

perception of the ingredient can impact perceived satiety. Yeomans and others
(2008) demonstrated with a flavored and unflavored porridge sweetened with either
sucrose or aspartame that participants demonstrated learned satiety by associating
the flavor of the porridge with post-ingestive effects of increased satiety from a higher
energy dense porridge sweetened with sucrose. Learned satiety would not be an
extrinsic factor as the food would have to be consumed in order for learned satiety
to affect repeat purchase intent. However, the concept of learned satiety is very
important in understanding how consumer expectation of satiety affects their repeat
purchase. Sweetness is usually a reliable predictor of energy intake because sugars
are caloric (Yeomans and others 2008); however, this relay mechanism for satiety
is not predictive of satiety from foods sweetened with noncaloric artificial sweeteners
(Swither and Davidson 2008). Swither and Davidson (2008) reported rats fed food
sweetened with artificial sweeteners, consumed more food than when sweetened
with sucrose and therefore gained more weight and became less sensitive to satiety
tests. If the consumer perceives an artificially sweetened food to be less satiating
than the same food sweetened with sucrose or another caloric sweetener, it is
possible that the consumer will consume more of the artificially sweetened food in
order to become sated. If a consumer perceives a lower calorie food to be less
satiating, the packaging size (portion size) may affect their frequency of purchase.
Yeomans and others (2005) reported participants ate a greater mass of low energy
dense porridge sweetened with aspartame than a high energy dense porridge
sweetened with sucrose. The increased consumption of the low calorie food will
affect the frequency of repeat purchase and the perceived satiety will be mitigated
by the portion size offered. If the food is to be consumed in larger portions because
the perceived satiety is low, the portion size offered at the time of purchase will be
important. Yeomans and others (2005) also reported than when the porridge was
flavored similarly in the high energy and low energy conditions, there was no
response in consumption amount and energy density, suggesting palatability (flavor)
was a more important predictor of the quantity consumed than perceived satiety.
This is not to suggest that satiety does not play a role in consumption amount, but
flavor may be the more important factor.

Perceived satiety of a food may also be dependent upon food type. Flood-Obbagy
and Rolls (2009) investigated the perceived satiety of different forms of apples
adjusted for energy density. Whole apple, applesauce, and juice with and without
additional fiber of similar weight and calorie density (except for the juice without
additional fiber) were evaluated by participants for perceived satiety after a timed
consumption period. Perceived hunger was lowest after consuming solid apples,
followed by applesauce, and both juices. Fullness was greatest for whole apples,
followed by applesauce and both juices. The results of this study suggest that food
form can affect perceived satiety. Purchase decisions based upon perceived satiety
will be different depending on perceived calorie density and food form. Overall, in a
DOL study, perceived satiety may be a critical attribute to assess and may be
measured by asking perceived satiety of a portion of the product or consumers’
satiety status before and after consumption followed by if it met their expectations.
Both visual analogue scales (VAS, line scales with anchors) and modified magnitude
estimation scales (Satiety Labeled Intensity Magnitude scales, SLIM) have been
proposed (Merrill and others 2002; Cardello and others 2005; Clegg and Shafat
2010).

Product Features—Brand and Labeling

Brand name

It is well known that branding affects consumer perceptions of a product (Vickers

1993; Deliza and MacFie 1996; Vranesevic and Stancec 2003; Wyma and others
2012; Kim and others 2013b; Rubio and others 2014). Brand name is an important
piece of information when consumers are deciding between competing products
(Deliza and MacFie 1996). Brand image has been defined as consumer's thoughts
and feelings about the brand (Roy and Banerjee 2007). Understanding consumer
perceptions of the brand image is of key importance for DOL because consumers
employ a product's brand image when perceiving the sensory attributes of a product
(Severi and Ling 2013). Private brands are often perceived as lower in quality than
national brands (Wyma and others 2012; Jervis and others 2012a; Rubio and others
2014). Private brands include store brands, private labels, distributor's brands,
reseller brands, middleman brands, own brands, dealer brands, and “generic”
products that are brand-free, no-name, house brands, and unbranded products
(Wyma and others 2012). However, consumers’ brand preferences may be a
reflection of their perceptions rather than from the actual product performance
(Bronnenberg and others 2007). Consumers who are quality conscious place more
trust in the performance of recognized brands and associate store brands as having
an increased risk of not performing up to expectations of that product (Paasovaara
and others 2012; Rubio and others 2014).

Rubio and others (2014) conducted a store intercept interview study with Spanish
consumers to understand consumer perception of branded and store branded
products. The more aware consumers were of a brand, the more the awareness
affected their perception of the product quality which in turn, increased the perceived
risk of purchasing a store brand product in that product category. Wyma and others
(2012) investigated consumer perceptions and preferences for private versus
national brand across a variety of food products (cooking oil, dry pasta, jam, ice
cream, rice, fruit juice, etc., n = 25 food products) in a grocery store intercept
questionnaire with South American consumers. Consumers surveyed showed
higher preferences for brand label products in all product categories surveyed with
the exception of cooking oil and dry pasta. Mueller and Szolnoki (2010) reported
informed liking, defined as the relationship between blind liking and acceptance of
extrinsic attributes of packaging, brand, and grape variety, to be the strongest driver
of purchase intent for wines. They suggested that extrinsic and intrinsic factors in
conjunction with each other were the highest predictors of liking for branded wines.
This was followed by price as the next highest driver of purchase intent. Paasovaara
and others (2012) investigated the perception of a national brand and a regionally
unfamiliar brand of drinkable yogurt. Liking of the national brand increased
significantly between a blind and unblinded tasting, however the regional unfamiliar
brand liking did not increase. This result highlights the increased quality perception
consumers place on foods with recognizable branding. Vranesevic and Stancec
(2003) investigated this effect with tin can pate’, a commonly consumed food in
Croatia. The study was conducted with 2 high-quality brands where only one of the
brands was well known to the consumers surveyed. The lesser known pate’ was
preferred under blinded conditions, but the well-known pate was preferred when the
brand was revealed during tasting, again demonstrating the quality perception
consumers attribute to well-known branded products. The effect of brand on product
perception may be a subconscious effect and therefore difficult to measure in any
direct survey situation. According to Mueller and others (2010), processing visual
cues (such as a brand label) are unconscious and unintentional, therefore asking
consumers about their preferences directly, may not result in accurate responses.
Studies have applied MRI imaging of consumers to demonstrate that different
regions of the brain were activated when consumers tasted products with and
without brand images (McClure and others 2004; Lindstrom 2008) also suggesting
the role of the subconscious in these decisions.
Vickers (1993) reported a branding effect dependent upon food type. Bower and
Turner (2001) investigated the effect of brand name, price, and intrinsic factors on
the purchase intent for crisp snack foods. In this study, the researchers evaluated
the effect of liking with and without brand label, and the effect of a well-known name
brand label compared to an economy label. The researchers reported intrinsic
effects to be the predominant drivers of purchase intent for crisp snack foods. The
researchers also reported a price effect where there was an increase in PI for higher
prices, and name brands performed better than economy products, indicating that
for crisp snack foods, consumers were willing to pay more for perceived higher
quality brand name. However, for crisp snack foods, liking was a stronger driver of
repeat purchase intent than brand label and price.

The effect of brand and other extrinsic factors can be assessed in a blinded tasting
environment with and without brand/label presentation, but can also be assessed in
other survey methods such as conjoint or KANO analysis. Conjoint analysis is a
choice-based study where the consumer acceptance of each individual component
of a product can be assessed conceptually (Orme 2009). Kano analysis is a question
format that assesses the impact of product attributes on consumer satisfaction or
dissatisfaction (Kano and others 1984; Xu and others 2009; Kim and others 2013b;
Li and others 2014). Jervis and others (2012a) evaluated consumer perception of
brand label sour creams in a conjoint comparative study. Store brands and unfamiliar
regional brands had a lower utility than national brands. Jervis and others (2012b)
conducted a joint study and reported that location of purchase was the most
important attribute over intrinsic attributes such as fat content, lightener type,
sweetener type, and flavor for coffee latte beverages. In this study, location was the
brand. Kim and others (2013b), in a thorough study of chocolate milk DOL, evaluated
product attributes in a conjoint study, KANO study, and in a blinded and unblinded
tasting condition. Conjoint results indicated brand names had a higher utility than
store and unfamiliar brand names although unlike latte-beverages, brand was not
the most important attribute. For chocolate milk, fat content followed by sugar
content were the most important attributes followed by brand. In blind and unblinded
tasting conditions, Kim and others (2013b) reported a brand effect for national brand
chocolate milks, an increase in purchase intent occurred when brand was known.
Also reported was a decrease in purchase intent when consumers were informed
that the milk was whole milk. The importance of brand image in perceived product
quality and purchase intent is clear, however the degree of importance is product
dependent. Therefore, in assessing the factors important for purchase and repeat
purchase of a product, the degree of importance of brand image cannot be ignored
and should be determined on a product-by-product basis.

Other labeling factors

Packaging and labeling of a food plays an important role at the point of purchase
(Deliza and others 2003). The extrinsic factors associated with the product
packaging and labeling are of primary importance for initial purchase intent and
setting up expectations for product liking and quality. Once consumer expectation is
matched, consumer satisfaction may occur and repeat purchase (Deliza and MacFie
1996). It is important to understand the factors that influence consumer expectation
as the weight of those factors may be as important as or more important than intrinsic
factors depending on the product. Moskowitz (2001) investigated image-related
attributes of margarine to taste/texture attributes and compared if they were
associated. The author reported liking, quality, similarity to butter, and healthfulness
to be related, demonstrating that nonsensory-related attributes can be important to
product liking on a product by product basis. Cardello (2003) investigated the
importance of processing technologies on product liking by assessing intrinsic and
extrinsic factors in a DOL study on chocolate pudding in a blinded and unblinded
condition referring to the knowledge of processing technology. Consumers 1st
evaluated 3 chocolate puddings in an unblinded condition, and then came back a
month later to taste the same 3 chocolate puddings with additional information on
processing technology used. The results compared the extrinsic factor of processing
condition to liking, however, no sensory attributes were also evaluated to have a
comparison of extrinsic and intrinsic factors. Cardello (2003) reported a greater effect
for processing technology for females than males. Liking scores were positively
influenced when the tasting was accompanied with a safety and benefit statement in
the unblinded condition.

Deliza and others (2003) investigated the utility of label attributes in a pictured
conjoint study of passion-fruit juice. Deliza reported 5 consumer segments all driven
by different labeling features. Each consumer segment employed different
packaging features when evaluated expected sensory liking of the pictured product.
The information a label or packaging provides to the consumer, like brand image,
affects the perceived quality and expectations of the consumer and will affect repeat
purchase if the intrinsic factors do not meet the expectations set by the extrinsic
factors. This degree and direction of this effect, (positively or negatively influence
hedonics) must be determined in DOL studies in order to understand what the actual
DOL will be in the purchasing environment. The role of these other labeling factors
can be assessed by a balanced crossover design, where in part one of the test, a
randomly selected half of the consumers (A) evaluate products without any
additional information, and the other half (B) evaluate the products with priming
statements regarding the targeted attributes. After a waiting period (suggested as 4
weeks), consumers return and re-evaluate the same products, group A now will have
the product information present, and group B will evaluate products without
information (Drake and others 2011; Li and others 2015).

Cost/Price—WTP

Product information raises sensory and hedonic expectations, and price can be an
indication of product quality (Cardello and Sawyer 1992; Grunert 2005; Rao 2005).
Price can subconsciously influence the perception of product quality and the
expectation of quality can then influence the actual product performance (Rao 2005).
This effect has also been observed with nonpricing information such as advertising
claims about product quality (Rao 2005). WTP is defined as “the maximum price a
buyer is willing to pay for a given quantity of a good” (Wertenbroch and Skiera 2002)
and purchase intent is most commonly used to indirectly assess WTP. Although
price and WTP are 2 different things, they are inextricably linked as the set price by
a manufacturer will be inevitably viewed in concordance with consumers WTP for
that product based upon all perceived quality and sensory cues. Price is the best
known extrinsic indicator of quality and in the absence of all other extrinsic variables,
a higher priced product will be judged as higher quality than the lower priced
alternative (Oude Ophuis and Van Trijp 1995). This statement is consistent with the
conjoint and label liking study conducted by Chrea and others (2011). They
investigated extrinsic factors that influenced purchase of Australian wines where
consumers have to rely on extrinsic factors as the product cannot be tasted prior to
purchase. Consumers relied on price followed by label and bottle cues for wine
purchase. Price was an indication of quality which was supported by the highest
price range having the highest utility across price levels.

The importance of price has been assessed in many other studies using surveys,
conjoint analysis, and auctioning. Grunert and others (2009) investigated 3
techniques for assessing WTP, contingent valuation, (which is a survey based
approach), experimental auctions, and choice-based conjoint analysis. The authors
reported conjoint analysis to yield lower estimates of WTP than contingent valuation
and auctions for a ready-made soup product. Bower and Turner (2001) reported
consumer responses to price labels in a survey were consistent with their responses
to price in a taste environment, suggesting surveying consumers before tasting may
be a reliable way to gain understanding of price as a DPI.

WTP and the importance of price are product specific, and are also largely influenced
by quality, nutritional information, and other properties and are greatly related to the
brand and labeling effects of a product. Xue and others (2010) investigated
consumers WTP for grass-fed beef after a blind evaluation of conventional and
grass-fed beef. Nutritional information for grass-fed beef was also evaluated before
or after tasting, or not at all, in 3 balanced consumer groups. Consumers who were
given nutritional information on grass-fed beef before or after tasting were WTP more
for grass-fed beef than the control group given no nutritional information. No
significant difference between nutritional information groups was reported. The
results of this study demonstrate the importance of nutritional information to
consumers WTP for grass-fed beef. Barreiro-Hurle and others (2008) investigated
consumer WTP for resveratrol-enriched red wine of consumers in Granada, Spain,
using a choice experiment. The authors reported consumers were willing to pay
more for a resveratrol-enriched red wine compared to a standard red wine. Janssen
and Hamm (2012) investigated consumer preferences and WTP for different organic
certification logos of apples and eggs. Overall consumers were WTP more for eggs
and apples with organic labels that included either a European Union logo,
government logo, or private logo. Consumers overall did not find products without a
logo as trustworthy and credible, which was why organic certification with a logo was
preferred. Van Loo and others (2011) conducted a similar experiment for organic
chicken breast using a choice experiment. Consumers were WTP more for an
organic label, however, an organic label with a United Stated Agricultural Dept. logo
had a higher premium than a general organic label.
Di Monaco and others (2005) investigated the effect of price on chocolate bars with
and without a health claim in a blinded tasting condition. The authors reported that
health claim and price did not affect liking but the increase in price reduced the
likelihood of purchase. Focus group results suggested chocolate gives pleasure, a
reward, therefore, healthfulness was not an important factor. Solheim and Lawless
(1996) investigated the effect of fat content and price on consumer purchase intent
of Cheddar cheese and reported a decreasing price increased purchase probability
and reported a gender effect for fat content effect on purchase probability.
Costanigro and others (2014) investigated consumer WTP for non-sulfited wine in a
choice experiment. The authors reported that consumers who suffer from sulfite-
induced headaches, were willing to pay a premium of $1.23 to avoid them, but were
only slightly more likely to purchase a wine that did not contain sulfites. Price and
perceived quality ultimately influenced purchase with little influence from organic and
sulfite labeling.

The primary reason that the role of price is so critical is that consumers may indicate
a strong hedonic preference or purchase intent for a product perceived as high
quality without actually buying the product once placed in a purchasing environment
(Lange and others 2000). One of the ways to assess consumer WTP for a product
is in a blind and unblinded tasting condition followed by an auctioning study that
determines hedonic preferences and confirms preferences in an actual purchasing
situation. The Vickrey auction method is a commonly applied auctioning method that
involves asking individuals to submit a sealed bid which corresponds to the
maximum price they would pay for a particular product. The winner is the highest
bidder who actually has to pay for the product at the 2nd highest bid (Vickrey 1961).
This strategy allows consumers to purchase a product at a price lower than their
maximum WTP, which motivates them to bid accurately. Lange and others (2002)
used a Vickrey auction method with a blind and unblinded tasting condition to assess
consumers WTP for Champagne. The results of the auctioning study discriminated
the Champagnes more so than hedonic ratings but both methods lead to the same
overall conclusions about the Champagnes.

Napolitano and others (2010) investigated the role of information about organic
product on beef liking and consumer WTP using a Vickrey auction. The type of
information significantly affected liking and WTP. Information on the organic
production system had a larger effect on liking than information on animal welfare,
and environmental pollution. Overall consumers responded to the fact that the beef
was organic rather than to positive messages in general. Consumers were also WTP
more for organic beef. Lawless and others (2012) used auctioning to determine
consumers WTP for a Concord grape and blackberry juice blend with a benefit
statement about anthocyanins before (or after) tasting depending on the study group.
Reading the health statement after tasting increased WTP more so than reading the
health statement prior to tasting. It was expected that additional beneficial
information would increase WTP as a primer for expected sensory attributes,
however for this product, this was not the case. Ginon and others (2009) investigated
consumer WTP for French baguettes with and without fiber information. High and
low fiber baguettes were evaluated in an auctioning study in the presence and
absence of health claims associated with fiber. Consumers were not WTP more than
a certain price for a baguette even when it had a high hedonic score. A ”source of
fiber” label of either brown flour or whole meal flour (product specific for the high fiber
baguettes) had a positive effect on WTP. The recommendation was to accompany
new nutraceutical products with in-store tasting sessions.

Chern and Lin (2012) investigated the importance of Country of Origin Labeling
(COOL) on Taiwanese products versus imported products with Taiwanese
consumers using an auctioning experiment with tasting. Freshness, safety, and
country of origin labeling were very important to Taiwainese consumers. In an
auctioning study, Taiwainese consumers were WTP more for charcoal-smoked
plums, and Oolong tea with a Taiwainese COO label. Lee and others (2012)
investigated the DOL of imported and domestic black olives. In a survey of
consumers, olive taste was the primary influence of purchase followed by price and
then country of origin. De Steur and others (2012) investigated WTP for genetically
modified (GM) rice in high-risk regions of China using an auctioning study. Chinese
female consumers of childbearing age were presented with a GM rice of higher folate
concentration, and non-GM rice with an additional folate supplement. Overall, female
participants were WTP more for GM rice than non-GM rice with a folate supplement.
However, WTP for GM rice was affected by consumer acceptance of GM foods.

The importance of price as a key factor in the purchase decision is clear. However,
WTP is heavily impacted by the nature, quality, and other additional product related
attributes information, the importance of price on purchase intent cannot be
assessed without assessing different WTP resulted from different product quality or
other informational cues. In its simplest form, purchase intent can be assessed or
consumers can be asked specific questions regarding WTP and provided with
pricing. A more realistic assessment of price relative to other product attributes can
be obtained by a conjoint study, or auction experiments can be applied to assess
WTP in an even more realistic scenario.

Decision Making and Emotion Impact

Consumers come into a purchase situation combining desires (utilities, goals,

personal values) and beliefs (expectations, knowledge, means; Angie and others
2011). The emotional experiences consumers have in the purchase situation are tied
up in their desires and beliefs as to what that product will do for them and if the
product meets/or does not meet their expectations upon consumption. Having an
understanding of a product's perception due to information from packaging/pricing
and sensory attributes is vital, but in order to truly understand the impact a product
has on consumer acceptance, the emotional experience and decision-making
process a consumer has with that product must also be understood.

In the shopping experience, consumers engage in generally 2 types of product

buying situations; low-involvement purchases and high-involvement purchase
(Kalnikaite and others 2013). A low-involvement purchase would be a product that a
consumer is already familiar with and purchases on a routine or more consistent
basis (for example, a favorite brand of carbonated soft drink). Because the consumer
is well familiar with the product/brand/satiety, and so on, less information and time
is needed for processing the purchasing decision (Kalnikaite and others 2013). A
high-involvement purchase would be purchasing a one-time or 1st time product that
may be of relatively high value and therefore require a deeper thought process
before purchasing (Kalnikaite and others 2013). For example, a bottle of wine to
serve at a dinner party, a refrigerated fruit smoothie with health benefits claims, or a
new enhanced water beverage might require more consumer thought. Real-world
decision making is often time limited and consumers employ simplifying heuristics
where they ignore most of the available information, focusing on the key attributes
of importance to them. Kalnikaite and others (2013) investigated the influence of
information in the form of smartphone applications in aiding consumers in purchasing
decisions and reported that for low-involvement purchases, consumers will generally
only engage trade-offs in 1 or 2 key attributes while ignoring other information. For
high-involvement products, consumers preferred cues to product quality, such as a
star rating, that summarized product information to facilitate their decision making.
Consumers want more information for these types of products, but also want the
information to be simply understood and summarized for easy decision making. The
implications on such thought processes are clear for the food scientist. For
commonly purchased household items such as sandwich bread, milk, meat, and so
on, it is imperative that the key purchase influencing attributes are known so that
efforts are focused on enhancing those attributes while keeping superfluous product
information available.

A consumer's emotional state prior to purchasing will affect their decision-making

processes. It is important to understand how basic emotions affect food choices.
Happiness has been shown to result in higher probability of choosing the safe option
more so than those experiencing sadness, whereas sadness has been associated
with the tendency to engage in thoughtful and more detail-oriented processing of
cognitive tasks (Angie and others 2011). Emotions such as anger or joy have a
greater impact on increased food intake compared to sadness or fear (Canetti and
others 2002). Macht (2008) suggested that high arousal/intense emotions (for
example, anger), suppress eating because of incompatible emotional responses,
while emotions of moderate arousal/intensity (for example, boredom) affect eating
depending on the motivation to eat (for example, negative emotions generally
increases intake of sweet and high-fat foods). Yeomans and Coughlan (2009)
reported a higher frequency of consumption of popcorn and raisins from a test
population with a reported negative mood after watching video clips intended to elicit
a negative emotional affect. This study only involved women and also reported that
women with a higher body mass index (BMI) consumed more snacks than women
with a lower BMI. Negative moods may decrease inhibitions and allow for overeating
(Herman and Polivy 1980). We eat because of cues that tell us sufficient time has
passed since our last meal, or to experience the sensory pleasure of foods, or
because of boredom (Prescott 2012). Consumer mood prior to purchase will affect
the type of food purchased and the quantity.
Several tools exist for analyzing consumer emotions to a product including
ScentMove™ (Porcherot and others 2015), EsSense Profile™, FrEmo™ (Gutjar and
others 2014), and more recently reported EmoSemio™ (Spinelli and others 2014).
This list is by no means exhaustive of the techniques available. Each of the product
attributes presented in this paper can be additionally analyzed for consumer
emotional response to the presence/absence each attributes. For example, the
emotional impact a brand name has on a product can be assessed by measuring
emotional responses to a food product with and without the brand name present in
a balanced, cross-over design to account for the branding effect. This process can
easily be extended for price, satiety, and any other labeling information. Brand, label,
and pricing perception can be easily assessed in multiple iterations without the need
to create real prototypes through the use of photographs. Jervis and others (2014)
demonstrated that pictures of crust and crumb of sliced sandwich bread modified to
represent a variety of attributes levels could be evaluated by adaptive choice based
conjoint to accurately predict consumer acceptance of real products in a consumer
acceptance test. Food and beverage companies can generate product concepts with
and without labeling information with an emotional response technique to determine
the emotional impact of that information. Utilizing trade-off scenarios by conjoint
analysis or a similar technique yields a great deal of useful information on what
product features have the greatest impact on consumer choice and when coupled
with emotion analysis, also measure the emotional impact. Several publications have
demonstrated the efficacy of conjoint analysis (Vickers 1993; Deliza 2003; Childs
and Drake 2009; Jansen and others 2009; Palacios and others 2009; Reisfelt and
others 2009; Kildegaard and others 2011; Jervis and others 2012a, 2012b; Olsen
and others 2012; Boquin and others 2014; Jervis and others 2014; Li and others
2014). Emotional analysis has also been a hot topic in food and beverage research
(Oliver and Wardle 1999; Canetti and others 2002; Desmet and Schifferstein 2008;
Macht 2008; Wallis and Hetherington 2009; Ferrarini and others 2010; King and
Meiselman 2010; King and others 2010; Porcherot and others 2010; Cardello and
others 2012; Spinelli and others 2014). Recently, Li and others (2014) utilized
conjoint and emotions analysis using the emotion word list generated by King and
Meiselman (2010) to determine the factors that affected parent's choice of chocolate
milk for their children. Consumer emotions related to products can also be assessed
by asking consumers how they feel when purchase/consume the products where the
options are choose all that apply (CATA). A list of emotional terms, such as happy,
joyful, angry, and so on (King and Meiselman 2010) are provided and consumers
indicate the related terms.

Conclusion

Ideally, when trying to assess product DOL for the purposes of understanding
product success in the market, extrinsic factors such as perceived satiety,
brand/labeling influence, price/WTP, and emotional impact should be assessed. This
information is important for product development but has the greatest impact when
a final product concept is developed and there is a level of understanding of the role
of product claims, brand identity, and pricing structure for the product. This type of
work can also be done with launched products to understand how the product is
perceived in the market and what can be done to improve product acceptance aside
from improving sensory attributes. Budget and timeline limitations will impact the
depth and complexity to which these factors can be addressed. In simplest form, a
few questions can be added to assess these items. Assessing this type of
information can be done more thoroughly through conjoint analysis and its
derivations, auctioning studies, and emotions assessments. The direct influence of
extrinsic attributes on sensory attribute DOL should be assessed in the presence
and absence of this information in a taste test with consumers that is balanced
across a blind (no information) and unblinded (information given) condition to assess
the effect of the information (price, brand, and so on). Given the sheer complexity of
information, the design of experiment in a taste test can be quite complex, therefore,
utilizing techniques that do not involve tasting such as conjoint analysis with pictures
and emotional analysis, can give additional insight into the influence of extrinsic
factors in a more efficient manner for iterative product development. WTP may be
the most important extrinsic factor since it has the ability to reflect the impact brought
by brand and labeling, emotional impact and can affect the initial and repeated
purchase along with perceived satiety. These types of tests should be leveraged by
sensory scientists to support product developers to produce successful food and
beverage products that will delight consumers.

Acknowledgments

Funding was provided in part by the Dairy Research Inst. (Rosemont, Ill., U.S.A.).
Use of trade names does not imply endorsement nor lack of endorsement by those
not mentioned.


Volume 82, Issue 2 February 2017 Pages 248–259

Concise Reviews and Hypotheses in Food Science

Aroma Glycosides in Grapes and Wine

J Liu, Xiao – Lin Shu, Niermat Ullah, Yong Sheng TAo,

Funding Information

This work was accomplished under the auspices of the National Natural
Science Foundation of China (Grant No. 31371724) and Fundamental
Research Funds of the Central Univ. of Ministry of Education of China (Grant
No. 2014YQ005).

Abstract

The major aroma components in grapes and wine include free volatile compounds
and glycosidic nonvolatile compounds. The latter group of compounds is more than
10 times abundant of the former, and constitutes a big aroma reserve in grapes and
wine. This review summarizes the research results obtained recently for the
identification of aroma glycosides in grapes and wine, including grape glycoside
structures, differences in aroma glycosides among grape varieties, hydrolysis
mechanisms, and the factors that influence them. It also presents the analytical
techniques used to identify the glycosidic aroma precursors. The operational
strategies, challenges, and improvements of each step encountered in the analysis
of glycosidic aroma precursors are described. This review intends to provide a
convenient reference for researchers interested in the methods used for the
determination of the aroma glucosides composition and the recognition of their
chemical structures.

Introduction

Aroma is an important sensory attribute of wine. In general, wine aromas can be

categorized into varietal, fermentation, and ageing aromas. Aromas from grape
berries determine the varietal typicality, whereas fermentation aromas can
supplement the varietal aromas (Peng and others 2013). Moreover, ageing aromas
arise from the development and optimization of fermentation and ageing aromas (Li
2006). The mainstream grape varieties for winemaking belong to Vitis vinifera L.
Surprisingly, grape berries of most Vitis vinifera L. have no scent, yet their wines are
characterized by their respective and unique varietal aromas. Aroma substances
exist as free volatiles and as bound aroma components in grapes and wine. The 2
most studied bound aroma substance classes are glycosides and cysteinylated
conjugates. Aroma glycosides in grape berries are much more abundant than
cysteinylated conjugates, and become volatiles after hydrolysis. Because
Cordonnier and Bayonove (1974) demonstrated the crucial role of glycosidic aroma
precursors from grapes in wine, much related research has been conducted. The
1st glycosidic bound compounds identified in grapes were terpene glycosides
(William and others 1982a). Then, Park and others (1991) reported that
monoterpene glycosides were found to be the most significant aroma precursors in
Alexandria Muscat grapes and wine, accounting for about 90% of the total
monoterpenes. Therefore, these aroma glycosides certainly constitute the potential
aroma reserve of Muscat grapes and wines (Noble and others 1987). In the last
decade, reviews related to aroma precursors reported the enzymatic hydrolysis of
aroma glycosides (Sarry and Günata 2004), the hydrolysis of terpene glycosides in
grape juice (Maicas and Mateo 2005), the formation of aroma glycosides in plants
(Bowles and others 2006), and glycosidic aroma compounds in grapes and wine
(Hjelmeland and others 2015). On the basis of an in-depth analysis of hundreds of
references published in the past 10 y, this review aims to inform the reader about
the research progress on aroma glycosides in grapes and wine with special attention
to the chemical analysis of aroma precursors.

Grape Aroma Glycosides

Grape aroma glycosides are aroma components (known as aglycones) linked to a

sugar moiety (known as glycones) in grapes and wine. Aroma glycosides are the
main aroma substances in grapes and the potential source of aroma volatiles
contributing to the wine aroma profile (Cabaroglu 2002; Arévalo-Villena and others
2007; Rodríguez-Bencomo 2011a,b). Each glycoside molecule contains a sugar
moiety, which renders itself nonvolatile. The aroma can be perceived when the free
volatile aroma compound is released by hydrolysis of the aroma glycoside.

Chemical structures of aroma glycosides

Thus far, all identified grape aroma glycosides have a glycone linked directly to the
free aroma compound (aglycone). The 1st one was identified by Winterhalter and
Skouroumounis (1997) as β-d-glucose. Another sugar molecule may be added at
the β-d-glucose moiety to form higher glycosides. Recent studies indicated that the
aroma glycosides in grapes are mainly disaccharide glycosides, to a lesser extent
monosaccharide glycosides, and that trisaccharide glycosides are less common
than the former 2 kinds (Williams and others 1982b; Strauss and others 1988;
Winterhalter & Skouroumounis 1997; Maicas & Mateo 2005; D'Ambrosio and others
2013; Schievano and others 2013). Moreover, rhamnose, arabinose, and apiose
have been identified as terminal glycones in disaccharide glycosides (Voirin and
others 1992). The aroma disaccharide glycosides in grapes and wine could therefore
be classified into 4 groups according to the sugar moieties: 6-O-β-d-glucosyl-β-d-
glucopyranosides, 6-O-β-d-apiofuranosyl-β-d-glucopyranosides, 6-O-α-l-
rhamnopyranosyl-β-d-glucopyranosides, and 6-O-α-l-arabinofuranosyl-β-d-
glucopyranosides (Mateo & Jiménez 2000; Sarry & Günata 2004; Pogorzelski &
Wilkowska 2007). In general, according to the difference concentration of terpenic
compounds in grape berries, most wine grapes of Vitis vinifera L. are classified as
Muscat varieties (>6 mg/L), non-Muscat varieties (1–4 mg/L), and neutral varieties
(<1 mg/L; Mateo and Jiménez 2000), which contain about 50% of 6-O-β-d-
apiofuranosyl-β-d-glucopyranosides, 6% to 13% of 6-O-α-l-rhamnopyranosyl-β-d-
glucopyranosides, and 4% to 9% of β-d-glucopyranosides. The content of 6-O-α-l-
arabinofuranosyl-β-d-glucopyranosides is often very low (Bayonove and others
1992; Sarry and Günata 2004; Maicas and Mateo 2005).

Terpene glycosides were the 1st to be recognized as an important subgroup of the

grape aroma glycosides (Williams and others 1981, 1982a; Günata and others 1985,
1988). Since then, studies reported more than 200 grape aroma glycosides, which
can be classified according to the chemical structure of their aglycones as terpenes,
C13-norisoprenoids, volatile phenols, aromatic phenols, C6 compounds, aliphatic
alcohols, aliphatic acids, benzenic, and phenolic acid derivatives, and so on. The
aglycones of the identified aroma glycosides are summarized in Table 1. The
chemical structures of typical aglycones are characterized by unsaturated bonds
except aliphatic alcohols (Figure 1).

Table 1. Main aglycones of the glycosidic aroma precursors in grapes

Categories Compounds

1. Note: references are represented by numbers. (1) López and others 2004;(2)
Genovés and others 2005; (3) Cabrita and others 2006; (4) Souid and others
2007; (5) Nasi and others 2008; (6) Genisheva & Oliveira, 2009; (7) Botelho
and others 2010; (8) Canosa and others 2011; (9) Pedroza and others 2010;
(10) Rodríguez-Bencomo and others 2011a; (11) Flamini and others 2014;
(12) Noguerol-Pato and others 2013; (13) Genovese and others 2013; (14)
Lamorte and others 2014.

Linalool oxide(2,3,6,8,10-12,14), Linalool(2,4-6,8-14), α-Terpineol(1,2,5,6,8-14),

β-Citronellol(2,12), Nerol(2-4,6,11-14), Terpinen-4-ol(6), Limonene(9),
Terpenes
Geraniol(2-6,8,9,11-14), Geranic acid(2,3,6,8,12,14), Citronellol(6,11),
Farnesol(1,9), Nerolidol(9), Epoxylinalool(14)

3-Hydroxy-β-damascenone(3,4,8), α-Ionone(9,10), β-Ionone(9,10,12,13),

C13-
β-Damascenone(1,3,5,7,9,10,13), 3-Oxo-α-ionol(3,4,8), Vomifoliol(3),
norisoprenoids
1,1,6-trimethyl-1,2-dihydronaphthalene (TDN)(1,3,7,13)
Categories Compounds

2,6-Dimethoxyphenol(1,10), 4-Ethylphenol(1,10,12), 4-
Ethylguaiacol (1,7,10,12), 2-Methoxy-4-vinylphenol , (10) 4-
Volatile phenols
vinylguaiacol phenol(3,7,8), m-Cresol(1,12), Guaiacol(1,7,8,10,12-14), 4-
Vinylphenol(1,3,10,12)

Aromatic Eugenol(1,3,7,8,10,12,14), Isoeugenol(1,10), Syringol(7,12,13), 3,5-

phenols dimethoxy-4-hydroxybenzaldehyde(13)

1-hexanol(2-5,8,9,12-14), trans-3-Hexen-1-ol(1,3,8,12,13), trans-2-hexen-

C6 compounds 1-ol(3,8,9,12,14), cis-3-hexen-1-ol(2-4,8,12-14), trans-2-
hexenal (1,2,8,9,12,14) , hexanal(1,7,8,12), hexanoic acid(1)

1-Octanol(1,2,8,12), 3-methyl-2-butanol(2), Isobutanol(2,12), Isoamyl

Aliphatic
alcohols(12), 1-Butanol(2,8,12), 3-pentanol(2,8), 2-pentanol(8), 2-
alcohols
heptanol(8)

Acetic acid(1,8), Octanoic acid(1,2,8), Isobutyric acid(12), Butanoic

Aliphatic acid
acid(12), Isovaleric acid(12), Capric acid(12)

2-Phenylethanol(1-4,7-9,12-14), Benzoic acid(1,12), Benzaldehyde(1-

Benzenic 4,7,8,12), Phenylacetaldehyde(12), Benzyl alcohol(3,4,8,10,12-14), Anise
compounds
alcohol(14)

Methyl vanillate(1,8,10,13,14), Ethyl vanillate(1,10,12),

Phenolic acid
Vanillylacetone(10), Vanillin(1,4,7,8,10,12,13), Methyl salicylate(3,8),
derivatives
Dihydroconiferyl alcohol(3), Acetovanillone(1,4,7,12)
Figure 1.

 Open in figure viewer

 Download Powerpoint slide

Typical aglycones of different chemical classes in grape aroma glycosides.

Differences in aroma glycosides among grape varieties

The profiles of aroma glycosides among different grape varieties or cultivars have
been explored to a certain extent (Rodriguez and others 2011a; Martínez-Gil and
others 2013; Panighel and others 2014). The proportion of glycosidic aroma
substances in the total content of aroma substances changes among grape varieties.
For example, the proportion of glycosidic-bound aroma substances in the total aroma
substances of Bical grape (a white Vitis vinifera variety in Portuguese Bairrada
appellations) was 54%, whereas the one of Maria Gomes (also known as “Fernao
Pires” and “Gaeiro,” another white Vitis vinifera variety in the Portuguese Bairrada
appellations) reached 67% (Rocha and others 2000). Berries from different grape
varieties may contain different aglycones in the aroma glycosides chemical
structures. For example, Pedroza and others (2010) identified the free and
glycosidically-bound aroma compounds in Merlot, Early Sugar (a white aromatic
table grape), and Petit Verdot grapes. The results showed that linalool glycoside was
detected only in Merlot and Early Sugar grapes, whereas geraniol glycoside was
detected in Petit Verdot grapes. Besides, nerolidol glycoside was detected only in
Early Sugar grapes, but not in the other 2 grapes. Rodríguez-Bencomo and others
(2011a) reported that Listán Blanco grapes, white-grape varieties grown on Canary
Islands in Spain, contained mainly terpene glycosides, although Gual grapes grown
on the Madeira Island in Portugal contained mainly glycosides of volatile phenols
and vinyl phenolics compounds. Additional relevant research results are presented
in Table 2.

Table 2. Major glycosidic aroma compounds in different grape varieties

Glycosides
Identification
Variety with higher Separation methods References
method
contents

Terpene Liquid–liquid Enzymatic

Maria Rocha and
glycosides continuous extraction hydrolysis,GC-
Gomes others 2000
(17%) with dichloromethane MS

Benzenic Liquid–liquid Enzymatic

Rocha and
Bical glycosides continuous extraction hydrolysis, GC-
others 2000
(17%) with dichloromethane MS
 Glycosides
Identification
Variety with higher Separation methods References
method
contents

Aliphatic
Enzymatic Genovés
Palomino alcohols,
Amberlite XAD-2 resin hydrolysis, GC- and others
Fino Grape Benzenic
MS 2005
glycosides

Trincadeira,
Enzymatic
Aragonez, Benzenic C18 solid-phase Cabrita and
hydrolysis, GC-
Moreto, C glycosides extraction column others 2006
MS
stelao

Benzenic, C13- Enzymatic

C18 solid-phase Cabrita and
Tinta Caiada norisoprenoids hydrolysis, GC-
extraction column others 2006
glycosides MS

Benzenic Enzymatic
Souid and
Khamri glycosides LiChrolut EN cartridge hydrolysis, GC-
others 2007
(53%) MS

C6 compounds Stirred with Acidic

Pedroza and
Petit-Verdot glycosides polydimethylsiloxane hydrolysis,
others 2010
(44%) coated stir bar SBSE-GC-MS

Terpene Stirred with Acidic

Merlot, Early Pedroza and
glycosides (56- polydimethylsiloxane hydrolysis,
Sugar others 2010
80%) coated stir bar SBSE-GC-MS

Caíño
Redondo Enzymatic
Benzenic Canosa and
Pedral LiChrolut EN cartridge hydrolysis, GC-
glycosides others, 2011
Mencía MS
Sousón

C13-
Enzymatic
norisoprenoids Canosa and
Espadeiro LiChrolut EN cartridge hydrolysis, GC-
glycosides others 2011
MS
(32%)

Rodríguez-
Benzenic, Acidic
Bencomo
Malvasía terpene Amberlite XAD-2 resin hydrolysis, GC-
and others
glycosides MS
2011a
 Glycosides
Identification
Variety with higher Separation methods References
method
contents

Benzenic, C6 Enzymatic Genovese

Uva di Troia, C18 reverse-phase
compounds hydrolysis, GC- and others
Aglianico solid phase cartridge
glycosides MS 2013

Strata-X polymeric Enzymatic Noguerol

Garnacha C6 compounds
reversed phase hydrolysis, GC- and others
Tintorera glycosides
cartridge MS 2013

Enzymatic
Benzenic C18 reverse-phase Lamorte and
Greco hydrolysis, GC-
glycosides solid phase cartridge others 2014
MS

Moreover, the contents of aroma glycosides are not the same in different parts of the
grape berry. The content of aroma glycosides in grape skin is generally higher than
those in the pulp and juice (López and others 2004; Cabrita and others 2006;
Rodríguez-Bencomo and others 2011b). For example, Cabrita and others (2006)
compared the aroma glycosides in the skins and juices of 10 grape varieties from
Portugal. The free aroma compounds released by enzymatic hydrolysis were
detected by gas chromatography-mass spectrometry (GC-MS). The results showed
that the glycosides of terpenes, benzenic compounds, C13-norisoprenoids, C6
compounds were more abundant in grape skin than in juice. In addition, Rodríguez-
Bencomo and others (2011a) found by GC-MS analysis of the aglycones after
acidolysis that the aroma glycosides of terpenes and benzenic compounds were
higher in the skin than in the juice of Marmajuelo and Malvasía grapes (2 white grape
varieties grown in Spain). Similar results have also been reported by Genovese and
others (2013) who studied the aroma glycosides in the skin and juice of Aglianico
and Uva di Troia (2 red grapes widely cultivated in Southern Italy). The GC-MS
results indicated that the contents of monoterpene aroma glycosides and certain
benzenic aroma glycosides were higher in skin than that in juice of grapes. However,
more of the bound C6 alcohols and furaneol were detected in the juice than in the
skin. Keeping in consideration the different content of the aroma compounds in skin
and juice of the grape, winemaking processes could be modified to improve the
characteristics of wine aroma. For example, Peng and others (2013) found that
prefermentative freezing could enhance the extraction of varietal compounds, such
as terpenols and norisoprenoids, from grape skins and thereby improve the wine
aroma intensity and complexity.

Furthermore, the profiles of aroma glycosides in berries from different parts of a

grape bunch may be different. Noguerol and others (2012a,b) found that Mouratón
and Brancellao grapes had more glycosidic aroma precursors, such as
monoterpenes and aromatic alcohols, at the tips of the grape bunch than at the
shoulders. The profile can be affected by climate conditions influencing biosynthesis
and ripening. The involvement of gene expressions in the formation of the glycosidic
aroma precursor was reported by Martin and others (2012). Further explorations are
required in this field.

The effects of vine cultivation conditions on the formation of grape aroma glycosides
have also been investigated by many researchers. For instance, Park and others
(1991) speculatively attributed the fluctuation of glycosidically-bound monoterpenes’
content in the skin and mesocarp of Alexandria grapes (Vitis vinifera L.) to the
influence of temperature during the development of the plant, without providing
further details on the temperature effect. Moreover, Bureau and others (2000)
pointed out that C13-norisoprenoidic and phenolic glycosides levels decreased in
shaded bunches. In contrast, Toci and others (2012) did not find any obvious
changes in the contents of glycoside compounds when regulating the water stress
to study the influence of soil management, such as the cover cropping and soil
tillage, on the wine aroma. Interestingly, Martínez-Gil and others (2013) sprayed oak
extracts on grape leaves at different veraison times and found that the oak extracts
could be assimilated and stored as glycosidic compounds in grapes.

Hydrolysis mechanism of aroma glycosides

Grape glycosides can spontaneously be hydrolyzed to release volatile aglycones as

grape berries are naturally acidic between pH 3 and 3.5 (Williams and others 1982c).
In addition, the hydrolysis of aroma glycosides can also be enzymatically catalyzed
by glycosidases, which are produced in the grape berries or secreted by yeasts
during alcohol fermentation.

Enzymatic hydrolysis

Aroma glycosides are usually enzymatically hydrolyzed during grape ripening and
alcohol fermentation, which causes the release of free aroma compounds.
Monosaccharide glycosides of aroma precursors are enzymatically hydrolyzed by
the endo- or exo-glucosidase, liberating directly the glucose moiety and the free
aromatic aglycone (Sarry and Günata 2004). The enzymatic hydrolysis of
disaccharide glycosides follows either a 1-step or a 2-step mechanism. In the 1-step
mechanism, the glycosidic linkage to the aglycone in the disaccharide glycoside is
broken by an endo-glycosidase enzyme, releasing the free aroma compound
(Ogawa and others 1997; Günata and others 1998). The 2-step mechanism involves
2 or more enzymes. According to the chemical structure of the grape aroma
disaccharide, a specific exo-glycosidase, such as α-l-arabinofuranosidase, β-d-
apiofuranosidase, and α-l-rhamnopyranosidase, breaks the bond between the
respective sugar moiety and the β-d-glucoside, releasing in the 1st step one
molecule of arabinose, apiose, or rhamnose, respectively. In the 2nd step, the
remaining glucoside is hydrolyzed by β-d-glucosidase to liberate the free aromatic
aglycone and 1 molecule of glucose (Sarry and Günata 2004).
In addition, the enzymatic hydrolysis of aroma glycosides highly depends on the
characteristics of the glycosidase. The determination of the chemical structures of
aroma glycosides usually involves their extraction from grape or wine by an
enzymatic hydrolysis, followed by purification, identification, and quantification of the
released aromatic aglycones by GC. Generally, routine sample preparations use the
commercially available purified glycosidases, such as almond β-glucosidase and the
enzyme AR2000 extracted from Aspergillus niger (Kang and others 2010; Tao and
others 2014). Besides, endo-glycosidases existing in some plant cells can directly
break the glycosidic linkage to liberate the aglycone regardless of the number of
sugar moieties. For instance, the concentration of β-d-glucosidase in grape berries
increases together with the ripening advancement of the fruit. Glycosidases are
adaptive to the natural acidic environment of grape and wine and have an optimal
catalytic activity at about pH 3.4. However, the enzyme is strongly inhibited at high
concentrations of glucose and ethanol (Günata and others 1990). Saccharomyces
cerevisiae also secretes glucosidases. However, these enzymes cannot maintain
efficient catalytic activity during the winemaking process conditions with high levels
of sugar, alcohol, and polyphenol and at low pH (Loscos and others 2007).
Consequently, most grape aroma glycosides cannot be hydrolyzed completely
during the winemaking process, especially when the grapes are not sufficiently ripe.

Therefore, increased attention has been paid to the development of exogenous

glycosidase, which are effective at hydrolyzing aroma glycoside even under
winemaking environment. For example, the commercial glycosidase AR2000
preparation obtained from Aspergillus niger, catalyzes efficiently the hydrolysis of
aroma precursors. However, it also contains a mixture of other enzymes, including
arabinosidase, apiosidase, and rhamnosidase (Cabaroglu and others 2003),
hydrolyzing aroma polysaccharides. Although the addition of the commercially
available glycosidase preparation enhances the aroma of the wine during
winemaking, it may generate off flavors from volatile phenols released by cinnamate
esterase (Dugelay and others 1993) and may hydrolyze, although at lower
specificity, anthocyanidins leading to the loss of wine color (Wightman and others
1997; Sarry and Günata 2004).

It has been reported that the major β-glucosidase producers among yeasts are the
non-Saccharomyces species (Miklosy and others 1995; Fernández and others 2000;
Strauss and others 2001). The β-glucosidases from non-Saccharomyces such as
Pichia anomala, Candida molischiana, and Candida wickerhamii are more tolerant
to the winemaking conditions and have a higher specificity toward glycosides than
other yeast species (Jutaporn and others 2009). For instance, Clemente and others
(2004) reported that Pichia fermentans has positive effects on the production of ethyl
caprilate and 2-phenyl ethanol, molecules that offer pleasant aroma to wine.
Meanwhile, Pichia fermentans produces acetaldehyde at a low level. Besides, the
Candida species can degrade malic acid (Dong-Hwan and others 2008) and
Hanseniaspora uvarum liberates both glycoconjugated forms of pyranic and furanic
oxides from linalool (Fernandez and others 2003). Recently, Hu and others (2016)
found a novel extracellular glycosidase preparation from an indigenous Rhodotorula
mucilaginosa strain selected from a winemaking region in China, which exhibits
strong tolerance towards winemaking conditions and shows high specificity for the
hydrolysis of glycosides containing benzenic derivatives and C13-norisoprenoids.
Furthermore, OIV Codex Oenologique International (2015) accepts the use of
exogenous enzyme preparations including glycosidase in the winemaking process.
The glycosidases preparation from non-Saccharomyces yeasts could be applied in
winemaking working synergistically with the glucosidases from Saccharomyces in a
mixed fermentation to enhance the release of aroma glycosides.

Chemical acidic hydrolysis

Aroma glycosides can be hydrolyzed under acidic conditions. Compared with

enzymatic hydrolysis, chemical acidic hydrolysis is much cheaper and less limited.
In theory, acidic hydrolysis is preferred to the enzymatic hydrolysis by
endoglycosidases, as the chemical process will not be inhibited by the high content
of glucose. Moreover, the products from a chemical acidic hydrolysis may reflect
more closely the natural states of the free aromatic compounds in grapes and wine
because wine is produced under acidic condition.

However, the natural acidic hydrolysis of aroma glycosides during wine fermentation
accounts for only a very small proportion of the overall hydrolysis (Loscos and others
2007). Besides, the chemical acidic hydrolysis, especially occurring at low pH
values, may rearrange the chemical structures of the released aglycones (Ugliano
and others 2006; Hampel and others 2014), which makes the identification of the
chemical structure of the glycosidic precursor very difficult. For example, Williams
and others (1982c) found that linalool and α-terpineol are the main chemically
hydrolyzed (at pH 3.2) products stemming from the glycosides of linalool, geraniol,
and nerol. Perestrelo and others (2012) also pleaded against the chemical acidic
hydrolysis by evaluating the catalytic capacities of chemical acidic and enzymatic
hydrolyses for grape aroma glycosides using released volatile compounds. They
reported that although the acid hydrolysis released more terpenoids and C 13-
norisoprenoids (59 free volatiles in total) than the enzymatic hydrolysis (42 free
volatiles in total), some volatile compounds (for example, monoterpene oxide) were
presumably artifacts formed in the sample preparation. In addition, the acidic
hydrolysis produced 17% of unknown ion fragments.

This characteristic of chemical acid hydrolysis is very much pronounced during the
formation of C13-norisoprenoids. When the glycosidic precursors of C13-
norisoprenoids are hydrolyzed, one odorless compound is 1st produced, which is in
turn converted via acid-catalyzed rearrangement to form a volatile aromatic
compound (Sefton and others 2011). For example, the β-damascenone, whose
precursors have been identified in fruit juices including grape juice (Skouroumounis
and others 1992), was formed during the acid hydrolysis or the acid-enzymatic
combined hydrolysis from the precursors (Figure 2). The last step in Figure 2 show
the molecule rearrangement occurring during the formation of β-damascenone
(Winterhalter and others 1990; Kinoshita and others 2010).
Figure 2.

 Open in figure viewer

 Download Powerpoint slide

Formation of β-damascenone from a glycosidic precursor.

In addition to the artifacts production during chemical acid hydrolysis, the

comparison of catalytic efficiency with enzymatic hydrolysis has been investigated
using different methods. For example, Loscos and others (2009) compared the
catalytic effects of the chemical acidic hydrolysis (pH 2.5, 100 °C, 1 h) to the
enzymatic hydrolysis (AR2000, pH 5.0, 40 °C, 16 h) for different grape aroma
glycosides. The results indicated that the enzymatic hydrolysis efficiency is higher
than the one of the chemical acidic hydrolysis. Furthermore, Hampel and others
(2014) compared acidic and enzymatic hydrolysis using a model mixture containing
free (linalool, ethyl decanoate, β-ionone) and glycosidically-bound (n-octyl-, n-
dodecyl-, phenyl-β-d-glucopyranoside) chemical standards. The results showed that
acidic hydrolysis released 20% to 60% of the glycosidically-bound compounds and
degraded more than 50% of the free compounds, whereas enzymatic hydrolysis
effectively released 90% to 100% of the glycosically-bound compounds.

Restated, enzymatic hydrolysis and acidic hydrolysis present their corresponding

shortcomings and advantages, and the method for hydrolysis should be selected
based on the experimental objectives.

Analysis of Glycosidic Aroma Precursors

Glycosidic aroma precursors account for the majority of grape and wine aromatic
compounds. To understand the formation of wine aroma and the influencing factors,
we have to determinate the grape and wine aroma profiles by accurate analysis of
the type and content of the glycosidic aroma precursors. The analysis of the
glycosidic aroma precursors is conducted by following 3 basic steps: preparation of
the sample, purification of the sample, and chemical and structural identification of
the precursors. Each step involves the use of some key operational methods
selected by the specific research objectives required. Figure 3 briefly illustrates these
successive steps and methods applied in this study.

 Open in figure viewer

 Download Powerpoint slide

Figure 3. Flow-chart of the analysis of glycosidic aroma precursors.

Preparation of the samples containing glycosidic aroma precursors

Extraction is a basic but crucial 1st step because the results will directly influence
the performances of the following steps, especially the liquid or gas chromatographic
analysis. The choice of the extraction process will strongly depend on the state of
the starting material. For example, a solid material should be 1st crushed and then
infused to extract the glycosides, although a liquid material like wine samples may
be directly enriched and purified just after a simple centrifugation.
The main physiological location of glycosidic aroma precursors was found in berry
skin, and most of the neutral varieties present only a minute amount of aroma
components in their flesh. Thus, the skins or seedless homogenates of grape berries
are more relatively suitable raw materials compared to grape juice, because of their
higher glycosidic-aroma component contents (López and others 2004). Generally,
the extraction methods are either solid–liquid extraction or homogenization–
centrifugation. The solid–liquid extraction method imitates the extraction process
during wine fermentation. Glycosidic aroma precursors diffuse from grape skin or
pulp cells to the extraction solvent such as ethanol, ethanol–water mixtures, or model
wine systems (Palomo and others 2006). This extraction process enables the
effective collection of aromatic glycosides while maintaining the integrity of that part
of the cell. In addition, solid–liquid extraction is currently a popular method due to its
operability and short time-consumption, and the preparation of the model wine plays
a pivotal role in the entire extraction process. For instance, Hernandez and others
(2008) selected a phosphate buffer (0.1 M Na2HPO4/NaH2PO4, pH 7.0, 13% [v/v]
ethanol) to extract the aroma precursors of 4 nonaromatic grape varieties
(Garnacha, Verdejo, Chardonnay, and Tempranillo). Alternatively, a model wine with
12% (v/v) ethanol at pH 3.2 was used to extract the glycosidic aroma precursors of
Fiano grapes (Ugliano and Moio 2008).

Homogenization followed by centrifugation ruptures the grape cell-wall and removes

macromolecular impurities from the cytochylema containing glycosidic aroma
precursors (Skouroumounis and others 1995; Singh and others 2011). This method
is also widely used as it offers the advantage of releasing more glycosidic aroma
precursors than the solid-liquid extraction through the destruction of the cellular
structure. However, the homogenization-centrifugation process only achieves a
partial separation of the aroma glycosides and the product still contains many
impurities. Therefore, an additional step involving filtration over a microporous filter
might be necessary to remove the impurities and reduce potential damage on the
solid-phase extraction (SPE) cartridge in the next step.

In summary, the separation of glycosidic aroma precursors could be achieved either

by solid–liquid extraction or homogenization-centrifugation. The 1st method is based
on the collection of target glycosides in an extraction solution, although the 2nd
method separates glycosides from cellular debris and macromolecular impurities
from the cytochylema of the raw materials. Recently, the 2 methods were combined
for the recovery of highly pure extracts (Hernandez-Orte and others 2008; Crupi and
others 2010; Rodríguez-Bencomo and others 2011a). A summary of the studies
reporting the preparation of grape or wine aromatic glycoside extracts and
corresponding analysis is given in Table 3.

Table 3. Different analytical methods of grape aroma glycosides

 Identificati
Extraction Referenc
Variety Buffer solution Glycosides on
method e
methods

Cortese,
Tartaric acid
Sauvignon Solid-liquid Terpenes, Enzymolysi Fernánde
buffer [pH 3.2, (5
Blanc, extraction norisoprenoids s, GC-MS z (2004)
g/l tartaric acid)]
Chardonnay

Chassag
Centrifugatio Enzymolysi ne and
Chardonnay —— Terpenes
n s, GC-MS others
(2005)

Glucosides,
apiosylglucosi Palomo
Water with 5 g/L
Muscat “a Solid-liquid des, Derivatizati and
tartaric acid (pH
petit grains” extraction rutinosides, on, GC-MS others
3.5)
apiosylglucosi (2006)
de

A buffer solution GC/MS, Hernand

Verdejo, Solid-liquid
(pH 7) (0.1 M LC/ESI- ez-Orte
Chardonnay, extraction, Terpenes,
Na2HPO4/NaH2 MS, and
Garnacha, homogenizat norisoprenoids
PO4, 13% MALDI- others
Tempranillo ion
ethanol v/v ) TOF-MS (2008)

Enzymolysi
Kang and
Centrifugatio Monoterpenes, s, HS-
Muscat —— others
n alcohols SPME -GC-
(2010)
MS

Chardonnay, Solid-liquid
HPLC- Crupi and
Merlot, extraction, Hexane/diethyl Norisoprenoid
DAD-MS others
Negroamaro, homogenizat ether (1:1) s
(ESI+) (2010)
Primitivo ion

Tartaric buffer of
Albarossa, pH=3.2 ( 8 g/L of Papini
Barbera, Solid-liquid tartaric acid, Terpenes, Enzymolysi and
Nebbiolo di extraction NaOH 1 N, norisoprenoids s, GC-MS others
Dronero double distilled , benzenes (2010)
water, 3 g/L
sodium
 Identificati
Extraction Referenc
Variety Buffer solution Glycosides on
method e
methods

metabisulfite, 20
% ethanol v/v)

A hydroalcoholic
Rodrígue
solution (0.1 M
Solid-liquid z-
NaH2PO4·H2O;
Gual, Listán extraction, Terpenes, Enzymolysi Bencomo
13% ethanol v/v;
blanco homogenizat norisoprenoids s, GC-MS and
pH 7.0 adjusted
ion others
with a NaOH
(2011a)
solution)

Martin
Gewürztrami Centrifugatio Enzymolysi and
Distilled H2O Terpenes
ner n s, GC-MS others
(2012)

Rodrígue
A buffer solution
Enzymolysi z-
(0.1 M
Solid-liquid s, HS- Bencomo
Muscat Na2HPO4/NaH2 Terpenes
extraction SPME-GC- and
PO4, pH 7, 13%
MS others
ethanol v/v)
(2013)

Martínez-
Filtration and
Syrah, Terpenes, Enzymolysi Gil and
centrifugatio ——
Chardonnay norisoprenoids s, GC-MS others
n
(2013)

pH=7 buffer Norisoprenoid Muñoz-

solution (0.1 M s, vanillins, Enzymolysi González
Liquid-liquid
Verdejo Na2HPO4/NaH2 Benzenoids, s, SPE-GC- and
extraction
PO4, 13% lipids MS others
ethanol v/v) derivatives (2014)

Enrichment and purification of glycosidic aroma precursors

GC-MS and liquid chromatography-mass spectrometry (LC-MS) are techniques

enabling the identification of individual glycosides, providing the samples have been
properly enriched and purified (Osborne and others 1993; Boido and others 2013).
During the enrichment and purification treatment, most impurities must be removed
and the appropriate concentrations of the target aroma glycosides must be achieved.
A typical enrichment-purification process of the crude glycosidic-extraction involves
3 successive steps comprising rotary evaporation, column chromatography, and
dissolution in a constant specific volume of a solvent. The chromatographic column
is usually packed with pretreated adsorption materials, such as XAD-2 resin
(Winterhalter and others 1998; López and others 2004; Cid and others 2009;
Rodríguez-Bencomo and others 2011a,2011b; Schievano and others 2013), XAD-4
resin (Gueguen and others 1996; Luan and others 2002; Benkwitz and others 2012),
or C18 column resin (Boido and others 2002; García and others 2011; Roland and
others 2011).

Glycosidic aroma precursors can adsorb on the column due to the interactions
between the polar groups of the precursors and the chemical groups on the surface
of the packing material in SPE cartridge. Then, distilled water and pentane (or
mixtures of pentane/methylene dichloride) are used to elute highly apolar
compounds and those of low polarity, respectively. Finally, the glycosidic substances
are washed-out by methanol, methanol-ethyl acetate, or mixtures of methylene
dichloride/methanol. Packed columns are extensively used on substantial amounts
of samples, crude extracts of aroma glycosides or grape homogenate supernatants
because they present high repeatability and few requirements for preconditioning.
The selection of the packing materials is critical as the performance of an SPE
column depends on the selectivity of the packing material over aroma glycosides.
The adsorption behaviors will influence the final purity and concentration of the
aroma glycosides and sequentially have an impact on the results of the quantitative
and qualitative analysis. For instance, Arévalo and others (2006) achieved
enrichment through the selective retention of terpenyl-β-d-glycosides when the
sample was passed through a C18 reverse-phase column. As a traditional method,
packed cartridges have been widely used in the extraction of aroma glycosides, but
on the other hand, specific packing materials can be very expensive, and much time
is spent in activating them. Moreover, there is a poor selectivity towards some aroma
glycosides presenting low polarity and difficulty to elute highly polar glycosides. In
order to obtain the desired recovery of the target aroma-glycosides from the
adsorption and desorption processes, it is necessary to select the suitable packing
materials to minimize the loss of the target compounds. The relation between the
chemical structure of aroma glycosides and the selective adsorption behavior of the
packing materials is to date difficult to elucidate.

The above-mentioned SPE technology has been applied generally for the
enrichment-purification process of glycosidic aroma precursors (Salinas and others
2012). The most popularly used C18-SPE cartridges for the purification of grape
aroma glycosides are the LiChrolut EN SPE cartridge (Ibarz and others 2006;
Hernandez and others 2008; Hernandez and others 2009; Martínez-Gil and others
2013) and Strata-X 33u Polymeric Reserved Phase SPE cartridge (Martínez-Gil and
others 2013). The SPE cartridges for column chromatography are fast and usually
provide efficient purifications. However, this method still presents the disadvantage
to require highly purified crude extracts prior to the loading on the SPE cartridge.
Moreover, due to the low treatment volume and limited service-life, SPE is not
suitable for the enrichment and purification of trace aroma glycosides contained in
large volume samples. Nevertheless, SPE is a powerful tool due to the high
selectivity observed for specific glycosidic aroma compounds.

After an SPE treatment, the target glycosidic aroma compounds are enriched and
most of the interfering impurities are washed out. However, further ultra-purification
is sometimes required for glycosides subjected to highly sensitive detection methods
such as high-performance liquid chromatography (HPLC) or GC. For example, to
avoid the influence of phenolic glycosides before quantifying the glycosidic aroma
precursors, polyvinlypolypyrrolidone (PVPP; García and others 2011; Martínez-Gil
and others 2013) and basic lead acetate (BLA; D'Ambrosio and others 2013;
Schievano and others 2013) can be used to absorb high levels of phenolic
compounds for removal in a subsequent centrifugation process. Moreover, to
quantify the total content of glycosidic aroma precursor, Salinas and others (2012)
used Fehling's reagent to remove traces of free glucose and fructose in the extracts
of aroma precursors before chemical acid hydrolysis. However, minute amounts of
glycosidic aroma precursors can be lost in the purifying steps such as adsorption of
the impurities, elution process, centrifugation, and membrane filtration before
detection. Therefore, in the case where the concentrations of the target glycosidic
aroma precursors are too low for the next step of analysis, a 2nd enrichment is
required (Table 4).

Table 4. Published methods of enrichment and purification of grape aroma

precursors

1st enrichment 2nd enrichment

Compo
unds to
Variet Final Referen
Column Eluant Column Eluant be
y solvent ce
analyze
d

Skourou
mounis
and
Multilay
Winterh
er coil “Analytic EtOAc/BuOH/
Norisop alter
counterc al” coil H2O (3:2:5) A
Riesli CHCl3/MeOH/ renoids (1994);
urrent flash pentane/ –
ng H2O (7:13:8) glycosid Winterh
chromat chromat EtOAc
es alter
ography ography gradient
and
MLCCC
Skourou
mounis
(1997)
 1st enrichment 2nd enrichment

Compo
unds to
Variet Final Referen
Column Eluant Column Eluant be
y solvent ce
analyze
d

Corte Methanol–
se, water, Fernánd
Water, Citrate- Aromati
Sauvi Methanol– ez and
C18-RP dichlorometh C18-RP phosphat c
gnon water, Di
cartridge ane, cartridge e buffer heterosi
Blanc, Methanol– Stefano
methanol (pH 5.0) des
Chard water, (2004)
onnay Methanol

Melon DOWEX
C18 Varietal Roland
B., 50WX4–
cartridge thiol and
Sauvi 100 ion Water Water, MeOH Water
(Sep- precurs others
gnon exchang
Pak) ors (2011)
Blanc e resin

Water, Water, Terpeni Rodrígu

pentane/dichl pentane/dichl c and ez-
Amberlit Lichrolut
Musc oromethane, oromethane, Milli-Q benzeni Bencom
e XAD-2 SPE
ut ethyl ethyl water c o and
resins cartridge
acetate/meth acetate/meth glycosid others
anol anol es (2013)

Water,
Aglycon Muñoz-
pentane:dichl 40 % Water
LiChrolu es of Gonzále
Verdej Amberlit oromethane, Methanol, dichloro
t EN aroma z and
o e XAD-2 ethyl dichlorometh methane
cartridge glycosid others
acetate:meth ane , pentane
es (2014)
anol

Besides the above-mentioned separation operation, countercurrent chromatography

(CCC), size exclusion chromatography, and preparative liquid chromatography have
also been used for the preliminary purification of glycosidic aroma precursors
(Winterhalter 1993). For example, CCC with the support from nuclear magnetic
resonance (NMR) analysis was used to purify C13-norisoprenoid glycosidic
precursors from grape leaves of Riesling (Skouroumounis and Winterhalter 1994).
Compared with the size exclusion chromatography and preparative liquid
chromatography, CCC may reduce the content of the artificial components brought
by active surfaces and the adsorption of solid sorbents (Winterhalter and
Skouroumounis 1997).

Prior to the instrumental analysis, evaporation of the organic solvent used during
purification can be performed on a vacuum rotary evaporator or a Vigreux column to
concentrate the target components (Palomo and others 2006, 2007; Martínez-Gil
and others 2013). In addition, evaporation of the organic solvent can also be
undertaken by applying a gentle nitrogen flow to prevent the oxidation of sensitive
glycosidic aroma precursors. Unfortunately, all the numerous published methods
can only detect part of the grape glycosidic aroma precursors because of their
diverse structures. Therefore, future research on enrichment-purification may focus
on the development of an improved efficient separation method for extracts
containing differently structured glycosides.

Quantification of the total aroma glycosides

Williams and others (1995) developed a glycosyl-glucose (G-G) method to quantify

the total glycosides based on released glucose as each grape aroma glycoside
molecule contains a glucose moiety. To estimate the content of the total aroma
glycosides, the aroma glycosidic fraction is 1st extracted and purified from grapes or
wine, then hydrolyzed to release glucose, and finally, the total glucose (TGG) content
is quantified using an enzymatic assay. However, the hydrolysis process is a
sensitive step as undesirable hydrolysis of other glycosylated phenolic derivatives,
such as anthocyanidins and flavonoids, may occur. Therefore, the removal of other
glycosides is necessary to obtain a correct G-G concentration that reflects the total
content of glycosidic aroma precursors. For example, the G-G method was adapted
to red grapes by Iland and others (1996). They have quantified the content of
anthocyanins by a spectrophotometric assay and subtracted the obtained value from
the “total G-G” (TGG) concentration in order to calculate the so-called “red free G-
G.” However, the red free G-G method is only suitable for grapes containing
monomeric pigments but not appropriate to wine in which the pigments are present
under a polymeric form.

Furthermore, the G-G method can be extended by subtracting the content of the
phenolic glycosides other than anthocyanins. Indeed, the content of phenolic
glycosides in gallic acid equivalents can be obtained by using the Folin-Ciocalteu
reagent. The “phenol free G-G” (PFGG) value is obtained by subtracting the content
of phenolic glycosides from the above-mentioned TGG (Zoecklein and others 2000).
Because the content of phenolic glycosides is estimated as Gallic acid equivalents,
it is not possible to completely exclude the phenolic glycosides from the PFGG value.
Nevertheless, the G-G method and related modifications remain valuable for the
quantification of the total aroma glycosides. These methods have the advantages of
being very quick and inexpensive, and do not require any sophisticated equipment.
It should be noticed that the G-G method is still under development. For instance,
after enriching the aroma glycosides by a C18 reverse phase adsorbent and releasing
the glucose molecules by hydrolysis, Michlmayr and others (2010) quantified
accurately the content of d-glucose by HPLC with a CarboPac PA 1 column coupled
to a pulsed amperometric gold electrode detector. Improvements of enrichment and
purification techniques may enable the accurate determination of the G-G content
reflecting the total content of aroma glycosides.

Analysis of glycosidic aroma precursors

Spectroscopy might be a suitable method to acquire a sketchy profile of semipurified

glycosidic aroma precursors. The chemical identification of the compounds is
performed by analyzing on the absorbance spectrum by the shape and wavelength
of the peaks that are directly related to the molecular structure of the compounds.
For instance, Boido and others (2013) exploited the potential of NIR spectroscopy
coupled with a chemometric technique by determining the concentrations of different
glycosylated aroma compounds present in Tannat grape juice and seedless
homogenates. However, the identification and quantification of the multiple
absorption peaks exhibited by the relevant chemical groups are challenging due to
the overlap of the glycosides peaks. Therefore, direct spectral analysis of the
glycosides extract is usually unable to identify individual glycosides. Yet, both HPLC
and GC could identify glycosidic aroma precursors of purified samples (Hjelmeland
and Ebeler 2014). The identification process for aroma glycosides is presented in
Figure 4.

 Open in figure viewer

  Download Powerpoint slide

Figure 4. Identification process of glycosidic aroma precursors.

GC analysis

GC analysis is suitable for volatile aroma-substances. These substances include

free aglycones released by acidolysis or enzymolysis of glycosides, and also volatile
glycosides stemming from derivatization. As discussed in the preceding sections,
the chemical structure of glycosides can be deduced by GC. However, chemical
acidic hydrolysis might lead to molecular rearrangements. It is therefore important to
select the optimal conditions for the acidolysis or enzymolysis step. After hydrolysis,
solid-phase microextraction (SPME) fibers and solvent bar microextraction (SBME)
could be employed to collect directly free volatile substances for GC analysis.
Reported major studies using the GC method for the identification of glycosides are
referred in Table 2 and 3. The GC technique has shown to be more sensitive since
it allows the detection of more compounds compared with HPLC. However, the
allosteric response caused by the hydrolysis of several aromatic glycosidic
precursors might reveal erroneous original aglycone structures (Nasi and others
2008).

In addition, specific glycosides can be submitted to derivatization for suitable GC

analysis in order to analyze the whole molecular structures. This chemical treatment
includes acetyl, formyl, and silicane derivatization. In fact, derivatization with
trifluoroacetyl and trimethylsilane (TMS) are frequently used methods.
Trifluoroacetyl derivatization is generally applied to disaccharide glycosides (Wang
and others 2000; Ugliano and others 2006). Derivatization with TMS might damage
the capillary column of the instrument as residual TMS might react with other
compounds running through it, reducing the service life of the capillary column.
Consequently, this reaction might lead to additional cleaning and maintenance after
each analysis of the derivatized compounds. Furthermore, the uncontrolled
silanization of the glycosides might lead to incorrect identifications of the precursors
(Little 1999). Therefore, the derivatization method has been rarely cited in recent
reports and the detection of the whole molecular structures of glycosides is
nowadays performed by liquid chromatography.

HPLC analysis

Compared with GC analysis, HPLC can detect aroma glycosides without the need
of a hydrolysis pretreatment. This technique avoids thus the destruction of target
compounds occurring during hydrolysis and permits the study of intact glycosides
molecular structures (Winterhalter and others 1998; D'Ambrosio and others 2013).
In particular, hydrophilic interaction chromatography allows the simultaneous
investigation on hydrolysates and their isomers or the detailed structures of aroma
glycosides. Classical HPLC analysis requires enriched and purified glycosidic aroma
precursors with a reverse LC column. Then, the elution is performed with a mixture
of acetonitrile and water. The glycosides with different chemical groups are
separated based on the differences of their partition coefficients between the liquid
and solid phases. Recently, the development of LC-MS technology was a
breakthrough in the identification of glycosides.

Different from the electron ionization of GC-MS, HPLC adopts a soft-ionization

technique. However, this technique cannot detect all substances present in traces.
In order to acquire more structural information, HPLC still needs to be coupled with
tandem mass spectrometry (MS/MS) or NMR. For example, Schievano and others
(2013) successfully separated and identified 12 important glycosidic aroma
precursors in Moscato Giallo grape by using preparative HPLC and NMR. Hayasaka
and others (2010a,b) applied the HPLC-MS/MS technique in grapes to recorded
mass fragments assigned to guaiacol-β-d-glucosides. However, this method can
only be used to tentatively identify target glycosides. It is difficult to distinguish
isobars or isomers exhibiting similar peaks in the MS/MS spectra. Furthermore, this
method requires the comparison with related synthesized standards (Hayasaka and
others 2010b). More examples with regard to the application of HPLC technology in
this field can be found in Table 3.

It is worth noting that HPLC is less sensitive than GC as HPLC requires higher
concentrations of target components in the extract than GC. Therefore, the varieties
of glycosidic aroma precursors identified by HPLC are usually fewer than those
identified by GC. Many compounds have unique and typical MS/MS fragments.
However, the MS/MS information on the glycosidic fragments is up to now scarce in
standard mass spectral libraries. Consequently, the interpretation of the MS results
in the analysis of aroma glycosides is yet challenging. GC is applied more often than
HPLC in the identification and quantification of glycosidic aroma precursors because
of the above-mentioned drawbacks. However, HPLC still keeps its advantage in the
identification of intact structure for aroma glycosides.

Conclusion

Glycosidic aroma precursors are important reserves of grape and wine aroma
components. Investigations with regard to the chemical structures and the hydrolysis
mechanisms of these precursors are helpful to the exploitation of wine aroma. GC
remains the main method for identifying free components stemming from the
hydrolysis of glycosidic aroma precursors. However, GC results may not reflect the
original chemical structure of the glycosides because molecular rearrangements
could happen during hydrolysis. Therefore, studies on molecular rearrangements of
aglycones are beneficial in revealing the structures of glycosides. More
understanding in the effects of enzyme specificity and hydrolysis conditions is also
needed. Further researches on the hydrolysis of various glycosides by different
glycosidases would be especially beneficial to the identification of grape aroma
glycosides.
 

Explore this journal >

1. Previous article in issue: In Memoriam: Betsy Baird

2. Next article in issue: Accelerated, Microwave-Assisted, and Conventional
Solvent Extraction Methods Affect Anthocyanin Composition from Colored
Grains

Volume 79, Issue 2 February 2014 Pages R129–R137

Concise Reviews in Food Science

Potential of Medicinal Plants as Antimicrobial and Antioxidant Agents in Food

Industry: A Hypothesis

Abstract

Many food preservation strategies can be used for the control of microbial spoilage
and oxidation; however, these quality problems are not yet controlled adequately.
Although synthetic antimicrobial and antioxidant agents are approved in many
countries, the use of natural safe and effective preservatives is a demand of food
consumers and producers. This paper proposes medicinal plants, traditionally used
to treat health disorders and prevent diseases, as a source of bioactive compounds
having food additive properties. Medicinal plants are rich in terpenes and phenolic
compounds that present antimicrobial and antioxidant properties; in addition, the
literature revealed that these bioactive compounds extracted from other plants have
been effective in food systems. In this context, the present hypothesis paper states
that bioactive molecules extracted from medicinal plants can be used as
antimicrobial and antioxidant additives in the food industry.

Introduction
The search of new safe substances for food preservation is being performed around
the world (Magnuson and others 2013). Synthetic food additives are passing a
difficult season in addition to the great deal of time and money that is required to
develop and approve new synthetic preservatives, especially in view of the public
pressure against them (Tajkarimi and others 2010). The excessive use of synthetic
preservatives, some of which are suspected because of their toxicity, increased
pressure on food manufacturers to either completely remove these agents or to
adopt natural alternatives for the maintenance or extension of a product's shelf life
(Seneviratne and Kotuwegedara 2009). Such obstacles provide new opportunities
for those seeking natural alternatives for new food preservatives.

Many plant occurring bioactive compounds can be considered as good alternatives

to synthetic antimicrobial and antioxidant food additives (Cowan 1999; Silva-
Espinoza and others 2013). These compounds are mostly derived from plants and
their antimicrobial and antioxidant in vitro testing have resulted in many publications
in the last decade (Nakatani 2000; Yanishlieva and others 2006; Krishnaiah and
others 2011; Martins and others 2013). The antimicrobial and antioxidant properties
of bioactive compounds are mainly due to their redox properties, ability to chelate
metals, and quenching reactive species of singlet oxygen (Krishnaiah and others
2011). However, the selection of the plant sources to extract these compounds must
be guided for the safe use of food additives. A possible alternative could be medicinal
plants that have been used for thousands of years to treat health disorders and
prevent diseases.

Medicinal plant parts (roots, leaves, branches/stems, barks, flowers, and fruits) are
commonly rich in terpenes (carvacrol, citral, linalool, and geraniol) and phenolics
(flavonoids and phenolic acids), and these compounds have been effective as food
additives (Cai and others 2004). For example, lemongrass is a medicinal plant
utilized as stomachic, antispasmodic, carminative, and antihypertensive agent (Naik
and others 2010); in addition, it is a source of terpenes like citral that has shown
antimicrobial activity against food pathogen and deteriorative bacteria, and
antioxidant effect avoiding lipid peroxidation in food matrices (Ahmad and others
2012; Masniyom and others 2012). Other medicinal plants that could be used to
sustain the idea of generating extracts with potential as food additives are:
Chenopodium ambrosioides rich in terpenes (used to control menses disorders,
fibroids, uterine hemorrhage, and parasitic diseases); Euphorbia stenoclada rich in
phenolics (used to control skin diseases, gonorrhoea, migraine, intestinal parasites
and wart cures); Geranium mexicanum rich in terpenes and phenolics (used as
remedy against tonsillitis, cough, whooping cough, urticaria, dysentery and
diarrhea); Gnaphalium oxyphyllum rich in phenolics (used to treat gripe, fever,
asthma, bronchitis, and cough); Helianthemum glomeratum rich in flavonoids (used
to treat bloody and mucoid diarrheas and for the relief of abdominal pain); Larrea
tridentata rich in phenolic compounds (used to treat respiratory infections as
tuberculosis); Marrubium vulgare rich in terpenes and phenolics (used mainly as an
expectorant); Peumus boldus rich in phenolics and alkaloids (regulator of the hepatic
function, colagogue, antispasmodic, digestive stimulant, and nervous sedative);
Eysenhardtia polystachya rich in flavonoids (used to treat kidney and bladder
infections, diuretic, antispasmodic and febrifuge). Previous studies have
demonstrated that other plant extracts rich in similar bioactive compounds are
effective in food systems. With this in mind, the present hypothesis paper describes
the premises to sustain that “antimicrobial and antioxidant agents having
potentialities in the food industries can be obtain from medicinal plants.” This
manuscript also contemplates the sensorial and toxicological shortcomings to
achieve this goal.

Premise I. Need of Natural Additives for Food Preservation

Most of the food systems are highly perishable products attributed to their intrinsic
characteristics and environmental conditions. Quality and safety of foodstuffs are
compromised by the loss of nutrients, sensorial attributes, and microbial growth.
Their composition rich in substrates for microbial and enzymatic reactions cause
these problems (Ayala-Zavala and others 2008). Food safety is an important
international issue that is compromised by the presence of pathogenic bacteria
(Godfray and others 2010). Food poisoning is associated with organisms such as
Staphylococcus aureus, Salmonella spp., Bacillus cereus, Clostridium botulinum,
Clostridium perfringens, Listeria monocytogenes, Campylobacter jejuni, E. coli
O157:H7, and Toxoplasma gondii (CDC 2011). Produce, seafood, poultry, and beef
are mostly linked to foodborne illness outbreaks and are reported to be responsible
for 53% of all the cases and 48% of illnesses (CSPI 2013). With food safety and
pathogens, the path forward is clear: Increase monitoring and enforcement, develop
and apply good hygiene and disinfection, and communicate risks broadly to the
consuming public. Besides microbial growth, oxidation is another problem in
numerous food products.

Lipids are the most susceptible food molecules to oxidative processes; however,
proteins and carbohydrates can also be affected. Food oxidation not only results a
rancid taste and an alteration in color, flavor or texture; in addition, it can result in
deterioration of the nutritional qualities and even compromise safety (McClements
and Decker 2006), for example, the vegetable oils are exposed to oxidizing
environment during food processing and get oxidized (Brewer 2013). Oxidative
deterioration can also occur in refrigerated, frozen, cooked, cured, freeze-dried, and
irradiated meat products; and natural antioxidants can be used to retard this process
(Pukalskas and others 2011). Food safety assurance and oxidation control are
challenging situations even for modern preservatives techniques. Additionally,
consumers demand safe and quality food preserved with natural additives (Brewer
2013). In this context, the search for effective natural substances to avoid food
quality decay and assure safety is a major challenge to the food scientist.

Premise II. Antibiotic and Antioxidant Activity of Medicinal Plants

As described before the use of natural compounds to avoid food quality deterioration
and assure microbial safety has been of interest for consumers, producers, and
scientists, in this context, medicinal plants should be considered as an alternative
source to generate food additives. This potential is justified in the similarity in
composition of these plants with other plant extracts that have successfully been
applied to maintain food quality and assure microbial safety, in addition to the
ancestral human uses of most of these plants.

Cymbopogon citratus (D.C.) Stapf. (Lemongrass)

C. citratus is an Indian native plant widely used in the kitchen for food and beverage
preparation; infusions or decoctions of its dried leaves have also been used against
various diseases as stomachic, antihypertensive, carminative, antispasmodic,
relaxant, and antidepressant agent (Figueirinha and others 2008). The essential oil
of C. citratus is composed mainly by geranial, neral, and myrcene, and these are of
major importance for the flavor and fragrance industries (Kasali and others 2001).
Infusion extract from C. citratus exhibited strong activity against DPPH radical
(Figueirinha and others 2008). In another study, C. citratus extract tested at 2
concentrations (33 and 50 μg/mL) were effective reducing the DPPH radical (40%
and 68%) and superoxide anion production (15% to 32%), and at 500 μg/mL of these
extracts the oxidation of erythrocytes membranes were kept in range of 19% to 71%
(Cheel and others 2005). Similarly, the tannin and flavonoid fractions from essential
oil-free infusion of C. citratus were effective in preventing the production of reactive
oxygen species. DPPH inhibition (IC50) concentrations of Cymbopogon
schoenanthus L. ranged from 6.8 to 26.4 mg/mL, depending on the extraction
solvent and sampled region (Khadri and others 2010). The same extracts showed
good antimicrobial activity against Streptococcus sobrinus at low concentration
(MIC = 4 mg/mL).

C. citratus can be used as a phytochemical agent incorporated into polymer coatings

(Seabrook 2005). C. citratus oil was effective to inactivate Salmonella enterica on
romaine, iceberg lettuces, and baby spinach (Moore-Neibel and others 2012). The
essential oil and its majors components (citral and geraniol) were most effective for
inhibiting Streptococcus agalactiae and B. cereus than S. aureus and E. coli; in
addition, these compounds showed antibiofilm activity against S. aureus (Aiemsaard
and others 2011). The same study concluded that the antibacterial activity depended
on the concentration, compositions, and cell target sites. The state in the art of the
composition and uses of C. citratus exemplified clearly the statement that a
medicinal plant can be used to generate extracts effective as food additives, more
details will be describe in the next premise.

Euphorbia stenoclada Baill. (Silver thicket)

Stem infusion from E. stenoclada is traditionally used to heal respiratory disorders

(Chaabi and others 2007). Two flavonoids, quercetin and quercitrin, were identified
in extracts of this plant and indicates that phenolic compounds could be their
bioactive constituents. The antiproliferative activity of these compounds were
attributable to their molecular structure (free hydroxyls groups, methylation,
glycosylation) (Chaabi and others 2007). The antioxidant effects of quercetin were
comparable to that of catechin in AAPH-initiated peroxidation of liposomal
suspension (Terao and others 1994). Quercetin is also effective to control lipid
peroxidation in phospholipids (Terao and others 1994). Guinea pigs treated either
with quercetin (142.9 mg/kg) or quercetrin (214.3 mg/kg) were resistant to Shigella
sp. infection (Vijaya and Ananthan 1996). Quercetin is also a potent antibacterial
agents against, S. epidermidis, S. aureus, Bacillus subtilis, E. coli, and Micrococcus
luteus (Rauha and others 2000). The activity of quercetin has been at least partially
attributed to the inhibition of DNA gyrase (Cushnie and Lamb 2005). This information
revealed that the antioxidant and antimicrobial activity of E. stenoclada could be
attributed partially to phenolic compounds.

Geranium mexicanum Kunt (Geranium)

The genus Geranium is formed by almost 400 species in tropical mountains and
temperate areas throughout the world (Alanís and others 2005). Some geranium
species are used in traditional medicine as antidiabetic, hemostatic,
antihemorrhoidal, antidiarrheic and in treatment of tonsillitis, cough, whooping
cough, urticaria, dysentery, kidney pain, and gastrointestinal ailments (Calzada and
others 2005). In previous studies, different types of compounds such as tannins
(Calzada and others 2001; Calzada and others 2005), flavonoids (Şöhretoğlu and
others 2009), lignans (Liu and others 2006) as well as essential oils (Toshkova and
others 2004), have been isolated from Geranium species. The extracts from G.
mexicanum (<8 mg/mL) possessed strong antibacterial activity against Salmonella
spp., Shigella spp., and E. coli (Alanís and others 2005). Epicatechin and 2 flavan-
3-ols were the active compound of G. mexicanum inducing the reduction of Giardia
lamblia and Entamoeba histolytica (Calzada and others 2005). Results revealed that
(–)-epicatechin therapy lowers lipid peroxidation in streptozotocin-induced diabetic
rat tissues (Quine and Raghu 2005). Some studies have shown that (–)-epicatechin
is a good hydroxyl radical scavenger, since it contains benzene ring and such
aromatic compounds are known to have very high rate constants for reactivity with
the hydroxyl radical (Rice-Evans 1999; Aruoma 2003).

Helianthemum glomeratum Lag (Clustered frostweed)

H. glomeratum is an endemic Mexican herb with medicinal properties (analgesic in

stomach pain, antidiarrheic, antiparasitic, and antidysenteric drug) (Calzada and
Alanís 2007). The antidiarrheal and antiamoebic effect of this plant is attributed to
tiliroside, quercitrin, kaempferol, epigallocatechin, and isoquercitrin (Calzada and
others 1995; Calzada and Alanís 2007). In addition, these compounds could be
effective as antioxidant, antibiotic, antiviral, antiallergenic, antigiardial, antiamoebic,
antiinflammatory, and protective against cancer and heart diseases (Barbosa and
others 2007). There is a patent proposing extracts of H. glomeratum as a dietary
supplement for regulating the appetite and weight loss (Teisen-Simony and Saaby-
Nielsen 2009). As shown in the literature, H. glomeratum is a rich source of phenolic
compounds; these could be used as additives in the food industry.

Gnaphalium oxyphyllum DC. (Gordolobo)

Gnaphalium spp. are plants widely used in traditional medicine for the relief of
swellings, stomach diseases, wounds, lumbago, antimalarial, prostatism, neuritis,
diuretic, angina ache, antipyretic, and for the lowering of blood pressure (Hocking
1997; Taddei-Bringas and others 1999; Rojas and others 2001; Campos-Bedolla
and others 2005). Constituents have been isolated as diterpenoids, flavonoids,
acetylenic compounds, and carotenoids (Meragelman and others 2003). Other
isolated constituents from this plant are ent-Kaur-16-en-19-oic acid, ent-3-
hydroxykaur-16-en-19-oic acid, zoapatlin, 13-epi-sclareol, 13-epi-cyclosclareol,
luteolin, 3-methoxyquercetin, 5-hydroxy-3,7- dimethoxyflavone, β-sitosterol and
stigmasterol (Villagómez-Ibarra and others 2001). Extracts from this plant have
proved to be effective as antibacterial agents. The hexane extracts of flowers or
leaves from G. oxyphyllum var. oxyphyllum, G. liebmannii var. monticola and G.
viscosum showed higher inhibition against both S. aureus and B. cereus, except the
G. viscosum leaf extract, which only inhibited B. cereus (Villagómez-Ibarra and
others 2001). The G. oxyphyllum flower extract, characterized by the highest content
of ent-kaur-16-en19-oic acid, showed a broader spectrum of activity, extended to S.
Typhimurium and E. coli. Also, the methanol extract of G. oxyphyllum flowers, in
which luteolin and 3-methoxyquercetin were found, significantly inhibited S. aureus
and B. cereus. G. oxyphyllum was effective to inhibit pathogens bacterial as S.
aureus, E. coli, Enterococcus faecalis, Candida albicans, Streptococcus pyogenes,
and Streptococcus pneumoniae (Rojas and others 2001). No reports of the
antioxidant activity of this plant were found, however, its phenolic composition is an
indicative of this potential property.

Chenopodium ambrosioides L (Wormseed)

C. ambrosioides is known in different countries as American wormseed, mastruz,

goosefoot, paico, or epazote. Aqueous leaves extracts and essential oil of this plant
are used traditionally as dietary condiments and in traditional medicine against
menses disorders, fibroids, uterine hemorrhage, parasitic diseases, and inhibits the
Ehrlich tumor growth (Ososki and others 2002; Nascimento and others 2006; Cruz
and others 2007). The essential oil from C. ambrosioides leaves at 100 μg/mL
showed fungicide activity against Aspergillus niger, A. flavus, A. fumigatus, Fusarium
oxysporum, Botryodiplodia theobromae, Sclerotium rolfsii, Cladosporium
cladosporioides, Macrophomina phaseolina, Pythium debaryanum, and
Helminthosporium oryzae (Kumar and others 2007). Essential oil from C.
ambrosioides has a significant in vitro and in vivo effect on macrophages and mice
infected with Leishmania amazonensis, inhibiting the progression of the infection
(Patrício and others 2008). The constituents of C. ambrosioides (flavonoids,
terpenes, and steroids) have shown antioxidant properties and its oil showed
scavenging activity against ABTS radical (Kumar and others 2007).

Larrea tridentata (DC.) Coville (Creosote bush)

L. tridentata grows in Sonora and Chihuahua deserts (Mexico), and Mojave (United
States); extracts from its stems and leaves are useful to treat tuberculosis, menstrual
pains, diabetes, and various cancers (Lambert and others 2005). Composition
studies indicated that flavonoids, triterpenes, and lignans (nordihydroguaiaretic acid,
NDGA) are the main constituents of this plant having efficacy to inhibit reactive
oxygen species and proliferation of human promyelocytic leukemia cells (Abou-
Gazar and others 2004; Jitsuno and Mimaki 2010). L. tridentata extracts were
effective reducing Fe by FRAP assay and inhibit the free radical DPPH; this
antioxidant activity might be explained by the high content of phenolic and NDGA
(Martins and others 2010). Three lignans (4-epi-larreatricin; dihydroguaiaretic acid;
3′-demethoxy-6-Oethylisoguaiacin) and 4 flavonoids (5,4′-dihydroxy-3,7,8-
trimethoxyflavone; 5,4′-dihydroxy-3,7,8,3′-tetramethoxyflavone; 5,8,4′-trihydroxy-
3,7-dimethoxyflavone; 5,4′-dihydroxy-7-methoxyflavone) from L. tridentata were
effective to inhibit the growth of S. aureus, Mycobacterium tuberculosis,
Enterobacter cloacae, E. faecalis, and E. coli (Martins and others 2013). L. tridentata
extracts have antifungal activity in vitro against at least 17 fungal pathogens of
economic importance; likewise, extracts and ground plant material incorporated into
the soil as powder inhibited 6 fungi that affect agricultural crops (Gamboa-Alvarado
and others 2003; Lira-Saldívar 2003). Flavonoid constituents from L. tridentata were
effective against viruses that cause polio, AIDS, and herpes (Graham and others
2000).

Marrubium vulgare L. (Horehound)

M. vulgare is a herb commonly known as “horehound” or “Marrubia,” this plant

possesses aromatic, tonic, diuretic, stimulant, diaphoretic, expectorant,
hypoglycemic, antidiabetics, stomachic, and antiparasites properties (Román and
others 1992; Moreno-Salazar and others 2008). The aqueous extracts from M.
vulgare elicit a high degree of antioxidant activity (VanderJagt and others 2002);
however, the ethyl acetate fraction was effective in reducing DPPH radical (50%) at
a low concentration (11.67 ± 1.51 Wg/mL) (Matkowski and others 2008). M. vulgare
leaves composition consisted of vitexin, vicenin II, luteolin 7-glucoside, apigenin 7-
(6”-p-coumaroyl) glucoside, apigenin, chrysoeriol, and luteolin (Nawwar and others
1989). The presence of these flavonoid compounds could be responsible for the
antioxidant activity of this plant. M. vulgare essential oil presented antibacterial
activity against Gram positive bacteria (S. epidermidis, S. aureus, Micrococcus
luteus, E. faecalis, E. cloacae, S. aureus, B. subtilis, and B. cereus) showing
inhibition zones ranging from 6.6 to 25.2 mm and MICs from 1120 to 2600 μg/mL,
while Gram negative bacteria (P. aeruginosa, K. pneumoniae, E. coli, and
Salmonella sp.) exhibited a higher resistance (Zied and others 2011). In addition,
this oil was effective against B. cinerea; however, A. niger, Penicillium digitatum, and
Fusarium solani were less sensitive (Zied and others 2011). Extracts from this plant
are used by Ricola® (Laufen, Switzerland), which is one of the manufacturers of
sweets candies made from M. vulgare in conjunction with other medicinal plants
(Pimpinella saxifrage, Veronica officinalis, Althaea officinalis, Alchemilla vulgaris,
Sambucus nigra Elderberry, Malva sylvestris, Mentha piperita, Salvia officinalis,
Achillea millefolium, Primula veris, Plantago lanceolata, and Thymus vulgaris) and
exports around 30 different types of candy and herbal teas to over 50 countries.
Horehound also serves as raw material for herbal extracts and beverage industries
as a substitute for hop in beer-breweries and can be used as an ingredient of cough
pastilles (Weel and others 1999).

Even when the antimicrobial and antioxidant activities of medicinal plants have been
studied with details, however, several research gaps were noticed: (i) the effect of
the solvent polarity in composition and antimicrobial and antioxidant efficacy of the
extracts should be contemplated; (ii) the antimicrobial effect of the evaluated plants
extracts is not generally directed to antibiofilm antibacterial tests, omitting that most
of the bacterial problems is caused by this organization of microbial growth; (iii)
despite the antioxidant and antimicrobial composition and uses of medicinal plants
in human health, the research about their potential use as food additives is few
compared with other plants with similar composition as spices, herbs, fruit, and
vegetable tissues.

Premise III. The Composition of Medicinal Plants is an Indicative of the

Potential Uses as Antimicrobial and Antioxidant Agents

As mentioned before, the clearest example of the uses of medicinal plants as food
additives is lemongrass. This plant has shown efficacy to inhibit food pathogen and
deteriorative bacteria, anticarcinogenic effects, and presents a fresh, juicy, citrus-
like aroma and taste (Table 1). Lemongrass and their major component (citral)
showed complete inhibition of fungal growth on tomatoes inoculated with B. cinerea
and Alternaria alternata (Plotto and others 2002). A 4 log reduction of E. coli
O157:H7 on fresh cut apples was achieved incorporating lemongrass essential oil
(0.7%) and citral (0.5%) into alginate coatings (Raybaudi-Massilia and others 2009).
Lemongrass oils and their active compound citral have also shown promising
antimicrobial and quality effects on fresh-cut melon (Raybaudi-Massilia and others
2009). In the last studies, this compound was used at low concentrations compared
to the lethal dose in rats (LD50 = 4960 mg/kg). In addition, this oil is listed as a
substance generally recognized as safe (21CFR182.20) (FDA 2013b). The
combination of C. citratus and C. zeylanicum essential oils (0.1 μL/mL), applied on
strawberry fruits inoculated with E. coli, reduced 5 log to 5 d of freeze storage (Duan
and Zhao 2009). The oil of this plant was used as antibacterial treatment against
Salmonella Newport inoculated on green leafy vegetables, where the greatest
reduction was observed in iceberg lettuce (4.3 log), followed by spinach, mature
spinach, and romaine lettuce; however, in this study the sensorial effect of the
treatment was not considered (Moore-Neibel and others 2012). On the other hand,
the mixture of C. citratus and turmeric acid applied in mussel reduced the microbial
counts and retarded lipid oxidation compared to controls, additionally the C. citratus
oil was found sensorially more acceptable than turmeric acid treatment to preserve
the odor and flavor of foods (Masniyom and others 2012). The addition of
lemongrass oil into gelatin film could enhance the antimicrobial and antioxidative
properties of sea bass slices (Ahmad and others 2012). This information reveals the
positive results of lemongrass, to preserve quality and assure food safety.
Table 1. Antimicrobial, antioxidant, odor/flavor, and toxicity of bioactive compound
from medicinal plants

Oral
Compo Antimicrobi Antioxidant Odor/flav Referenc
Plant toxicity
und al effect effect or notes es
(LD50)

1. NDA, no data available.

Spicy,
cooling,
(DPPHIC50 = (Zeytinogl
thymol-
448.05, 433), u and
like, herbal
(ABTS-+IC50 = others
and
1.77), (β- 2003;
camphore
C. carotene/linolei Safaei-
Carvacr E. coli, S. ous with 810 mg/k
ambrosio cIC50 = 50.18, Ghomi
ol aureus smoky g in rats
ides 8.13. and
nuances/S
CytotoxicityIC50 others
picy,
= 60 μg for 2009;
herbal,
CO25 Öztürk
phenolic,
myoblast cells 2012)
medicinal
and woody

Fresh,
juicy,
Iemon
peel, with a
sweet (Aiemsaa
tangy rd and
E. coli, B. Cytotoxic
green others
C. cereus, S. activity against 4960 mg/
Citral nuance/Le 2011;
citratus aureus, S. HeLa cell line kg in rats
mon peel, Das and
agalactiae. (100 μg/mL)
citrus, others
juicy, 2013)
green,
lime,
woody and
herbal

E. coli, S. (β- Citrus, >5610 mg (Belsito

C. aureus P. carotene/linolei orange, /kg in and
Linalool
citratus aeruginosa, cIC50 = 2.7). floral, rabbits others
B. cereus, K. Cytotoxic terpy, waxy and 2008;
 Oral
Compo Antimicrobi Antioxidant Odor/flav Referenc
Plant toxicity
und al effect effect or notes es
(LD50)

pneumoniae, activity against and 2790 mg/ Ebrahima

S. HeLa cell line rose/Citrus kg in rats badi and
epidermidis, (100 μg/mL) , orange, others
S. lemon, 2010;
dysenteriae, floral, Das and
P. vulgaris, S. waxy, others
paratyphi-A aldehydic 2013)
serotype, C. and woody
albicans, A.
niger

Floral,
sweet,
rosey,
fruity and
(Belsito
E. coli, S. citronella-
>5000 mg and
entérica, S. like with a
/kg in others
aureus, S. citrus
C. rabbits 2008;
Geraniol agalactiae L. nuance/Flo
citratus and Aiemsaar
mononocytog ral, rosy,
3600 mg/ d and
enes, B. waxy and
kg in rat others
cereus perfumey
2011)
with a
fruity,
peach-like
nuance

Terpy,
herbaceou
s, woody
with a rosy
E. coli, S. celery and (Aiemsaa
β- Tiphymurium, carrot
C. 5000 mg/ rd and
Myrcen S.agalctiae, nuance/W kg in rats
citratus others
e S. aureus, B. oody, 2011)
cereus vegetative,
citrus,
fruity with a
tropical
mango and
 Oral
Compo Antimicrobi Antioxidant Odor/flav Referenc
Plant toxicity
und al effect effect or notes es
(LD50)

slight leafy
minty
nuances

Harsh
chemical,
woody and
terpy-like
with an
oxidized
citrus
lemon
note. It has
spicy
nuances
(DPPHIC50 = reminiscen (Cosentin
>1000), t of cumin, o and
E. coli, S. +
p- C. (ABTS- IC50 = oregano 4750 mg/ others
aureus, B.
Cymene citratus >1000), (β- and kg in rats 1999;
cereus
carotene/linolei cilantro/Te Öztürk
cIC50 = 388 rpy and 2012)
rancid with
slightly
woody
oxidized
citrus
notes. It
has spice
nuances of
green
pepper and
oregano

B. cereus, S. (DPPHIC50 = 980 mg/k (Rao and

G. aureus, M. 35.6 μM, g in rats others
Kaempf mexican luteus, L. 14.3 μM). and 2007;
erol um, L. Monocytogen cytotoxic H460 3200 mg/ Santas
tridentata es, P. cell apoptosis kg in and
and G. aeruginosa, to 80 μM, rabbits others
cytotoxic 2010;
 Oral
Compo Antimicrobi Antioxidant Odor/flav Referenc
Plant toxicity
und al effect effect or notes es
(LD50)

oxyphyllu activity in cells Gao and

m jurkatIC50 = others
48.2 μM 2011)

L. (DPPHIC50 = (Begum
Apigeni tridentata E. coli, S. 115.3 μM), and
n and M. aureus (ABTS-+IC50 = Prasad
vulgare 107.8 μM, 2012)

(Drewno
wski and
Gomez-
Carneros
2000;
Rao and
others
(DPPHIC50 =
2007;
C. 19.98 μg/mL,
Ramadan
perfringens, 3.07μmol/L,
and
B. subtilis, B. 4.9 μM), (β-
M. others
cereus, S. carotene/linolei
vulgare, 2009;
aureus, S. cIC50 = 3.64
E. Céspede
Querceti lutea, M. μmol/L, 6.67 Odorless/b 161 mg/k
polystach s and
n flavus, E. coli, μmol/L). itter g in rats
ya and E. others
P. cytotoxic
stenocla 2010;
aureuginosa, activity in cells
da Hirai and
E. coli, L. jurkatIC50 =
others
Monocytogen 8.4 μM, No
2010;
es, toxic to
Santas
5 μg/mL
and
others
2010;
Radovan
ović and
others
2012)
 Oral
Compo Antimicrobi Antioxidant Odor/flav Referenc
Plant toxicity
und al effect effect or notes es
(LD50)

(Drewno
wski and
Gomez-
Carneros
2000;
B. subtilis, S. Katalinić
G. >10000 m
aureus, S. and
mexican g/kg in
lutea, M. (FRAP = 2.0), others
Catequi um, E. Odorless/b rats and
flavus, M. (DPPH%inihibition 2004;
n polystach itter >10000 m
catarrhalis, = 94) Cueva
ya and P. g/kg in
Streptococcu and
boldus mice
s sp. others
2012;
Radovan
ović and
others
2012)

(Drewno
wski and
Gomez-
G.
M. Carneros
mexican
catarrhalis, 2000;
um E.
Epicate Streptococcu (DPPHIC50 = Odorless/b Rao and
polystach
chin s sp., S. 7.4 μM) itter others
ya and G.
agalactiae, S. 2007;
oxyphyllu
pneumonia Cueva
m
and
others
2012)

(Heron
Nordihy No toxic and
dro- L. A. flavus, A. DCFHIC50 = to Yarnell
NDA
guaiaret tridentata parasiticus 0.7 μg/mL 300 mg/k 2001;
ic acid g Vargas-
Arispuro
and
 Oral
Compo Antimicrobi Antioxidant Odor/flav Referenc
Plant toxicity
und al effect effect or notes es
(LD50)

others
2005)

(Konrath
and
others
TRAP = 10 μM >2000 mg 2008;
Marrubii M. Odorless/b
NDA Trolox /kg in Paula de
n vulgare itter
equivalents mice Oliveira
and
others
2011)

IC50 =
0.66 mg/ (Falé and
Boldine P. boldus NDA NDA NDA mL for others
HeLa 2012)
cells

Based in the composition of other medicinal plants, it is possible to establish a

connection with their potential efficacy as food preservatives. As stated before, some
of the most common bioactive compounds of medicinal plants are terpenes and
phenolics, these constituents found in other plant tissues have shown to be effective
inhibiting microbial growth and/or oxidation reactions in several food matrices. This
premise can support the statement that extracts obtained from medicinal plants, with
similar composition, could be effective as food additives.

Carvacrol is a bioactive compound with antimicrobial and antioxidant capacity that

can be found in C. ambrosioides, a medicinal plant used to treat menses disorders,
fibroids, uterine hemorrhage, and parasitic diseases. Carvacrol (LD 50 = 810 mg/kg
in rats) is listed as a flavor and fragrance substance that may be safely used in food
according to FDA (2013a) and the European Union Regulation (2008) (Nr
1334/2008). This flavor agent presents spicy, cooling, thymol-like, herbal, and
camphoreous with smoky odor notes; and a spicy, herbal, phenolic, medicinal, and
woody flavor notes. A mixture of carvacrol and 1, 8-cineol in different proportions
were effective reducing native microflora and inoculated bacteria (L.
monocytogenes, Aeromonas hydrophila, and P. fluorescens) on fresh cut-
vegetables (de Sousa and others 2012). In addition, in the same study the
vegetables exposed to the mixture of carvacrol and 1,8- cineole obtained sensorial
scores significantly higher than the control after 48 h of storage. The addition of
carvacrol controlled the fungal growth of B. cinerea in grapes without affecting the
tissue integrity; however, this study did not contemplate the odor and flavor effect of
the treatment (Martínez-Romero and others 2007). Natural microflora of kiwifruit was
reduced (4.6 log CFU) by the treatment with carvacrol (0.15 μL/mL) after 21 d of
storage at 4 °C; though, the efficacy was lower (2 log CFU reduction) on honeydew
melon stored 5 d at 4 °C, this variation was attributed to the difference in fruits pH,
this study did not contemplate the sensorial impact (Roller and Seedhar 2002). Tsao
and Zhou (2000) found that sweet cherries treated with the combination of carvacrol,
thymol, and methyl jasmonate was effective to control fungal growth of Monilinia
fructicola. Chinese bayberries were treated with carvacrol (1 μL/L), reducing fruit
decay significantly and increasing total phenolic, anthocyanin, and individual
flavonoid compounds (Jin and others 2012). The panelists’ rankings for sensory
acceptability of carrot treated with carvacrol, thymol, and cinnamaldehyde indicated
a negative impact, due to their strong flavors; the addition of carvacrol or thymol as
food preservatives was unacceptable to the panelists (Valero and Giner 2006). This
information shows the attention that carvacrol is receiving as food additive; even
when more research is needed regarding the sensorial impact of its use. Considering
the presence of this compound in C. ambrosioides, extracts from this plant can
present a similar potential.

Phenolic compounds are receiving attention to be considered as food additives, and

they are a well distributed group in medicinal plants. For example, E. polystachya,
P. boldus, G. mexicanum, and E. stenoclada contain phenolic compounds as
catechin (LD50 = >10000 mg/kg in rats and <10000 mg/kg in mice) and quercetin
(LD50 = 161 mg/kg in rats), and these compounds have been reported as food
additives in other tissues. Catechin oligomers from apple tissues were reported to
be effective inhibitors of cholesterol oxidation in commercial meat products such as
pork sausage, raw and roast ham, bacon, and hamburgers (Osada and others 2000).
Similarly, other flavonoids (for example, quercetin) found in a lot of plant tissues,
including medicinal plants, are successfully employed to inhibit fish oil oxidation
(Valenzuela and others 1991).

Medicinal plants have varied tastes predominantly sweet, bitter, pungent, or

astringent, some of which are particularly complex as they have several active
constituents. For example, there is an invention claiming the use of gallic acid
(monophenolic compound) as sweetness inducer, it has distinct advantages over
other sweet inducers. The advantage of gallic acid over other sweeteners is that the
induced sweetness is noncaloric, long-lasting and does not have any nonsweet
aftertaste (for example, aspartame has a metallic aftertaste). The initial taste of gallic
acid itself is only mildly sour, in contrast to the unpleasant bitter taste of catechins
(Giza and others 2002). Whereas lower-molecular-weight phenolic compounds tend
to be bitter, higher-molecular-weight polymers are more likely to be astringent
(Drewnowski and Gomez-Carneros 2000). Products, which are predominantly bitter
or astringent, such as phenolics, need to be taste-masked in order to improve
compliance. To achieve a more effective treatment of food products (sensorially
accepted) using extracts with compounds from medicinal plant several technologies
must be contemplated: nanoemulsions, nanocapsules, vapors, edible films to
optimize the responses (antimicrobial, antioxidant, and sensorial). The generation of
this information will be profound on the antimicrobial and antioxidant knowledge and
effective uses of plant extracts on food matrices

Conclusions and Future Research

Despite the antioxidant and antimicrobial composition of medicinal plants in human

health, the research about their potential use as food additives is negligibly
compared with other plants of similar composition as spices, herbs, fruit, and
vegetable tissues. These functional properties may be due to the terpenes and
phenolic contents that can act as the principal contributors of the antioxidant and
antimicrobial power of botanical materials. This hypothesis manuscript supports the
idea that some medicinal plants may be good source of antimicrobial and antioxidant
agents for the food industries. The research around this topic reflects that several
studies have explored the extraction, antioxidant, and antimicrobial efficacy of
medicinal plant tissues; however, the practical use of these sources of bioactive
compounds as food additives has been overlooked. Future research is suggested
for exploring different medicinal plants traditionally used as antimicrobial and/or
antioxidant treatment. Different extraction procedure could be performed to
characterize the composition of the medicinal plants and special attention should be
paid to this process considering that new molecules could be identified, even when
the focus should be in those structures indicating a safe use in food systems.
Different technologies, such as nanoemulsions, edible films or coatings, controlled
release from encapsulation systems, among others, can be used to apply medicinal
plant extracts in food matrices. Moreover, the impact of the treatments on the
sensorial characteristic of the treated products should be contemplated. Finally, the
conception of functional foods could be considered through the addition of the
medicinal plants extracts; however, several experimental approaches could be
considered to prove this statement. This proposal looks forward to promoting the use
of natural extracts and fulfilling consumer demands for healthier foods.

Explore this journal >

Early View

1. Previous article in Early View: Preparation of Chitosan-Coated

Nanostructured Lipid Carriers (CH-NLCs) to Control Iron Delivery and Their
Potential Application to Food Beverage System
2. Next article in Early View: Next-Generation Sequencing Analyses of Bacterial
Community Structures in Soybean Pastes Produced in Northeast China

Concise Reviews & Hypotheses in Food Science

Vitamin Fortification of Fluid Milk

Eileen B. Yeh, David M. Barbano and MaryAnne Drake

Volume 79, Issue 4April 2014 Pages R428–R441

Abstract

Vitamin concentrates with vitamins A and D are used for fortification of fluid milk.
Although many of the degradation components of vitamins A and D have an
important role in flavor/fragrance applications, they may also be source(s) of off-
flavor(s) in vitamin fortified milk due to their heat, oxygen, and the light sensitivity. It
is very important for the dairy industry to understand how vitamin concentrates can
impact flavor and flavor stability of fluid milk. Currently, little research on vitamin
degradation products can be found with respect to flavor contributions. In this review,
the history, regulations, processing, and storage stability of vitamins in fluid milk are
addressed along with some hypotheses for the role of vitamin A and D fortification
on flavor and stability of fluid milk.

Practical Application

Many of the degradation components of vitamins A and D have flavor/fragrance

applications, but there is little published research on their possible flavor
contributions to fluid milk. Vitamin concentrates can impact flavor and flavor stability
of fluid milk. Proposed mechanisms of off-flavor development and changes in flavor
stability of fluid milk are discussed in this review.
Introduction

Current fluid white milk consumption in the United States has steadily declined in the
last 50 years (International Dairy Foods Association 2008; USDA 2014). As a result,
the level of dietary vitamin D provided by fluid milk in the U.S. diet has also declined
(Moore and others 2004; Whiting and Calvo 2011; Looker and others 2011).
Increased fortification of vitamin D in fluid milk and fortification of a larger variety of
products with vitamin D have been recommended (Calvo and Whiting 2003; Dietary
Guidelines Advisory Committee 2015). Milk is commonly displayed under
fluorescent or light-emitting diode (LED) lighting in retail dairy cases. Longer storage
periods increase the risk of light exposure, which increase the risk of altering milk
quality before purchase (Johnson and others 2015). Both fluorescent and LED
lighting sources deliver light energy in UV and visible wavelengths regions that cause
light oxidation of photosensitive molecules such as riboflavin in milk. Light oxidation
is the most common source of flavor and nutritional problems in milk. Another
possible source of flavor variability in fluid milk is vitamin fortification.

Vitamin fortification has a long history in fluid milk in the United States to reduce
rickets in children, and the FDA mandated in the 1990s that fortified fluid milks must
be within 100% to 150% of label claims to address documented variability in vitamin
amounts (Public Health Service 1940). Reduced fat and skim milks are required to
be fortified with vitamin A at a minimum of 2000 International Unit (IU) and this is
optional for whole milk. All fluid pasteurized milk must be fortified with vitamin D at a
minimum of 400 IU. Vitamin fortification is a standard procedure for pasteurized fluid
milks in the United States and vitamin concentrates are added to milk before
pasteurization (PMO 2015). There are 2 types of vitamin concentrates: oil soluble
and water dispersible formulations (Murphy and Newcomer 2001). Recent work has
addressed vitamin D fortification of cheesemilk (Wagner and others 2008; Ganesan
and others 2011; Tippets and others 2012), processed cheese (Upreti and others
2002), and yogurt (Hanson and Metzger 2010) but the role of vitamin concentrate on
flavor and flavor stability of fluid milk has not been addressed. Vitamin A is sensitive
to light exposure and rapidly degrades. This degradation process occurs more
quickly in skim milk compared to milks with fat in them (Senyk and Shipe 1981;
Whited and others 2002). The 2 previous studies suggested that vitamin A
concentrate imparted an off flavor to skim milk, but the specific role of the actual
vitamin source and the carrier on off flavor or light oxidized flavor was not evaluated.
It is of key importance for the dairy industry to strategically position vitamin
fortification and enhance fluid milk quality. Understanding the possible role of vitamin
fortification on off flavor and light oxidized flavor in fluid milk is of great importance.

The hypotheses of this review are as follows: (1) vitamin concentrates can contribute
flavor to fluid milk, and (2) vitamin concentrates can increase light oxidized off flavors
in fluid milk and/or contribute off flavor(s) to fluid milk.

Origin of Vitamin Fortification in Fluid Milk

Fortification is defined as the process of adding micronutrients such as essential
vitamins to food (Alvarez 2009). The fortification of food products has been practiced
for more than 80 y. Vitamin deficiencies that lead to rickets, ariboflavinosis, and
pellagra in the U.S. are reduced by the consumption of foods fortified with vitamin D,
vitamin B2, and niacin, respectively (West and others 2002; Bishai and Nalubola
2002). Vitamin D fortification of milk and milk products began in the 1930s. This
practice was recommended by the American Medical Association Council on Foods
and Nutrition to assist in reducing rickets in children (Stevenson 1955). Various
methods such as animal feed supplementation and direct irradiation were applied to
increase vitamin D content in milk, but the direct addition of vitamin concentrates
proved to be most reliable and has become the accepted industry practice
(Roadhouse and Henderson 1950). The wide acceptance of milk fortification with
vitamin D led to the fortification of milk with vitamin A, which was initiated in the
1940s (Public Health Service 1940). The goal of fortifying milk with vitamin A was to
ensure that Americans have a constant source of vitamin A because milk provided
10% of American consumers’ food energy.

The Need for Vitamin Fortification

Vitamin D

Vitamin D is essential for calcium absorption and is involved in the mineralization

process required for bone growth. Deficiency of vitamin D causes rickets (softening
of bones) in children and osteomalacia in adults (Ceglia 2009). Recent studies also
suggest that vitamin D plays a role in prevention of prostate, breast, and colorectal
cancers (Grant and others 2007; Schwartz and Skinner 2007; Garland and others
2006; Bouillon and others 2006). The major vitamin D forms responsible for human
health benefits are ergocalciferol (D2; Figure 1) and cholecalciferol (D3; Figure 2).
These vitamin D forms are considered to be inactive until they are converted to their
biologically active form, 1,25-dihydroxy vitamin D3, in the liver or kidney (Holick
1995).
Figure 1.

 Open in figure viewer

 Download Powerpoint slide

Chemical structure of ergocalciferol or vitamin D2.

Figure 2.

 Open in figure viewer

 Download Powerpoint slide

Chemical structure of cholecalciferol or vitamin D3.

Vitamin D is an essential vitamin that is made when the body is exposed to sunlight
(Holick 1994). However, no photosynthesized vitamin D is produced in the skin for
several months of the year for those who live in northern latitudes during winter, and
supplementation of vitamin D is required to prevent deficiency (Calvo and others
2004; Weaver and Fleet 2004). Vitamin D3 is found mainly in fish products and fish
liver oils (Kutsky 1981). However, most foods typically consumed by humans are low
in vitamin D content. Fish contains approximately 120 to 500 IU of vitamin D3 per 3-
oz serving, which is 50% to 200% of the recommended daily intake level as opposed
to less than 25% in unfortified grains, meats, vegetables, and breakfast cereal
(Holden 2009). Vitamin D3 can also be produced synthetically by purification of 7-
dehydrocholesterol from animal products and converted to vitamin D3 by irradiation
(Kutsky 1981). This synthetic form of vitamin D3 is added to many foods, particularly
milk products.

Vitamin A
Vitamin A is needed for normal growth, vision, reproduction, and differentiation of
epithelial cells. Vitamin A deficiency results in night blindness, xerophthalmia
(progressive blindness caused by drying of the cornea of the eye), keratinization
(accumulation of keratin in digestive, respiratory and urinary-genital tract tissues)
and finally exhaustion and death (Zile and Cullum 1983). Vitamin A is a group of
compounds that includes retinoids and carotenoids (Zile and Cullum 1983). Vitamin
A from animal sources is already in a form of retinol that can be easily absorbed by
the human body, whereas vitamin A from plant sources is a carotenoid that the body
can transform into a retinol (Zile and Cullum 1983). Vitamin A palmitate (Figure 3) is
a form of vitamin A found naturally in animal sources and also produced
synthetically. Whole milk, butter, cheese, and eggs are also important dietary
sources of vitamin A palmitate (Zile and Cullum 1983). Vitamin A fortification in
reduced and fat-free milks is required because whole milk contains some vitamin A
palmitate; however, vitamin A levels in reduced and fat free milks are much lower
because fat soluble vitamin A palmitate is removed with fat. Therefore, fortification
of vitamins in dairy products has been one of the approaches to address vitamin A
deficiencies (Parish and Richter 1979).

Figure 3.

 Open in figure viewer

 Download Powerpoint slide

Chemical structure of vitamin A palmitate.

Vitamin Regulations in Fluid Milk

In 1935, vitamin fortification of milk was addressed in the U.S. Public Health Service
Milk Ordinance and Code, and in 1939, Vitamin D Milk was defined as milk in which
the vitamin D content was increased by a method and in an amount approved by the
regulating health official (Public Health Service 1940). The Milk Ordinance and Code
did not specify any fortification levels, but did recommend monitoring vitamin D
concentrations by bioassays in a laboratory approved setting as required by the
health officer. In 1953, the Milk Ordinance and Code established a level of at least
400 IU per quart (qt) for vitamin D milk fortification. This document also addressed
the option of fortifying milk with other vitamins and minerals under “Fortified Milk and
Milk Products,” although no concentrations were specified for other nutrients. The
Milk Ordinance and Code continued to be revised and became the Grade “A”
Pasteurized Milk Ordinance (PMO) in 1965 (Public Health Service 1965).

According to the U.S. Food and Drug Administration (FDA) regulations as specified
in the PMO, the acceptable fortification concentrations were 80 to 120% of the label
claims for both vitamins A and D. Over fortification of fluid milk with vitamin A and D
can cause intoxication, soft tissue damage, and kidney failure (Jacobus and others
1992; Blank and others 1995). Based on FDA regulatory mandate (Public Health
Service 1965), fluid milk products with levels of vitamin D over 800 IU and vitamin A
over 6000 IU in fluid milk are considered a public health threat and should be
prohibited from sale and distribution. In 1978, the PMO required that each processor
monitor vitamin concentrations in fortified products at least once a year in a FDA
certified laboratory using standard test methods such as high performance liquid
chromatography (HPLC) methods (Public Health Service 1978).

New regulatory standards for vitamin fortification in the 1990s arose due to issues
with variability in the vitamin content of retail milk. Over fortification of milk with
vitamin D in one processing plant resulted in human illness from consumption of
these products (Jacobus and others 1992). Several studies also found that
numerous fortified milk products failed to meet label claims (Brown and others 1992;
Holick and others 1992; Nichols 1991). Murphy and others (2001) reported that 53%
to 55% of fortified milk products were out of compliance with the label contents for
vitamin A and for vitamin D. Milk fortification practices were not entirely consistent
with the Nutritional Labeling and Education Act of 1990. As a result, the FDA revised
the accepted fortification levels and mandated that vitamin concentrations for
fortified fluid products must be within 100% to 150% of the label claims (Nichols
1992), which equates to 400 to 600 IU per quart for vitamin D and 2000 to 3000 IU
per quart for vitamin A (PMO 2015). If fluid milk products are found below 100% or
above 150% of the required values or label claims, they should be resampled and
the source of the problem determined (PMO 2015).

Fortification was not specified for low-fat or skim milks. Low-fat milk was first defined
in the 1965 PMO as containing not less than 0.5% and not more than 2% fat. Skim
milk was defined as milk with fat removed to a content of less than 0.5%. In the 2015
PMO, 21 CFR 130.10 states

“That nutrients must be added to the food to restore nutrient levels so that the product
is not nutritionally inferior to the standardized food for products which combine a
nutrient content claim, i.e., low-fat, non-fat, or reduced fat, with a standardized term,
i.e., milk, sour cream, eggnog.”

Therefore, vitamin A should be added to dairy products from which fat has been
removed, in an amount necessary to replace the amount of these vitamins lost in the
removal of fat. Unfortified whole milk is not considered to be a significant source of
vitamin D because natural vitamin D concentrations in whole milk range from 0.34
to 0.84 IU per gram of fat (McBean and Speckmann 1988). Thus, removal of fat does
not render milk nutritionally inferior for vitamin D, and therefore fortification with
vitamin D is optional for all milks. If added, vitamin D must be present at 400 IU per
quart (PMO 2015). Most fluid milk in the United States is fortified with vitamin D due
to the importance of vitamin D in human nutrition (Holick and others 1992; McBean
and Speckmann 1988; Miller and others 2000).

The fortification of milk with vitamin D almost eliminated the public health concern of
rickets in the 19th century. However, vitamin D deficiency has reemerged as a global
health concern. Current low intakes of vitamin D and reduced time in sunlight
exposure (increased time spent indoors) have resulted in inadequate vitamin D
status (Dietary Guidelines Advisory Committee 2015). There has been interest in
fortifying dairy products with higher levels of vitamin D. Patterson and others (2010)
found that on average, vitamin D3 in 2% milk was higher in 2007 compared with the
vitamin D3 levels in 2001, with a trend toward more samples of whole milk having
greater than 150% of the labeled content. Hanson and Metzger (2010) reported that
increasing the fortification of vitamin D from 100 to 250 IU per serving was stable
over the shelf lives of HTST-processed 2% milk, UHT-processed 2% chocolate milk,
and low-fat yogurt, and no change in sensory characteristics was found in these
products. They concluded that increasing the fortification of vitamin D in milk was a
feasible strategy to increase vitamin D supplementation.

Fortifying products with higher amounts of calcium or vitamin D is permitted by FDA

provided that the product and any label claims comply with FDA regulations including
standards of identity (U.S. Food and Drug Administration, 2013). The addition of
vitamin D to milk and yogurt is optional. Unfortified cheese and cultured dairy
products are not considered good sources of vitamin D. If added, the minimum
amount of vitamin D in each quart of the product must not be less than 400 IU (PMO
2015). The acceptable range for vitamin D is 400 IU to 600 IU per quart of milk (PMO
2015).

Vitamin A can be found in significant amounts in unfortified whole milk because it is

primarily associated with the fat phase of the milk at 37.7 IU per gram of fat (McBean
and Speckmann 1988). However, milk fat removal results in vitamin A reduction in
low-fat and skim milks. As demand for low-fat and skim milk products increased in
the United States, there was a nutritional concern for reduction of vitamin A present
in these products. This concern was addressed in the 1978 PMO (Public Health
Service 1978), which required low-fat and skim milks to be fortified with vitamin A to
the nutritional equivalence of the general milk standard-containing not less than 2000
IU per quart. Vitamin A fortification is optional for whole milk, but if added, the
concentration must not be less than 2000 IU per quart.

To ensure proper fortification of dairy products, it is important to effectively monitor

product concentrations. The latest revision of the PMO (2015) mandated that each
manufacturer have representative milk products assayed at least annually in a
laboratory certified by the FDA with methods that are acceptable to the FDA.
Currently HPLC methods are the testing procedures of choice for vitamins A and D
in milk. FDA is involved in the standardization of testing procedures among
laboratories and has implemented a certification program for vitamin testing
laboratories. Current listings of certified laboratories can be found in the IMS
Quarterly Report. Samples for vitamin testing should be sent to an approved
laboratory within a few days of processing, kept cold (≤4 °C) without freezing and
protected from light. Once received at the laboratory, analyses should begin as soon
as possible.

Vitamin Premixes and Addition in Fluid Milk Processing

In the United States, vitamin D is generally added to fluid milk as synthetic vitamin
D3, whereas vitamin A is added as synthetic retinyl palmitate (Public Health Service
1965). Synthetic vitamin D3 is made from irradiation of animal fat, usually from
lanolin, the waxy secretions from sheep skin (Budvari 1996; Holick 1999; Smith
2016). Retinyl palmitate is made from combining vitamin A acetate with methyl
palmitate in the presence of sodium hydroxide (Budvari 1996; Smith 2016). There
are 2 different forms of vitamin premix: oil based and water dispersible formulations.
These are available for the various dairy processing systems (van Deutekom 2015).
In general, oil-based vitamin premix has to be added into the flow of milk after the
cream separator. The water dispersible vitamin premix can be added to the flow of
milk before the separator, or anywhere in the milk flow (van Deutekom 2015). Water
dispersible vitamins are not water soluble, only oil soluble. An emulsifier
(polysorbate) is added into the vitamin premix to make it water dispersible. Water
dispersible vitamin premix (water as the main ingredient) has a specific gravity >1.00,
whereas oil-based premix (oil as the main ingredient) has a specific gravity <1.00
(van Deutekom 2015). Most vitamin premix neither contains nor is manufactured
from any genetically modified materials (van Deutekom 2015). Oil-based vitamin
premix with corn oil as the carrier may be manufactured with commodity corn oil and
therefore cannot be certified as GMO free (van Deutekom 2015). Vitamin premixes
contain vitamin D3 and/or vitamin A palmitate in a carrier generally consisting of a
combination of any of the following: sunflower oil, corn oil, water, polysorbate 80,
propylene glycol, and glycerol monooleate. Antioxidants and/or preservatives may
also be added (Murphy and Newcomer 2001).

It is also important to note that vitamin concentrate potency will degrade with time.
Therefore, concentrates should be stored in accordance with manufacturer's
recommendation to maintain label potency (PMO 2015). According to Sensory
Effects (Bridgeton, Mo., U.S.A.) and International Food Products (IFP; Fenton, Mo.,
U.S.A.), 2 commercial suppliers, vitamin premix product should be stored at room
temperature (10 to 27 °C) in a dry, dark place. Product should be used within 1 year
from date of manufacture. IFP does not make any natural claims for their vitamin
premix products (van Deutekom 2015). Sensory effects has an organic label claim
on some of their vitamin premix products that contain sunflower oil as the vitamin
carrier (Payne 2015).

Vitamins can be added into the pasteurizing vat, the HTST balance tank, or on a
continuous basis into the pipeline after standardization. The addition of vitamins
usually occur after separation and fat standardization, and before pasteurization.
Homogenization will then take place after pasteurization to allow the vitamins to be
distributed evenly throughout the milk. Two vitamin addition procedures can be used:
the batch addition or addition with metering pumps. The batch procedure requires
accurate measurement of the fortified milk volume, accurate measurement of the
vitamin concentrate, and proper mixing. The metering pump procedure requires the
pumps in the HTST unit to be activated only when the unit is in forward-flow (PMO
2015).

Under or over fortification can occur when vitamins are added before separation and
standardization, resulting in low fat product being under fortified and high fat product
being over fortified. This occurs because vitamin A and D are fat-soluble, they will
gradually become more concentrated in the milk fat portion of the milk. Therefore, it
is recommended to add the vitamins after separation and standardization (PMO
2015).

Methods of Vitamins A and D Analysis

A list of the official AOAC (2007), CEN (2000), and ISO (2000) methods available
for determining fat soluble vitamins has been reported (Blake 2007). These
procedures involve mostly liquid chromatography (LC), but also include
spectrophotometric and gas chromatographic (GC) techniques. In the various official
procedures for fat soluble vitamins, extractions are usually made either by
saponification or by direct solvent extraction. Saponification removes the fat portion
of milk and facilitates extraction by releasing carotenoids, retinoids, tocopherols, and
vitamin D compounds from the matrix. Saponification is generally performed with the
addition of antioxidants such as ascorbic acid, pyrogallol, butylated hydroxyl toluene,
or hydroquinone to reduce oxidation losses, along with nitrogen flushing (Blake
2007). After saponification, extraction takes place with organic solvents such as
hexane, petroleum ether, ethyl ether or mixtures of these substances(Table 1).
Vitamin D can also be directly extracted with organic solvents without saponification
process. These organic solvents are methyl dichloride alone, or mixed with methanol
and hexane alone, or mixed with ethyl ether, or chloroform (Kazmi and others 2007).
However, vitamin D recoveries from direct extraction were lower than those obtained
by saponification (Delgado and others 1992; Hagar and others 1994). Vitamin D3 is
insoluble in water, but soluble in 95% ethanol, acetone, fats, and oils, and readily
soluble in chloroform and ether. Retinyl palmitate is also insoluble in water, but
soluble in chloroform, ether, and vegetable oil (corn oil), and slightly soluble in
alcohol.

Table 1. Vitamin analysis methods

 Product
Author Detection Compounds
Extraction method evaluate Sensitivity
name method separated
d

Detection
limits:
0.003
μg/100 mL
for vitamin
Cholecalcifer
D3, <1
Salo- Saponification ol (vitamin
μg/100 mL
Vaananen follow by extraction D3),
Fluid milk for beta-
and others with hexane-ethyl HPLC-UV tocopherols,
and fish carotene,
2000 (Food acetate (8:2), beta-
and 2
Chemistry) stream N2 carotene, all-
μg/100 mL
trans-retinol
for
tocopherols
and all-
trans-
retinol

Saponification Vitamin D3
Upreti and
followed by content of
others 2002 Cholecalcifer
extraction with Processe 2.5 μg/100
(Journal of HPLC-UV ol (vitamin
petroleum d cheese mL for
Dairy D3)
ether/diethyl ether, processed
Science)
stream N2 cheese

Milk,
liquid
Staffas and
infant Vitamin D3
Nyman
saponification formula, content
2003 Cholecalcifer
followed by cooking reported in
(Journal of HPLC-UV ol (vitamin
extraction with oil, milk: 0.4 to
AOAC D3)
heptane, stream N2 margarin 0.418
Internationa
e, infant μg/100 mL
l)
formula,
fish oil

Direct extraction 95% to 97%

Kazmi and with Cheese, Cholecalcifer recovery for
others 2007 methanol/chlorofor HPLC-UV yogurt, ol (vitamin cheese,
(Internation m, stream N2 ice cream D3) 99% to
102% for
 Product
Author Detection Compounds
Extraction method evaluate Sensitivity
name method separated
d

al Dairy yogurt,
Journal) 98% to 99%
for ice
cream

Canned
salmon,
vitamin
D3
Phillips and Vitamin D
fortified
others 2008 Extraction with ethyl content
skim milk,
(Journal of ether/petroleum Cholecalcifer reported in
HPLC-UV orange
Food ether and ol (vitamin skim milk:
and MS juice,
Compositio separatory funnel, D3) 1.03 μg/100
ready-to-
n and stream N2 mL to 1.15
eat
Analysis) μg/100 mL
breakfast
cereal,
processe
d cheese

Skim
milk,
orange Vitamin D
Byrdwell
Extraction with ethyl juice, content
2009
ether/petroleum cereal, Cholecalcifer reported in
(Journal of HPLC-UV
ether and salmon, ol (vitamin skim milk:
Agricultural and MS
separatory funnel, spiked D3) 1.08 μg/100
and Food
stream N2 peanut mL to 1.14
Chemistry)
oil, μg/100 mL
processe
d cheese

Saponification
Chen and Detection
followed by All-trans-
Reddy 2015 Skim, 2% limits:
extraction with retinyl
(Journal of fat, and
diethyl ether in HPLC-UV palmitate and 0.0022
AOAC whole
hexane and cholecalcifero μg/mL for
Internationa milk
separating funnel, l (vitamin D3) vitamin A
l) and 0.0008
stream N2
 Product
Author Detection Compounds
Extraction method evaluate Sensitivity
name method separated
d

μg/mL for
vitamin D3

Sensitivity
was higher
for UPLC
than for
HPLC (that
Chauveau-
Saponification is for
Duriot and Compariso Forages,
follow by extraction Carotenoids, retinol: 0.8
others 2010 n between bovine
with hexane-ethyl retinol, and 1.4 ng
(Analytical HPLC & plasma,
acetate (8:2), tocopherols per
Bioanal UPLC and milk
stream N2 injection vs.
Chem)
1.3 and 2.0
ng per
injection,
respectivel
y)

Skim, 1%
fat milk,
AOAC
2% fat All-trans-
Official Extraction with
LC-UV milk, and retinyl
method hexane, stream N2
1% fat palmitate
2002.06
chocolate
milk

Fluid milk,
infant
Saponification
formula,
AOAC followed by
gruel, Cholecalcifer
Official extraction with
LC-UV margarin ol (vitamin
method heptane and
e, D3)
2002.05 separatory funnel,
cooking
stream N2
oil, and
fish oil

For over 20 years, HPLC has been the method of choice for the determination of
total fat-soluble vitamins such as vitamins A and D. However, HPLC methods raise
environmental and economic concerns due to the large amount of organic solvents
used. Thus, the continued search to improve HPLC further have led to ultra-high
performance liquid chromatography (UHPLC). In comparison with HPLC, UHPLC
uses narrow bore, shorter columns, less run time, lower flow rate, lower injection
volume and solvent volume/sample, smaller particle size, and much higher back
pressure (Bohoyo-Gil and others 2012; Rivera and Canela-Garayoa 2012). Thus,
the UHPLC method has several advantages over conventional chromatography,
which include faster analyses due to shorter retention times, narrower peaks giving
increased signal-to-noise ratio, higher resolution and sensitivity (Rivera and Canela-
Garayoa 2012). UHPLC also saves at least 80% of mobile phase compared to HPLC
(Chen and Kord 2009). Similar to the HPLC methods, UHPLC methods also have to
undergo special attention to method standardization, validation, sampling, and
sample preparation to guarantee data reliability, because the smaller the analytical
sample, the more difficult it is to guarantee a representative sample. In milk, the
UHPLC method provided similar concentrations of vitamin A to that obtained by an
HPLC method (Chauveau-Duriot and others 2010).

In all of these procedures, it is important to avoid vitamin losses due to light oxidation.
Significant errors may result from poor use of glassware in vitamins analysis.
Therefore, the use of fluorescent lighting fitted with an appropriate UV filter in the
laboratory is recommended. Low-actinic glassware (amber colored glassware which
is used to protect contents from light) can also be used to reduce vitamin losses
(Castanheira and others 2006).

Vitamin Destruction by Photooxidation

The light visible spectrum is the portion of the range of all possible frequencies of
electromagnetic radiation that is visible to the human eye. The human eye can
typically respond to wavelengths from 390 to 700 nm (Starr 2005). Light exposure at
wavelengths below 500 nm causes the destruction of light-sensitive vitamins such
as riboflavin, vitamin A, and vitamin C, induces chemical reactions that affect milk
proteins and lipids, and results in the development of unpleasant flavors in fluid milk
(Fanelli and others 1985; Sattar and others 1977; Schroder and others 1985). Dairy
products are very sensitive to light oxidation because of the presence of riboflavin
(vitamin B2). Milk contains large amounts of riboflavin, a water-soluble vitamin with
an average concentration between 1.36 and 1.75 mg/L (Dimick 1982; Zygoura and
others 2004). Light oxidation dramatically alters the structure of riboflavin and
reduces its level in milk (Bekbolet 1990). Riboflavin is the most studied
photosensitizer in milk (Sattar and others 1977; Webster and others 2009), and this
strong photosensitizer is able to absorb visible and UV light and transfer this energy
into highly reactive forms of oxygen known as singlet oxygen (Boff and Min 2002).
Excitation of riboflavin occurs when exposed to light at 250, 270, 370, 400, 446, and
570 nm wavelengths (Kyte 1995). The cascade of oxidation reactions lead to
significant losses of vitamins (vitamin A, B2, C, D, and E) and amino acids, and also
leads to lipid oxidation and formation of strong off-flavors (Borle and others 2001).
As a result, off-flavors may be linked to the decrease in nutritional value of milk.
Therefore, light protection of photosensitive molecules in milk is needed to protect
milk flavor and nutrient quality. Despite vitamin loss and established changes in milk
flavor which are unpleasant to consumers, HTST milk in the U.S. is still routinely
packaged in clear HDPE jugs (Brotherson and others 2016; Potts and others 2017).

Studies have shown that whole milk products have slower rates of vitamin A losses
compared to low-fat or skim milk products (Senyk and Shipe 1981; deMan 1981;
Gaylord and others 1986; Whited and others 2002). Senyk and Shipe (1981)
indicated that light at wavelengths of 400 to 500 nm penetrated 40% to 50% deeper
into skim milk than into whole milk. Measureable vitamin A losses occurred at 2, 4,
and 16 hours at 2000 1× fluorescent light for nonfat, reduced fat, and whole milk,
respectively (Whited and others 2002). Vitamin A losses were also distinct between
fluorescent light and LED light exposure (Brotherson and others 2016).

Natural vitamin A in whole milk was more stable to light than added vitamin A due to
natural vitamin A is found in milk fat globules whereas added retinyl palmitate is
dispersed in the water phase of milk, which was more prone to oxidation due to
greater contact with oxygen (Thompson and Erdody 1974). Vitamin D loss occurred
at a rate much slower than other vitamins in milk. Using the same system of exposing
milk samples in test tubes at the same light intensity of 300 ft-c, Gaylord and others
(1986) reported the rate constant for riboflavin loss was 0.0616 per hour and retinyl
palmitate was 0.0298 per hour, whereas Renken and Warthesen (1993) reported the
rate constant for vitamin D loss was 0.0009 per hour. Chocolate milk products also
have reduced vitamin A degradation, either due to protection by carrageenan alone
or in combination with chocolate color and/or flavor. Vitamin A protection in chocolate
milk is due to increased light scattering by the additional particles. Chocolate flavor
components of chocolate milk can also reduce development of light-oxidized off-
flavors (Chapman and others 1998). These findings demonstrate that exposure of
fluid milk to light can adversely affect both flavor quality and nutritional value of fluid
milk products.

Vitamin Stability and Storage

Vitamin D

Several studies have been conducted on the stability of vitamin D in milk and other
dairy products (Banville and others 2000; Kazmi and others 2007; Wagner and
others 2008; Hanson and Metzger 2010; Tippetts and others 2012). These studies
have all indicated that vitamin D is stable during processing and storage. Vitamin D
in fortified homogenized whole milk is very stable and is not affected by
pasteurization or other processing procedures (PMO 2015). Vitamin D 3 appears to
be stable in cheese during both short-term (Banville and others 2000) and long-term
storage (Kazmi and others 2007; Wagner and others 2008). Vitamin D was also
stable over the shelf life of HTST-processed 2% fat milk, UHT-processed 2% fat
chocolate milk, and low-fat strawberry yogurt (Hanson and Metzger 2010). Tippetts
and others (2012) suggested incorporation of vitamin D3 as an emulsion using milk
proteins as emulsifier to improve retention of vitamin D3 in the cheese curd. Vitamin
D3 is also stable in yogurt and ice cream stored for 4 weeks, with high retention of
95% to 100% and 98% to 100%, respectively (Kazmi and others 2007).
Vitamin D itself is susceptible to degradation by oxygen, heat, and light once freed
from the protection of food matrix. Degradation products resulting from UV light
reactions include lumisterol, tachysterol, isotachysterol, and suprasterols (Bouillon
and others 1998; Grady and Thakker 1980). Other degradation products of vitamin
D also include trans-vitamin D (from oxidative reactions), pyrovitamin D, and
isopyrovitamin D (from thermal reactions; Grady and Thakker 1980). Octanoate and
decanoate ester of vitamin D3 were also identified as degradation products of vitamin
D3 in a thermally stressed tablet containing vitamin D3 (Ballard and others 2007).
Disagreement with respect to vitamin D stability due to other environmental factors
has been reported. Cremin and Power (1985) stated that vitamin D was unstable to
oxidation, light, and acid. Kutsky (1981) reported that vitamin D was unstable to
oxidation and light but stable to acid and alkali, whereas Kreutler (1980) reported
that vitamin D was remarkably stable to light, heat, and oxygen, and these factors
did not affect its activity.

Vitamin A

As stated previously, vitamin A is generally added to fluid milk as retinyl palmitate

(Public Health Service 1965). Retinyl palmitate is the ester of retinol and palmitic
acid. The stability of added retinyl palmitate may be affected by heat, light, or the
presence of acids which may cause degradation or conversion of 11-cis-retinal to
all-trans-retinal, resulting in lowered biological activity (Mousseron-Cadet 1971).
Ultraviolet light causes isomerization and degradation of retinoid compounds in
solution. Under more intense light, other transformations can take place such as
dimerization or chemical reaction between 2 monomers of retinyl esters (Mousseron-
Cadet 1971). In addition, large losses of vitamin A activity can occur during
processing, transportation and storage of fortified foods (Dary and Mora 2002).
Hartman and Dryden (1974) reported that procedures such as pasteurization,
sterilization, spray and roller drying or evaporation caused little loss of vitamin A in
milk products. However, prolonged heating of milk, butter, or butterfat at high
temperatures in the presence of oxygen can decrease vitamin A activity.

Despite reported vitamin A off-flavor in fluid milk products, degradation products of

vitamin A palmitate in fluid milk have not been previously reported in the literature.
However, in nonfat dry milk, a distinct hay-like flavor was detected from oxidation
products of vitamin A palmitate, the major oxidation products attributed to the off
flavor were beta-ionone (Figure 4) and dihydroactinidiolide (Suyama and others
1983). In another study conducted with corn flakes fortified with vitamin A palmitate,
2 vitamin A palmitate isomers, 9-cis and 13-cis were found from degradation of
vitamin A palmitate during elevated (45 °C) temperature storage (Kim and others
2000). This study also showed that the loss of vitamin A was reduced in the presence
of other vitamins including B1, B6, B12, C, and D (Kim and others 2000). Different
forms of vitamin premix also have an effect on vitamin A stability when exposed to
light. Triangle test results indicated that light-induced off flavor could be distinguished
in skim milks with water based vitamin A at 2000 IU after 6 h of light exposure, while
off flavor in skim milks with oil based vitamin A at 2000 IU could not be distinguished
until after 24 h of light exposure (Fellman and others 1991). Furthermore, oil matrices
have been shown to have protective effects on vitamin A stability by delaying its
oxidative degradation (Dary and Mora 2002; Loveday and Singh 2008).

Figure 4.

 Open in figure viewer

 Download Powerpoint slide

Chemical structure of beta-ionone.

Effect of Vitamin Addition on Fluid Milk Flavor

There are many factors that can influence milk flavor, including milk handling,
processing, and storing (Strobel and others 1953; Nursten 1997). Other factors that
may contribute to off-flavors in milk are enzymatic degradation of milk fat and protein
related to increasing milk somatic cell count, bacterial growth, chemical composition
of milk, chemical changes, and addition of foreign material (Bodyfelt and others
1988; Barbano and others 2006). Pasteurization of milk imparts a cooked flavor,
especially immediately after processing (Badings and others 1981; Boelrijk and
others 2003). In general, proteins, lipids and carbohydrates are the precursors of the
aroma compounds that contribute these milk flavors. However, vitamin degradation
may also contribute aroma-active compounds. Several studies have suggested that
added vitamin A concentrate imparted a detectable off-flavor, particularly in skim and
low-fat milk, and occasionally in whole milk products (Weckel and Chicoye 1954;
Whited and others 2002). Consumers have reported that vitamin A fortified milk has
an oily, haylike flavor (Fellman and others 1991). However, there are no published
studies, to our knowledge, that have directly evaluated the role of vitamin preparation
addition on flavor and flavor stability of fluid milk.

Future Work

Vitamin fortification has an established history and role in fluid milk. Ongoing interest
in maximizing milk flavor and appeal and interest in increasing vitamin D fortification
levels demands that the role of vitamin fortification on fluid milk flavor and flavor
stability be clarified. Sensory experiments need to be conducted on vitamin fortified
skim and 2% fat milk because they represent the majority of vitamin A and D fortified
milks purchased and fat plays a protective role on vitamin degradation, light
oxidation, and sensory thresholds of off flavors. If vitamin fortification results in off
flavors, it will be more likely detected in reduced fat or skim milks where sensory
thresholds for nonpolar compounds are reduced compared to whole milk or higher
fat products. Aroma active compounds in pure vitamin concentrates need to be
identified with gas chromatography-mass spectrometry (GC-MS) and gas
chromatography-olfactometry (GC-O) to pinpoint if vitamin premixes are the source.
Representative vitamin concentrates can then be selected to fortify skim and 2% fat
milk for sensory testing. Additional experiments with skim and 2% fat milk also need
to be conducted with light exposure to determine if vitamin concentrates accelerate
or contribute light oxidized flavors in fluid milk.

Conclusions

Off-flavors in fluid milk can negatively impact milk consumption and consumer
product acceptability. Established sources of off-flavor in fluid milk are thermal
degradation, enzymes associated with high milk somatic cell count, microbial
contamination, and exposure to light. The most common source of flavor problems
encountered in milk is exposure to light. Milk exposure to light can result in vitamin
destruction. Another possible source of off-flavor in fluid milk is vitamin fortification.
Several studies have addressed the influence of light oxidation on fluid milk flavor
and stability, and the effect of light exposure on vitamin stability. No studies have
directly addressed the effect of vitamin addition on flavor and flavor stability of fluid
milk. Understanding the impact of vitamin addition and degradation in fluid milk will
help the dairy industry to better position vitamin fortification and enhance fluid milk
quality. Identifying the aroma-active compounds in milk resulting from vitamin
fortification can also help identify sources of off flavors in fluid milk and milk products
that may negatively impact milk flavor quality and consumer product acceptability.

Acknowledgments

Funding was provided in part by the National Dairy Council (Rosemont, Ill., U.S.A.).
 

Explore this journal >

1. Previous article in issue: Industrial Applications of Selected JFS Articles

2. Next article in issue: Isolation and Characterization of Collagen from the Body
Wall of Sea Cucumber Stichopus monotuberculatus

Volume 80, Issue 4April 2015 Pages R665–R670

R: Concise Reviews in Food Science

XXXVI. Sensory Perception, Nutritional Role, and

Challenges of Flavored Milk for Children and
Adults
Abstract

Milk and milk products provide essential nutrients for both adults and children.
However, overall milk consumption of both adults and children does not meet the
recommendations from Dietary Guidelines for Americans. Flavored milk can
increase milk consumption for children and adolescents, but the added sugar content
raises concern. Since the removal or reduction of flavored milk decreases milk
consumption for children, it is important to understand all aspects of flavored milk in
order to increase milk consumption while minimizing sugar intake. This review will
address adult and children perception of flavored milk as well as its nutrition,
regulations in school meal programs, and challenges. Understanding the sensory
perception of both adults and children for flavored milk can help food developers and
manufacturers to address attractive attributes while reducing the sugar content to
meet the needs of a healthy diet.

Introduction

Milk is a liquid produced by the mammary glands of mammals and has been regularly
consumed by human for more than centuries as an important food (Varnam and
Sutherland 2001). Since 1611, when the 1st cows arrived in the United States, milk
and milk products have had an important position in American history and are
frequently consumed in the United States (Intl. Dairy Foods Assn. 2012). In 2013
alone, there was 91.3 million metric tons of milk produced (NASS and USDA 2014).
By modification of milkfat or water content of milk, milk may be adjusted to other
varieties such as cream, skim milk, concentrated milk, concentrated skim milk, dry
whole milk, or nonfat dry milk (US FDA 2013a). Milk products can have added
sweeteners and/or other ingredients that result in products such as sweetened
condensed milk, ice cream, frozen custard, sherbets, or gelato; with fermentation,
milk products also include cultured milk, sour cream, yogurt, or cheese and whey
products (Intl. Dairy Foods Assn. 2012). For regular fluid milk with no other claims,
optional addition of ingredients such as vitamin A (2000 to 3000 Intl. Units per quart
[946.4 mL]) and D (400 to 600 Intl. Units per quart [946.4 mL]), fruit and fruit juice,
natural and artificial food flavorings can be added to create various fluid milk
products (US FDA 2013a; USDHHS, US PHS, and US FDA 2013).

Vitamin D fortification of milk was initiated in the 1930s to prevent infantile rickets,
as a recommendation by the American Medical Association's Council on Foods and
Nutrition. The fortification of vitamin A started in the 1940's in vitamin–mineral
formulations as vitamin D fortification became common (Dairy Practices Council
2001). Since 1978, vitamin A fortification has been required by the Pasteurized Milk
Ordinance for low fat and nonfat milks since vitamin A levels are reduced with fat
reduction, and vitamin D fortification has remained optional. However, for fluid milks
that make vitamin addition claims, the fortification levels have to meet the FDA
requirements as described above (Dairy Practices Council 2001).

Studies have shown that the consumption of milk is beneficial to the health of
children and adolescents (Black and others 2002; Spence and others 2011), and
milk and dairy products consumption is linked to calcium intake (Jensen and
Kesavan 1993). Dietary Guidelines for Americans suggested that bone health in
children and adolescents is related to the intake of milk and milk products (USDA
and USDHHS 2010). Milk and milk products contain numerous essential nutrients
such as omega-3 fatty acids, short and medium chain fatty acids, vitamins (A, E,
B12), minerals (that is, calcium, magnesium, zinc, and so on), and bioactive
compounds (Haug and others 2007). Quann and Adams (2013) confirmed that the
consumption of dairy falls short for children based on the recommended daily
consumption suggestions of Dietary Guidelines for Americans (3 servings/d of fat-
free or low-fat milk or equivalent milk for adults and children ages 9 to 18 y, 2.5
servings/d for children ages 4 to 8 y, and 2 servings/d for children ages 2 to 3 y;
USDA and USDHHS 2010). For adults >20 y, the percentages of fluid milk consumed
decreased significantly between 1977 to 1978 and 2005 to 2006 (USDA ARS 2010).
Furthermore, the fluid milk consumption frequency decreased for adolescents and
adults from 1977 to 1978 to 2007 to 2008. Consumers who drank milk 3 or more
times per day decreased from 13% to 4% and consumers who did not drink fluid milk
on a given day increased from 41% to 54% (Stewart and others 2013).

Flavored milk can increase milk consumption among both adults and children and
also provides essential nutrients like plain milk and other milk products (Anonymous
2003; Murphy and others 2008). However, the high sugar content of flavored milk
has raised health concerns and caused some schools to remove or reduce flavored
milk from their lunch programs (Quann and Adams 2013). It is important to find a
balance between trying to increase milk consumption and the control of sugar intake.
The objective of this review is to review the nutritional role of flavored milk for adults
and children, and the challenges that flavored milk is facing due to the controversial
attitudes toward its healthfulness.

Flavored Milk and Milk Drinks

Sir Hans Sloane, a British physician, was believed to 1st invent chocolate milk by
mixing cocoa with milk to make the drink more palatable when he was introduced to
cocoa as a local favorite drink in Jamaica in the late 1600s, and it was 1st sold as a
medicine in England (Natural History Museum, U.K. 2014). In modern America,
flavored milk and milk drinks are milk-based drinks with caloric additions such as
chocolate- and fruit-flavored milk, milk shakes, hot chocolate/cocoa, malted milk,
lattes and similar coffee beverages, and eggnog (USDA ARS 2010). In 2011, there
were 2.09 billion kg sales in flavored milk and milk drinks, which decreased by 2.8%
compared to 2010 (Intl. Dairy Foods Assn. 2011). Among the different flavors of
flavored milk (for example, chocolate, strawberry, and vanilla), chocolate milk is the
most popular milk flavor for both children and adults in the United States (Boor 2001;
Thompson and others 2004; Thompson and others 2007; NDC 2010). Ready-to-
drink refrigerated chocolate milk is often processed with cocoa powder. Stabilizers
such as carrageenan or other thickening agents are added to prevent separation
and to improve the viscosity and texture (Varnam and Sutherland 2001). Artificial
and natural flavors may also be added to chocolate milk to improve the flavor profile.
Chocolate milk can also be created by consumers at home by adding commercially
available cocoa powder or chocolate syrup to plain milk. The flavors of other flavored
milks are achieved by the addition of natural and/or artificial flavors.

Nutritive sweeteners such as sugar and other nonnutritive sweeteners (artificial and
natural) are often used to sweeten flavored milk (NDC 2009). The amount of sugar
that is added to flavored milk varies due to different formulas and manufactures
procedures (NDC 2009). On average, 28 g/serving (8 oz, 236.6 mL) total sugar
which amounts to 16 g/serving added sugar is present in ready-to-drink reduced fat
milk beverages (flavored and sweetened, with added calcium, vitamins A and D).
For chocolate-flavored drinks (whey and milk based), 21 g/serving total sugar was
present; for chocolate milk (with added vitamin A and D) specifically, 24 to 25
g/serving total sugar was present (USDA ARS 2013). Due to the added sweeteners
and/or added solids, flavored milk is pasteurized at a temperature that is 3 ºC higher
than regular fluid milk (USDHHS, US PHS, and US FDA 2013).

Flavored Milk for Children

Nutrition of flavored milk

Flavored milk provides the same 9 essential nutrients as plain milk (unflavored):
protein, calcium, potassium, phosphorus, vitamins A, D, and B12, riboflavin, and
niacin (NDC 2011). For children >2 y, fluid milk is the number one single food
contributor of calcium, potassium, phosphorous, and vitamin D (Rafferty and Heaney
2008). The replacement of dairy servings (302.3 mg calcium per serving) by calcium-
equivalent servings (such as fortified soy beverage [1.1 servings], bony fish [1.2
servings], or leafy greens [2.2 servings]) will not sufficiently replace other nutrients
including protein, potassium, or phosphorus (Fulgoni and others 2011). The dairy
consumption of Americans falls short for the daily intake recommendations of Dietary
Guidelines for Americans and is at least in part due to the increase of consumption
of soft drinks and juices (Wyshak 2000; Rampersaud and others 2003; Forshee and
Storey 2003; USDA ARS 2010; NDC 2011; Quann and Adams 2013; Stewart and
others 2013). The dairy consumption of Americans, especially for children, is very
important.

Flavored milk can improve dairy consumption and lower the intake of soft drinks and
fruit drinks, and increase calcium intake (Johnson and others 2002). Johnson and
others (2002) also found that the added sugar intake did not increase with flavored
milk consumption, possibly due to the result of lower intake of soft drinks and fruit
drinks. Similarly, Frary and others (2004) reported that sweetened dairy products,
including flavored milks, increased dairy consumption and calcium intake for children
and adolescents. In that study, sweetened dairy products were also associated with
added sugar intakes for children but not for adolescents. Murphy and others (2008)
reported similar or indifferent intake of added sugar between flavored, plain and
nonmilk drinkers for children ages 6 to 11 y and 12 to 18 y, and the total milk intake
of flavored milk consumers was significantly higher than plain milk consumers.
Higher intakes of potassium, calcium, and phosphorus were also observed among
flavored milk consumers compared to plain milk consumers in some age-sex groups.
Milk consumption (flavored and plain) positively influenced nutrient intakes and did
not have a negative influence on body mass index (BMI; Murphy and others 2008).
This finding was supported by Vanselow and others (2009) in which chocolate milk
was not associated with changes in BMI over a 5-y period for adolescents. Nicklas
and others (2013) reported that among total energy consumed by children, flavored
and white milk only contributed 2% to 6%, and contributed a higher percentage of
essential nutrients compared to the percentage of added sugar consumption.
Noel and others (2013) reported that nonconsumers of flavored milk had higher
intakes of dietary fiber but lower intakes of several nutrients compared to flavored
milk consumers for children 10 to 13 y but did not differ for other foods intake
including fruits, vegetables, beverages, ready-to-eat breakfast cereals, and
sweets/cookies. Flavored milk consumers also had higher calcium intake. However,
in this study, flavored milk consumers had almost 150 kcal more energy intake
compared to nonconsumers in that study and Noel and others (2013) hypothesized
that it was possible that flavored milk may already be part of an unhealthy lifestyle.
Noel and others (2013) reported that over a 2-y period, overweight children that were
flavored milk consumers had less favorable changes in body fat and body weight.
They suggested that overweight children should reduce consumption of flavored
milk, however, they also noted that consumption of flavored milk was unlikely to be
associated with body fat or weight gain for normal weight children. Collectively, these
studies suggest that flavored milk nutritional benefits outweigh apprehensions
regarding sugar intake or obesity concerns, particularly when coupled with a
concerted industry effort to reduced added sugar in flavored milk.

Flavored milk in schools

Currently, the USDA requires the Natl. School Lunch Program (NSLP) and School
Breakfast Program (SBP) to meet the meal pattern and nutrition standards based on
the latest Dietary Guidelines for Americans (USDA FNS 2012b, 2013). By 1994,
participating schools of NSLP were required to offer whole milk and/or milk with other
varieties of fat contents (SNA and NDC 2008). In 1994, the Healthy Meals for Healthy
Americans Act (US 103rd Congress 1994) authorized that participating schools of
NSLP “shall offer students a variety of fluid milk consistent with prior year
preferences unless the prior year preference for any such variety of fluid milk is less
than 1 percent of the total milk consumed at the school.” This allowed many schools
to gradually phase out whole milk (SNA and NDC 2008). In 2004, the Child Nutrition
and WIC Reauthorization Act (US 108th Congress 2004) stated that participating
schools “shall offer students fluid milk in a variety of fat contents” which removed the
whole milk requirement completely (SNA and NDC 2008). According to this act,
schools could also offer flavored and unflavored fluid milk. USDA regulated schools
were to only offer low-fat (1%) and fat-free flavored milk due to the recommendations
of the 2010 Dietary Guidelines which recommended that children over 2 y should
consume low-fat (1%) or fat-free fluid milk (USDA FNS 2011). In January of 2012,
the USDA issued a final rule for the Nutrition Standards in the Natl. School Lunch
and Breakfast Program of 2012, and required that the NSLP and SBP could only
offer flavored fluid milk if it was fat-free. This requirement became effective July 1,
2012, the beginning of school year 2012 to 2013 (USDA FNS 2012a).

For the milk that children choose to drink in the NSLP, nearly 70% is flavored (NDC
2011). However, the concerns over added sugar and calorie content of flavored milk
caused several schools to reduce or remove flavored milk from school meals
(Patterson and Saidel 2009; NDC 2011; Quann and Adams 2013). The removal of
flavored milk caused unintended negative consequences where the consumption of
milk by children decreased significantly. Patterson and Saidel (2009) reported that
in a school district in Connecticut, milk sales decreased 37% to 63% in all K-12
grades in a 3-mo period after flavored milk was removed.

Similarly, Quann and Adams (2013) reported that the removal of flavored milk would
negatively influence milk consumption in a study that included 49 schools. In that
study, compared to when school offered both flavored milk and white milk, milk sales
and milk consumption were significantly lower when the school only offered white
milk (P < 0.001). The percentage of purchased milk discarded was also significantly
higher when only white milk was offered compared to when both white milk and
flavored milk were offered (P < 0.001). Quann and Adams (2013) also investigated
the replacement of nutrient deficits from decreased milk consumption (calcium,
vitamin A and D, potassium, magnesium, and phosphorus), and found that in order
to sufficiently replace the nutrients, 3 to 4 additional foods would be needed. And
this resulted in additional calorie (16 to 141 kcal) intake and additional cost ($2200
to $4600 annually per 100 students). Hanks and others (2014) also reported in a
pilot study that banning chocolate milk in school cafeterias had a negative effect on
milk consumption. Milk consumption data in 11 Oregon elementary schools were
compared before and after the ban of chocolate milk in this study. Despite the
appeared benefit of calorie and sugar reduction, children consumed less milk overall
and more white milk was wasted (leftover). Therefore, unintended negative
consequences from removing flavored milk from school need to be addressed.

The final rule of Nutrition Standards in the Natl. School Lunch and Breakfast Program
(USDA FNS 2012a) established minimum and maximum calorie limits for meal
patterns. For lunch, it is 550 to 650, 600 to 700, and 750 to 850 kcal for grades K-5,
6 to 8 and 9 to 12, respectively. For breakfast, it is 350 to 500, 400 to 550, and 450
to 600 kcal for grades K-5, 6 to 8, and 9 to 12, respectively. However, the final rule
does not have a calorie limit for flavored milk. If school meal programs can reduce
added sugar in children's diet through other approaches, such as introduction of
calorie-reduced flavored milk, and include flavored milk in a balanced diet and
reduce calorie and sugar intake from other foods as well, the unintended
consequences brought by the removal of flavored milk in the school meal program
could be avoided.

Calorie-reduced flavored milk

Like white milk, flavored milk also contains 12 g/serving (240 mL) natural sugar,
mainly lactose. Added sweeteners can be nutritive such as sucrose, HFCS, or
nonnutritive sweeteners (NDC 2009). Concerns over HFCS have cost removal of
chocolate milk from some schools in the past (Patterson and Saidel 2009). However,
there are conflicted opinions on if HFCS contributes to childhood obesity. Bray and
others (2004) reported that HFCS in beverages may play a role in the epidemic of
obesity, but the American Medical Assn. has reviewed that due to the similar
metabolic way of digestion, HFCS is not likely more associated with obesity than
sucrose (American Medical Assn. 2008). Efforts have been made to reduce the
calories of flavored milk. From 2006 to 2010, calories have decreased from 165.9 to
154. Zero per 8 oz (236.6 mL; reduced fat and reduced sugar; Van Horn and others
2010). However, the higher cost associated with reduced-sugar flavored milk, since
it costs more to produce, may sometimes cause it not to be selected by school
nutrition directors (NDC 2009; Van Horn and others 2010).

Yon and others (2012) reported that in a plate waste study, there was no significant
difference in milk consumption between standard flavored milk and reduced-calorie
flavored milk. In that study, standard flavored milk (160 to 170 kcal/serving) was 1%
fat flavored milk with 25 to 27 g sugar content per serving (8 oz to 236.6 mL), and
reduced-calorie flavored milks (150 kcal/serving) were either fat-free or 1% fat
flavored milk with 22 to 27 g sugar per serving. Ninety nine percent of the flavored
milk in this study was chocolate milk which is consistent with previous studies that
chocolate milk is the most popular flavored milk (Boor 2001; Thompson and others
2004; Thompson and others 2007; NDC 2010). Yon and others (2012) reported that
reduced-calorie flavored milk was still well accepted, however, the reformulated
milks in that study did not have a huge calorie reduction such that flavor differences
were likely small and possibly not detected by the children. Furthermore, the study
was done prior to the final rule of Nutrition Standards in the Nat. School Lunch and
Breakfast Program (USDA FNS 2012a) which has more strict regulations on the fat
content of flavored milk.

Artificial sugar substitutes can be used to reduce the sugar content of chocolate milk.
However, Dairy Management, Inc.TM and the Natl. Dairy Council have reported that
mothers and health professionals had limited acceptability for sugar substitutes
although they are proved to be safe for adults and children (Ragalie 2007). In that
study, all sugar substitutes were artificial, including saccharine, aspartame,
acesulfame-k, neotame, and sucralose. In a German choice experiment study, the
majority of parents also showed lower preference for artificial sweeteners as well
(Christoph and others 2011). Li and others (2014) also recently reported that parents
preferred natural nonnutritive sweeteners over nutritive sweeteners in chocolate
milk, and preferred artificial nonnutritive sweeteners the least. Therefore,
investigating other natural nonnutritive sweeteners is needed for reformulating
reduced sugar chocolate milk for children. Efforts have been made to sweeten
chocolate milk by monk fruit juice. Monk fruit is the fruit of Siraitia grosvenorii
(Swingle; also known as Monordica grosvenorii), belongs to the family
Cucurbitaceae, and is most commonly known as Luo Han Guo to Chinese (DuBois
and Prakash 2012; Pawar and others 2013). Monk fruit has been consumed for
centuries as traditional Chinese medicine for coughs, sore throat, and other maladies
(DuBois and Prakash 2012). The major sweet component of Monk fruit was identified
as mogroside V (Takemoto and others 1983). In 2009, PureLo®, monk fruit juice
concentrate manufactured by BioVitorria was affirmed as GRAS (GRAS Notice No.
301), and in 2011, powdered monk fruit extract was affirmed as GRAS (GRAS Notice
No. 359). Schools in Kansas City, Missouri and Omaha, Nebraska regions have
started to offer fat-free chocolate milk with only 3 g added sugar that is also
sweetened by monk fruit juice (Anonymous 2013), and Lawrence public schools food
service (Kansas) will soon include similar products (Wilson 2013). The innovation of
calorie reduced chocolate milk with natural sweeteners should help increase milk
consumption with concurrent sugar reduction.
The Intl. Dairy Foods Assn. and the Natl. Milk Producers Federation (NMPF) filed a
petition in 2009 to ask FDA to remove the “reduced calorie” or “reduced sugar” label
from the chocolate milk package if the milk was manufactured with “any safe and
suitable” nonnutritive sweeteners (US FDA 2013b). Existing regulations require
manufacturers to include a nutrient content claim on the package if an ingredient that
is not in a standard product was included. The argument was that nutrient content
claims such as “reduced calorie” are not appealing to children and milk without such
claims can let consumers “more easily identify its overall nutritional value” (US FDA
2013a). FDA requested interested persons to submit comments and information in
order to evaluate the need for the amendments (US FDA 2013c). In a small study (n
= 34) of young adolescents from 11 to 14 y, >50% participants self-reported that
Nutrition Facts labels frequently or sometimes influenced their knowledge on the
healthfulness of a food, but none of the participants used the Nutrition Facts label
daily (Hawthorne and others 2006). Education on nutrition labels can help children
to identify healthier and more nutritious foods (Hawthorne and others 2006; Katz and
others 2011). If children are educated to read front-of-package labels and/or Nutrition
Facts labels, labels like “reduced calorie” should be able to help children to choose
more healthful food and beverages.

Dental caries and flavored milk

In 2011, dental visits for Americans 2 to 17 y increased from 78.9% to 81.4%

compared to 2010, and untreated dental caries decreased from 16.2% (2005 to
2008) to 15.6% (2007 to 2010; HHS 2013), which indicated that people pay more
attention to dental health nowadays. Unregulated sugar consumption, such as sugar
from sugary liquids and snacks can increase dental caries for early childhood
(American Dental Assn. 2000). The biofilm that accumulates on teeth from numerous
bacteria is called dental plaque (Rosan and Lamont, 2000), and plaque acidity from
sugar exposure contributes to early dental caries (Lingström and others 2000). Milk
lowers the plaque pH in the mouth through the metabolism of lactose content into
organic acid by plaque organisms (Levine 2001). However, previous research has
shown that the decrease in plaque pH by milk is negligible and cow's milk is
noncariogenic due to its high buffering capacity, high calcium and phosphate
content, milk proteins, and enzymes (Jenkins and Ferguson 1966; Levine 2001).

Flavored milk is often sweetened with sucrose. Studies have confirmed that the
addition of sucrose to plain milk will increase cariogenicity (Thompson and others
1984; Bowen and Pearson 1993). Dunning and Hodge (1971) evaluated the
influence of cocoa and sugar in milk on cariogenicity. In that study, over a 2-y period,
children who consumed conventional chocolate milk with cocoa and sugar had a
“small, but nonsignificantly higher” increase in caries compared to plain milk
consumers. Chocolate milk consumers also had a “borderline significant” increase
in caries compared to consumers with cocoa and an artificial sweetener group. Petti
and others (1997) also reported that in an Italian schoolchildren study, children with
lower sucrose consumption had better dental health. Milk and sucrose consumption
had a negative significant association with dental health, but sucrose had a much
higher significance compared to milk which had a weak negative association.
Optimal oral hygiene practices could substantially reduce the incidence of dental
caries. More research is still needed to fully understand and to establish the
relationship between flavored milk, especially chocolate milk, and dental caries.

Children's perception of fluid milk

Various studies have been conducted on children's perception of plain fluid milk.
Monneuse and others (1991) reported in a study for 10 to 19 y consumers, gender
and age had an impact on intensity ratings and preference of sugar and fat in dairy
products. Chapman and Boor (2001) reported that gender and age did not influence
the liking of milk for children between age 6 to 11 y. In this study, the liking of milk
was significantly affected by how much children liked plain milk in general. Children
who really liked milk in general reported higher liking scores compared to children
who liked milk “somewhat,” followed by children who did not like milk. For chocolate
milk, the studies conducted by Hough and others (1997) and Hough and Sanchez
(1998) on powdered chocolate milk for both children and adults showed that there
was variation in preferences between adults and children. Thompson and others
(2007) evaluated Hispanic children perception of flavored milk. Chocolate milk was
the most popular flavor among participants but strawberry and vanilla were popular
too. For the majority of the children who consumed chocolate milk, it was consumed
every day. Children preferred chocolate milk because of the flavor and color, and
they liked both ready-to-drink chocolate milk and powdered mixes. In this study,
children indicated that they preferred chocolate milk that had good chocolate flavor
and was not watery. And children were generally less discerning of chocolate milks
compared to adults when presented with multiple chocolate milks and in general
liked all chocolate milks. Children also indicated that they had tried coconut and
banana flavored milk, and thought cherry flavor and tropical flavors such as mango,
kiwi, and pineapple would be good too. Not surprisingly, studies demonstrate that
children like flavored milks in general and that flavored milks provide an attractive
alternative to other sweetened beverages and drinks.

Flavored Milk for Adults

Flavored milk as a recovery drink

Athletes had improvements in endurance performance following consumption of a

carbohydrate beverage (Coggan and Coyle 1991) and carbohydrate beverages are
also used to increase glycogen recovery speed after exercise to benefit sequential
workouts (Pritchett and others 2009). Research has shown that the addition of
protein to a carbohydrate beverage (such as chocolate milk) ingested post exercise
may have a positive influence on exercise recovery (Saunders and others 2004;
Skillen and others 2008; Valentine and others 2008).

Despite the pressure it receives in school meal systems, chocolate milk as a

recovery drink after exercise and workouts has increased in popularity among adults.
It is currently served at the end of Ironman races and marathons across the United
States (Yang 2014). Chocolate milk can be used as an effective alternative for post
exercise recovery due to the high carbohydrate and protein content. Karp and others
(2006) compared time to exhaustion, average heart rate, rating of perceived
exertion, and total work between 9 male athletes who ingested chocolate milk, fluid
replacement drink, or carbohydrate replacement drink. Athletes who ingested
chocolate milk had better performance compared to those with carbohydrate
replacement drink. Similarly, Thomas and others (2009) also investigated the effect
of chocolate milk consumption on endurance capacity of 9 male athletes compared
to carbohydrate replacement drink and fluid replacement drink. After consuming
chocolate milk, skeletal muscle protein synthesis and whole-body protein recovery
of athletes were significantly increased compared to the carbohydrate only
beverage. Gilson and others (2010) conducted a randomized cross-over study on
intercollegiate soccer players to investigate the effect of chocolate milk consumption
after soccer training on muscle recovery. The ingestion of chocolate milk post
exercise showed similar effects compared to carbohydrate beverage, but future
studies with greater increases in training volumes and longer periods of time are
needed to determine if functional significance exists between treatments. Lunn and
others (2012) also reported in a study with male runners that after endurance
exercise, chocolate milk consumption had favorable effects compared to
carbohydrate only beverages. Chocolate milk popularity among adults as a
postexercise recovery drink could be another way to increase adult milk
consumption.

Sensory perception of flavored milk

Not surprisingly, flavor is a key attribute of flavored milks. Folkenberg and others
(1999) reported that mouthfeel in instant cocoa milk drinks was related to both flavor
and viscosity properties, and consumers in this study preferred sweet and milky
products that were not too thick. Campbell and others (2003) reported that the
addition of chocolate flavor could increase the liking of conjugated linoleic acid
fortified fluid milk. For chocolate milk specifically, Thompson and others (2004)
reported that cocoa aroma was the major driver of liking for chocolate milks for
adults. Consumers in this study were segmented into 3 groups, cocoa aroma and
flavor was a universal driver while 2 of the groups were also positively influenced by
malty or cooked/eggy flavors in chocolate milk. Participants indicated that they
consumed chocolate milk mainly as a snack, and most preferred chocolate milk with
moderate sweet taste and thickness. Thompson and others (2007) reported that both
Caucasian and Hispanic adult participants indicated “natural” chocolate flavor was
the most important attribute for flavored milk (Thompson and others 2007). Harwood
and others (2012) investigated the rejection and detection threshold of bitterness in
chocolate milk with milks spiked with sucrose octaacetate. In this study, consumers
with self-reported preference for dark chocolate had a higher rejection threshold, but
the same detection threshold of bitterness compared to consumers who preferred
milk chocolate, which suggested that personal preferences for dark or milk chocolate
had an impact on the rejection threshold of bitterness in chocolate milk and thus
impacted specific preferences for chocolate flavor in milk (e.g., dark compared with
milk chocolate flavor).
Kim and others (2013) reported that adult consumer purchase of chocolate milk was
influenced primarily by the labeled fat content on the package, followed by sugar
content and brand. Interestingly, fat-free chocolate milk was not preferred by most
adults consumers, instead reduced fat (1% or 2% milkfat) was preferred, perhaps
further emphasizing the role of desirable flavor as the primary reason that adults
consume chocolate milk. In an online adaptive conjoint study, Li and others (2014)
reported that the majority of parents preferred reduced fat chocolate milk sweetened
by natural nonnutritive sweeteners or sucrose for their children, however, no actual
chocolate milks were tasted in this study. Future studies could evaluate if labeling of
“exercise recovery” or “workout drink” would impact adult consumer acceptance,
especially for athletic consumers.

Conclusions

Flavored milk plays a significant role in increasing milk consumption to ensure

children's essential nutrients intake as well as providing nutrition for adults. The
controversy on the sugar content and calories of chocolate milk has influenced and
caused many regulatory and social opinion changes. Numerous studies indicate that
chocolate milk, and by association, other flavored milks, are consumed for their
desirable flavor and as such, desirable flavor must be maintained while controlling
sugar and caloric content. The wealth of nutrition studies suggest that the benefits
of moderated flavored milk consumption outweigh any possible negative effects,
particularly as efforts are made to minimize sugar content. Understanding the
nutritional values and regulations of flavored milk is important for its current status
and future perception. Future studies need to be conducted on sensory acceptance
of calorie reduced flavored milks, especially chocolate milk, for both young adults
and children to obtain sugar reduction and still maintain acceptability for both
marketing and health values.

Acknowledgment

Funding provided in part by the Dairy Research Inst. (Rosemont, IL).

 

1. Next article in issue: Enzymatic Properties of β-1,3-Glucanase from

Streptomyces sp Mo.

Volume 78, Issue 4April 2013 Pages R495–R501

R: CONCISE REVIEWS/HYPOTHESES IN FOOD SCIENCE

Microbial Contamination in Sprouts: How Effective Is Seed Disinfection

Treatment?

Abstract

Microbial contamination of sprouts by Salmonella and Escherichia coli O157 : H7 has

been a common cause of foodborne diseases and a continuing challenge to the
sprout industry. Seed disinfection treatment has been recommended as a major
intervention step in a multihurdle approach to reduce the risk of illness associated
with contaminated sprouts. U.S. Food and Drug Administration cited 20000 ppm
calcium hypochlorite as an example treatment in its recommendation for seed
treatment and this treatment has been considered the reference standard for seed
disinfection treatment for over a decade. However, promising new disinfection
treatments have emerged in recent years. In this study, we summarized published
data and compared the efficacies of different disinfection methods in the reduction
of microbial contamination on seeds. Our findings suggest that while biological
interventions such as competitive exclusion and certain chemical treatments appear
to be similar to 20000 ppm calcium hypochlorite for seed disinfection, physical
methods especially high pressure may be more effective than the reference standard
regardless of the type of bacteria or seed. The combination of 2 or more treatments,
sequentially or simultaneously, may further improve disinfection results. Since
treatments with high levels of chemical disinfectants, especially 20000 ppm calcium
hypochlorite, can pose environmental and worker safety risks, alternative
intervention approaches should be considered. Additional studies to confirm the
greater efficacy of certain physical and combined seed disinfection treatments and
to identify other effective management strategies are needed to further improve
sprout safety.

Introduction

Multiple outbreaks have been linked to consumption of raw and lightly cooked
sprouts throughout the world (Taormina and others 1999; Ben Chapman 2012;
Centers for Disease Control and Prevention Foodborne Outbreak Online Database
2012). Outbreaks have been associated with a wide variety of sprout types, including
alfalfa, mung bean, clover, cress, soybean, and radish spouts. The infected
population can be as few as single number (Taormina and others 1997) and as large
as several thousands including dozens of deaths, as in the catastrophic public health
crisis in Japan in 1996 (Ministry of Health and Welfare of Japan 1997) and more
recently in Germany (Robert Koch Inst. 2011). As sprouts have been popular due to
their nutritive value (Kurtzweil 1999) and other perceived benefits such as
anticarcinogenic and anticholesterolemic properties (Donaldson 2004), many
consumers could be exposed to disease if the sprouts are contaminated.

Unlike other fresh produce, sprout production involves a unique seed germination
process that can support the growth of pathogens, if present. Contamination of
sprouts by microbial pathogens is a major concern and poses a challenge to the
industry because the optimal conditions for sprout germination are also ideal for
bacterial proliferation. A variety of microbes comprise the normal microbial
community in sprouts (Loui and others 2008). Salmonella and E. coli O157 : H7 are
the 2 most frequent pathogenic contaminants causing sprout-associated outbreaks
in the United States (Figure 1).
Figure 1.

 Open in figure viewer

 Download Powerpoint slide

Sprouts-associated outbreaks in the U.S.A. Data represent the confirmed outbreaks

by Centers for Disease Control and Prevention.

To help industry minimize microbial hazards in sprouts, the U.S. Food and Drug
Administration (FDA) asked the National Advisory Committee on Microbiological
Criteria for Foods (NACMCF) to review the science behind sprout outbreaks and
suggest recommendations to enhance sprout safety. In 1999, FDA released
guidelines, based largely on the work of NACMCF, which included 5 basic
recommendations: (1) growing seed for sprouting using good agricultural practices,
(2) conditioning and storing seed under sanitary conditions, (3) following GMPs, as
appropriate, at sprouting facilities, (4) applying a disinfection treatment to seed
immediately before sprouting, and (5) in-process testing of spent sprout irrigation
water for pathogens of concern before finished product enters market channels (U.S.
Food and Drug Administration 1989, 1999a,b, 1999; National Advisory Committee
on Microbiological Criteria for Foods 1999). As an example, FDA cited 20000 ppm
calcium hypochlorite for seed disinfection treatment.

Sprout associated outbreaks appeared to decline after release of FDA's sprout

guidelines in 1999 (Figure 1). However, the extent of implementation of these
guidelines is a continuing concern (Thomas and others 2003). In addition, the
efficacy of the seed disinfection treatment itself has been found to be highly variable
(Montville and Schaffner 2004). Since the introduction of these guidelines in 1999,
sprout-associated outbreaks have not been satisfactorily controlled (Figure 1). It is
important to examine the effectiveness of the different seed disinfection treatments
as one of the control strategies. Not only the 20000 ppm calcium hypochlorite, but
also other emerging options that might work better for reducing microbial
contamination in sprouts should be assessed. To that end, we summarized and
compared the efficacies of different seed disinfection treatments that have been
reported in the literature (Table 1).

Table 1. Studies of disinfection treatments on sprout seeds

 Physical and biological
Chemical Combination
treatment

Disinfect Reduc Disinfect Reduc Autho Disinfecti Reduc

Author,
ion tion ion tion r, on tion

(log (log (log

Author, conditio conditio condition
CFU/g Year CFU/g Year CFU/g
Year n† n† †
) ) )

1. C: Temperature (°C), s: second, min: minute, h: hour, d: day.

2. † Disinfection treatments for different pathogens or seeds under same
conditions were listed separately.

20000 ppm Ca(OCL)2 Heat Chemical + Chemical

20000
ppm
Bari, Bari, Bari, Ca(OCL)2
20 min 2.50 50 C, 1 h 1.00 2.91
2010a 2003 2010a + Chlorine
(2000
ppm)

20000
ppm
Bari, Bari, Bari, Ca(OCL)2
20 min 2.70 50 C, 1 h 0.90 3.21
2010a 2003 2010a + Chlorine
(2000
ppm)

Malic acid
(10%) +
Fransi
Buchholz, Bari, Thiamine
15 min 1.13 50 C, 1 h 1.73 sca, 3.41
2010 2003 dilauryl
2012a
sulfate
(1%)

Malic acid
(10%) +
90 C + 0 Fransi
Charkows Bari, Thiamine
15 min 2.50 C, 90 s + 4.34 sca, 3.02
ki, 2001 2008 dilauryl
30 s 2012b
sulfate
(1%)
 Physical and biological
Chemical Combination
treatment

Disinfect Reduc Disinfect Reduc Autho Disinfecti Reduc

Author,
ion tion ion tion r, on tion

(log (log (log

Author, conditio conditio condition
CFU/g Year CFU/g Year CFU/g
Year n† n† †
) ) )

Malic acid
(10%) +
90 C + 0 Fransi
Fransisca, Bari, Thiamine
20 min 3.41 C, 90 s + 5.08 sca, 1.88
2012 2008 dilauryl
30 s 2012b
sulfate
(1%)

Ca(OH)2
Holida
Fransisca, Bari, 50 C, 24 (1%) +
20 min 1.36 3.50 y, 3.96
2012 2009a h Tween 80
2001
(1%)

Ca(OH)2
Holida
Fransisca, Bari, 50 C, 24 (1%) +
20 min 2.16 5.00 y, 3.39
2012 2009a h Tween 80
2001
(1%)

Ca(OH)2
Holida
Gandhi, Bari, 50 C, 24 (1%) +
10 min 4.98 5.00 y, 3.12
2003 2009a h Span 20
2001
(1%)

Ca(OH)2
Holida
Holiday, Bari, (1%) +
10 min 2.96 50C,24h 2.50 y, 2.99
2001 2009a Span 20
2001
(1%)

Lactic
acid
75 C + 0
Holiday, Bari, Lang, (2.5%) +
10 min 3.19 C, 20 s + 4.38 3.70
2001 2009b 2000 Chlorine
20 s
(2000
ppm)
 Physical and biological
Chemical Combination
treatment

Disinfect Reduc Disinfect Reduc Autho Disinfecti Reduc

Author,
ion tion ion tion r, on tion

(log (log (log

Author, conditio conditio condition
CFU/g Year CFU/g Year CFU/g
Year n† n† †
) ) )

Lactic
75 C + 0 acid (5%)
Lang, Bari, Lang,
15 min 6.90 C, 20 s + 5.80 + Chlorine 3.40
2000 2009b 2000
20 s (2000
ppm)

Ca(OCL)2
(200
Bari, Kim,
Liao, 2009 45 min 2.80 85 C, 10 s 2.82 μg/mL) + 1.50
2010a 2010
Drying (24
h)

CLO2 (200
Bari, Kim, μg/mL) +
Kim, 2003 10 min 4.75 85 C, 10 s 3.29 3.80
2010a 2010 Drying (24
h)

Emery
658 +
Kumar, Bari, Pierre,
20 min 1.14 85 C, 40 s 3.69 Glycerol 4.60
2006 2010a 2006
monolaur
ate

Emery
658 +
Kumar, Bari, Pierre,
20 min 0.51 85 C, 40 s 3.84 Glycerol 6.90
2006 2010a 2006
monolaur
ate

Levulinic
acid
Rajkowski Feng Zhao,
20 min 0.00 55 C, 8 d 1.17 (0.5%) + 6.40
, 2009 ,2007 2010
SDS
(0.05%)
 Physical and biological
Chemical Combination
treatment

Disinfect Reduc Disinfect Reduc Autho Disinfecti Reduc

Author,
ion tion ion tion r, on tion

(log (log (log

Author, conditio conditio condition
CFU/g Year CFU/g Year CFU/g
Year n† n† †
) ) )

Levulinic
acid
Zhao, Feng, Zhao,
20 min 6.30 55 C, 8 d 7.75 (0.5%) + 5.60
2010 2007 2010
SDS
(0.05%)

Chemi
Zhao,
20 min 6.30 Hu, 2004 55 C, 4 d 5.00 cal +
2010
Heat

ClO2(500
μg/mL) +
Bang, drying (25
Other chemicals Hu, 2004 55 C, 5 d 3.00 4.80
2011 C, 2 h) +
Heat (55
C, 36 h)

50 C +
NaOCl, Electrolyz
Beuchat, Neetoo, 65 C, 10 Bari,
2,000 3.90 4.70 ed 4.02
1997 2011 d 2003
ppm, 30s oxidizing
water

50 C +
Electrolyz
Beuchat, H2O2, Neetoo, 65 C, 10 Bari,
3.57 4.50 ed 1.52
1997 6%, 30 s 2011 d 2003
oxidizing
water

50 C +
High Electrolyz
Beuchat, Ethanol, Bari,
3.76 pressur ed 2.64
1997 80%, 30 s 2003
e oxidizing
water
 Physical and biological
Chemical Combination
treatment

Disinfect Reduc Disinfect Reduc Autho Disinfecti Reduc

Author,
ion tion ion tion r, on tion

(log (log (log

Author, conditio conditio condition
CFU/g Year CFU/g Year CFU/g
Year n† n† †
) ) )

Peroxyac 50 C +
Buchholz, Neetoo, 600 MPa, Bari,
etic acid, 1.77 5.70 Sanitizer, 5.00
2010 2008 2 min 2009a
1% 17 h

Peroxyac 50 C +
Buchholz, Neetoo, 500 MPa, Bari,
etic acid, 1.34 3.50 Sanitizer, 5.00
2010 2008 2 min 2009a
3% 17 h

Monocap
600 MPa, 50 C +
Chang, rylin, 75 Neetoo, Bari,
7.05 2 min, 20 3.70 Sanitizer, 4.50
2010 mM, 10 2009a 2009a
C 17 h
min

Monocap
Presoake 50 C +
Chang, rylin, 75 Neetoo, Bari,
7.11 d + 600 5.00 Sanitizer, 5.00
2010 mM, 10 2009a 2009a
MPa 17 h
min

85 C (10
Caprylic
550 MPa, s) +
Chang, acid, 75 Neetoo, Bari,
7.05 2 min, 40 4.90 Chlorine 3.53
2010 mM, 10 2009b 2010a
C (2000
min
ppm, 2 h)

85 C (10
Caprylic
500 MPa, s) +
Chang, acid, 75 Neetoo, Bari,
7.11 2 min, 45 5.80 Chlorine 3.29
2010 mM, 10 2010 2010a
C (2000
min
ppm, 2 h)

85 C (40
Himathon Ammonia 500 MPa, s) +
Neetoo, Bari,
gkham, , 180 3.00 2 min, 45 5.20 Chlorine 3.69
2010 2010a
2001 ppm, 22 h C (2000
ppm, 2 h)
 Physical and biological
Chemical Combination
treatment

Disinfect Reduc Disinfect Reduc Autho Disinfecti Reduc

Author,
ion tion ion tion r, on tion

(log (log (log

Author, conditio conditio condition
CFU/g Year CFU/g Year CFU/g
Year n† n† †
) ) )

85 C (40
Himathon Ammonia s) +
Wuytack 300 MPa, Bari,
gkham, , 180 2.00 6.00 Chlorine 3.84
, 2003 15 min 2010a
2001 ppm, 22 h (2000
ppm, 2 h)

85 C (40
Himathon Ammonia s) +
Wuytack 300 MPa, Bari,
gkham, , 180 5.50 6.00 Chlorine 4.69
, 2003 15 min 2010b
2001 ppm, 22 h (2000
ppm, 2 h)

85 C (40
Himathon Ammonia s) +
Irradiati Bari,
gkham, , 180 5.00 Chlorine 4.84
on 2010b
2001 ppm, 22 h (2000
ppm, 2 h)

H2O2, Heat +
Holiday, Nei,
8%, 10 3.48 1.5 kGy 2.80 Irradia 4.56
2001 2010
min tion

50 C +
Holiday, H2O2,8%, Saroj, 2 kGy, on Bari,
3.25 4.60 Irradiation
2001 10min 2007 ice 2003
(2.0 kGy)

Ca(OH)2, 50 C +
Holiday, Thayer, Bari,
1%, 10 3.65 2 kGy 2.00 Irradiation 4.85
2001 2003 2003
min (1.5 kGy)

Ca(OH)2, 50 C +
Holiday, Thayer, Bari,
1%, 10 3.15 2 kGy 3.30 Irradiation 5.49
2001 2003 2003
min (2.0 kGy)
 Physical and biological
Chemical Combination
treatment

Disinfect Reduc Disinfect Reduc Autho Disinfecti Reduc

Author,
ion tion ion tion r, on tion

(log (log (log

Author, conditio conditio condition
CFU/g Year CFU/g Year CFU/g
Year n† n† †
) ) )

SC-CO2,
50 C +
Jung, 10 MPa, Biologic Bari,
2.48 Irradiation 4.43
2009 5 min, 45 al 2003
(1.5 kGy)
C

SC-CO2,
Pseudom 50 C +
Jung, 15 MPa, Fett, Bari,
3.51 onads 2 4.98 Irradiation 5.69
2009 10 min, 2006 2003
to 79 (2.0 kGy)
35 C

Electrolyz 50 C +
Kocharu
ed Bari, Irradiation
Kim, 2003 1.66 nchitt, Phage 1.00 5.00
oxidizing 2009a (1.0 kGy),
2009
water 17 h

50C +
Deionize Pseudom
Matos, Bari, Irradiation
Kim, 2003 d water, 0.57 onads 2 4.22 4.50
2005 2009a (1.0 kGy),
10 min to 79
17 h

50 C +
Chlorine Bacterioc
Nandiwa Bari, Irradiation
Kim, 2003 water, 10 1.70 in 3.00 5.00
da, 2004 2009a (0.25
min (colicin)
kGy), 17 h

50 C +
NaOCl,
Kumar, E.asburia Bari, irradiation
200 ppm, 3.50 Ye, 2010 5.56 5.00
2006 e 2009a (0.25
28 C
kGy), 17 h

50 C +
NaOCl,
Kumar, Bari, Irradiation
200 ppm, 3.50 Ye, 2010 Phage 3.41 5.00
2006 2009a (0.25
28 C
kGy), 17 h
 Physical and biological
Chemical Combination
treatment

Disinfect Reduc Disinfect Reduc Autho Disinfecti Reduc

Author,
ion tion ion tion r, on tion

(log (log (log

Author, conditio conditio condition
CFU/g Year CFU/g Year CFU/g
Year n† n† †
) ) )

Lactic
Lang, acid, 5%, Other
3.00
2000 10 min, s
42 C

Acetic 65 C (12
Neeto
Lang, acid, 5%, h) + 600
2.40 o, 5.00
2000 10 min, MPa (2
2011
42 C min, 35 C)

Acidified 65 C (12
Neeto
sodium h) + 600
Liao, 2009 4.00 o, 4.40
chlorite, MPa (2
2011
800 ppm min, 35 C)

Acidified
Acidified
chlorite,
Nei, chlorite +
Nei, 2010 200 to 2.90 4.20
2010 Irradiation
1200
(1.5 kGy)
ppm

Fatty- Ozone +
Rajko
Pierre, acid- 1%
5.45 wski, 4.12
2006 based Peroxyac
2009
sanitizer etic acid

Fatty-
Pierre, acid- Ye, E.asburia
5.62 6.72
2006 based 2010 e + Phage
sanitizer

Electrolyz
Sharma, ed Ye, E.asburia
1.56 7.62
2003 oxidizing 2010 e + Phage
water
 Physical and biological
Chemical Combination
treatment

Disinfect Reduc Disinfect Reduc Autho Disinfecti Reduc

Author,
ion tion ion tion r, on tion

(log (log (log

Author, conditio conditio condition
CFU/g Year CFU/g Year CFU/g
Year n† n† †
) ) )

Electrolyz
Stan,
ed 2.04
2003
oxidizing

Chemical Disinfection

As a classical tool to combat microbial contamination, chemicals have been widely

studied for disinfection treatment of sprout seeds. Although their efficacy could be
limited by the degree of exposure of the pathogens to chemical agents and additional
approaches to ensure an effective exposure such as presoak, optimized chemical-
to-seed ratio, stirring, and vacuum might be critical, the convenience and affordability
of chemical treatments promote them as the primary choice for sprout seed
disinfection. A representative chemical treatment, calcium hypochlorite used at a
concentration of 20000 ppm, is cited by the FDA in its guidance recommending seed
disinfection (U.S. Food and Drug Administration 1999a). However, the sprout
industry has not fully welcomed this treatment, largely due to concerns about worker
safety and its unfavorable impact on the environment. In fact, some argue that the
referenced 20000 ppm calcium hypochlorite disinfection treatment for sprout seeds
is not consistent with the production of “organic” sprouts and that its use is not
allowed in some countries, such as Germany (Weiss and others 2007). Some sprout
firms and countries are adopting alternative treatments. For example, in Japan, one
of the major sprout-consuming countries, a heat treatment for disinfection of mung
beans is used (Bari and others 2010a).

20000 ppm calcium hypochlorite

Cited by the FDA and considered as the reference standard for seed disinfection,
20000 ppm calcium hypochlorite has been reported to effectively reduce the
microbial load on seeds. Seed treatment with 20000 ppm calcium hypochlorite at
room temperature for 10 to 15 min produces an average of 3.08 log CFU/g reduction
in bacterial population (Figure 2). However, there is a high level of variability in the
efficacy reported in the literature (Table 1). Increasing the treatment time did not
improve the level of microbial reduction but did decrease the germination rate of the
seeds (Kim and others 2003). Although the variation in efficacy could be attributed
to many factors, including different experimental protocols, the inability of the
chlorine to reach the bacteria in contaminated seeds could be a major cause of the
inconsistency. Greater microbial reductions were observed on smooth seeds than
on scarified seeds (Holliday and others 2001) and different seed surface coating
conditions and topology were likely the key contributors to the variation of
accessibility to the pathogens (Fransisca and Feng 2012). This treatment, under
typical conditions (10 to 15 min), does not significantly affect the germination rate,
and the disinfection effects are similar on different types of bacteria.
Figure 2.

 Open in figure viewer

 Download Powerpoint slide

Box plot showing comparison of seed disinfection efficacies by various treatments,

reported by 44 published articles. Bar in the middle of the box is mean and the
dashed line represents the mean reduction of reference treatment, 20000 ppm
Ca(OCl)2. Statistical analyses were summarized in the table. General linear model
was performed using SAS 9.2 for Windows software. The symbol (*) denotes
statistical significance (P < 0.05).

Other chemicals

The need to decontaminate seeds in a way that is efficacious, cost-effective, safe,

and environmentally friendly, without a negative impact on seed germination, has
prompted researchers to continue testing alternative seed treatments. Many other
commonly used chemical agents such as H2O2, ethanol, lactic acid, peroxyacetic
acid, and fatty acid have been tested for seed disinfection (Beuchat 1997; Lang and
others 2000; Kim and others 2003; Liao 2009; Buchholz and Matthews 2010; Chang
and others 2010) but their efficacies vary (Table 1). In addition, few of these
treatments have been the subject of repeated studies, making an accurate estimate
of efficacy of individual treatments very difficult. Although the average efficacy of
some of these chemical treatments may be similar to that of 20000 ppm calcium
hypochlorite (Figure 2), and the maximal reported efficacy could reach a microbial
reduction as high as 7.11 log CFU/g, the majority of these treatments can only
achieve a moderate bacterial reduction (<3.50 log CFU/g). The used treatment
conditions also vary significantly between different chemicals. The treatment time
could be as short as several seconds and as long as days at either ambient or
elevated temperatures. Because of the inconvenience of the applications due to the
varying treatment conditions, inadequate validation of efficacies, as well as the
influence of FDA guidance for seed disinfection, these treatments are less adopted
than the 20000 ppm calcium hypochlorite.

Physical Inactivation

Physical methods, especially heat and high-pressure treatments, have been used
for microbial inactivation in applications by other food industry groups, for example,
juice processing. The recent adoption of this strategy for sprouts appears promising.
Unlike chemical treatments, which may be limited by inaccessibility to pathogens
sheltered in scarified surfaces and the interior of the seeds, physical treatments have
better penetration characteristics for reaching bacteria in those places. In addition,
physical methods such as heat and high pressure are more environmentally friendly
and have been actively promoted by researchers in countries where the 20000 ppm
calcium hypochlorite treatment is not an option.
Heat

Heat treatment has been a primary alternative to chemical treatment, especially in

Japan. Treating seeds at a relatively warm temperature in the range of 50° C for
hours or days have been reported, with varying disinfection efficacies (Bari and
others 2003, 2009a, 2010; Hu and others 2004; Feng and others 2007; Neetoo and
Chen 2011). A quick treatment at an elevated temperature of 90° C for about 90 s
followed by rapid (30 s) cooling to 0° C improved the disinfection compared to the
treatments at lower temperatures for longer duration (Bari and others 2008).
However, due to the decreased germination rate resulting from the high temperature,
this method is not applicable for use on seed for sprouting, although a similar
procedure at 75° C appears promising (Bari and others 2009b). Recent work on
using heat for disinfection involves treating seeds in hot water at 85° C for 10 s (Bari
and others 2010a), which was reported to consistently achieve a bacterial reduction
of approximately 3.0 log CFU/g. Increasing the treatment time to 40 s has been
shown to further improve treatment efficacy by more than 0.5 log without a significant
reduction of germination rate (Bari and others 2010a, 2008b).

High pressure

The application of high pressure for seed disinfection also appears to be very
promising (Table 1 and Figure 2). Seed treatment at 500 to 600 MPa for 2 min at
room temperature can achieve a reduction in bacterial load of 3.50 log CFU/g or
more (Neetoo and others 2008; 2009b). Coupled with presoaking or higher
temperature, the high-pressure seed disinfection treatment can be even more
effective (Neetoo and others 2009a,b, 2010). Treatment at a medium pressure (300
MPa) for a longer time (15 min) produced similar disinfection results, although a
delay in germination was observed (Wuytack and others 2003). On average, the
high-pressure treatment can achieve a microbial reduction of 5.09 log CFU/g, which
is the most effective method so far among all available seed disinfection treatments
and is significantly better than the reference standard of 20000 ppm calcium
hypochlorite (Figure 2). The variation in the efficacy of high-pressure treatment
between different studies is small, which could be explained by the relatively few
number of researchers who have conducted the high-pressure studies.
Nevertheless, additional studies should be performed to verify efficacy under
different treatment conditions and seed types and to determine if the method is
commercially viable.

Irradiation

Irradiation at different dosages has been tested and approved for seed
decontamination by the FDA (U.S. Food and Drug Administration 2000). Although
the maximal dose for seed treatment approved by the FDA is 8 kGy (U.S. Food and
Drug Administration Memorandum 1999), the majority of studies targeted 2 kGy or
below as a generally acceptable practice for the treatments of seeds destined for
sprouting with desirable germination and quality. Since there are relatively few
studies available, we combined studies conducted at 2.0 kGy with those conducted
at 1.5 kGy to summarize treatment efficacy for this review (Rajkowski and Thayer
2001; Schoeller and others 2002; Thayer and others 2003; Bari and others 2004;
Saroj and others 2007; Nei and others 2010). Irradiation at this dose range can
consistently produce an average of 3.18 log CFU/g reduction in microbial
contamination (Figure 2), and it is almost equally effective in seeds and final sprout
products (data not shown); however, its impact on length, yield, and appearance of
sprouts (Rajkowski and others 2003) and potential nutrient loss (that is, vitamin C,
Bari and others 2004) are issues that currently prevent this method from becoming
widely acceptable as a disinfection treatment.

Biological Inhibition

Competitive exclusion from normal microbial flora has received some attention for
the control of pathogens in food, and this strategy may also work in sprouts to a
certain extent. Multiple strains of bacteria, bacteriocins, as well as bacteriophages
have been tested to inhibit the growth or multiplication of Salmonella or E. coli
O157 : H7 during sprout production (Nandiwada and others 2004; Matos and
Garland 2005; Fett 2006; Kocharunchitt and others 2009; Ye and others 2010).
Although 2 strains of lactic acid bacterium have been shown to be able to effectively
control (> 6 log CFU/g inhibition) the growth of either Salmonella or E. coli O157 : H7
in culture broth (Wilderdyke and others 2004), tests on inoculated seeds reveal less
satisfactory results. In general, results from the limited number of studies with seeds
suggest that the inhibition of microbial contamination by these biological agents
could potentially achieve a similar effect as 20000 ppm calcium hypochlorite in
reduction of microbial populations. However, due to the complexity of its application,
uncertainty about its efficacy on an industrial scale, and the concern of potential
adverse health effects, whether this strategy will become a major control option in
sprout production remains to be determined.

Combined Treatment

As discussed above, seed disinfection by a single type of treatment can significantly

reduce microbial populations; however, combination treatments could be more
effective, and synergistic effects may be achieved by applying 2 or more methods,
either sequentially or simultaneously.

Different combinations of treatments (Table 1) have been studied to search for better
seed disinfection methods (Lang and others 2000; Holliday and others 2001; Bari
and others 2003; 2009, 2010a,b; Kim and others 2003, 2010; Pierre and Ryser 2006;
Nei and others 2010; Ye and others 2010; Zhao and others 2010; Bang and others
2011; Fransisca and others 2012; Fransisca and Feng 2012). Although some
combined treatments are not effective as expected, they usually produce a greater
microbial reduction than can be achieved by the individual treatments alone. Not
surprisingly, the overall disinfection efficacy of the combined treatments is
significantly better than 20000 ppm calcium hypochlorite alone (Figure 2). While
identifying an optimal combination might be challenging due to the complexity
introduced by the application of multiple treatments and to issues related to practical
implementation, combination treatments, once established, could provide the best
control strategy.

Conclusions

Seed sanitation is an important part of a multihurdle risk management strategy in the

production of sprouts. Numerous studies have evaluated different approaches for
pathogen reduction in sprouting seeds. The accumulated evidence is rich so that a
fair estimate of the efficacy of different treatments can be generated, as summarized
in Table 1. While the majority of the published studies considered here successfully
avoided significant losses in seed germination, disinfection efficacies vary. Our
results suggest that although 20000 ppm calcium hypochlorite is currently
considered to be the reference standard for seed disinfection, it may be equivalent
to other approaches, or even less efficient than certain newly emerged treatments
(Figure 2). Physical treatments, especially high pressure, show promise for
producing better disinfection results than that obtained by 20000 ppm calcium
hypochlorite treatment. The improvement may be attributable partially to the greater
ability of physical treatments to reach and affect bacteria both inside and outside of
seeds. Although irradiation is at least as effective, this treatment has not been
accepted by the industry (or sprout consumers) for a number of reasons, including
the impact of this treatment on sprout quality and yield. Certain combined treatments
also appear to be more effective than 20000 ppm calcium hypochlorite. More studies
are needed to confirm the efficacy of physical and combined seed disinfection
treatments. The current reference standard may not be the best sanitation treatment
for sprout production, because the high levels of chemicals can pose environmental
and worker safety risks. Alternative intervention approaches should be considered.
Given the “green” nature of physical treatments, further investigations of this type of
approach are warranted.

It is worth mentioning that the majority of articles included in this review report studies
involving varying experimental designs and procedures and the use of artificially
inoculated seeds with significant higher pathogen load than naturally contaminated
seeds. In addition, although it appears that seed disinfection could achieve a similar
effect on Salmonella and E. coli O157 : H7, studies on other common or emerging
pathogens such as L. monocytogenes and E. coli O104 : H4 are lacking. Therefore,
it is difficult to recommend specific treatments to the sprout industry based on the
published literature. Risk reduction measures should be validated for the specific
conditions in a production facility. It also should be pointed out that even though
satisfactory seed disinfection can be achieved, it is possible that the final products
can still be contaminated due to the growth of pathogens survived in the treated
seeds during sprouting. It is unlikely that the seed disinfection alone will eliminate
the microbial contamination in sprout production, and thus, additional risk
management options, such as a microbial sampling and testing program, should also
be implemented in order to minimize microbial hazards associated with sprouts. The
change of the current industrial practice in sprout production based on new science-
based evidence will be necessary to improve sprout safety.
Acknowledgments

This work was supported by the FDA Commissioner's Fellowship Program. We

thank Dr. Mary Lou Tortorello and Dr. Absar Alum for critically reading and editing
this manuscript.

R901–R909, X.E. Li, S.M. Jervis and M.A. Drake no se sabe

1. Previous article in issue: Composition-Based Prediction of Temperature-

Dependent Thermophysical Food Properties: Reevaluating Component
Groups and Prediction Models
 2. Next article in issue: The Role of Wheat and Egg Constituents in the
Formation of a Covalent and Non-covalent Protein Network in Fresh and
Cooked Egg Noodles

Volume 82, Issue 1January 2017 Pages 16–23

Concise Reviews and Hypotheses in Food Science

XXXVII. Novel Umami Ingredients: Umami Peptides and

Their Taste
1. Yin Zhang, Chandrasekar Venkitasamy, Zhongli Pan, Wenlong Liu and
Liming Zhao

Abstract

Umami substances are very important for food seasoning and healthy eating. In
addition to monosodium glutamate and some nucleotides, recent investigations have
revealed that several peptides also exhibit umami taste. In recent years, 52 peptides
have been reported to show umami taste, including 24 dipeptides, 16 tripeptides, 5
octapeptides, 2 pentapeptides, 2 hexapeptides, 1 tetrapeptide, 1 heptapeptide, and
1 undecapeptide. Twenty of these peptides have been examined for the present of
umami taste. In this review, we have listed these umami peptides based on their
category, source, taste, and threshold concentration. The evidence for peptides
showing umami taste, the umami taste receptors on the human tongue, and the
peptides whose umami taste is controversial are also discussed.

Introduction

Umami taste has been widely accepted as the 5th basic form of taste, along with the
other 4 basic tastes of sweet, sour, salty, and bitter. This acceptance has been
mainly attributed to the identification of G protein–coupled receptors for glutamate
such as mGluR4 (Chaudhari and others 2000) and the heteromeric T1R1+T1R3
receptor (Nelson and others 2002). Umami ingredients are very important for food
seasoning and are widely used in food production. They also show many health
benefits, including reducing fat deposition, weight gain, and plasma leptin levels in
rats (Kondoh and Torii 2008; Nakamura and others 2008). Umami ingredients were
found to regulate gastrointestinal functions (Nakamura and others 2008) and to
decrease the risk of stroke and coronary heart disease in adults by reducing sodium
intake in their diets (Yamaguchi 1998; Aburto and others 2013). These food-
seasoning and health-improving functions of umami ingredients have provoked more
investigations to find new umami substances and evaluate their taste properties
(Kunishima and others 2000; Masic and Yeomans 2014).

Monosodium glutamate (MSG) was the 1st molecule reported to have umami taste
(Ault 2004). Then, in 1967, ribonucleotides such as guanosine monophosphate
(GMP) and inosine monophosphate (IMP) were found to have synergistic effects
with MSG (Yamaguchi 1967; Zhang and others 2013). Later, succinic acid (Med.
1974), theanine, gallic acid, theogallin (Kaneko and others 2006), pyroglutamic acid
(Buckholz and Scharpf 1994), N-glycosides (Schlichtherle-Cerny and others 2002),
pyroglutamyl peptides (Amado and Schlichtherle-Cerny 2003), N-acetylglycine
(Grigorov and others 2003), succinoyl amides of amino acids (Frerot and BENZI
2004), and glycopeptides (Iwasaki and others 2004) were all reported to have umami
taste. Alapyridaine, which is a product of the Maillard reaction (Soldo and others
2003a), and morelid, which is found in Morel mushrooms (Rotzoll and others 2005),
were also found to enhance umami taste (Ley and others 2006). In addition to these
natural and synthesized umami compounds that have been proven to possess or
enhance umami taste (Kondoh and Torii 2008; Nakamura and others 2008), recent
investigations have demonstrated that a few peptide molecules produced from
hydrolysates of fish protein, beef bouillon, or other foods, have umami taste, as do
some synthesized peptides (Arai and others 1973; Tamura and others 1989b; Winkel
and others 2008). However, umami peptides are widely questioned for their taste
characteristics. Several researchers believe that umami peptides have umami taste,
but others disagree. There is an ever-increasing demand for natural food products
and ingredients (Winkel and others 2008). Umami peptides have also been found to
be desirable natural ingredients, with a high demand for their full development and
application in food products. This article provides a review of research carried out on
umami peptides and their taste characteristics; umami taste receptors; and the
disputes regarding the umami taste of umami peptides.

Peptides Showing Umami Taste

Umami taste is called a meaty, broth-like, or savory taste, as it has been used to
describe the taste of savory and meat broth foods (Lioe and others 2010; Coulier
and others 2011). One of the earliest reports on umami-taste peptides was on the
glutamyl umami oligopeptides (3 dipeptides and 1 tripeptide; Table 1), which were
separated and purified from α-chymotrypsin-modified soybean protein hydrolysate
(Arai and others 1972). Arai and others (1973) investigated the correlation between
the chemical structures and taste characteristics of l-glutamyl oligopeptides and
reported that the highly acidic (hydrophilic) l-glutamyl oligopeptides might have a
umami taste that contributes to the favorable flavor of food protein hydrolysate. They
thought that interactions of the cationic amino and anionic carboxyl groups induced
by the 5-member l-glutamyl ring structure were the reason for the brothy taste.
Fujimaki and others (1973) used 5 proteases (pepsin, rapidase, papain, pronase,
and bioprase) to hydrolyze the protein concentrate of fish and found that the acidic
oligopeptide fraction (molecular weight lower than 1000) tasted brothy and had an
amicable after-taste. They isolated 4 dipeptides and tripeptides from the enzymatic
hydrolysate that had a flavor resembling that of MSG (Table 1; Noguchi and others
1975). Yamasaki and Maekawa (1978) isolated H-Lys-Gly-Asp-Glu-Glu-Ser-Leu-
Ala-OH from the gravy of papain-treated beef meat (beef umami peptide, BMP).
When BMP was reported, there were many arguments about its taste, and no new
findings on umami peptides or peptides contributing to the umami taste of
hydrolysate were published for the next several years. Twenty-four years after BMP
was 1st reported, Schlichtherle-Cerny and Amadò (2002) found 4 MSG-like
pyroglutamyl peptides from flavourzyme-hydrolyzed deamidated wheat gluten.
Winkel and others (2008) studied the structure, taste threshold level, and solubility
of these proglutamyl peptides and opined that binding of the GMP/IMP or glutamate
to an umami receptor might be the reason for the umami taste of these pyroglutamyl
peptides. Recently, Rhyu and Kim (2011) found that low molecular weight acidic
peptides (F-IV; 1000> MWP500) were the component compounds that contributed
to the umami taste of doenjang water extract. Su and others (2012) found 2 novel
umami peptides, an octapeptide and an undecapeptide, from peanut hydrolysate.
Bagnasco and others (2013) reported that medium-to-small size polypeptides
contributed to the umami taste of hydrolysate of rice middlings. Dang and others
(2014) isolated 2 umami peptides from water-soluble extractions of 2 kind of hams
(Table 1).

Table 1. Peptides reported to have umami taste

Amino
acid Reported
Type and
sequenc Taste as threshold
number of Source Authors
e of reported (pH) concentratio
peptides
umami n of umami
peptide

1. “–”, not reported.

Oka and
24 Dipeptides Asp-Ala Soy sauce Umami – Nagata
(1974)

Ohyama
Synthesize
Ala-Asp Bitter > Umami 13 mM and others
d
(1988)

Ohyama
Synthesize
Ala-Glu Umami (neutral) 1.5 mM and others
d
(1988)

Tamura and
Synthesize Salty/umami
Asp-Asp 4.79 mM others
d (6.0)
(1989a)

Tamura and
Synthesize
Asp-Glu Salty/umami(6.0) 1.25 mM others
d
(1989)
 Amino
acid Reported
Type and
sequenc Taste as threshold
number of Source Authors
e of reported (pH) concentratio
peptides
umami n of umami
peptide

Ohyama
Synthesize
Asp-Leu Umami (neutral) 2.5 mM and others
d
(1988)

Proteinase
Arai and
-modified
Glu-Asp Brothy – others
soybean
(1972)
protein

Arai and
Synthesize
Brothy (6.0) 200 mg% others
d
(1973)

Fish
Noguchi and
protein
MSG-like 200 mg% others
hydrolysat
(1975)
e

Tamura and
Synthesize Salty/umami
3.14 mM others
d (6.0)
(1989)

Proteinase
Arai and
-modified
Glu-Glu Brothy – others
soybean
(1972)
protein

Arai and
Synthesize
brothy (6.0) 1% solution others
d
(1973)

Fish
Noguchi and
protein
MSG-like (6.0) 150 mg% others
hydrolysat
(1975)
e
 Amino
acid Reported
Type and
sequenc Taste as threshold
number of Source Authors
e of reported (pH) concentratio
peptides
umami n of umami
peptide

Tamura and
Synthesize
Salty/umami(6.0) 2.73 mM others
d
(1989)

Maehashi
Synthesize
Umami 1% (g/mL) and others
d
(1999)

Ohyama
Synthesize
Glu-Leu Umami 3 mM and others
d
(1988)

Tamura and
Synthesize
Glu-Lys Umami (6.0) 3.12 mM others
d
(1989)

Tamura and
Synthesize
Glu-Orn Umami/sour 3.12 mM others
d
(1989)

Proteinase
Arai and
-modified
Glu-Ser Brothy – others
soybean
(1972)
protein

Arai and
Synthesize Weak brothy
– others
d (6.0)
(1973)

Fish
Noguchi and
protein
MSG-like (6.0) 200 mg% others
hydrolysat
(1975)
e

Arai and
Synthesize
Glu-Thr Brothy taste – others
d
(1973)
 Amino
acid Reported
Type and
sequenc Taste as threshold
number of Source Authors
e of reported (pH) concentratio
peptides
umami n of umami
peptide

Maehashi
Synthesize
Glu-Val Umami/sweet 1% (g/mL) and others
d
(1999)

Ohyama
Synthesize
Gly-Asp Umami 6 mM and others
d
(1988)

Ohyama
Synthesize
Gly-Glu Umami > bitter 0.8 mM and others
d
(1988)

Ohyama
Synthesize
Leu-Glu Umami > bitter 1.5 mM and others
d
(1988)

Tamura and
Lys- Synthesize Salty/umami
1.22 mM others
Gly·HCl d (6.0)
(1989)

Tamura and
Orn- Synthesize
Umami 1.5 mM others
Orn·2HCl d
(1989)

Tamura and
Orn- Synthesize
Salty/Umami 1.25 mM others
Ala·HCl d
(1989)

Deamidate
Schlichtherl
d wheat
e-Cerny and
pGlu-Pro gluten MSG-like –
Amadò
hydrolysat
(2002)
e

Fish
Noguchi and
protein
Thr-Glu MSG-like 300 mg% others
hydrolysat
(1975)
e
 Amino
acid Reported
Type and
sequenc Taste as threshold
number of Source Authors
e of reported (pH) concentratio
peptides
umami n of umami
peptide

Ohyama
Synthesize
Val-Asp Bitter>umami 25 mM and others
d
(1988)

Ohyama
Synthesize
Val-Glu Umami>Bitter 1.5 mM and others
d
(1988)

Ohyama
Ala-Asp- Synthesize
16 tripeptides Umami>Bitter 3 mM and others
Ala d
(1988)

Ala-Glu- Synthesize Ohyama,

Umami (neutral) 0.8 mM
Ala d et al. (1988)

Fish
Noguchi and
Asp-Glu- protein
MSG-like 300 mg% others
Ser hydrolysat
(1975)
e

Fish
Noguchi and
Glu-Asp- protein
MSG-like 300 mg% others
Glu hydrolysat
(1975)
e

Fish
Noguchi and
Glu-Gln- protein
MSG-like 200 mg% others
Glu hydrolysat
(1975)
e

Frerot and
Glu-Glu- Synthesize
Umami – Escher
Leu d
(1998)

Proteinase
Arai and
Glu-Gly- -modified
brothy – others
Ser soybean
(1972)
protein
 Amino
acid Reported
Type and
sequenc Taste as threshold
number of Source Authors
e of reported (pH) concentratio
peptides
umami n of umami
peptide

Fish
Noguchi and
protein MSG-
200 mg% others
hydrolysat like(neutral)
(1975)
e

Ohyama
Gly-Asp- Synthesize
Umami(neutral) 1.5 mM and others
Gly d
(1988)

Ohyama
Gly-Glu- Synthesize
Umami = Bitter 1.5 mM and others
Gly d
(1988)

Frerot and
Leu-Glu- Synthesize
Umami – Escher
Glu d
(1998)

Deamidate
Schlichtherl
d wheat
pGlu- e-Cerny and
gluten MSG-like –
Pro-Gln Amadò
hydrolysat
(2002)
e

Deamidate
Schlichtherl
d wheat
pGlu- e-Cerny and
gluten MSG-like –
Pro-Glu Amadò
hydrolysat
(2002)
e

Deamidate
Schlichtherl
d wheat
pGlu- e-Cerny and
gluten MSG-like –
Pro-Ser Amadò
hydrolysat
(2002)
e
 Amino
acid Reported
Type and
sequenc Taste as threshold
number of Source Authors
e of reported (pH) concentratio
peptides
umami n of umami
peptide

Fish
Noguchi and
Ser-Glu- protein
MSG-like 200 mg% others
Glu hydrolysat
(1975)
e

Ohyama
Val-Asp- Synthesize
Umami 13 mM and others
Val d
(1988)

Ohyama
Val-Glu- Synthesize
Umami (neutral) 1.5 mM and others
Val d
(1988)

Yamasaki
1 Glu-Ser- Synthesize Sour > astringent and
–
Tetrapeptide Leu-Ala d > umami > bitter Maekawa
(1980)

Yamasaki
2 Glu-Glu-
Synthesize Sour > astringent and
Pentapeptide Ser-Leu- –
d > umami > bitter Maekawa
s Ala
(1980)

Glu-Glu- Nakata and

Synthesize Sour/umami
Asp-Gly- 1.25 mM others
d /sweet
Lys (1995)

Yamasaki
Asp-Glu- Sour > astringent
2 Synthesize and
Glu-Ser- > umami > sweet –
Hexapeptides d Maekawa
Leu-Ala > bitter
(1980)

Cys-Cys- Dang and

Jinhua
Asn-Lys- Umami – others
hams
Ser-Val (2014)
 Amino
acid Reported
Type and
sequenc Taste as threshold
number of Source Authors
e of reported (pH) concentratio
peptides
umami n of umami
peptide

Ala-His-
Dang and
1 Ser-Val- Parma
Umami – others
Heptapeptide Arg-Phe- hams
(2014)
Tyr

Lys-Gly- Yamasaki
The gravy
5 Asp-Glu- and
of beef Delicious taste –
Octapeptides Glu-Ser- Maekawa
meat
Leu-Ala (1978)

Yamasaki
Synthesize Umami, sour, and
–
d sweet Maekawa
(1980)

Tamura and
Synthesize
Umami/sour 1.41 mM others
d
(1989)

Lys-Gly-
Nakata and
Ser-Leu- Synthesize Sour/umami/swe
0.78 mM others
Ala-Asp- d et
(1995)
Glu-Glu

Ser-Leu-
Nakata and
Ala-Asp- Synthesize
Umami/sour 1.25 mM others
Glu-Glu- d
(1995)
Lys-Gly

Ser-Leu-
Nakata and
Ala-Lys- Synthesize
Sour/umami 1.50 mM others
Gly-Asp- d
(1995)
Glu-Glu

Peanut Su and
Ser-Ser- hydrolysat Umami – others
Arg-Asn- e (2012)
 Amino
acid Reported
Type and
sequenc Taste as threshold
number of Source Authors
e of reported (pH) concentratio
peptides
umami n of umami
peptide

Glu-Gln-
Ser-Arg

Glu-Gly-
Ser-Glu-
1 Peanut Su and
Ala-Pro-
Undecapeptid hydrolysat Umami – others
Asp-Gly-
e e (2012)
Ser-Ser-
Arg

Table 1 enumerates a total of 52 peptides that were reported to show umami taste,
along with their category, source, taste, and threshold concentration. These 52
peptides include 24 dipeptides, 16 tripeptides, 5 octapeptides, 2 pentapeptides, 2
hexapeptides 1 tetrapeptide, 1 heptapeptide, and 1 undecapeptide. Comparison of
the taste of the synthesized and naturally formed dipeptides and tripeptides in Table
1, such as Glu-Ser, Glu-Asp, Glu-Glu, and Glu-Gly-Ser, shows that both dipeptides
and tripeptides have similar tastes. From Table 1, the threshold concentration of
MSG is 1.5 mM (Ohyama and others 1988; Soldo and others 2003b), which shows
that the umami taste of most dipeptides and tripeptides is weaker than that of MSG,
but the umami taste of Gly-Glu (0.8 mM) and Ala-Glu-Ala (0.8 mM) is stronger than
that of MSG. The tastes of tetrapeptides, pentapeptides, hexapeptides, and
heptapeptides were found to be different. The synthesized hexapeptides and
heptapeptides showed weaker umami taste than the naturally formed hexapeptides
and heptapeptides. The naturally formed octapeptides showed umami taste, but the
synthesized octapeptides showed not only umami taste but also sour and sweet
tastes.

Comparing the taste of all naturally formed and synthesized peptides in Table 1, it
can be found that the peptides from hydrolysates showed umami taste, but their
tastes changed after they were synthesized. This tendency became more obvious
as the size of the umami peptide increased. For all reported umami peptides, taste
confirmation of most of the naturally formed umami peptides was performed using
the synthesized peptide. There is a high possibility that this practice of using
synthesized peptides to confirm the taste of naturally formed peptides might be the
cause of many disputes about the taste of umami peptides.

Disputes about the Taste of Umami Peptides

There are 20 peptides whose tastes are in controversy, including 14 dipeptides, 5
tripeptides, and 1 octapeptide (Table 3). Some peptides were reported to show
umami taste when they were separated from hydrolysate but became controversial
when the peptide was synthesized to confirm the taste. In 1978, Yamasaki and
Maekawa (1978) stated that BMP had umami potency that was higher than
glutamate itself. Later, they synthesized BMP and compared it with isolated BMP,
finding that the taste of the synthesized BMP was savory, sour, and sweet even
though the synthesized and isolated BMPs were identical in many aspects
(Yamasaki and Maekawa 1980). Both Tamura and others (1989a) and Spanier and
others (1995) synthesized BMP, investigated its taste and found that BMP tasted
umami, but Tamura and others (1989a) did not perceive the sweet taste of BMP.
However, van Wassenaar and others (1995) did not agree with the findings of
Tamura and others (1989a), and they experimentally claimed that BMP had no
savory taste. They synthesized BMP and characterized it. BMP was tasted by a
trained flavor panel, and it was found that the synthesized BMP and a few peptide
fragments did not have any taste, umami, or otherwise. van Wassenaar and others
(1995) opined that impurities in the synthesized BMP might have caused disturbance
in the taste evaluation of the BMP synthesized by Tamura and others (1989a) and
Spanier and others (1995). They speculated that the impurities might have originated
from side reactions, racemization, or truncation of sequences (van Wassenaar and
others 1995). van den Oord and van Wassenaar (1997) re-examined 12 dipeptides
and 4 tripeptides that Tamura and others (1989a), Arai and others (1973), Ohyama
and others (1988), and Noguchi and others (1975) reported as having umami or
savory tastes. They analyzed the taste of 16 peptides at pH 4.0 and 6.0, with or
without 0.6% sodium chloride, and discovered that not any of the peptides had a
distinct umami taste, even at concentrations much higher than the reported
thresholds. To test possible synergy between the peptides and ribonucleotides, van
den Oord and van Wassenaar (1997) mixed 5.4 mM Glu-Glu at pH 6.0 with 0.6 mM
IMP and reported that the taste of the mixture was perceived to be the same as that
of IMP alone under the same conditions. However, Maehashi and others (1999)
performed similar experiments and found that there was an enhancement of taste
between 1% Glu-Glu solution and 0.02% IMP. van den Oord and van Wassenaar
(1997) attributed the discrepancy between their results on the taste of peptides and
those of other researchers to the following 2 reasons. (1) The discrepancy might be
caused by the presence of bitter peptides and derivatives of amino acids. These
chemicals might have influenced the taste evaluation. A similar problem has reported
when evaluating the salty taste of ornithyl peptides (Tada and others 1984; Tuong
and Philippossian 1987). (2) The discrepancy might have originated from the actual
tasting procedures and cultural differences of the taste panelists. However, they
concluded that the actual tasting procedures and cultural differences had little
influence on umami taste analysis by comparing the perception difference of umami
taste between panelists of American and Japanese, and between panelists of Dutch
and Japanese.

In our opinion, taste disputes might be caused by the following 4 other factors. The
1st reason is that the method of preparation of umami peptides might influence the
taste analysis of the peptides. The investigation of Maehashi and others (1999) is
taken as an example; they reexamined the taste of 11 umami peptides obtained from
hydrolysate. They synthesized the 11 umami peptides and found that few showed
an independent umami taste. The second reason might be that the isomeric
structural differences of the peptides might also have influenced their flavor
properties. For example, MSG exists in 2 forms, namely, D and L. The L form is the
naturally occurring form, and only the L form possesses flavor activity. Similarly,
there are 3 possible isomeric forms of nucleotides, namely, 2′, 3′, and 5′, and only 5′
nucleotides showed umami enhancing effect (Maga and Yamaguchi 1983). The third
reason could be that differences in the spatial structure of the umami peptides from
hydrolysate and synthesized ones might have disturbed the taste analysis. Figure 1
shows the spatial structures of the natural BMP and BMP with one D-form amino
acid. It demonstrates that if one D-form amino acid is used during the synthesis of
BMP, it will result in an obvious difference in its spatial structure. According to an
umami receptor investigation (Chaudhari and others 2000; Nelson and others 2002),
the spatial structure of an umami substance is a key to the recognition of umami
taste. Therefore, the spatial structural changes resulting from the isomeric forms of
amino acids during the synthesis of a peptide may influence the umami taste of the
peptide. The 4th reason might be that the interactions of umami peptides with other
compounds might have disturbed the proper assessment of their taste.

Figure 1.

 Open in figure viewer

 Download Powerpoint slide

Spatial structures of natural BMP and BMP with one D-form amino acid.

Interactions between Umami Peptides and Other Compounds

The taste interactions of newly found umami peptides and other flavor compounds
are characterized as shown in Table 2. Recent investigations have demonstrated
that there are 3 types of interactions of umami peptides, the interactions of peptide
with peptide, peptide with nucleotide, and peptide with cation.

Table 2. Peptides that show umami taste when mixed with other ingredients

Peptide and Threshold

Interaction
interaction Flavor concentration Author
category
ingredients reported (mM)

1. a

–”, not reported.

Tamura
Peptide with Lys-Gly + Asp-Glu-
Umami/sour 1.41 and others
peptide Glu + Ser-Leu-Ala
(1989a)

Orn-β-Ala + Asp- Tamura

Glu-Glu + Ser-Leu- Umami/sour 1.41 and others
Ala (1989)

Tamura
Lys-Glu + Glu-Glu
Umami/sour 0.94 and others
+ Ser-Leu-Ala
(1989)

Maehashi
Peptide with 0.5% Glu-Glu +
umami/salty/sweet – and others
nucleotide 0.02% IMP
(1999)

Maehashi
0.5% Glu-Val +
Umami/sour – and others
0.02% IMP
(1999)

Maehashi
0.5% Ala-Asp-Glu
Umami/bitter – and others
+ 0.02% IMP
(1999)

Maehashi
0.5% Ala-Glu-Asp
Umami/sour – and others
+ 0.02% IMP
(1999)

Maehashi
0.5% Asp-Glu-Glu
Umami/salty – and others
+ 0.02% IMP
(1999)
 Peptide and Threshold
Interaction
interaction Flavor concentration Author
category
ingredients reported (mM)

Maehashi
0.5% Ser-Pro-Glu
Umami/bitter/salty – and others
+ 0.02% IMP
(1999)

0.5% Glu-Glu +
0.5% Glu-Val + Maehashi
0.5% Asp-Glu-Glu Umami/sour/salty – and others
+ 0.5% Glu-Glu- (1999)
Asn + 0.02% IMP

0.5% Asp-Glu-Ser
Maehashi
+ 0.5% Glu-Glu +
Sour/umami/bitter – and others
0.5% Asp-Glu-Glu
(1999)
+ 0.02% IMP

Nakata and
Peptide salt Asp-Asp + Na+ Salty/umami 3.33 others
(1995)

Nakata and
Asp-Asp + K+ Indistinct umami 3.24 others
(1995)

Nakata and
Asp-Glu + Na+ Umami/salty 3.32 others
(1995)

Nakata and
Asp-Glu + K+ Indistinct umami 3.72 others
(1995)

Nakata and
Glu-Asp + Na+ Salty/umami 1.74 others
(1995)

Nakata and
Glu-Asp + K+ Indistinct umami 1.71 others
(1995)
 Peptide and Threshold
Interaction
interaction Flavor concentration Author
category
ingredients reported (mM)

Nakata and
Glu-Glu + Na+ Umami/salty 2.22 others
(1995)

Nakata and
Glu-Glu+K+ Indistinct umami 3.62 others
(1995)

Nakata and
Asp-Asp-Asp +
Salty/umami 3.23 others
Na+
(1995)

Nakata and
Asp-Asp-Glu+Na+ Umami/salty 2.06 others
(1995)

Nakata and
Asp-Glu-Asp + Na+ Umami/salty 2.68 others
(1995)

Nakata and
Glu-Asp-Asp + Na+ Salty/umami 3.23 others
(1995)

Nakata and
Asp-Glu-Glu + Na+ Umami/salty 2.09 others
(1995)

Nakata and
Glu-Asp-Glu + Na+ Salty/umami 3.09 others
(1995)

Nakata and
Glu-Glu-Asp + Na+ Umami/salty 3.47 others
(1995)

Nakata and
Glu-Glu-Glu + Na+ Umami/salty 2.09 others
(1995)
 Peptide and Threshold
Interaction
interaction Flavor concentration Author
category
ingredients reported (mM)

Peptide with Lys-Gly-Asp-Glu- Wang and

other taste Glu-Ser-Leu-Ala + Umami increased – others
ingredients MSG (1996)

Lys-Gly-Asp-Glu- Wang and

Glu-Ser-Leu-Ala + Umami increased – others
NaCl (1996)

Table 3. Controversial umami peptides

Taste (pH)
Type and
Taste as Reference perceived
number of Peptide Authors
reported (pH) reported in present
peptides
work

van den
Not umami, Oord and
14 Ohyama and
Ala-Glu Umami (neutral) no other van
Dipeptides others (1988)
taste Wassenaar
(1997)

Maehashi
Oka and
Asp-Ala Umami Tasteless and others
Nagata (1974)
(1999)

van den
Not umami,
Oord and
Salty/umami Tamura and no other
Asp-Asp van
(6.0) others (1989a) taste at
Wassenaar
either level
(1997)

van den
Not umami,
Oord and
Salty/umami Tamura and no other
Asp-Glu van
(6.0) others (1989) taste at
Wassenaar
either level
(1997)

Not umami,
Salty/umami Nakata and no other van den
Asp-Glu
(6.0) others (1995) taste at Oord and
either level van
 Taste (pH)
Type and
Taste as Reference perceived
number of Peptide Authors
reported (pH) reported in present
peptides
work

Wassenaar
(1997)

van den
Not umami, Oord and
Ohyama and
Asp-Leu Umami (neutral) no other van
others (1988)
taste Wassenaar
(1997)

Noguchi and
Ohyama and
Asp-Leu Umami (neutral) Bitter(6.0) others
others (1988)
(1975)

van den
Not umami,
Oord and
Arai and no other
Glu-Asp Brothy(6.0) van
others (1973) taste at
Wassenaar
either level
(1997)

MSG-like Noguchi and

(neutral) others (1975)

Tamura and
Salty/umami(6.0)
others (1989)

Nakata and
Salty/umami(6.0)
others (1995)

van den
Not umami,
Oord and
Arai and slightly
Glu-Glu Brothy(6.0) van
others (1973) bitter at any
Wassenaar
level
(1997)

Noguchi and
MSG-like (6.0)
others (1975)

Tamura and
Salty/umami(6.0)
others (1989)
 Taste (pH)
Type and
Taste as Reference perceived
number of Peptide Authors
reported (pH) reported in present
peptides
work

van den
Not umami,
Oord and
Ohyama and no other
Glu-Leu Umami (neutral) van
others (1988) taste at
Wassenaar
either level
(1997)

Arai and
Ohyama and
Umami (neutral) bitter (6.0) others
others (1988)
(1973)

van den
Not umami,
Oord and
Tamura and no other
Glu-Lys Umami (6.0) van
others (1989) taste at any
Wassenaar
level
(1997)

van den
Not umami, Oord and
Weak brothy Arai and
Glu-Ser no other van
(6.0) others (1973)
taste Wassenaar
(1997)

Noguchi and
MSG-like (6.0)
others (1975)

Maehashi and Arai and

Glu-Val Umami/sweet flat
others (1999) others 1973

van den
Not umami,
Oord and
Salty/umami Tamura and slightly
Lys-Gly van
(6.0) others (1989) bitter at any
Wassenaar
level
(1997)

van den
Not umami,
Oord and
Ala-Glu- Ohyama and no other
5 Tripeptides Umami (neutral) van
Ala others (1988) taste at any
Wassenaar
level
(1997)
 Taste (pH)
Type and
Taste as Reference perceived
number of Peptide Authors
reported (pH) reported in present
peptides
work

Maehashi
Asp-Glu- Noguchi and
MSG-like Sour, salty and others
Ser others (1975)
(1999)

van den
Not umami,
Oord and
Glu-Glu- Noguchi and no other
MSG-like (6.0) van
Glu others (1975) taste at
Wassenaar
either level
(1997)

van den
Not umami,
Oord and
Gly-Asp- Ohyama and no other
Umami (neutral) van
Gly others (1988) taste at any
Wassenaar
level
(1997)

van den
Not umami,
Oord and
Val-Glu- Ohyama and no other
Umami (neutral) van
Val others (1988) taste at any
Wassenaar
level
(1997)

Yamasaki and
Maekawa
Lys-Gly- (1978); van
Not umami
1 Asp-Glu- Umami, sour, Yamasaki and Wassenaar
or other
Octapeptide Glu-Ser- sweet Maekawa and others
taste
Leu-Ala (1980); (1995),
Tamura and
others (1989)

Tamura and others (1989a) examined the taste of a mixture containing N-terminal
dipeptide (Lys-Gly), acidic tripeptide (Asp-Glu-Glu), C-terminal tripeptide (Ser-Leu-
Ala), and salty dipeptide (Orn-β-Ala). They found that all the combinations produced
an umami taste with the same character and almost the same strength as BMP.
According to the sensory analysis results of Tamura and others (1989a), the tastes
of Lys-Gly, Ser-Leu-Ala, Asp-Glu-Glu, and Orn-β-Ala are umami, sour, bitter, and
bitter/sweet, respectively. The taste of the mixture was similar to the umami taste of
BMP, indicating that there were interactions that enhanced, suppressed or masked
taste, as well as synergistic interactions that affected taste.

Maehashi and others (1999) found that one synthesized peptide (Glu-Val) and 5
synthesized tripeptides (Ala-Asp-Glu, Ala-Glu-Asp, Asp-Glu-Glu, Ser-Pro-Glu, and
Glu-Glu-Asn) showed umami taste when they were mixed with 0.2% IMP. The taste
of peptide mixtures (Glu-Glu+Glu-Val+Asp-Glu-Glu+Glu-Glu-Asn and Asp-Glu-
Ser+Glu-Glu+Asp-Glu-Glu) combined with 0.02% IMP also showed umami taste.
Among these peptides, Glu-Glu, Glu-Val and Asp-Glu-Ser were found to elicit weak
umami taste (Noguchi and others 1975), but their tastes were enhanced when they
were mixed with IMP. Other di- or tripeptides did not show umami taste on their own,
but they showed umami taste when they were mixed with 0.02% IMP. These results
indicated that there are complicated interactions between IMP and peptides.

Nakata and others (1995) investigated the interactions of acidic peptides with Na+
and K+ and found that as pH levels increased to 6 using NaOH and KOH, the sodium
salts of the acidic dipeptides showed both meaty and salty tastes. The potassium
salts of each peptide showed a blurred taste that could not be categorized as umami
or salty. The sensory analysis of the salts of acidic peptides suggested that the
sequence of Glu and Asp in the peptide might be a key factor establishing salty and
umami tastes.

Winkel and others (2008) compared the threshold limit values of the taste of MSG,
GMP, lactoyl GMP, and acetyl GMP in water, sodium chloride solutions, and model
broths. They found that the taste threshold value of MSG in water (5 ppm) > GMP in
0.5% NaCl and 0.05% MSG solution (0.5 ppm) > lactoyl GMP in bouillon (0.03 ppm)
> acetyl GMP in bouillon (0.01 ppm). These results implied that the media used to
dissolve the umami peptide might also have affected their taste assessment.

The results of taste evaluation of peptides are impacted by other peptides,

nucleotides, and cations, and possibly by the dissolved media. These substances
might disturb the accurate evaluation of umami peptides. The interactions of umami
peptides with sour, sweet, or bitter substances and the dissolved media should be
further investigated. However, only limited investigations have been performed in
recent years on these aspects. The possible reasons might include (1) controversy
on the taste of umami peptides, preventing more detailed investigations on the
interactions of umami peptides with other taste substances; and (2) umami peptides
are zwitterionic, which makes the interactions between peptides and other flavor
substances too complex to differentiate the taste characteristics. To further prove
the existence of umami taste and prevent disturbance caused by other factors during
umami taste evaluation, recent studies have focused on umami receptors.

Umami Receptors

Taste plays an important role in assessing the nutritional value of food and prevents
the absorption of toxic substances (Chandrashekar and others 2006). Taste receptor
cells are gathered into taste buds. Sweet, bitter, sour, salty, and umami ingredients
stimulate the taste buds and are recognized by different cells containing specialized
receptors. The obtained stimulating signals will be transformed into neural signals.
Thus, the taste quality of taste ingredients will be felt (Chandrashekar and others
2006). Because umami taste was established as one of the 5 basic tastes, several
discoveries of umami receptors in the taste buds have been reported (Uneyama and
others 2009).

The 1st umami receptor, called taste-metabotropic glutamate 4 (mGluR4), was

discovered in 2000 (Chaudhari and others 2000). Taste-mGluR4 is an unusual
dimeric G protein–coupled receptor (GPCR; Kunishima and others 2000), and a
truncated version of the famous glutamate receptor mGluR4. The 2nd umami
receptor, T1R1+T1R3, was found in 2002 (Li and others 2002; Nelson and others
2002), and the 3rd, an unusual mGlu receptor, which is relevant to the brain
glutamate receptor mGluR1, was found in 2005 (San Gabriel and others 2005).
Among these receptors, T1R1+T1R3 is broadly considered as the major receptor for
umami stimuli (Behrens and others 2011). The umami receptor T1R1+T1R3 is
special in that it belongs to the class C GPCR family of proteins and has 7
transmembrane sections (Temussi 2009). Most of the class C GPCRs are
homodimers, but the umami receptor T1R1+T1R3 is a heterodimer (Nelson and
others 2002). The typical model of the umami receptor T1R1+T1R3 was reported by
Temussi (2012) and Chandrashekar and others (2006). The monomer of T1R1 is
considered critical for sensing umami taste (Mouritsen and Khandelia 2012). It
makes up of an extracellular Venus flytrap domain (VFTD). The VFTD was found as
the ligand-binding site within homologous proteins such as mGluRs (Kunishima and
others 2000). In mGluRs, 2 lobes of the VFTD could keep open or close together
with the conformations of the protein (Kunishima and others 2000; Tsuchiya and
others 2002; Ahmed and others 2007; Muto and others 2007). Taking Glu and 5’-
ribonucleotides as examples, Glu stabilizes the closed conformation (Mouritsen and
Khandelia 2012), and this makes Glu have umami taste. IMP and GMP bind to the
outer garment of the VFTD and markedly strengthen the response of the mGluRs to
glutamate (Zhang and others 2008). Therefore, both of them intensify the taste of
Glu.

The umami taste of Asp-Glu-Ser was doubted by Maehashi and others (1999), and
the umami tastes of Glu-Asp, Glu-Glu, and Glu-Ser were questioned by van den
Oord and van Wassenaar (1997). However, Kim and others (2015) presented 5
umami peptides (Glu-Asp, Glu-Glu, Glu-Ser, Asp-Glu-Ser, and Glu-Gly-Ser) with a
bitter substance (salicin) in a Ca2+-flux signaling assay. They found that these umami
peptides markedly reduced the salicin-induced intracellular calcium influx with time
increase. They reported that the results provided evidence to prove that umami
peptides restrain bitter taste by bitter taste receptor(s).

A few recent studies have suggested that other receptors might also be participated
in umami taste sensation. Damak and others (2003) reported that T1R3-knockout
mice retained the taste response to monosodium glutamate. Maruyama and others
(2006) found that taste buds lacking T1R3 mice still showed markedly glutamate-
evoked Ca2+ responses. Single-unit recordings on taste sensory neurons indicated
that umami taste responses did not need T1R3-containing receptors (Yoshida and
others 2009). The receptor type GPR92 is activated by peptide hydrolysates and
free amino acids (Choi and others 2007a, 2007b), and it was expressed in
enteroendocrine cells of the gastric mucosa and in G cells, which secreted gastrin
upon stimulus with protein hydrolysate (Haid and others 2012). These results
indicated that umami cells might react to both amino acids and peptides in protein
hydrolysates (Haid and others 2013). Similar to the umami receptor T1R1+T1R3,
the sweet receptor T1R2+T1R3 is also a heterodimer (Chandrashekar and others
2001; Li and others 2002). T1R3 is a common component of both the T1R1+T1R3
and the T1R2+T1R3. However, the sweet receptor T1R2+T1R3 recognizes not only
sweet amino acids (glycine and D-tryptophan) but also a sweet dipeptide
(aspartame) and sweet proteins such as monellin, brazzein, and thaumatin (Hatada
and others 1985; Li and others 2002; Temussi 2002). The large cavity on the T1R3
promoter and the molecular structure of the sweet protein are the key factors for the
recognition of the sweet protein (Temussi 2002, 2011). Molecular mechanics
demonstrates that T1R1+T1R3 binds points in a comparatively mini binding hole
(Zhang and others 2008). These investigations of umami and sweet receptors
suggested the existence of umami peptides and that their spatial structure might
greatly influence their flavor. Therefore, it will be worth further investigating the
interactions of peptide spatial structures with umami receptors and the taste
characteristics of these peptides.

Based on the results of the reported investigations, it is suggested that more

research on umami peptides and their tastes should be required, focusing on (1) the
preparation of natural peptides using peptide gene expression methods to
investigate differences in taste with synthesized peptides; (2) the spatial structure of
synthesized and naturally produced umami peptides should be characterized in their
taste confirmation, as changes in the spatial structure might influence the
interactions of peptides with umami receptors; (3) the clarification of the interactions
of umami peptides themselves or with other taste ingredients; and (4) the ability to
increase the purity of the synthesized peptides, as the maximum purity of
synthesized peptides currently reported was >98%.

Conclusions

There are 52 peptides that have been reported to show umami taste, but 20 of them
have been questioned. New umami peptides and peptides contributing to the umami
taste of hydrolysate have been continuously reported. Investigation of umami
receptors has suggested that umami peptides might show umami taste. Therefore,
more investigation should be conducted to prove the taste of umami peptides. In
particular, new methods should be adopted to produce the natural peptides and then
confirm their taste. This will be beneficial to finding new umami substances and
supplying sound materials for the investigation of umami peptide receptors. It will
also be useful to investigate the flavor interactions with taste receptor responses and
other physiological responses.

Acknowledgment
The authors thank the financial support provided by the National Natural Science
Foundation of China (No. 31501505).


 1. Previous article in issue: Diet Effects in Gut Microbiome and Obesity
2. Next article in issue: Antioxidant and Anti-inflammatory Activities of Methanol
Extracts of Tremella fuciformis and Its Major Phenolic Acids

Volume 79, Issue 4April 2014 Pages R452–R459

R: Concise Reviews in Food Science

XXXVIII. The Distribution of Fat in Dried Dairy Particles

Determines Flavor Release and Flavor Stability
C.W. Park and M.A. Drake

Abstract

Dried dairy ingredients are utilized in various food and beverage applications for their
nutritional, functional, and sensory properties. Dried dairy ingredients include milk
powders of varying fat content and heat treatment and buttermilk powder, along with
both milk and whey proteins of varying protein contents. The flavor of these
ingredients is the most important characteristic that determines consumer
acceptance of the ingredient applications. Lipid oxidation is the main mechanism for
off-flavor development in dried dairy ingredients. The effects of various unit
operations on the flavor of dried dairy ingredients have been investigated. Recent
research documented that increased surface free fat in spray dried WPC80 was
associated with increased lipid oxidation and off-flavors. Surface free fat in spray-
dried products is fat on the surface of the powder that is not emulsified. The most
common emulsifiers present in dried dairy ingredients are proteins and
phospholipids. Currently, only an association between surface free fat and lipid
oxidation has been presented. The link between surface free fat in dried dairy
ingredients and flavor and flavor stability has not been investigated. In this review,
some hypotheses for the role of surface free fat on the flavor of dried dairy
ingredients are presented along with proposed mechanisms.

Practical Application

Dried dairy ingredients are utilized in various food and beverage applications for their
nutritional, functional, and sensory properties. Lipid oxidation is the main mechanism
for off-flavor development in dried dairy ingredients, and the distribution of fat may
play a critical role in flavor and flavor stability. Some hypotheses for the role of
surface free fat on the flavor of dried dairy ingredients are presented along with
proposed mechanisms.

Introduction

Bovine milk contains 88% water along with the macronutrients fat, protein, and
carbohydrates. Due to the high water content and available macronutrients for
microbial growth the shelf life of milk is relatively short. Dehydration is used to extend
the shelf life of milk and milk-derived products. Milk and milk-derived products are
generally dehydrated in 2 ways, by spray drying or by roller drying. The resulting
product is a powder with very low moisture content. Spray drying is the most common
method used in the dairy industry due to the intense heat treatment involved in roller
drying. These dried dairy ingredients extend the shelf life and provide functional
benefits and convenience.

Dried dairy ingredients are generally classified by their physical and compositional
properties and can be divided into 2 general groups, milk powders and protein
powders. All milk powders are defined as having less than 5% moisture (USDEC
2005). Nonfat dry milk (NFDM) and skim milk powder (SMP) are very similar in that
both are produced from pasteurized skim milk and have less and 1.5% fat by weight
(USDEC 2005). SMP must have at least 34% protein by weight, which can be
regulated by addition of milk permeate, whereas NFDM does not have a legal
definition in regards to protein (USDEC 2005). The cumulative thermal treatments
for low-heat, medium-heat, and high-heat SMP and NFDM are 70 °C for 2 min, 70
to 78 °C for 20 min, and 88 °C for 30 min, respectively (USDEC 2005). Whole milk
powder (WMP) is produced from pasteurized whole milk and has between 26% and
40% fat by weight (USDEC 2005). Buttermilk powder is produced from buttermilk
during butter manufacture and contains greater than 4.5% milk fat (USDEC 2005).
In 2012, over 1 million metric tons of dry milk products were produced in the United
States with the majority as NFDM (USDA 2013).

Concentrated protein ingredients are available in the dairy industry due to advances
in membrane filtration as well as ion exchange chromatography. The membrane
filtration processes commonly used are microfiltration (MF), ultrafiltration (UF),
nanofiltration, and reverse osmosis. Milk protein can be concentrated by UF of skim
milk to remove lactose and minerals and produce milk protein concentrates (MPCs)
with protein concentrations of 40% to 90% of the total solids. Using a combination of
MF and UF, whey proteins can be concentrated to produce whey protein concentrate
(WPC) with 25% to 89% protein or whey protein isolate (WPI) with greater than 90%
protein of the total solids. Whey proteins that are removed prior to cheese making
are called serum proteins and can be further concentrated to make serum protein
concentrate (Nelson and Barbano 2005; Evans and others 2009, 2010). Production
of WPC and WPI in the United States reached 230000 metric tons in 2012 while
production of dry MPC reached 46000 metric tons (USDA 2013). Other dried dairy
protein ingredients include caseins (rennet or acid) and caseinates.
Dried dairy ingredients are used in numerous applications due to their nutritional and
functional properties (Kenny and others 2000; Foegeding and others 2002; Anema
and others 2006; Davis and Foegeding 2007; Raikos 2010) but the most important
factor in consumer acceptance of dried dairy ingredient applications is flavor (Caudle
and others 2005; Drake 2006; Childs and others 2007). In order to characterize the
flavor of dried dairy ingredients, flavor lexicons have been developed (Drake and
others 2003; Carunchia Whetstine and others 2005; Drake and others 2009). Off-
flavors resulting from lipid oxidation in dried dairy ingredients result in decreased
consumer acceptance of dried dairy ingredient applications (Caudle and others
2005; Lloyd and others 2009b; Evans and others 2010). Raw milk quality has a
substantial impact on the off-flavors in milk powders and is affected by animal feed,
season, or microbiological quality (Celestino and others 1997; Coulon and Priolo
2002; Croissant and others 2007). Stapelfeldt and others (1997) reported that both
water activity and storage temperature were important in reducing off-flavors in
WMP, and Lloyd and others (2009a, 2009b) confirmed that temperature and oxygen
levels were crucial to minimize lipid oxidation. Lipid oxidation is also a primary
contributor to loss of shelf life in SMP although shelf stability is substantially longer
than WMP (Drake and others 2006). Sources of lipid oxidation off-flavors in dried
protein ingredients have been attributed to various unit operations such as starter
culture, storage, bleaching, agglomeration, and instantization (Croissant and others
2009; Wright and others 2009; Campbell and others 2011a, 2011b; White and others
2013).

The physical properties of dairy powders impact various functional characteristics

and may also impact the sensory properties. Because lipid oxidation is the source of
many off-flavors in dried dairy ingredients, an understanding of the effect that
distribution and physical characteristics of lipids in dried dairy ingredients have on
flavor and flavor stability is of great importance to the dairy industry. The goal of this
manuscript is to investigate the influence of the distribution of fat on the flavor of
dried dairy ingredients. A link between fat distribution in dried dairy ingredients and
flavor and flavor stability has yet to be investigated.

Flavor Deterioration in Dairy Products

It is generally recognized that the main reactions that deteriorate the flavor of dried
dairy ingredients are lipid oxidation and Maillard browning (Karagul-Yuceer and
others 2001; Farkye 2006). Generally, lipids alone do not contribute to the flavor of
foods due to their low volatility but products formed during the decomposition of lipids
can impact the flavor significantly (McClements and Decker 2008). Hydrolytic
rancidity and autoxidation are the 2 main types of decomposition of lipids. Hydrolytic
rancidity refers to the liberation of free fatty acids from the glycerol backbone
whereas autoxidation involves a complex sequence of chemical changes due to the
interaction of unsaturated lipids with oxygen (Frankel 1998a; McClements and
Decker 2008). Hydrolytic rancidity in milk is mostly attributed to endogenous
lipoprotein lipase enzymes (Deeth 2006). Lipid autoxidation decomposes fatty acids
into volatile compounds that are generally aldehydes, ketones, carbonyls, alcohols,
and acids (Frankel 1998b) and these are the primary source of off-flavors in dried
dairy ingredients.

Lipid oxidation in milk powders has been studied extensively. Lloyd and others
(2009a) demonstrated that common off-flavors in WMP produced in the United
States were grassy and painty. These off-flavors increased with storage time and
were correlated to an increase in various lipid oxidation products. Aldehydes and
ketones are among the main volatile compounds responsible for off-flavors in WMP
and SMP (Shiratsuchi and others 1994; Karagul-Yuceer and others 2001; Karagul-
Yuceer and others 2002; Carunchia Whetstine and others 2007; Lloyd and others
2009a, 2009b). Lipid oxidation and flavor of both SMP and WMP can be influenced
by many factors including light exposure, anti-oxidant addition, preheating treatment,
storage temperature, nitrogen flushing, moisture content, and relative humidity (Hall
and Lignert 1984; McCluskey and others 1997; Stapelfeldt and others 1997; Hardas
and others 2002; Lloyd and others 2009b). The concentration of unsaturated fatty
acids plays a role in the oxidation stability of milk powders (Romeu-Nadal and others
2007). The physical distribution of fat in the powders also affects lipid oxidation in
milk powders and will be discussed in subsequent sections of this manuscript.

Lipid oxidation is the primary contributor to a decrease in shelf life and an increase
in off-flavors in WPC34, WPC80, and WPI (Carunchia Whetstine and others 2005;
Wright and others 2009; Evans and others 2009, 2010). In sweet whey powder, a
combination of lipid oxidation and Maillard reactions contribute to off-flavors
(Mahajan and others 2004; Sithole and others 2005). Off-flavors in dried whey
proteins are often associated with different processing steps. The use of starter
culture increases lipid oxidation in liquid whey due to their hydrolytic enzymatic
activity, which can then carry through into WPC or WPI (Campbell and others
2011a). Mesophilic starter cultures impact the oxidative stability of WPC more than
thermophilic starter cultures (Liaw and others 2011). Because the orange colorant
annatto used in Cheddar cheese manufacture is also found in liquid Cheddar whey,
it must be bleached to obtain a colorless powder. Bleaching of Cheddar whey
increases off-flavors and lipid oxidation (Croissant and others 2009; Listiyani and
others 2011; Jervis and others 2012; Kang and others 2012). Other unit operations
that increase lipid oxidation include storage of liquid whey or retentate,
agglomeration, and instantization (Wright and others 2009; Whitson and others
2011; Campbell and others 2011b; White and others 2013). Because lipid oxidation
is responsible for off-flavors and loss of shelf life in both milk powders and dried
protein powders, increased lipid oxidation due to increased surface free fat in the
powders could be detrimental to the flavor and flavor stability.

Flavor Binding

In order for flavor perception to occur, flavor compounds must be volatile in the food
system as well as in the mouth. Interactions between flavor compounds and
constituents in dried dairy ingredients can impact flavor release. Thus, flavor quality
of dried dairy ingredients could be improved if volatile compounds responsible for
off-flavors were less volatile, therefore increasing the sensory detection threshold of
the volatile compounds. Milk proteins that contain nonpolar amino acids are of
interest in flavor binding due to the nonpolar nature of flavor compounds. The
measurement of flavor binding of proteins is generally done either by headspace
analysis or by equilibrium dialysis (O'Neill 1996).

Roberts and Pollien (2000) investigated the influence of milk components on the
volatility of different flavor compounds. The main factor in flavor retention in milk was
the milk fat content with milk fat concentrations up to 1.5% in their experimental
design. This was due to the lipophilic nature of many flavor compounds. The volatility
of some compounds (diacetyl, 2,3-pentanedione, guiacol) was not affected by the
concentrations of the different milk components and some (3-methyl butanal, 2-
methylpropanal, 4-ethylguiacol) decreased with decreased milk fat content. Volatility
of other compounds (β-damascenone and 1-octen-3-one) decreased with
decreasing milk solids-not-fat suggesting that there was also binding with protein. In
another study, the effects of lipid type and solid fat content on volatile compound
release were investigated. Lipid type did not have a significant effect but an increase
in solid fat content increased the volatile compound release in milk-based emulsions
(Roberts and others 2003). These studies demonstrate that fat in milk products
influences volatile compound release whether it is due to concentration or physical
state. Given the large impact that fat has on volatile compound release, it is probable
that the distribution and emulsification of fat in dried dairy powders has a strong
influence on the volatile compound release and overall flavor.

Hansen and Booker (1996) investigated the influence of casein and whey protein on
the binding of flavor compounds used in ice cream mixes. Their work demonstrated
that whey proteins reduced flavor compound intensities more than casein. It was
hypothesized that the high thermal treatment denatured more of the whey proteins
than caseins, which exposed more nonpolar regions to bind the flavor compounds
(Hansen and Booker 1996; O'Neill 1996). However, other studies have observed a
decrease in flavor binding by β-lactoglobulin and WPI upon heating above
denaturation temperatures (O'Neill and Kinsella 1988; McNeill and Schmidt 1993).
This has been suggested to be due to structural changes and aggregation that
occurred during the heat treatment (Kuhn and others 2006). Sodium caseinate also
decreased vanillin concentrations in dairy protein beverages (McNeill and Schmidt
1993; Li and others 2000) but WPC decreased vanillin flavor intensity more than
sodium caseinate (Hansen and Heinis 1991). Extensive research has been done on
the flavor binding properties of various milk proteins and has been reviewed by Kuhn
and others (2006). In WPI, the major protein responsible for binding flavor
compounds was β-lactoglobulin (Kuhn and others 2007). Collectively, these studies
provide strong evidence for the binding of flavors by milk-derived proteins. The
encapsulation of fat in dried dairy powders by proteins capable of binding flavors
could increase the sensory quality by reducing the volatility of the compounds
responsible for off-flavors.

Surface Free Fat

The term free fat is defined as fat that is no longer emulsified (Palanuwech and
others 2003). In dried dairy ingredients, proteins and phospholipids are the most
common emulsifiers. Free fat can be an indicator of damage to the milk fat globule
membrane (MFGM) (Kim and others 2002). A more complete definition of free fat in
dairy powders is fat that is not entirely coated by amphiphilic molecules or protected
by a matrix of carbohydrates and proteins during drying (Vignolles and others 2007).
When free fat is on the surface of the powder particles, it is referred to as surface
free fat. The surface free fat in milk powders can alter important properties of the
dried milk powder such as: oxidative stability, wettability, dispersability, solubility,
flowability, and ability to be used in chocolate-processing applications (Vignolles and
others 2007).

Free fat is most commonly extracted using an organic solvent such as hexane or
petroleum ether (Vignolles and others 2007). The use of polar solvents is avoided
because they can lead to the extraction of total fat (Buma 1971). During a free fat
extraction, a fixed amount of organic solvent is added to a fixed amount of powder
and swirled gently for a given amount of time. The solvent is then filtered and the fat
is measured gravimetrically after evaporation of the solvent. Because of this, free fat
can also be referred to extractable fat. Increasing extraction time and temperature
increased the amount of free fat that was extracted (Buma 1971; Kim and others
2002).

Because it can take more time to extract the free fat from the interior of the particle,
there is the ability to extract different fractions of free fat, whether from the surface
or the interior (Kim and others 2005b). Kim and others (2005b) were able to separate
3 different fractions of fat: surface free fat, inner free fat, and encapsulated fat. Only
minor changes in fatty acid composition were found in the fat extracted from different
fat fractions with the high melting saturated fatty acids being slightly more
represented in the surface free fat than the interior free fat or the encapsulated fat.
In both the surface free fat and the inner free fat, the oleic, linoleic, and linolenic acid
composition accounted for approximately 22% of the total fatty acid profile. These
unsaturated fatty acids are highly susceptible to lipid oxidation (Frankel 1998a).
These results demonstrate that the surface free fat is rich in unsaturated fatty acids
and suggest that surface free fat could be susceptible to lipid oxidation. Truyen and
Orsi (1977) observed greater concentrations of unsaturated fatty acids in surface
free fat in milk powders and increased concentrations of polar lipids in emulsified fat,
fat not solvent extractable. In the manuscript, it was not specified whether the free
fat was surface or inner free fat but it is assumed to be surface free fat due to the
extraction being very similar to the surface free fat procedure described by Kim and
others (2005b). The extraction time was 10 min whereas the extraction time for inner
free fat by Kim and others (2005b) was 48 h.

High levels of unsaturated fatty acids such as oleic, linoleic, and linolenic acid make
lipid oxidation a concern in milk and milk-derived products because they are among
the most common unsaturated fatty acids to undergo lipid oxidation (Frankel 1998a).
Surface free fat may be more susceptible to oxidation than emulsified fat because
emulsified fat is encapsulated with proteins and phospholipids, which have anti-
oxidant properties and will be discussed further.

Surface free fat in dairy powders can be highly influenced by processing and storage.
In general, parameters that can be manipulated include inlet air temperature, outlet
air temperature, feed solids concentration, and atomizing conditions. Elevated inlet
air temperatures increase the particle size because the crust on the particle surface
is formed more quickly, leaving less time for the particle to shrink (Birchal and others
2005; Nijdam and Langrish 2006). Larger particles can encapsulate more fat, thus
decreasing the surface free fat content (Buma 1971; Beristain and others 2001).
Increased inlet air temperatures decreased surface free fat in WMP whereas
increasing outlet air temperatures increased free fat (De Vilder and others 1976;
Kelly and others 2002). Increased inlet temperatures and increased feed solids
concentration increased the surface free fat in spray-dried WPC80 (Park and others
2014). Lactose crystallization increased general free fat by damaging the MFGM and
proteins that encapsulate the fat droplets (Aguilar and Ziegler 1994).

Surface Composition of Dairy Powders

The way that dairy powders are produced can greatly influence the composition on
the surface of the dried particles. During spray drying, the concentrate feed is
sprayed into small droplets, which are mixed with hot air to evaporate the water,
leaving a dry particle with a low moisture content (<0.5%). The drying of particles in
the spray dryer can be classified into 2 different periods. The 1st period is when the
bulk of the water is evaporated. During this period, the water can move freely to the
surface of the droplet and thus keeps the surface saturated with water. The
temperature of the droplet during this time is prevented by rising above the wet bulb
temperature due to the cooling that occurs when water evaporates (Fellows 2009).
The wet bulb temperature during drying is generally no greater than 60 °C (Schuck
2013). During the 2nd period, enough moisture has been removed from the droplet
that the surface is no longer saturated with water. At this point, a crust made of solid
particles forms and the amount of water that is evaporated decreases. Because the
water evaporation rate decreases, the temperature of the dried particles increases
(Birchal and others 2006; Kim and others 2009b).

The solids composition of dried dairy ingredients is primarily made up of fat, protein,
and carbohydrate (lactose). The distribution of these components on the surface of
dried powder particles can affect different functional properties. Kim and others
(2009a) demonstrated that the surface composition of SMP, WMP, and instantized
WMP was determined solely by the spray drying process and not by subsequent
fluidized bed drying. During spray drying, the fat, protein, and lactose reorient
themselves where the fat and protein migrate to the surface due to their
hydrophobicity and the lactose migrates to the center due to its hydrophilic nature.
This makes the surface composition of the powder different from the composition of
the entire powder including the interior, or bulk composition. In order to analyze the
surface composition of dairy powders, scanning electron microscopy and a
technique called electron spectroscopy for chemical analysis (ESCA) are utilized
(Kim and others 2009a). In ESCA, the milk powder is assumed to be made of protein,
fat, and lactose. By analyzing the elemental composition of the surface, mainly
carbon, oxygen, and nitrogen, the relative percentages of fat, protein, and lactose
on the surface can be calculated (Kim and others 2009a, 2009b). A more in depth
explanation of this technique has been described by Faldt and others (1993).

Kim and others (2002) investigated the bulk and surface compositions of
commercially produced SMP, WMP, cream powder, and WPC. In SMP, the bulk
composition of lactose, protein, and fat was 58%, 41%, and 1% respectively and the
surface composition was 36%, 46%, and 18%, respectively. In WPC, the bulk
composition of lactose, protein, and fat was 8%, 86%, and 6%, respectively, with the
surface composition much different, 6%, 41%, and 53%, respectively. In WMP and
cream powder, fat represented 98% and 99% of the surface composition,
respectively. Gaiani and others (2007) observed that a native caseinate powder with
0.4% lipid had a surface lipid content of 6%. It was hypothesized that the high spray
drying outlet temperature (90 °C) was above the melting temperature of the powder
lipids. As a consequence, the lipids were in the melted state and had increased
mobility throughout the particle (Kim and others 2005a, 2005b; Nijdam and Langrish
2006). The fact that fat is overrepresented at the surface of dairy powders can have
implications for different functional properties such as flowability, particle stickiness
and solubility (Kim and others 2005a; Nijdam and Langrish 2006). Free fat on the
surface of dairy powders could be more susceptible to lipid oxidation due to greater
access to oxygen. Thus, decreased surface free fat could reduce off-flavors and
increase flavor stability.

The presence of fat on the surface decreased the wettability of spray-dried

emulsions stabilized by both whey and milk proteins due to the hydrophobic nature
of the fat (Faldt and Bergenstahl 1996; Millqvist and others 2001). Increasing
concentrations of lactose on the particle surface increased the wettability. When
stored in a humid environment, the fat was redistributed to the surface at the
expense of lactose (Faldt and Bergenstahl 1995; Faldt and Bergenstahl 1996). This
observation was confirmed by Shrestha and others (2007) who observed in SMP
that fat and protein were more likely to migrate to the powder particle surface than
lactose. Kim and others (2005a) reported that surface fat inhibited the flowability of
dairy powders. SMP with low surface fat was observed to flow better than powders
with high surface fat coverage (WMP, cream powder, WPC). Higher surface fat has
been correlated to increased oxidation in dairy powders (Granelli and others 1996).
Lloyd and others (2009a, 2009b) did not observe a correlation with free fat and flavor
stability in WMP produced in the United States. The range of surface free fat in the
U.S. WMP was 1.1% to 7.7% and international WMP ranged from 2.8% to 6.7%. A
possible reason for the lack of correlation was that the WMP were made at 4 different
manufacturing facilities, confounding the effect that surface free fat alone would have
on flavor stability in WMP.

Kim and others (2009a) reported that spray drying was the most important
manufacturing process in determining the surface composition of spray-dried milk
powders. Fluidized bed drying had no significant effect on the surface composition
of the milk powders. As the particle dries, the Peclet number and the initial saturation
of the concentrate to be dried influence the particle formation (Vehring and others
2007). The Peclet number is defined as the ratio between the diffusion coefficient of
the solute and the evaporation rate. As the particle dries, the shape, size, and
surface composition are determined by the ability of the components to reposition
themselves due to the droplet viscosity or the presence of precipitates (Vehring and
others 2007). This was observed by Nijdam and Langrish (2006) with the drying of
milk powder. Increasing fat content in milk powders increased the surface fat
coverage with the most dramatic increase seen in powders ranging from 0% to 5%
fat. Spray drying at increased inlet temperatures favored the accumulation of lactose
on the surface rather than protein. The theory proposed was that higher
temperatures lead to accelerated formation of the surface crust, leaving less time for
larger molecules such as proteins to reach the surface. The increased viscosity of
the droplets would also reduce the amount of fat able to migrate to the surface. An
increase in concentrate viscosity during spray drying was observed to decrease
solubility of SMP (Baldwin and others 1980). Kim and others (2002) observed that
of the milk components, fat migrated to the surface more than lactose or protein.
Their results also showed a dramatic increase in surface free fat in dairy powders
with less than 6% fat, a WPC with 6% fat bulk composition had 53% of the particle
surface covered in free fat. These results along with those of Nijdam and Langrish
(2006) suggest that the bulk fat content of the powder has a significant effect on
surface free fat when the bulk fat content is low. As the percentage of bulk fat in the
powder increases, its effect on surface free fat diminishes significantly.

Particle size distribution of dairy powders is also of importance. Dairy powders are
often agglomerated to increase the average particle size and porosity. Larger sized
particles are more soluble because they are more porous and thus allow for an
increase in wetting ability. Nijdam and Langrish (2006) observed that regardless of
the fat content, milk powders spray dried at elevated inlet temperatures resulted in
an increased average particle size. Increased fat content in the milk powders
decreased the particle size and increased surface free fat. Elevated inlet
temperatures also increased the particle size of spray-dried WPC80 but increased
feed solids concentration increased the particle size to a greater extent (Park and
others 2014). In the spray-dried WPC80, a decrease in surface free fat was observed
in WPC80 that had a larger particle size. Fitzpatrick and others (2004) documented
the particle sizes of 26% fat milk powders with varying free fat contents. Although
not a major focus of the study, it is of interest to note that for the powders with the
lowest free fat content, increased particle size reduced the free fat content. These
results suggest that particle size also plays a large role in the surface free fat content
in dried dairy ingredients. As the particle size decreases, the surface area per unit
mass increases, leaving more particle surface to be covered by free fat and less to
be encapsulated (Buma 1971).

Free Fat and Lipid Oxidation

The 2 main classes of compounds responsible for encapsulating fat in dried dairy
ingredients are phospholipids and proteins. Due to their close proximity to fat during
emulsification, both are of interest in regards to their ability to promote or inhibit lipid
oxidation. Also, due to their hydrophilic and hydrophobic properties, they may be
able to slow the migration of fat to the surface by interacting with both the fat and the
hydrophilic lactose interior. Phospholipids are amphiphilic and contain 2 hydrophobic
acyl chains and a hydrophilic portion (Rombaut and others 2006). Phospholipids
contain 2 fatty acids esterified on the glycerol backbone at the sn-1 and sn-2
positions with a phosphoric acid on the sn-3 position through a phosphate ester bond
(Rombaut and Dewettinck 2006).

Phospholipids account for about 1% of the total bovine milk lipids and about 60% of
these come from the MFGM (Gallier and others 2010). The phospholipid content of
various dairy products derived from milk are shown in Table 1. During processing
steps (heating, agitation, homogenization, and aeration) the MFGM is ruptured and
the phospholipids enter into the aqueous phase (Rombaut and Dewettink 2006). This
is why phospholipids can be found in products that are not fat rich. In many cases,
the proportion of phospholipids to total fat is higher in products with little fat.

Table 1. Polar lipid content of various dairy products

Dairy product g polar lipid/100 g total lipid

1. a

Adapted from Boyd and others (1999).

2. b

Adapted from Rombaut and Dewettink (2006). Whole milk was calculated
from 100 g of product using 4.0% total fat content.

3. c

Adapted from Vaghela and Kilara (1995).

4. d

Adapted from Gaiani and others (2007).

5. e

Adapted from Morr and Foegeding (1990).

Whey powdera 21.1 to 41.7

Creamb 0.35 to 0.86

Butterb 0.20 to 0.27

Dairy product g polar lipid/100 g total lipid

Whole milkb 0.36

Skim milkb 19.06

Buttermilkb 21.7 to 33.1

Cheddar cheeseb 0.47

Cheddar wheyb 5.32

Cottage cheeseb 5.30

WPC34c 17.53

WPC75c 23.6

Native phosphocaseinated 67.7

WPCe 10.8 to 45

It is unclear whether phospholipids native to dried dairy products are pro- or anti-
oxidants. It appears that their effect on lipid oxidation is product specific.
Phospholipids can negatively affect the flavor of dairy products through oxidation of
lipids due to their high levels of unsaturated fatty acids (Sattar and deMan 1975;
Sessa 1985). Phospholipids also act as pro-oxidants by lowering the surface tension
of the lipid, allowing oxygen from the headspace to diffuse to the oil and thus,
increase lipid oxidation (Choe and Min 2006). In WPC, phospholipids are
concentrated along with the protein. In WPC75, phospholipids were as much as 23%
of the total lipid content (Vaghela and Kilara 1995). This renders proteins as an
important part in the encapsulation of fat in dried dairy ingredients. The concentrated
phospholipids associated with the MFGM increase the potential for lipid oxidation to
occur in dried whey ingredients. Wright and others (2009) observed that WPI that
was instantized with lecithin, a phospholipid, had decreased shelf life due to
increases in lipid oxidation products and off-flavors.

Phospholipids have also been reported to have anti-oxidant activity, which can
impact the flavor of foods (Chen and Nawar 1991). The mechanism for the anti-
oxidative effects of phospholipids is still not clear. The phospholipids with polar
groups containing nitrogen are effective anti-oxidants under most conditions (Choe
and Min 2006). Phospholipids also chelate metals, which decrease lipid oxidation. If
the concentration of the phospholipids is too high then the phospholipids act as pro-
oxidants. Yoon and Min (1987) reported that phospholipids were only anti-oxidants
when Fe2+ was present and chelated. Sources of Fe2+ in dried dairy ingredients
include metalloproteins such as lactoferrin, serum transferrin, and ovotransferrin and
are found in many dried dairy ingredients (Jervis and Drake 2013). It is possible that
in dried dairy ingredients with increased fat emulsified by native phospholipids that
decreased off-flavors would result due to anti-oxidant properties of the native
phospholipids. Phospholipids also could reduce lipid oxidation by encapsulating the
fat and preventing it from migrating to the surface due to their hydrophilic properties
and the hydrophilic nature of the lactose rich interior of the spray-dried particles.

Milk proteins are the other main class of compounds involved in the encapsulation
of fat in dried dairy ingredients. Anti-oxidant properties of native milk proteins that
encapsulate and reduce surface free fat may also in turn reduce lipid oxidation and
off-flavors in dried dairy products. The ability to unfold at the oil/water interface and
expose the hydrophilic and hydrophobic amino acids allows proteins to encapsulate
fat globules. Casein has been demonstrated to have greater anti-oxidant effects than
whey proteins (Allen and Wrieden 1982; Hu and others 2003). It was hypothesized
that differences in anti-oxidant properties among the proteins was due to interfacial
thickness, chelating properties, and free-radical scavenging amino acids. The anti-
oxidant activity can come from the electrical charge proteins impart on the fat
droplets, repelling pro-oxidant metals away from fat droplets (Donnelly and others
1998). Others have stated that the effectiveness of casein to prevent lipid oxidation
is due to the ability to bind copper and other pro-oxidant metals and to unfold and
surround the fat globule membrane (Frankel 1998b). β-lactoglobulin was observed
to be a mild anti-oxidant and a loss in anti-oxidant activity was attributed to structural
changes during heating (Liu and others 2007). The anti-oxidant or pro-oxidant
properties of proteins are highly dependent on the food system and unexpected
activity can arise from different interactions between food components.

Milk proteins may also decrease lipid oxidation in dried dairy ingredients by
encapsulating milk fat and preventing it to reach the surface of the powders. Kim and
others (2002) observed that fat that was encapsulated by protein was preferentially
located beneath the layer of surface free fat. They also demonstrated that oxygen
uptake for powders with higher levels of surface free fat was greater than in powders
with lower levels of surface free fat. Because oxygen is key to lipid oxidation and an
increase in oxygen content correlates to greater lipid oxidation, this demonstrated
that powders with higher surface free fat were more susceptible to lipid oxidation and
potentially off-flavor development. De Vilder and others (1976) observed a positive
correlation between surface free fat and particle porosity. Nitrogen was able to
penetrate the milk powder particles with increased surface free fat. This suggests
that decreased surface free fat could also decrease lipid oxidation by limiting oxygen
exposure to the interior of the powder particles. Hardas and others (2000) observed
increased oxidation in surface free fat compared to encapsulated under the surface
in emulsions made with milk fat. In their study, surface free fat had a greater increase
in peroxide value and hexanal over time and a greater decrease in linoleic and
linolenic acid contents compared to the encapsulated fat. These studies suggest that
efforts should be made to reduce the amount of free fat on the surface of dried dairy
ingredients to improve flavor and flavor stability.

Implications for Flavor and Flavor Stability

The protein-binding properties, surface composition, and surface free fat are all
important characteristics that influence the flavor of dried dairy ingredients. Because
spray drying and storage of dried dairy ingredients influence these properties, an
emphasis should be placed on parameters during these unit operations to improve
the flavor and flavor stability. A reduction in surface free fat could reduce off-flavors
in dried dairy ingredients due to the properties of the emulsifying native proteins and
phospholipids. Emulsifying proteins in dried dairy ingredients can bind more off-
flavors, decrease oxygen permeability, and decrease lipid oxidation and native
phospholipids potentially have anti-oxidant effects. Park and others (2014) observed
this effect directly in WPC80 where decreased surface free fat corresponded to
decreased off-flavor intensity and associated lipid oxidation compounds. Lower
levels of surface free fat were observed in WPC80 spray dried at increased feed
solids concentration. Higher feed solids concentration during spray drying in WPI
was demonstrated to increase whey protein denaturation (Anandharamakrishnan
and others 2007). The denatured whey proteins would have a greater ability to
encapsulate the fat due to their exposed hydrophobic regions. This is of importance
because preheat treatment is an important unit operation in the manufacture of dried
dairy ingredients. While it is possible that preheat treatment could in theory reduce
surface free fat due to greater milk fat encapsulation, its effect on flavor would be
more difficult to predict since many thermally induced flavor compounds would result
and confound any benefits of reduced surface free fat. Keogh and O'Kennedy (1999)
observed higher levels of fat oxidation on the surface of spray-dried whey
protein/milk fat emulsions. A summary of research related to the influence of process
parameters on characteristics of dairy powders related to flavor and surface free fat
is shown in Table 2.

Table 2. The role of various process parameters on surface free fat and/or flavor of
dairy powders

Process
parameters

Powder investigated Effect Author

1. WMP, whole milk powder; WPC, whey protein concentrate; WPI, whey
protein isolate; SPC, serum protein concentrate.

Increased nozzle size and

Spray dryer nozzle
outlet air temperature
size, outlet air
increased surface free fat
WMP temperature, feed Kelly and others (2002)
and increased feed solids
solids
concentration decreased
concentration
surface free fat
 Process
parameters

Powder investigated Effect Author

The use of a 2-stage

drying process involving
One-stage and 2- spray drying and fluidized
WMP stage drying and bed drying decreased De Vilder (1980)
homogenization surface free fat.
Homogenization also
decreased surface free fat

Increasing lactose
Lactose crystallization with high
crystallization by shear and elevated
high shear and temperature increased the
WMP Koc and others (2003)
elevated amount of free fat in WMP
temperature in a to almost 80% compared
mixer to low shear and
decreased temperature.

WMP that was classified

as low-heat WMP was
consistently higher in lipid
Preheat treatment Stapelfeldt and others
WMP oxidation throughout
prior to spray drying (1997)
extended storage than
medium or high-heat
WMP.

Increased inlet
temperature and feed
Spray dryer inlet
solids concentration
temperature, feed
WPC80 decreased off-flavors Park and others (2014)
solids
along with surface free fat
concentration
while increasing particle
size.

The heat used during the

WPC34 spray drying process had
Freeze drying and little effect on the flavor of Evans and others
and
spray drying WPC34 and SPC34 (2009)
SPC34
because no consistent
differences in flavor were
 Process
parameters

Powder investigated Effect Author

observed between freeze

drying and spray drying.

Increasing the lactose:

WPC concentration
Homogenization reduced free fat but not
WPC/WPI conditions and surface fat. The higher Keogh and O'Kennedy
Emulsion composition of the levels of fat on the surface (1999)
feed of powder particles
increased the level of
oxidation during storage.

Increased outlet air

Spray dryer outlet
temperature and feed
air temperature, Anandharamakrishnan
WPI solids concentration
feed solids and others (2007)
increased the amount of
concentration
whey protein denaturation.

Future Work

Future experiments should be conducted to investigate the effect of decreased free

fat content on the flavor and oxidative stability of various dried dairy ingredients, both
milk powders and protein ingredients of differing fat contents. Although surface free
fat has been studied extensively in WMP, a link between surface free fat, flavor, and
flavor stability has only been recently proposed in WPC80. In particular, spray drying
parameters and processing steps prior to spray drying should be optimized for the
sensory properties of the dried powders. These parameters should include
homogenization pressures, concentrate solids concentration, inlet and outlet air
temperatures, and atomization conditions. The effect of surface free fat in various
dried dairy ingredients on flavor stability over time would be very useful to the dairy
industry as a whole.

Conclusion

Dried dairy ingredients continue to be a focus for the dairy industry due to their
reduction in shipping costs, versatility, and extended shelf life. Because flavor is the
limiting factor in consumer liking of dried dairy ingredient applications, improved
flavor and flavor stability by altering powder characteristics is of great importance.
Advances in spray drying technology and research have resulted in discoveries
regarding the influence that spray-drying parameters have on protein binding,
particle surface composition, and free fat. Understanding how these powder
characteristics relate to the flavor and flavor stability of dried dairy ingredients will
help the dairy industry produce products with improved flavor and increase their
implementation in the food industry.

Acknowledgments

Funding provided in part by the Dairy Research Inst. (Rosemont, Ill., U.S.A.). The
use of trade names does not imply endorsement nor lack of endorsement by those
not mentioned.

XXXIX. Additions of Glutathione or Specific Glutathione-

rich Dry Inactivated Yeast Preparation (DYP) to Sauvignon
blanc Must: Effect on Wine Chemical and Sensory
Composition
M. Gabrielli1, J.L. Aleixandre-Tudo2, P.A. Kilmartin3, N. Sieczkowski4, W.J. du
Toit2*
(1) Department of Food, Environmental and Nutritional Sciences, Università degli
Studi di Milano, Via G. Celoria 2, 20133 Milano, Italy
(2) Department of Viticulture and Oenology, Stellenbosch University, Private Bag
X1, Matieland, 7602, South Africa
(3) Wine Science Programme, School of Chemical Sciences, The University of
Auckland, Private Bag 92019, Auckland, New Zealand
(4) Lallemand SAS, R&D Oenology Division, 31700 Blagnac, France
Submitted for publication: June 2016
Accepted for publication: October 2016
Key words: Sauvignon blanc, glutathione, inactivated yeast, aroma, sensory analysis

Although the addition of pure glutathione (GSH) is not allowed under current
regulations, the concentration of this compound can be increased in wine
through the addition of glutathione-enriched dry yeast preparations (DYP).
These preparations have been observed to have antioxidant properties and
could thus influence wine aroma and sensory characteristics. The main aim of
this study was to investigate the effect of DYP and GSH juice additions on the
sensory and chemical composition of Sauvignon blanc wine. Four juice
additions were performed and compared against a control treatment: 5.5 mg/L
of GSH; 0.4 g/L of DYP; 80 mg/L of GSH; 0.4 g/L of DYP plus 80 mg/L of GSH.
After three months of bottling, the volatile and sensorial composition was
investigated. The addition of DYP preparations to must increased the
concentration of certain wine volatile compounds, with increased attributes of
riper tropical fruit aromas, which was not always observed with the GSH
additions. The addition of DYP influenced the concentrations of some volatile
compounds, which modified the white wine aroma. The release of compounds
other than GSH by the yeast products is proposed as the reason for these
changes. The results observed in this study can assist winemakers to modify
the aroma profile of Sauvignon blanc wines.
INTRODUCTION
The tripeptide L-glutamyl-L-cysteinyl-glycine, or glutathione (GSH), is the major low-
molecular-mass thiol compound in plants and animals (Kritzinger et al., 2013a). GSH
plays an important role in the biosynthesis of many aroma precursors present in the
berries of several cultivars. An example of this is given by the varietal’s thiol
compounds that contribute to typical Sauvignon blanc aromas of flower, boxwood,
blackcurrant (mercapto-4-methylpentan-2- one, 4MMP), citrus, guava and passion
fruit (3-mercapto-hexan-1-ol, 3MH, 3-mercaptohexyl acetate, 3MHA), which
originate from the precursors [S-3-(hexan-1-ol)-L-cysteine, S-4-(4-methylpentan-2-
one)-L-cysteine, [S-3-(hexan-1-ol)- glutathione and S-4-(4-methylpentan-2-one)-
glutathione (Coetzee & Du Toit, 2012). GSH conjugates are thus found in the juice
and can be a source of the aroma compounds in the wines (Fedrizzi et al., 2009;
Roland et al., 2010). The concentration of GSH in grapes and juice depends on the
cultivar and climatic conditions. Lavigne-Cruege and Dubourdieu (2002) reported
GSH levels in different white cultivars to be between 17 and 114 mg/kg (56 to 372
μmol/ kg). The GSH content in grape juice is influenced by factors such as exposure
to oxygen, tyrosinase activity and grape skin maceration during pre-fermentation
(Cheynier et al., 1989; Du Toit et al., 2007), with reported ranges varying from 10 to
100 mg/L (Cheynier et al., 1989). Additionally, several authors have reported
conflicting results on the trend of GSH during alcoholic fermentation (Du Toit et al.,
2007; Andujar-Ortiz et al., 2012; Kritzinger et al., 2013b). The concentration of GSH
in wine is lower than in the juice and grapes, and values from 1 to 20 mg/L typically
are found (Du Toit et al., 2007). Saccharomyces cerevisiae can also affect GSH
content in wine during alcoholic fermentation and lees ageing (Kritzinger et al.,
2013b), with a maximum reported increase of around 20 mg/L for a particular yeast
strain (Lavigne-Cruege et al., 2007).
The biological importance of GSH is due to the nucleophilic and reducing properties
of the thiol group in the cysteine residue (Elskens et al., 1991). GSH is involved in
many vital functions of S. cerevisiae, such as maintaining the integrity of
mitochondrial membranes, responding to oxidative stress, xenobiotics and
endogenous toxic metabolite detoxification, and responding to deficiencies in
sulphur and nitrogen (Penninckx, 2002; Tominaga & Dubourdieu, 2003). Moreover,
GSH plays an important role in the oxidative phenomena of white musts and wines.
Enzymatic oxidation occurs in musts and is correlated with the amounts of
hydroxycinnamic acid esters (caftaric acid and coutaric acid) and flavan-3-ols (Li et
al., 2008). Phenolic compounds, particularly those with an ortho-diphenol group, are
mainly responsible for oxidative browning (Betés- Saura et al. 1996). The reaction
involves the oxidation of catechols to form unstable quinones, which are responsible
for the wine browning (Oliveira et al., 2011). GSH limits this phenomenon through
the trapping of the unstable quinone molecule (Cheynier et al., 1990; Rigaud et al.,
1991). GSH has also been proposed as an alternative antioxidant to SO2 that
potentially could lead to a decrease in the required sulphur levels in finished wines
(Makhotkina et al., 2014).
In addition, exposure to oxygen during white wine ageing can profoundly change
the colour and aromatic profile of a wine (Simpson, 1978). In a young wine, the
oxidative degradation of volatile compounds leads to a loss in fruity and floral
aromas. Moreover, it has been reported that, in cases where the concentration of
GSH in white wine exceeds 6 to 10 mg/L, both colour and aroma were better
preserved during ageing and storage (Lavigne-Cruege & Dubourdieu, 2004). A
number of authors have reported a positive effect of GSH against the oxidation of
varietal thiols during the ageing and storage of white wine (Blanchard et al., 2004;
Dubourdieu & Lavigne-Cruege, 2004; Ugliano et al., 2011, Coetzee & Du Toit 2012;
Coetzee et al., 2013).
Despite the observed effects of GSH in wine, the addition of GSH prior to bottling
was recently approved by the OIV, as long as doses are never higher than 20 mg/L
(OIV, 2015). In addition, GSH can be added to the must or wine through the addition
of other permitted additives (DYP), which allow the amount of GSH to be increased
(Pozo-Bayón et al., 2009; OIV, 2013). DYPs are manufactured from the thermal
inactivation of S. cerevisiae cultivated under specific conditions in which the intra-
cellular accumulation of GSH is stimulated (Kritzinger et al., 2013a). In the work of
Pozo- Bayón et al., (2009), four classification categories for DYPs were proposed:
inactive yeast, yeast autolysates, yeast hulls or walls, and yeast extracts. The
products referred to as GSH-rich inactivated dry yeast preparations (DYPs) are
added during alcoholic fermentation and are expected to increase the wine GSH
content, either by the liberation of GSH into the wine, or by allowing the yeast to
assimilate GSH precursors during alcoholic fermentation (Kritzinger et al., 2012).
The addition of DYPs has been recommended by the producers in order to prevent
aroma and colour losses and to increase the availability of fermentative nutrients in
wines (Pozo-Bayón et al., 2009). Andujar-Ortíz et al. (2012) and Kritzinger et al.
(2012) reported an increase in the concentration of GSH after the addition of DYPs
to Grenache and Sauvignon blanc juice. Moreover, Andujar- Ortíz et al. (2014)
reported an increase in the intensity of typical fruity attributes (banana, strawberry)
in young Grenache rosé wines made after the addition of DYPs. These authors
suggest that the GSH released from DYP could have caused a protective effect
against the oxidation of wine aroma compounds. Other authors have also observed
the protection of white and rosé wines against oxidation after the addition of a
specific GSH-enriched inactivated DYP during alcoholic fermentation. Better
protection of the chromatic and organoleptic properties, as well as a higher presence
of some of the particular volatile thiols, was observed (Aguera et al., 2012).
The evolution of the GSH content in GSH- or DYP-supplemented must during the
fermentation process currently is poorly investigated. It is also not well known
whether changes brought about in the sensorial and volatile composition of wines
treated with DYPs are due to the GSH contained in these products or to other
constituents being released into the must or wine. The aim of this work was thus to
study the levels of extracellular GSH during alcoholic fermentation after the addition
of different dosages of pure GSH, or by adding varying concentrations of a specific
glutathione-rich inactivated dry yeast preparation (DYP). Moreover, the study also
reports and discusses the influence of GSH and DYP additions on the volatile
composition and sensory characteristics of Sauvignon blanc white wines.
MATERIALS AND METHODS
Chemicals and reagents
The organic yeast rehydration nutrient (GO-FERM PROTECT EVOLUTION™) and
the specific GSH-rich inactivated yeast (OPTI-MUM WHITE EVOLUTION™ (DYP))
were obtained from Lallemand Blagnac, France. N2 gas and CO2 gas were
purchased from Afrox, Cape Town, South Africa. Ascorbic acid, sulphur dioxide as
K2S2O5, methanol, reduced GSH, formic acid, acetonitrile, tris-EDTA, meta-
phosphoric acid, ethanol, sodium sulphate anhydrous, p-hydroxymercurybenzoate,
butylated hydroxyanisole, dichloromethane and diethyl ether were purchased from
Sigma Aldrich (St Louis, MI, USA). Standards for the quantification of esters,
alcohols, fatty acids and terpenes were also obtained from Sigma Aldrich. HPLC-
grade water was obtained from a Milli-Q system (Millipore Filter Corp., Bedford, MA).
Juice and winemaking treatments
Clear Sauvignon blanc juice (2013 vintage, 22.5°Brix, pH 3.3 and total acidity 5.5
g/L) was collected from Neil Ellis Wines, based in Stellenbosch, and divided into 10
L stainless steel canisters. Nitrogen in gas form was blown inside the canisters to
replace oxygen before juice transfer. The juice was divided into five treatments: the
control (C), and additions of 5.5 mg/L of GSH (GSH 5.5), 80 mg/L of GSH (GSH 80),
0.4 g/L of DYP (DYP), and 80 mg/L of GSH plus 0.4 g/L of DYP (DYP+GSH 80).
DYP was added at levels according to Lallemand’s recommendation (DYP
treatment). Finally, the maximum dose of GSH was combined with DYP addition at
0.4 g/L to investigate a possible synergistic effect. An aliquot of the juice was used
to ensure complete dissolution of the GSH or DYP additions into the juice matrix.
Treatments were supplemented just before yeast inoculation according to the
supplier’s recommendations (www.lallemand.com). Three fermentation replicates
per treatment were performed. The yeast strain Saccharomyces cerevisiae
LALVIN® QA23 (Lallemand) was rehydrated in GO-FERM PROTECT
EVOLUTION™ at 0.3 g/L (Lallemand) and inoculated into the juice at 0.3 g/L.
Fermentation took place under controlled temperature at 15°C. The progress of
fermentation was monitored by measuring weight loss. At the end of fermentation,
when no more weight loss had occurred for three consecutive days, 60 mg/L of
sulphur dioxide was added to all the treatments and the wines were stored at -4°C
for ten days to carry out tartrate stabilisation. The wine was then racked with N2 from
the yeast lees and bottled in green 750 mL screw-cap wine bottles under CO2 gas.
Standard winemaking analyses
Routine wine analyses were performed after alcoholic fermentation (residual
reducing sugar, total acidity (TA), alcohol, volatile acidity (VA), malic acid and pH)
using a Grapescan™ FT 120 instrument (Foss Electric, Denmark).
Sampling procedure
Samples destined for UPLC-MS/MS analyses were taken just after the juice was
collected in the 10 L stainless steel canisters. The sampling was carried out before
and after the addition of DYP and GSH, and at difference stages throughout the
fermentation, namely after 25%, 50%, 75% and 100% completion of alcohol
fermentation. Twenty mL of juice were transferred from the canisters into 20 mL
plastic vials. To completely inhibit residual phenol oxidase or laccase activity, SO2
and ascorbic acid is raised to 1 000 mg/L and 500 mg/L respectively. The plastic
vials were saturated with CO2 before and after the juice was transferred into the
vials. Additional CO2 was blown over the headspace, and the vials were sealed
hermetically. The vials were then frozen at -20°C until analysis. The analysis of
aroma compounds and sensory analysis were performed three months after bottling.
Preparation of glutathione samples
Samples (4 mL) were thawed and centrifuged (Centrifuge 5415 D, Eppendorf,
Hamburg, Germany) at 1 600 rpm for 5 min at 20°C. The supernatant was then
diluted five times with a solution containing 1 000 mg/L of SO2 and 500 mg/L of
ascorbic acid. Finally, the samples were filtered through a 0.45 μm syringe filter and
injected onto the UPLC-MS/MS instrument (Kritzinger et al. 2012).
Quantification of reduced GSH released from the specific glutathione-rich
inactivated yeast (DYP)
The GSH levels released from the DYP were evaluated in a model solution
consisting of 5 g/L of tartaric acid adjusted to pH 3.3 with 1 M NaOH. In order to
decrease oxygen in the model solution, N2 gas was bubbled for 30 min prior to DYP
addition, after which 0.4 g of DYP was added to 100 mL of model solution and stirred
for 10 min. A sample was then drawn and injected within the next half an hour. The
quantification was performed in triplicate.
Glutathione analysis
GSH concentrations were determined using an UPLC-MS/ MS method described by
Kritzinger et al. (2012). A Waters Acquity UPLC (Milford, MA, USA) connected to a
Waters Xevo triple-quadrupole mass spectrometer was used for GSH quantification.
The separation was achieved using a Waters Acquity BEH phenyl column (100 mm
x 2.1 mm x 1.7 μm). For solvents A and B, 0.4% trifluoroacetic acid and acetonitrile
were used respectively. Solvent gradient has been reported elsewhere (Kritzinger et
al., 2012). A cone voltage of 18 V in combination with a multiple reaction monitoring
transition of 308.1 > 179.1 at a collision energy of 17 eV was used as optimised
setting. The injection volume was 3 μL and a dilution of five times in water was used
to achieve the best accuracy.
Thiol analysis
Thiol compounds were analysed using the extraction method reported by Tominaga
et al. (1998), with some modifications as reported by Coetzee et al. (2013). Initially,
5 mL of 1 mM p-hydroxymercurybenzoate (p-HMB), and then 0.5 mL of 2 nM
butylated hydroxyanisole (BHA) solution, were added to 50 mL of wine. After stirring,
50 μL of a deuterated internal standard solution (supplied by Auckland University,
New Zealand) containing 22 mM of 3-mercapto-hexanol (3MH), 2.8 mM of 3-
mercapto-hexyl acetate (3MHA) and 2.5 nM of 4-methoxy-2-methyl-2-
mercaptobutane (4M2M2SB) used to quantify 4-mercapto-4-methylpentan-2-one
(4MMP) was incorporated into the wine. This was followed by the injection of 2 μL of
a concentrated sample onto an Agilent Gas Chromatograph 6890N coupled to an
Agilent 5973 mass-selective detector (Agilent, Santa Clara, CA, USA). A HP-
Innowax column (60 m x 252 μm x 0.25 μm) with helium as carrier gas was used for
compounds separation. Flow rates, oven temperatures and standard curves used
for thiol quantification have been reported elsewhere (Coetzee et al., 2013).
Ester, alcohol, fatty acid and terpene analyses
Ester, alcohol and fatty acid concentrations were determined using the GC-Flame
ionisation detector (FID) method described by Coetzee et al. (2013). The volatile
compounds were extracted from the wine (5 mL) using a liquid-liquid extraction by
sonicating the diethyl ether (1 mL) wine mixture for 5 min. 4-Methyl-2-pentanol (100
μL of 0.5 mg/l solution in a 12% ethanol–water mixture) was used as internal
standard. The mixture was then centrifuged at 4 000 RPM for 3 min before the ether
layer was removed and dried by adding Na2SO4. The extracts were injected in
duplicate in a Agilent 6890 Plus GC (Little Falls, Wilminghton, DE, USA) equipped
with a split/splitless injector and FID detector using a J&W DB-FFAP capillary GC
column (Agilent, Little Falls, Wilmington, DE, USA) with the following dimensions: 60
m x 0.32 mm x 0.5 μm. A 15:1 split ratio was used, with a split flow rate of 49.5
mL/min and a column flow rate of 3.3 mL/min using hydrogen as a carrier. Method
validation, quantification as well as separation conditions have been reported
elsewhere (Coetzee et al., 2013).
Monoterpene analysis was performed following a solid phase extraction in a Visiprep
SPE vacuum manifold 20-21 Effect of port model from Supelco (Bellefonte,
Pennsylvania, USA). A volume of 50 μl of 2,6-dimethyl-6-hepten-2-ol (25 mg/L in
ethanol) was used as internal standard and was added to 50 mL of wine.
Dichloromethane (4 mL), methanol (4 mL) and, finally, a 12% ethanol–water solution
(4 mL) were used to condition the cartridges (Strata SDB-L; Phenomenex, Torrance,
CA, USA). Vacuum suction was used to rinse the wine through the cartridge, and 4
mL of Milli-Q-Water were later used to clean the cartridges. Samples were then dried
under vacuum for 15 min. Finally, dichloromethane (2 mL) was used to elute the
terpenic compounds from the cartridge. Sodium sulphate crystals were added to
remove any traces of water. Each sample was injected in duplicate in splitless mode
on a Hewlett-Packard (Palo Alto, CA, USA) 5890 Series II gas chromatograph
equipped with a 60 m x 0.32 mm x 0.5 mm fused DB-FFAP capillary column (J&W
Scientific, Folsom, CA, USA), and FID. Separation conditions have been reported
elsewhere (Coetzee et al., 2013) and quantification was done by comparison with a
calibration curve of pure standards.
Sensory analysis
Wines corresponding to the different treatments were analysed by means of general
descriptive sensory analysis (DA). A trained panel consisted of two males and eight
females, with ages ranging from 22 to 65 years old (with an average age of 40.1).
Panellists were trained three times a week. Each training session lasted two hours,
with a 15-minute break. The first three training sessions were used to establish the
lexicon needed to describe the wines. During the first training session, the five wines
were presented and the panel was asked to generate sensory attributes to describe
them. A list of attributes was compiled and the corresponding reference standards
were prepared and presented in the following session. The list of attributes used to
describe the set of wines was reduced from 21 attributes to 13 by means of
discussion amongst the panellists during the following two training sessions (Table
1). The last four training sessions were used to establish consensus amongst the
panellists with regard to the attributes and the wines’ ratings on a 10 cm line scale
anchored at none on the left-hand side and intense on the right-hand side. Although
a 10 cm line scale was used, the results of the data analysis are expressed out of
100. The wines were evaluated by the panellists in triplicate on the same day.
Panellists were forced to take breaks longer than 10 minutes between replicates.
The sensory evaluation was conducted in off-white tasting booths located in a
sensory laboratory. The sensory laboratory was quiet, odour free and had controlled
air conditioning (20 ± 2°C). Twenty-five millilitres of each wine sample were
presented in black ISO tasting glasses covered with petri dishes coded with pseudo
random three-digit codes generated by an online research randomiser. In order to
limit biases, each panellist received the samples in a different order according to a
balanced Williams Latin square design, computed using the design of experiments
(DOE) function of the MX package of XLSTAT (Microsoft, Redmond, WA, USA).
Data was captured on paper ballots, line scales were measured with rulers and data
was captured using Microsoft Excel 2010 (Microsoft, Redmond, WA, USA).
Panel performance was assessed using the workflow suggested by Tomic et al.
(2010). Repeatability (p*MSE plots and F-plots), discriminability and panel
consensus (Tucker-1) were tested using PanelCheck 1.4 (Nofima Mat, Tromsø,
Norway). DA data was analysed by means of two-way ANOVA with panellists and
wine samples as main effects. Wine sample effect was tested by the regular F test
to identify significant (p < 0.05) attributes. Only significant attributes were used for
further statistical analysis. Principal component analysis (PCA) was conducted on
the correlation matrix of the average intensity scores to obtain a multivariate sensory
map of the DA data.
Data analysis
After quantification, descriptive statistical measurements, including mean and
standard deviation, were calculated using Microsoft Excel 2007 (Microsoft
Corporation, www.microsoft.com). ANOVA was performed with

TABLE 1
Sensory attributes and corresponding reference standards used during general
descriptive sensory analysis training.
Attribute Reference standard composition
Fresh pineapplea,c three 1 cm2 pieces of fresh pineapple
Ripe pineapplea,c three 1 cm2 pieces of canned pineapple (Koo)
Fresh tropicalb 30 mL of tropical fruit juice (Ceres), 5 mL fresh
homogenised pineapple supernatant
Canned tropicalb 30 mL tropical fruit juice (Ceres), 5 mL canned
guava syrup (Koo)
Green guavab one 2 cm2 piece of guava with a green skin
Ripe guavab one 2 cm2 piece of guava with a yellow/pink
skin
Grapefruita,c one 3 cm2 grapefruit peel, one 3 cm2
grapefruit flesh
Passion fruita,c one 2 cm2 piece of skin and four pips of fresh
passion fruit
Bananaa,c one 3 cm2 piece of fresh banana (without the
peel)
Stonefruitb one 3 cm2 piece of fresh yellow cling peach,
20 mL apricot juice (Liqui-Fruit)
Cooked vegetablesa 2 mL asparagus brine (Koo), 5 mL green bean
brine (Koo)

statistica 10 (Statsoft Inc., www.statsoft.com), followed by a post-hoc Fisher’s least

significant difference (LSD) analysis, to determine significant differences in the
chemical and sensory analyses (p < 0.05). Principal component analysis (PCA) was
performed with the same software for sensory data analysis.
RESULTS AND DISCUSSION
Wine fermentations followed similar performance in the assayed treatments, being
completed within 16 days, with residual sugar levels lower than 4 g/L. The standard
oenological parameters (alcohol, volatile acidity (mg/L), pH, total acidity (mg/L) and
residual sugar) showed non-significant differences between treatments (results not
shown).
GSH released from DYP
The release of reduced GSH from DYP was performed in a synthetic wine as
reported by Andujar-Ortiz et al. (2012). The amount of reduced GSH released from
0.4 g/L of GSH-rich inactivated yeast preparation into a model solution under our
conditions was found to be 5.48 ± 0.42 mg/L. Based on this result, 5.5 mg/L of GSH
was added to the juice in one of the studied treatments. Kritzinger et al. (2012)
reported a GSH release of between 1.45 and 2.53 mg/L for a 0.3 g/L addition of
DYP. Moreover, the results observed in this study are close to those reported by
Andujar-Ortiz et al. (2012), in which 0.4 g/L of eight different DYPs were tested. The
reduced GSH release was found to be between 2.03 and 0.46 mg/L. The differences
found in the levels of GSH could be attributed to variations in the manufacturing
processes. The availability of different nitrogen sources and other nutrients during
the growth of the yeast culture (Andujar- Ortíz et al., 2012), or possible batch
differences, could explain the reported results. Furthermore, Li et al. (2004) and Wen
et al. (2004) reported cysteine level as a limiting factor for GSH biosynthesis. In
addition, Tirelli et al. (2010) and Andujar-Ortiz et al. (2012) reported a decrease in
the levels of GSH present in DYPs, possibly due to the thermal treatment involved
in the drying process.
GSH concentrations during alcoholic fermentation
The assayed treatments presented different trends in GSH levels during alcoholic
fermentation. Before GSH and/or DYP additions (ba), GSH concentrations in the
juices ranged from 2.4 to 3.2 mg/L (Fig. 1), but these were non-significant. After
additions (aa, Fig. 1), the treatments GSH 80, which was selected based on the
highest GSH concentration reported in a South African grape juice (Du Toit et al.,
2007), and DYP + GSH 80 showed the highest levels of GSH, whereas, as expected,
the control (C) wines showed the lowest levels. After addition (aa), the GSH 5.5 and
DYP treatments showed similar levels of GSH, with 8.8 and 8.3 mg/L respectively.
The difference in GSH levels between the control (C) and the DYP treatment (5.7
mg/L) was consistent with the results reported by Andujar-Ortiz et al. (2012). During
alcoholic fermentation, the GSH levels decreased until 50% completion of the
fermentation was reached, and from this point onwards an increase was observed
until the end of the fermentation. Several authors have observed contradictory
results, with either an increase (Park et al., 2000; Fracassetti et al., 2010; Kritzinger
et al., 2013a) or a decrease in GSH concentrations (Du Toit et al., 2007; Coetzee et
al., 2013) during alcoholic fermentation. The GSH 5.5 and DYP treatments showed
similar levels of GSH during the alcoholic fermentation, with no significant
differences at the end of this process (11.8 and 13.2 mg/L respectively). Moreover,
the GSH 80 and DYP+GSH 80 wines showed similar concentrations (67.3 and 69.8
mg/L respectively), which were significantly higher when compared to the other
treatments.
The GSH assimilation during alcoholic fermentation might be carried out by cellular
transporters, which have already been characterised by Miyake et al. (1998) and
Bourbouloux et al. (2000). GSH could be assimilated and secreted by yeast during
alcoholic fermentation; therefore, the amount of GSH might be affected by factors
such as yeast strain, oxygen level, initial GSH concentrations and nutriment status
of the juice (Kritzinger et al., 2013b). Kritzinger et al. (2013b) found that the addition
of glutathione-enriched inactive dry yeast preparations led to increased levels of
GSH at the end of alcoholic fermentation. However, it is not clear whether these
results are due to increases in GSH levels due to GSH contained in the DYP, GSH
synthesis by the yeast from free amino acids, peptides, etc. in the DYP, or
preferential uptake of nutrients supplied by the DYP over those supplied by native
GSH (Kritzinger et al., 2013b). One of the aims of this study was to investigate
whether increased GSH levels in the wine are due to the production of GSH from
constituents in DYP, or simply due to the addition of GSH contained in the DYP.
After addition (aa), the DYP additions led to increases in GSH levels in the juice, but
this higher trend did not occur at the latter stages of alcoholic fermentation, with no
significant differences between the DYP and corresponding GSH additions. This
work could not provide a clear answer to this question, which requires further
investigation.
Volatile compounds

Table 2 shows the levels of volatile compounds quantified in the Sauvignon blanc
wines. Esters, fatty acids and higher alcohols have a fermentative origin, while
terpenes and certain volatile thiols are classed as varietal aromas (Tominaga &
Dubourdieu, 2003; Swiegers at al., 2005; Polásková et al., 2008). Concentrations
found in the wines made in this study are in accordance with previously reported
research (Francis & Newton, 2005; Benkwitz et al., 2012; Jouanneau et al., 2012).
Moreover, differences were found for the aroma composition between the different
wine treatments (Table 2). The volatile compounds ethyl butyrate, ethyl hexanoate,
ethyl octanoate, ethyl decanoate, isoamyl acetate and 2-phenylethyl acetate were
all found above their perception thresholds (p.t.) and were influenced to a different
extent in some treatments compared to the control wines. The ethyl decanoate levels
were also significantly higher in the DYP+GSH 80 wines compared to the control.
Ethyl acetate levels were also increased by the DYP addition. Levels of hexyl acetate
were increased in all the treatments – except the GSH80 treatment – when
compared to the control. The increases in certain esters found in some of the treated
wines could be related to the extra supplementation of nitrogenous

TABLE 2

Concentration of esters (mg/L), fatty acids (mg/L), alcohols (mg/L), terpenes (μg/L)
and thiols (ng/L) (mean ± standard deviation) detected in the control wines (C) and
in the wines

supplemented with GSH-DYP and GSH. The quantification of aroma compounds

was performed at the end of alcoholic fermentation. Letters indicate significant
difference (p < 0.05).

Data are reported as mean values (n = 3) ± standard deviation.

GSH and GSH-SIY treatments

Aroma compounds Odour quality Perception

threshold Control GSH 5.5 GSH 80 DYP DYP+GSH 80

Ethyl acetate Pineapple 12.3Ω 64.4 ± 1.1b 68.0 ± 2.6ab 65.1 ± 1.2b 69.4 ± 3.9a 67.2
± 1.5ab

Esters (mg/L)

Ethyl butyrate † Apple 0.020π 1.02 ± 0.01b 1.11 ± 0.04a 1.06 ± 0.01ab 1.12 ± 0.05a
1.07 ± 0.06ab

Isoamyl acetate † Banana 0.030Ω 4.80 ± 0.13b 5.76 ± 0.25a 5.46 ± 0.22ab 5.90 ±
0.74a 5.73 ± 0.62a

Ethyl hexanoate † Apple-peel-fruit 0.014π 0.80 ± 0.07b 1.42 ± 0.07a 1.35 ± 0.01ab
1.41 ± 0.10a 1.39 ± 0.17a

Hexyl acetate Fruit-herb 0.67α 0.12 ± 0.01b 0.20 ± 0.02a 0.17 ± 0.02ab 0.21 ± 0.05a
0.20 ± 0.04a

Ethyl lactate Fruit 155β 11.25 ± 0.16ab 11.26 ± 0.23ab 11.08 ± 0.13b 11.52 ± 0.12a
11.48 ± 0.23a
Ethyl octanoate † Fruit-fat 0.005π 1.54 ± 0.04b 1.90 ± 0.05a 1.74 ± 0.04ab 1.87 ±
0.20a 1.84 ± 0.20a

Ethyl-3-hydroxybutanoate - 67μ 0.51 ± 0.01b 0.53 ± 0.01ab 0.52 ± 0.02b 0.53 ±

0.02ab 0.55 ± 0.02a

Ethyl decanoate † Apple-peel-fruit 0.20π 0.68 ± 0.03b 0.75 ± 0.75ab 0.70 ± 0.03ab
0.75 ± 0.05ab 0.78 ± 0.08a

Diethyl succinate Wine-fruit 200β 0.34 ± 0.01b 0.35 ± 0.01b 0.34 ± 0.01b 0.36 ±
0.01a 0.36 ± 0.01a

Ethyl phenylacetate Fruit-sweet - 0.01 ± 0.00a 0.10 ± 0.06a 0.02 ± 0.03a 0.07 ±
0.02a 0.06 ± 0.06a

2-Phenylethyl acetate † Fruit-sweet 0.25Ω 0.68 ± 0.02b 0.74 ± 0.01a 0.73 ± 0.01ab
0.76 ± 0.05a 0.74 ± 0.03a

Fatty acids (mg/L)

Acetic acid † Vinegar 200Ω 420 ± 26b 463 ± 22a 419 ± 14b 454 ± 21ab 427 ± 28ab

Propionic acid Soy-pungent-rancid 8.1Ω 1.70 ± 0.13ab 1.89 ± 0.15a 1.61 ± 0.02b
1.72 ± 0.08ab 1.64 ± 0.12b

Isobutyric acid Rancid-butter-cheese 2.3Ω 1.27 ± 0.09c 1.51 ± 0.06ab 1.38 ± 0.06bc
1.52 ± 0.04a 1.48 ± 0.10ab

Butyric acid † Rancid-sweat-cheese 0.17Ω 1.86 ± 0.06c 2.17 ± 0.05ab 2.02 ± 0.03b
2.18 ± 0.06ab 2.20 ± 0.16a

Isovaleric acid † Sweat-fat-rancid 0.033β 1.04 ± 0.06c 1.14 ± 0.03ab 1.09 ± 0.04bc
1.19 ± 0.03a 1.18 ± 0.06a

Valeric acid - - 0.46 ± 0.04a 0.51 ± 0.02a 0.34 ± 0.06b 0.50 ± 0.02a 0.49 ± 0.02a

Hexanoic acid † Resin-flower-green 0.42Ω 5.05 ± 0.15b 6.10 ± 0.25a 5.72 ± 0.14ab
5.99 ± 0.5a 5.84 ± 0.62a

Octanoic acid † Sweat-cheese 0.50Ω 8.44 ± 0.21b 9.78 ± 0.36a 9.31 ± 0.37ab 9.47
± 0.82ab 9.37 ± 1.02ab
FIGURE 1

Extracellular GSH trend in Sauvignon blanc wine during alcoholic fermentation.

Letters indicate significant difference (p < 0.05). ba: before additions of GSH and/or
DYP, aa: after additions of GSH and/or DYP. Mean values (n = 3) and standard
deviation (vertical bars) are reported

compounds in DYP, as GSH and alpha amino acids released from GSH and/or DYP
might have increased the production of these compounds during the alcoholic
fermentation.
Fatty acids have been described as having fruity, fatty, cheese, sweat, butter and
rancid notes (Francis & Newton, 2005). The treatments also led to changes in the
fatty acid concentrations compared to the control wines, depending on the treatment
and compound investigated (Table 2). In the case of butyric acid, for instance, all
GSH and DYP additions led to higher levels of this fatty acid compared to the control
wines, with varying results observed for hexanoic, octanoic and isobutyric acids.
Regarding hexanoic acid, significantly higher concentrations were found in the GSH
5.5, DYP and DYP+GSH 80 wines when compared to the control. Moreover,
isovaleric acid showed higher levels in the DYP and DYP+GSH 80 compared to the
GSH 5.5 wines, in which the levels were also significantly higher than in the control
wines. Finally, variable results were observed for octanoic and valeric acids.
Higher alcohols and esters are formed during alcoholic fermentation and play an
important role in wine flavour (Valero et al., 2002). The higher alcohols are
characterised by the aromas of rose, spice, lilac, alcohol, ripe fruit resin, herbaceous
and whiskey (Francis & Newton, 2005). At concentrations above 400 mg/L they are
considered as negative quality factors (Rapp & Versini, 1991), while at low
concentration (up to 300 mg/L) they are thought to increase the aromatic complexity
of the wine (Ribéreau-Gayon et al., 2006).
The amounts of total higher alcohols observed in this study ranged from 220 mg/L
(C) to 251 mg/L (DYP). Hexanol, isoamyl alcohol and 2-phenylethanol were detected
at concentrations higher than their perception thresholds, and at higher levels in the
DYP-treated wines than those observed in the control wines. Hexanol and isoamyl
alcohol levels were also significantly higher in the wines treated with GSH 5.5, DYP
and DYP+GSH 80 than in the control wines. Moreover, the levels of 2-phenyl ethanol
were significantly higher in DYP and DYP+GSH 80 when compared to the control
wines. Interestingly, methanol levels were slightly higher in the wines made from
juice to which GSH had been added at different levels (GSH 5.5 and GSH 80).
Finally, no clear trends were observed in the assayed wines for propanol levels.
Andújar-Ortiz et al. (2014) also found increased levels of isoamyl acetate, hexyl
acetate, 2-phenylethyl acetate and certain long-chain ethyl esters, as well as
hexanoic acids to which DYP had been added, which corroborates our results.
Limited data is available on the effect of only GSH additions on ester, fatty acid and
higher alcohol production during alcoholic fermentation, which this work
investigated.
The monoterpene alcohols feature prominently in the aroma of many wine varieties,
such as Muscat, Gewurztraminer and Weisser Riesling (Kritzinger et al., 2013a).
Terpenes have been described as having rose, lavender, geranium, lychee and
green notes (Francis & Newton, 2005). Due to the oxidation phenomena during wine
ageing, monoterpene alcohols, such as geraniol and linalool, decrease to form
dihydric alcohols and terpene dioxides, which have higher perception threshold
(Simpson, 1978; Simpson & Miller, 1983). The volatile compound citronellol, the
sesquiterpene β-farnesol 1, linalyl acetate and nerol were significantly increased by
DYP addition (DYP and DYP+GSH 80 treatments). Moreover, the levels of nerol
were also significantly higher in the GSH 80 wines compared to the control and GSH
5.5 treatments. Other researchers have also found higher levels of certain terpenes
in wines exposed to DYP treatments (Andujar-Ortiz et al., 2012; Rodríguez-
Bencomo et al., 2014). As the wines in this study were analysed only three months
after bottling, the anti-oxidant activities of DYP in preserving certain terpenes, as
reported by Rodríguez-Bencomo et al. (2014), probably also had such an effect.

Sauvignon blanc wines are characterised by the presence of highly odorous thiol
compounds, such as 4-mercapto-4- methylpentan-2-one (4MMP), 4-mercapto-4-
methylpentan- 2-ol (4MMPOH), 3-mercaptohexan-1-ol (3MH) and

TABLE 3. Wine sensory data (out of 100) generated with descriptive analysis for the
different treatments. Letters indicate significant differences (p < 0.05).
Sensory C GSH 5.5 GSH 80 DYP DYP+80
descriptor GSH
Cooked 1.3 ± 4.0a 1.4 ± 4.4a 2.5 ± 5.9a 3.3 ± 7.0a 2.5 ± 5.3a
vegetable
Overall 35.7 ± 14.8a 43.6 ± 12.2b 47.7 ± 12.7b 56.7 ± 10.1c 56.2 ± 8.5c
tropical
Canned 25.9 ± 10.4a 34.8 ± 18.0b 49.4 ± 13.6c 62.3 ± 10.6d 63.6 ± 10.9d
tropical
Banana 2.2 ± 7.2a 11.0 ± 20.7b 23.2 ± 15.4c 32.1 ± 15.9d 33.6 ± 19.3d
Grapefruit 26.3 ± 18.4b 30.1 ± 18.9b 12.1 ± 15.6a 11.7 ± 9.8a 15.7 ± 15.4a
Green guava 11.5 ± 15.6b 11.3 ± 15.0b 3.1 ± 7.5a 2.4 ± 10.4a 2.9 ± 9.8a
Ripe guava 1.0 ± 3.9a 5.0 ± 13.0a 12.8 ± 16.2b 19.6 ± 15.6bc 21.1 ± 16.7c
Fresh 36.7 ± 15.3c 29.53 ± 19.9b 6.3 ± 17.3a 0.0 ± 0.0a 1.2 ± 5.0a
pineapple
Canned 0.8 ± 4.6a 11.8 ± 22.2b 29.5 ± 19.5c 41.4 ± 15.4d 39.4 ± 17.1d
pineapple
Passion fruit 8.8 ± 10.9bc 11.5 ± 9.8c 3.0 ± 6.9a 4.2 ± 6.7a 5.9 ± 9.8ab
Stone fruit 4.1 ± 9.4a 6.5 ± 12.4a 14.73 ± 10.0b 19.6 ± 11.5b 14.3 ± 11.6b
Fresh tropical 74.1 ± 10.4d 65.2 ± 18.0c 50.6 ± 13.6b 37.7 ± 10.6a 36.4 ± 10.9a
3-mercaptohexyl acetate (3MHA) (Darriet et al., 1995, Tominaga et al., 1998).
3MHA and 3MH are characterised by the mixed aroma of boxwood flowers,
grapefruit and passion fruit (Tominaga et al., 2006). 3MHA (p.t. 4 ng/L) and 3MH (p.t.
60 ng/L) were always present at concentrations higher than their perception
thresholds (Table 2). 3MH and 3MHA levels were significantly higher in the wines
treated with the DYP. Surprisingly, the addition of GSH, even at relatively high
concentrations of 80 mg/L, did not increase the levels of these two important volatile
thiols (Table 2). Glutathione-rich inactivated yeast (DYPs) might stimulate the
production of varietal thiols through the release of GSH or/and cysteine, although
the former mechanism seems unlikely, since the additions of pure GSH did not lead
to higher levels of these compounds. In addition, higher levels of glutamic acid
released from DYP can lead to higher levels of 3MH (Šuklje et al., 2016).
Contradictory results are found in the literature regarding the influence of GSH on
thiol release, with some authors reporting an increase and others finding decreases
(Patel et al., 2010; Roland et al., 2010; Makhotkina et al., 2014). Moreover, GSH has
been used as an antioxidant at harvest, complementing a moderate SO2 addition.
Higher thiol levels were observed, likely through an antioxidant effect (Makhotkina
et al., 2014). Should GSH or the DYPs support a more reductive environment and
greater H2S levels, the production of 3MH, and subsequently 3MHA, might also be
favoured, although this might be a minor route under normal winemaking conditions.
However, Harsch et al. (2013) demonstrated that supplying grape juice with an
external source of hydrogen sulphide drastically increases thiol formation in the
finished wines.
GSH-enriched DYPs seemed to influence higher alcohols, fatty acids and ester
levels, but this effect was less pronounced in the GSH-treated wines. These results
point to a possible positive effect of GSH on these specific volatile compounds that
might be related to its activity as a strong antioxidant. However, these mechanisms
seem unlikely, as the addition of pure GSH at the same level as that found in the
DYP, as well as a high addition (80 mg/L), often did not lead to the same increases.
In the specific case of terpenes and volatile thiols, only DYP additions showed higher
levels of these volatile compounds. Multiple reasons, such as the modification of by-
products during fermentation, the release of volatiles from DYPs or the interaction of
wine volatiles with compounds released by DYPs, could explain the differences
observed in this study; and these need further investigation.
Descriptive analysis
In order to determine which sensory attributes were influenced by the addition of
GSH or GSH-DYP, descriptive analysis was performed on the Sauvignon blanc
wines. The results of the descriptive analysis are reported in Table 3. The statistical
analysis (ANOVA) of each sensory descriptor showed significant differences
between treatments (Table 3). The DYP and DYP+GSH 80 treatments were
described by the panel as having higher levels of overall tropical, canned tropical,
banana, stone fruit, ripe guava and canned pineapple attributes, whereas the GSH
80 sample presented significantly higher levels of the abovementioned compounds
when compared against the control wines, but at lower intensities (except for the
banana and stone fruit attributes) than those found for the DYP additions. The aroma
of the control wine (C) was characterised by a sensory profile associated with fresh
tropical and fresh pineapple notes. The GSH 5.5 wines had similar intensities of
grapefruit, passion fruit and green guava compared to the control, and these
intensities were higher than those observed in the other three treatments. The DYP
additions thus influenced the volatile profile through increases of notes associated
with ripe tropical fruits. On the other hand, adding 5.5 mg/L GSH to the must before
fermentation did not seem to be sufficient to alter the sensorial profile of the wine
compared to that of the control wine. However, our sensory results also correlate
with those of Andújar-Ortiz et al. (2014), who also found increases in banana
character in the DYP-treated wines, with higher levels of peaches The sensorial
differences observed in this study, along with those of Andújar-Ortiz et al. (2014) and
Šuklje et al. (2016), are thus probably not due to the GSH released from the DYP
products, but due to the yeast forming different volatiles, brought about by the amino
acids and peptides released from these products..
CONCLUSION
The addition of GSH or DYP to Sauvignon blanc grape must has been shown to
increase the concentrations of certain volatile compounds, which included thiols and
some monoterpenes in the case of the DYP-added treatments. The addition of DYP
can also influence the sensory characteristics of the wine, leading to a change in the
wine’s aroma profile. The amount of GSH released into the wines when DYP was
added, however, seems to be limited, and the increases in GSH, volatile thiols and
other compounds are probably due to a change in the chemical composition of the
must. Nitrogen-based compounds and other components present in the DYP
preparations are more likely responsible for the changes observed than the GSH
added to the must due to DYP addition. However, this needs to be investigated
further using other cultivars and DYP preparations.
LITERATURE CITED
Aguera, E., Samson, A., Caillé, S., Julien-Ortiz, A., Sieczkowski, N. & Salmon, J.M.,
2012. Apport de levures inactivées riches en glutathion en cours de fermentation
alcoolique: un nouvel outil pour la protection des vins blancs et rosés contre
l’oxydation. Rev. Fr. Oenol. 250, 2-10.
Andújar-Ortiz, I., Chaya, C., Martín-Álvarez, P.J., Moreno-Arribas, M.V. & Pozo-
Bayón, M.A., 2014. Impact of using new commercial glutathione enriched inactive
dry yeast oenological preparations on the aroma and sensory properties of wines.
Int. J. Food Prop. 17, 987-1001.
Andujar-Ortiz, I., Pozo-Bayón, M.A., Moreno-Arribas, M.V., Martín-Álvarez, P.J. &
Rodríguez-Bencomo, J.J., 2012. Reversed-phase high-performance liquid
chromatography–fluorescence detection for the analysis of glutathione and its
precursor γ-glutamyl cysteine in wines and model wines supplemented with
oenological inactive dry yeast preparations. Food Anal. Methods 5, 154-161.
Benkwitz, F., Tominaga, T., Kilmartin, P.A., Lund, M., Wohlers, C. & Nicolau, L.,
2012. Identifying the chemical composition related to the distinct flavour
characteristics of New Zealand Sauvignon blanc wines. Am. J. Enol. Vitic. 63, 62-
72.
Betés-Saura, C., Andrés-Lacueva, C. & Lamuela-Raventós, R.M., 1996. Phenolics
in white free run juices and wines from Penedès by high-performance liquid
chromatography: Changes during vinification. J. Agric. Food Chem. 44, 3040-3046.
Blanchard, L., Darriet, P. & Dubourdieu, D., 2004. Reactivity of 3-mercaptohexanol
in red wine: Impact of oxygen, phenolic fractions, and sulfur dioxide. Am. J. Enol.
Vitic. 55, 115-120.
Bourbouloux, A., Shahi, P., Chakladar, A., Delrot, S. & Bachhawat, A.K., 2000.
Hgt1p, a high affinity glutathione transporter from the yeast Saccharomyces
cerevisiae. J. Biol. Chem. 275, 13259-13265.
Cheynier, V., Rigaud, J. & Moutounet, M., 1990. Oxidation kinetics of trans-
caffeoyltartrate and its glutathione derivatives in grape musts. Phytochem. 29, 1751-
1753.
Cheynier, V., Souquet, J.M. & Moutounet, M., 1989. Gluthatione content and
gluthatione to hydroxycinnamic acid ratio in Vitis vinifera grapes and musts. Am. J.
Enol. Vitic. 40, 320-324.
Coetzee, C. & Du Toit, W.J., 2012. A comprehensive review on Sauvignon blanc
aroma with a focus on certain positive volatile thiols. Food Res. Int. 45, 287-298.
Coetzee, C., Lisjak, K., Nicolau, L., Kilmartin, P. & Du Toit, W.J., 2013. Oxygen and
sulfur dioxide additions to Sauvignon blanc must: Effect on must and wine
composition. Flavour Frag. J. 28, 155-167.
Darriet, P., Tominaga, T., Lavigne-Cruege, V., Boidron, J.N. & Dubourdieu, D., 1995.
Identification of a powerful aromatic component of Vitis vinifera L. var. Sauvignon
wines: 4-mercapto-4-methylpentan-2-one. Flavour Frag. J. 10, 385-392.
Dragone, G., Mussatto, S.I., Oliveira, J.M. & Teixeira J.A., 2009. Characterization of
volatile compounds in an alcoholic beverage produced by whey fermentation. Food
Chem. 112, 929-935.
Dubourdieu, D. & Lavigne-Cruege, V., 2004. The role of glutathione on the aromatic
evolution of dry white wine. Wine Internet Technical Journal, 02, 2.
Du Toit, W.J., Lisjak, K., Stander, M. & Prevoo, D., 2007. Using LC–MS to assess
glutathione levels in South African white grape juices and wines made with different
levels of oxygen. J. Agric. Food Chem. 55, 2765-2769.
Elskens, M.T., Jaspers, C.J. & Penninckx, M.J., 1991. Glutathione as an
endogenous sulphur source in the yeast Saccharomyces cerevisiae. J. Gen.
Microbiol. 137, 637-644.
Fedrizzi, B., Pardon, K.H., Sefton, M.A., Elsey, G.M. & Jeffery, D.W., 2009. First
identification of 4-S-glutathionyl-4-methylpentan-2-one, a potential precursor of 4-
mercapto-4-methylpentan-2-one, in Sauvignon blanc juice. J. Agric. Food Chem. 57,
991-995.
Fracassetti, D., 2010. Investigation on cysteinyl thiol compounds from yeast affecting
wine properties. Ph.D. Thesis, Technology and Biotechnology, University of Milan,
Italy.
Francis, I.L. & Newton, J.L., 2005. Determining wine aroma from compositional data.
Aust. J. Grape Wine Res. 11, 114-126.
Gómez-Míguez, M.J., Cacho, J.F., Ferreira V., Vicario, I.M. & Heredia, F.J., 2007.
Volatile components of Zalema white wines. Food Chem. 100, 1464- 1473.
Harsch, M.J., Benkwitz, F., Frost, A., Colonna-Ceccaldi, B., Gardner, R.C. & Salmon,
J.M., 2013. New precursor of 3-mercaptohexan-1-ol in grape juice: Thiol-forming
potential and kinetics during early stages of must fermentation. J. Agric. Food Chem.
61, 3703-3713.
Jouanneau, S., Weaver, R.J., Nicolau, L., Herbst-Johnstone, M., Benkwitz, F. &
Kilmartin, P.A., 2012. Sub-regional survey of aroma compounds in Marlborough
Sauvignon blanc wines. Aust. J. Grape Wine Res. 18, 329- 343.
Kritzinger, E.C., Bauer, F.F. & Du Toit, W.J., 2013a. Role of glutathione in
winemaking: A review. J. Agric. Food Chem. 61, 269-277.
Kritzinger, E.C., Bauer, F.F. & Du Toit, W.J., 2013b. Influence of yeast strain,
extended lees contact and nitrogen supplementation on glutathione concentrations
in wine. Aust. J. Grape Wine Res. 19, 161-170.
Kritzinger, E.C., Stander, M. & Du Toit, W.J., 2012. Assessment of glutathione levels
in model solution and grape ferments supplemented with glutathione-rich inactivated
dry yeast preparations using a novel UPLC-MS/MS method. Food Addit. Contam.
Part A. 30(1), 80-92.
Lavigne-Cruege, V. & Dubourdieu, D., 2002. Role of glutathione on development of
aroma defects in dry white wines. In 13th International Enology Symposium, Trogus,
H., Gafner, J. and Sutterlin, A. Eds. International Association of Enology: Montpellier,
France, pp 331-347
Lavigne-Cruege, V. & Dubourdieu, D., 2004. Affinamento sulle fecce e freschezza
dei vini bianchi. Vigne Vini 31, 58-66.

Lavigne-Cruege, V., Pons, A. & Dubourdieu D., 2007. Assay of gluthatione in musts
and wines using capillary electrophoresis and laser-induced fluorescence detection.
Changes in concentration in dry wines during the alcoholic fermentation and aging.
J. Chromatogr. A 1139, 130-135.

Li, H., Guo, A. & Wang H., 2008. Mechanisms of oxidative browning of wine. Food
Chem. 108, 1-13.
Li, Y., Wei, G. & Chen, J., 2004. Glutathione: A review on biotechnological
production. Appl. Microbiol. Biotechnol. 66, 233-242.
Makhotkina, O., Araujo, L.D., Olejar, K., Herbst-Johnstone, M., Fedrizzi, B. &
Kilmartin, P.A., 2014. Aroma impact of ascorbic acid and glutathione additions to
Sauvignon blanc at harvest to supplement sulphur dioxide. Am. J. Enol. Vitic. 65,
388-393.
Mateo, J.J. & Jimenez, M., 2000. Monoterpenes in grape juice and wines. J.
Chromatogr. A. 881, 557-567.
Miyake, T., Hazu, T., Yoshida, S., Kanayama, M., Tomochika, K., Shinoda, S. & Ono,
B., 1998. Glutathione transport systems of the budding yeast Saccharomyces
cerevisiae. Biosci. Biotechnol. Biochem. 62, 1858-1864.
Moreno, J.A., Zea, L., Moyano, L. & Medina, M., 2005. Aroma compounds as
markers of the changes in sherry wines subjected to biological ageing. Food Control.
16, 333-338.
Oliveira, C.M., Ferreira A.C.S., De Freitas, V. & Silva, A.M.S., 2011. Oxidation
mechanisms occurring in wines. Food Res. Int. 44, 1115-1126.
OIV (Organisation of Vine and Wine), 2013. Resolution OIV-OENO 459- 2013. 11th
OIV General Assembly, Bucharest, Romania.
OIV (Organisation of Vine and Wine), 2015. Resolutions OIV-OENO 445- 2015 and
OIV-OENO 446-2015. 13th General Assembly, Mainz, Germany.
Park, S.K., Boulton, R.B. & Noble, A.C., 2000. Formation of hydrogen sulphide and
glutathione during fermentation of white grape musts. Am. J. Enol. Vitic. 51, 91-97.
Patel, P., Herbst-Johnstone, M., Lee, S.A., Gardner, R.C., Weaver, R., Nicolau, L. &
Kilmartin, P.A., 2010. Influence of juice pressing conditions on polyphenols,
antioxidants and varietal aroma of Sauvignon blanc microferments. J. Agric. Food
Chem. 58, 7280-7288.
Peinado, R.A., Mauricio, J.C. & Moreno, J., 2006. Aromatic series in sherry wines
with gluconic acid subjected to different biological aging conditions by
Saccharomyces cerevisiae var. capensis. Food Chem. 94, 232-239.
Penninckx, M.A., 2002. An overview on glutathione in Saccharomyces versus non-
conventional yeasts. FEMS Yeast Res. 2, 295-305.
Plotto, A., Barnes, K.W. & Goodner, K.L., 2006. Specific anosmia observed for β-
ionone, but not for α-ionone: Significance for flavour research. J. Food Sci. 71, 401-
406.
Polásková, P., Herszage J. & Ebeler, S.E., 2008. Wine flavour: Chemistry in a glass.
Chem. Soc. Rev. 37, 2478-2489.
Pozo-Bayón, M.A., Andújar-Ortiz, I. & Moreno-Arribas, M.V., 2009. Scientific
evidences beyond the application of inactive dry yeast preparations in winemaking.
Food Res. Int. 42, 754-761.
Rapp, A. & Versini, G., 1991. Influence of nitrogen compounds in grapes on aroma
compounds of wine. In: Rantz, J. (ed.). Proc. Internat. Symp, Nitrogen in Grapes and
Wines of the American Society for Enology and Viticulture, Month? 1991, Seattle,
WA. pp. 156 – 164.
nd
Ribéreau-Gayon, P., Glories, Y., Maujean, A. & Dubourdieu, D., 2006 (2 ed).
Handbook of enology: The chemistry of wine stabilization and treatments, vol 2. John
Wiley & Sons Ltd, Chichester, UK.
Rigaud, J., Cheynier, V., Souquet, J.M. & Moutounet, M., 1991. Influence of must
composition on phenolic oxidation kinetics. J. Sci. Food Agric. 57, 55-63.
Rodríguez-Bencomo, J.J., Andújar-Ortiz, I., Moreno-Arribas, M.V., Simó, C.,
González, J., Chana, A., Dávalos, J. & Pozo-Bayón, M.A., 2014. Impact of
glutathione-rich inactivated dry yeast preparation on the stability of terpenes during
model wine aging. J. Agric. Food Chem. 62, 1373-1383.
Roland, A., Vialaret, J., Razungles, A., Rigou, P. & Schneider, R., 2010. Evolution
of S-cysteinylated and S-glutathionylated thiol precursors during oxidation of melon
B. and Sauvignon blanc musts. J. Agric. Food Chem. 58, 4406-4413.
Simpson, R.F., 1978. Aroma and compositional changes in wine with oxidation,
storage and ageing. Vitis 17, 274-287.
Simpson, R.F. & Miller, G.C., 1983. Aroma composition of aged Riesling wine. Vitis.
22, 51-63.
Šuklje, K., Antalick, G., Buica, A., Coetzee, Z.A., Brand, J., Schmidtke, L.M. & Vivier,
M.A., 2016. Inactive dry yeast application on grapes modify Sauvignon blanc wine
aroma. Food Chem. 197, 1073-1084.
Swiegers, J.H., Bartowsky, E.J., Henschke, P.A. & Pretorius, I.S., 2005. Yeast and
bacteria modulation of wine aroma and flavour. Aust. J. Grape Wine Res. 11, 139-
173.
Tirelli, A., Fracassetti, D. & De Noni, I., 2010. Determination of reduced cysteine in
oenological cell wall fractions of Saccharomyces cerevisiae. J. Agric. Food Chem.
58, 4565-4570.
Tomic, O., Luciano, G., Nilsen, A., Hyldig, G., Lorensen, K. & Naes, T., 2010.
Analysing sensory panel performance in a proficiency test using the PanelCheck
software. Eur. Food Res. Technol. 230(3), 497-511.
Tominaga, T. & Dubourdieu, D., 2003. Recent advance of aroma research of wines
made from non-aromatic grape varieties – an example of Sauvignon blanc in
Bordeaux. J. Brew. Soc. Jpn. (Nippon Jozokyokai Shi) 98, 628-637.
Tominaga, T., Furrer, A., Henry, R. & Dubourdieu, D., 1998. Identification of new
volatile thiols in the aroma of Vitis vinifera L. var. Sauvignon blanc wines. Flavour
Frag. J. 13, 159-162.
Tominaga, T., Niclass, Y., Frérot, E. & Dubourdieu, D., 2006. Stereoisomeric
distribution of 3-mercaptohexan-1-ol and 3-mercaptohexyl acetate in dry and sweet
white wines made from Vitis vinifera (var. Sauvignon blanc and Semillon). J. Agric.
Food Chem. 54, 7251-7255.
Ugliano, M., Kwiatkowski, M., Vidal, S., Capone D., Siebert, T., Dieval, J.B.,
Aagaard, O. & Waters, E.J., 2011. Evolution of 3-mercatohexanol, hydrogen sulfide,
and methyl mercaptan during bottle storage of Sauvignon blanc wines: effect of
glutathione, copper, oxygen exposure, and closure-derived oxygen. J. Agric. Food
Chem. 59, 2564–2572.
Valero, E., Moyano, L., Millan, M.C., Medina, M. & Ortega, J.M., 2002. Higher
alcohols and esters production by Saccharomyces cerevisiae. Influence of the initial
oxygenation of the grape must. Food Chem. 78, 57- 61.

Wen, S., Zhang, T. & Tan, T., 2004. Utilization of amino acids to enhance glutathione
production in Saccharomyces cerevisiae. Enzyme Microb. Technol. 35, 501-507.
 

1. Previous article in issue: Modeling the Fate of Escherichia coli O157:H7 and
Salmonella enterica in the Agricultural Environment: Current Perspective
2. Next article in issue: Diet Effects in Gut Microbiome and Obesity

Volume 79, Issue 4April 2014 Pages R428–R441

R: Concise Reviews in Food Science

GRASr2 Evaluation of Aliphatic Acyclic and Alicyclic Terpenoid Tertiary

Alcohols and Structurally Related Substances Used as Flavoring Ingredients

Volume 79, Issue 4, April 2014, Pages: R428–R441, Lawrence J. Marnett, Samuel
M. Cohen, Shoji Fukushima, Nigel J. Gooderham, Stephen S. Hecht, Ivonne M.C.M.
Rietjens, Robert L. Smith, Timothy B. Adams, Maria Bastaki, Christie L. Harman,
Margaret M. McGowen and Sean V. Taylor

Abstract

This publication is the 1st in a series of publications by the Expert Panel of the Flavor
and Extract Manufacturers Assoc. summarizing the Panel's 3rd re-evaluation of
Generally Recognized as Safe (GRAS) status referred to as the GRASr2 program.
In 2011, the Panel initiated a comprehensive program to re-evaluate the safety of
more than 2700 flavor ingredients that have previously met the criteria for GRAS
status under conditions of intended use as flavor ingredients. Elements that are
fundamental to the safety evaluation of flavor ingredients include exposure,
structural analogy, metabolism, pharmacokinetics, and toxicology. Flavor
ingredients are evaluated individually and in the context of the available scientific
information on the group of structurally related substances. Scientific data relevant
to the safety evaluation of the use of aliphatic acyclic and alicyclic terpenoid tertiary
alcohols and structurally related substances as flavoring ingredients are evaluated.
The group of aliphatic acyclic and alicyclic terpenoid tertiary alcohols and structurally
related substances was reaffirmed as GRAS (GRASr2) based, in part, on their rapid
absorption, metabolic detoxication, and excretion in humans and other animals; their
low level of flavor use; the wide margins of safety between the conservative
estimates of intake and the no-observed-adverse effect levels determined from
subchronic studies and the lack of significant genotoxic and mutagenic potential.

Introduction

For over 50 y, the Expert Panel of the Flavor and Extract Manufacturers Assoc.
(FEMA) has served as the primary, independent body evaluating the safety of flavor
ingredients. A key part of the FEMA Generally Recognized as Safe (GRAS) program
is the cyclical re-evaluation of the “generally recognized as safe” status of flavor
ingredients determined to be GRAS by the FEMA Expert Panel during the course of
its operations. The Panel has previously completed 2 re-evaluations of all FEMA
GRAS flavor ingredients. This summary represents the 1st in a series of publications
summarizing the Expert Panel's 3rd re-evaluation of GRAS status referred to as the
GRASr2 program. While these summaries contain only critical data relevant to the
safety evaluation, the Expert Panel has reviewed all of the available data during the
course of its safety evaluation.

Currently, the FEMA Expert Panel has determined that over 2700 flavor ingredients
have met the criteria for GRAS status under conditions of intended use as flavor
ingredients. The FEMA GRAS program is ongoing; Panel decisions on individual
flavor ingredients are provided to the U.S. Food and Drug Administration and are
published on a regular basis on the FEMA Web site and in Food Technology.
Elements that are fundamental to the safety evaluation of flavor ingredients include
exposure, structural analogy, metabolism, pharmacokinetics, and toxicology (Smith
and others 2005a).

Flavor ingredients are evaluated individually and in the context of the available
scientific information on the group of structurally related substances. The group of
aliphatic acyclic and alicyclic terpenoid tertiary alcohols and structurally related
substances was reaffirmed as GRAS (GRASr2) based, in part, on their self-limiting
properties as flavoring substances in food; their rapid absorption, metabolic
conversion, and excretion in humans and experimental animals; their low level of
flavor use; the wide margins of safety between the conservative estimates of intake
and the no-observed-adverse effect levels (NOAEL) determined from subchronic
and chronic studies and the lack of genotoxic and mutagenic potential. This evidence
of safety is supported by the fact that the intake of aliphatic acyclic and alicyclic
terpenoid tertiary alcohols and structurally related substances as natural
components of traditional foods is greater than their intake as intentionally added
flavoring substances.

Chemical Identity

As part of the GRASr2 program, the FEMA Expert Panel (FEXPAN) evaluated a
group of 44 flavor ingredients that fit into the group of aliphatic acyclic and alicylic
terpenoid tertiary alcohols and structurally related flavoring ingredients. This group
includes 14 aliphatic terpene tertiary alcohols, 8 alicylic terpene alcohols, 6 esters of
terpene tertiary alcohols, 3 alicyclic tertiary alcohols and 2 esters of alicyclic tertiary
alcohols, 2 aliphatic tertiary alcohols, 3 phenyl tertiary alcohols, and 6 esters of
phenyl substituted aliphatic tertiary alcohols.

In 1995, the FEXPAN re-evaluated 23 members of this chemical group and all were
determined to be GRAS reaffirmed (GRASr) under conditions of use. In the
intervening time period, there have been substantial increases in the volume of use
of some members of this group as flavor ingredients, as reported in the 2005 FEMA
Poundage Update Survey (Gavin and others 2008). Additionally, 20 new substances
have been evaluated and determined to be GRAS since the GRASr for this group.

Exposure

From the most recent annual poundage surveys, of the 44 substances in this review
7 have reported volumes of greater than 500 kg/y (see Table 1). More than 76% of
the total annual volume of production for this group is accounted for by linalool and
linalyl acetate (Gavin and others 2008). The volumes of production reported for
linalool and linalyl acetate are 200% and 166%, respectively, greater than the annual
volumes reported in the 1995 volume of use survey (Lucas and others 1999; Gavin
and others 2008) and 840% and 560%, respectively, greater than the annual
volumes evaluated in the original GRASr of the group (National Academy of Science,
NAS 1989). Consequently, the estimated daily per capita intakes using maximized
survey-derived intake (MSDI) for these 2 materials have increased 5- and 15-fold,
respectively, as compared to their original GRASr evaluations. Other substances
with high volumes are α-terpineol and the acetate and butyrate esters of α,α-
dimethylphenethyl alcohol (see Table 1).

Table 1. Identity and exposure data for aliphatic acyclic and alicyclic terpenoid
tertiary alcohols and structurally related substances used as flavoring ingredients

Daily per Annual

Most
capita volum
recen
intakec e in
t
FEM DT (“eaters natural
Flavoring CAS Nr. and annu Consump
A clas only”) ly
ingredient structure al tion ratioe
Nr. sa occurri
volu μg/k ng
me, μg/
g foods,
kgb d
bw/d kgd

1. aDT,
Decision Tree; Cramer and others (1978).
2. bFromNAS (1989); Lucas and others (1999); Gavin and others (2008). Values
greater than zero but less than 0.1 kg were reported as 0.1 kg.
3. cMSDI (Tg/person/d) calculated as follows:

[[(annual volume, kg) × (1 × 109 Tg/kg)]/[population × survey correction

factor × 365 d]], where population (10%, “eaters only”) = 31 × 106 for the
 Daily per Annual
Most
capita volum
recen
intakec e in
t
FEM DT (“eaters natural
Flavoring CAS Nr. and annu Consump
A clas only”) ly
ingredient structure al tion ratioe
Nr. sa occurri
volu μg/k ng
me, μg/
g foods,
kgb d
bw/d kgd

United States, where the correction factor = 0.8 for the Gavin and others
(2008), and Lucas and others (1999) and 0.6 for NAS (1989) and anticipated
volumes, representing the assumption that only 80% or 60% of the annual
flavor volume, respectively, was reported in the poundage surveys (NAS
1989; Lucas and others 1999; Gavin and others 2008; private
communication to FEMA). [(Tg/person/d)/body weight], where body weight =
60 kg. Slight variations may occur from rounding.

4. dQuantitative data for the United States reported by Stofberg and Grundschober
(1987).
5. eThe consumption ratio is calculated as follows:

(annual consumption via food, kg)/(most recent reported volume as a

flavoring substance, kg).

6. fNA, not available; +, reported to occur naturally in foods (TNO 2013), but no
quantitative data; –, not reported to occur naturally in foods.
7. gAt the 51st meeting of the Joint FAO/WHO Expert Committee on Food Additives,
methyl 1-acetoxycyclohexylketone (Nr. 23) was not supported by a relevant
NOAEL. Data for structurally related substance, 6-acetoxydihydrotheaspirane (Nr.
36) provided a NOAEL of 3 mg/kg bw/d in a 90-d study in rats (Griffiths 1979) which
is at least 100000 times the daily per capita intake of methyl 1-acetoxycyclohexyl
ketone when used as a flavor ingredient.

263 54
1. Linalool I 78-70-6 49000 90 71619 1
5 13

306 0.0 0.00

2. Tetrahydrolinalool I 78-69-3 0.4 −f NA
0 4 1
 Daily per Annual
Most
capita volum
recen
intakec e in
t
FEM DT (“eaters natural
Flavoring CAS Nr. and annu Consump
A clas only”) ly
ingredient structure al tion ratioe
Nr. sa occurri
volu μg/k ng
me, μg/
g foods,
kgb d
bw/d kgd

264
3. Linalyl formate I 115-99-1 24 3 0.04 + NA
2

263 24
4. Linalyl acetate I 115-95-7 21900 40 18567 1
6 19

264
5. Linalyl propionate I 144-39-8 100 11 0.2 + NA
5

263
6. Linalyl butyrate I 78-36-4 5 1 0.01 + NA
9

264
7. Linalyl isobutyrate I 78-35-3 73 8 0.1 − NA
0
 Daily per Annual
Most
capita volum
recen
intakec e in
t
FEM DT (“eaters natural
Flavoring CAS Nr. and annu Consump
A clas only”) ly
ingredient structure al tion ratioe
Nr. sa occurri
volu μg/k ng
me, μg/
g foods,
kgb d
bw/d kgd

264
8. Linalyl isovalerate I 1118-27-0 13 1 0.02 − NA
6

264 0.00
9. Linalyl hexanoate I 7779-23-9 4 0.4 − NA
3 7

264 0.0 0.00

10. Linalyl octanoate I 10024-64-3 0.5 − NA
4 6 09

304 69
11. α-Terpineol I 98-55-5 6300 12 21451 3
5 6

305 0.0 0.00

12. Terpinyl formate I 2153-26-6 0.1 + NA
2 1 02
 Daily per Annual
Most
capita volum
recen
intakec e in
t
FEM DT (“eaters natural
Flavoring CAS Nr. and annu Consump
A clas only”) ly
ingredient structure al tion ratioe
Nr. sa occurri
volu μg/k ng
me, μg/
g foods,
kgb d
bw/d kgd

304
13. Terpinyl acetate I 8007-38-0 82 9 0.2 61 1
7

14. Terpinyl 305

I 80-27-3 30 3 0.06 + NA
propionate 3

304 0.00
15. Terpinyl butyrate I 2153-28-8 5 0.6 − NA
9 9

16. Terpinyl 305 0.00

I 7774-65-4 5 0.5 − NA
isobutyrate 0 8

17. Terpinyl 305

I 1142-85-4 5 0.7 0.01 − NA
isovalerate 4
 Daily per Annual
Most
capita volum
recen
intakec e in
t
FEM DT (“eaters natural
Flavoring CAS Nr. and annu Consump
A clas only”) ly
ingredient structure al tion ratioe
Nr. sa occurri
volu μg/k ng
me, μg/
g foods,
kgb d
bw/d kgd

18. p-Menth-3-en-1- 356 0.00

I 586-82-3 2 0.3 3 2
ol 3 5

19. 4- 224
I 562-74-3 640 71 1 38968 61
Carvomenthenol 8

20. p-Menth-8-en-1- 356

I 138-87-4 320 47 0.8 1792 6
ol (β-Terpineol) 4

21. 2-Ethyl-1,3,3-
349 0.0 0.00
trimethyl-2- II 18368-91-7 0.1 − NA
1 1 02
norbornanol

323
22. 4-Thujanol I 546-79-2 34 4 0.06 9200 271
9
 Daily per Annual
Most
capita volum
recen
intakec e in
t
FEM DT (“eaters natural
Flavoring CAS Nr. and annu Consump
A clas only”) ly
ingredient structure al tion ratioe
Nr. sa occurri
volu μg/k ng
me, μg/
g foods,
kgb d
bw/d kgd

23. Methyl 1-
370
acetoxycyclohexyl I 52789-73-8 11 2 0.03 − NA
1
ketoneg

24. 2,3,4-Trimethyl- 390

I 3054-92-0 5 0.7 0.01 − NA
3-pentanol 3

25. (+/-)-2,4,8-
421 0.00
Trimethyl-7-nonen- I 437770-28-0 1 0.1 − NA
2 2
2-ol

26. (E)- and (Z)-

421 0.00
2,4,8-Trimethyl-3,7- I 479547-57-4 1 0.1 − NA
1 2
nonadien-2-ol

277
27. Nerolidol I 7212-44-4 181 20 0.3 300000 1657
2
 Daily per Annual
Most
capita volum
recen
intakec e in
t
FEM DT (“eaters natural
Flavoring CAS Nr. and annu Consump
A clas only”) ly
ingredient structure al tion ratioe
Nr. sa occurri
volu μg/k ng
me, μg/
g foods,
kgb d
bw/d kgd

28. 1-Phenyl-3- 288 0.00

I 10415-87-9 1 0.1 − NA
methyl-3-pentanol 3 2

29. p-α, α-
324 0.0 0.00
Trimethylbenzyl I 1197-01-9 0.1 500000 5000000
2 1 02
alcohol

30. (±)-Ethyl 2-
426 0.11
hydroxy-2- I 77-70-3 50 7 + NA
8 7
methylbutyrate

31. α,α-
239
Dimethylphenethyl I 100-86-7 95 10 0.2 + NA
3
alcohol

32. α,α-
239
Dimethylphenethyl I 10058-43-2 3 0.4 0.01 − NA
5
formate
 Daily per Annual
Most
capita volum
recen
intakec e in
t
FEM DT (“eaters natural
Flavoring CAS Nr. and annu Consump
A clas only”) ly
ingredient structure al tion ratioe
Nr. sa occurri
volu μg/k ng
me, μg/
g foods,
kgb d
bw/d kgd

33. α,α-
239 51
Dimethylphenethyl I 151-05-3 4700 9 − NA
2 9
acetate

34. α,α-
239 91
Dimethylphenethyl I 10094-34-5 8300 15 − NA
4 7
butyrate

35. α,α-
238 0.0 0.00
Dimethylbenzyl I 7774-60-9 0.4 − NA
8 4 07
isobutyrate

36. 6-
365 0.01
Acetoxydihydrothea II 57893-27-3 5 0.7 − NA
1 17
spirane
 Daily per Annual
Most
capita volum
recen
intakec e in
t
FEM DT (“eaters natural
Flavoring CAS Nr. and annu Consump
A clas only”) ly
ingredient structure al tion ratioe
Nr. sa occurri
volu μg/k ng
me, μg/
g foods,
kgb d
bw/d kgd

37. 6-
354 0.0 0.00
Hydroxydihydrothea II 65620-50-0 0.4 − NA
9 4 07
spirane

38. Dimethylbenzyl 440 0.00

I 93762-34-6 1 0.1 − NA
carbinyl crotonate 3 2

39. Dimethylbenzyl 440 0.0 0.00

I 891781-90-1 0.2 − NA
carbinyl hexanoate 4 3 05

40. Caryophyllene 441 0.01

I 56747-96-7 5 0.7 + NA
alcohol 0 17
 Daily per Annual
Most
capita volum
recen
intakec e in
t
FEM DT (“eaters natural
Flavoring CAS Nr. and annu Consump
A clas only”) ly
ingredient structure al tion ratioe
Nr. sa occurri
volu μg/k ng
me, μg/
g foods,
kgb d
bw/d kgd

449 0.11
41. Cubebol I 23445-02-5 50 7 + NA
7 7

450
42. (–)-Sclareol I 515-03-7 39 6 0.1 + NA
2

450
43. (+)-Cedrol I 77-53-2 10 2 0.03 + NA
3

466
44. α-Bisabolol I 23089-26-1 227 33 0.55 + NA
6
 Daily per Annual
Most
capita volum
recen
intakec e in
t
FEM DT (“eaters natural
Flavoring CAS Nr. and annu Consump
A clas only”) ly
ingredient structure al tion ratioe
Nr. sa occurri
volu μg/k ng
me, μg/
g foods,
kgb d
bw/d kgd

Aliphatic acyclic and alicyclic terpenoid tertiary alcohols and structurally related
substances often have a sweet floral rose to a fruity citrus green organoleptic profile.
The majority impart a balsamic rose or bergamot citrus flavor and are used as flavor
ingredients at average usual use levels of 1 to 10 parts per million (ppm) in a wide
range of food categories, but most commonly in baked goods, beverages, candies,
chewing gum, frozen dairy and gelatins, and puddings. The few with specialized
flavor profiles, such as cool or minty, are used at much higher levels, with average
maximum use levels in excess of 1000 ppm in chewing gum.

Twenty-two of the 44 flavor ingredients in this group have been reported to occur
naturally, and can be found in chamomile, cocoa, coffee, a variety of fruits and
especially citrus fruit varieties and vegetables, lemon juice, black and green teas,
calamus, soybean, pepper, strawberry guava, beer and wine (TNO 2013). For some
substances within this group, quantitative data are available that indicate that their
consumption as naturally occurring constituents of food is far greater than their
intake as flavor ingredients (Stofberg and Grundschober 1987).

Annual volumes of production for each ingredient plus the daily per capita intake
values calculated using MSDI and reported natural occurrence are summarized in
Table 1.

Absorption, Distribution, Metabolism, and Elimination

As described in the previous GRAS affirmation and GRASr group summaries,

aliphatic acyclic and alicyclic terpenoids and related esters undergo efficient
metabolism. Based on the results of studies under a wide variety of conditions,
including aqueous buffered media, simulated gastric juice, simulated human
intestinal fluid, blood plasma, whole hepatocytes and liver microsome preparations,
terpene esters formed from tertiary alcohols (for example, linalool), and simple
aliphatic carboxylic acids are expected to undergo hydrolysis. Although differences
in the rates of hydrolysis occur under in vitro conditions in gastric juice and intestinal
fluids, ready hydrolysis is observed in tissue preparations that have an abundant
concentration of carboxylesterases (CES), especially the liver (Hosokawa 2008;
Fukami and Yokoi 2012). The most important class of these enzymes is the B-
esterases, which are members of the serine esterase superfamily. Generally, CES
enzymes are ubiquitous throughout mammalian tissues and are found at the highest
levels in hepatocytes (Hosokawa 2008; Fukami and Yokoi 2012). CES have been
classified into 5 major subgroups based on their catalytic capabilities. For the present
discussion, the 2 subgroups of interest are CES1, which hydrolyzes esters
composed of a large carboxyl group and a small alcohol group, and CES2 with
typical substrates composed of a small carboxyl group and a large alcohol (Fukami
and Yokoi 2012).

In general, esters are hydrolyzed to their corresponding alcohol and carboxylic acid.
It is expected that the tertiary aromatic alcohols will undergo direct conjugation of the
hydroxyl group with glucuronic acid (Williams 1959), while the tertiary terpenoid
alcohols formed as a result of hydrolysis are rapidly absorbed and converted to the
glucuronic acid conjugates which are excreted in the urine, or are further oxidized to
CO2 that is subsequently expired (Phillips and others 1976; Diliberto and others
1988).

Linalyl esters (Nr. 3 to 10) or those derived from α-terpineol (Nr. 12 to 17) are
expected to be hydrolyzed in humans to yield the parent terpenoid alcohol (linalool
or α-terpineol, respectively) and the corresponding saturated aliphatic carboxylic
acid. Methyl 1-acetoxycyclohexyl ketone (Nr. 23) is expected to hydrolyze to acetic
acid and methyl 1-hydroxycyclohexyl ketone. Similarly it is expected that esters of
α,α-dimethylphenethyl alcohol (Nr. 32 to 35, 38 and 39) will be rapidly hydrolyzed to
yield α,α-dimethylphenethyl alcohol and the corresponding acid.

Linalyl acetate was facilely hydrolyzed in water and simulated gastric and pancreatic
fluids in an in vitro hydrolysis study. The mean half-lives for linalyl acetate hydrolysis
were 5.5 and 52.5 min in gastric and pancreatic fluids, respectively (Hall 1979). In
neutral gastric juice, linalyl acetate is slowly hydrolyzed (t½ = 121 min) to a mixture
of linalool and the ring-closed isomer α-terpineol (see Figure 1). In simulated gastric
juice, linalyl acetate is rapidly hydrolyzed (t½ < 5 min) to yield linalool, and this rapidly
rearranges into α-terpineol (Hall 1979; Buck and Renwick 1998a). However, when
incubated with intestinal fluid in the presence and absence of pancreatin, linalyl
acetate was slowly hydrolyzed (t½ = 153 to 198 min) (Buck and Renwick 1998b).
Hydrolysis studies have shown that the hydrolysis of linalyl acetate is slower when
incubated with homogenates of rat intestinal mucosa, blood, and liver (rate constants
k = 0.01 to 0.0055/min) compared to the hydrolysis rate in acidic gastric juice (k >
5/min). Based on the in vitro hydrolysis data, it can be concluded that linalyl acetate
and other linalyl esters are hydrolyzed in gastric juice to yield linalool, which
undergoes rapid ring-closure to yield α-terpineol. Both linalool and α-terpineol may
then be either conjugated and excreted or oxidized to more polar excretable
metabolites.
Figure 1.

 Open in figure viewer

 Download Powerpoint slide

Metabolism of linalool in mammals.

In humans and animals, the principle route of elimination for tertiary alcohols is
through formation of glucuronic acid conjugates and excretion in the urine and feces
(Williams 1959; Parke and others 1969, 1974b; Horning and others 1976; Ventura
and others 1985). Unsaturated terpenoid tertiary alcohols have the potential to
undergo allylic oxidation to form polar diol metabolites, which may be excreted either
free or conjugated. Those diols containing a primary alcohol function may undergo
further oxidation to the corresponding carboxylic acid (Horning and others 1976;
Ventura and others 1985; Madyastha and Srivatsan 1988).

The metabolic fate of the aliphatic tertiary alcohol linalool (Nr. 1) has been studied in
mammals (Figure 1). Linalool undergoes rapid oxidation (t½ for linalool = 11 min) by
CYP-450 in a rat liver homogenate metabolic activation system (with added NADP
and glucose-6-phosphate) (Buck and Renwick 1998c). In an earlier study, male
Wistar rats orally administered a single dose of 500 mg/kg body weight (bw) 14C-
linalool excreted 55% of the radiolabel in the urine as the glucuronic acid conjugate,
while 23% was excreted as CO2 in expired air, and 15% was excreted in the feces
within 72 h of dose administration. After 72 h nearly all of the radiolabel had cleared
the body with only 3% of the radioactivity detected in tissues (Parke and others
1974a). Reduction metabolites such as dihydro- and tetrahydrolinalool were also
detected in significant amounts in the urine, either free or as their conjugates, after
administration of a single dose of linalool to rats, although further details of this study
have not been reported (Rahman 1974) (see Figure 1).

Weak induction of cytochrome P-450 (CYP-450) activity has been reported at

relatively high doses of linalool. In a repeat-dose study, male rats (IISc strain) were
administered 800 mg/kg bw/d of linalool for 20 d. A notable increase in allylic
oxidation products 8-hydroxylinalool and 8-carboxylinalool was seen in the urine. A
statistically significant but transient induction of 50% in CYP-450 activity in the liver
microsomes was also noted after 3 d of treatment, followed by a return to control
values after 6 d (Chadha and Madyastha 1984). In a similar study, induction of CYP-
450 was not observed until the 30th d of administration of 500 mg linalool/kg bw/d to
Wistar rats (Parke and others 1974b).

The data suggest that glucuronic acid conjugation and excretion is the primary route
of metabolism of linalool. Allylic oxidation becomes an important pathway only after
repeated dosing at relatively high doses. It has been suggested that the
biotransformation of the diol metabolite of linalool to the corresponding aldehyde via
the action of the NAD+-dependent enzyme alcohol dehydrogenase is inhibited due
to the bulky nature of the neighboring alkyl substituents and the substrate specificity
of the enzyme (Eder and others 1982).

In a repeat-dose study, male albino rats (IISc strain) were administered 600 mg/kg
bw/d of α-terpineol for 20 d. Metabolite analysis showed oxidation at the allylic methyl
group to the corresponding carboxylic acid, of which a small fraction was
hydrogenated to yield the corresponding saturated carboxylic acid (Madyastha and
Srivatsan 1988) (see Figure 2). In addition, α-terpineol resulted in a statistically
significant induction of liver microsomal CYP-450 content, with 52% to 104%
increase over the content in untreated rats that peaked on day 2 of treatment, and
in a moderate increase (up to 43%, on day 2) in the activity of NADPH-cytochrome
c reductase in treated rats (Madyastha and Srivatsan 1988), suggesting that
oxidation is mediated by CYP-450.
Figure 2.

 Open in figure viewer

 Download Powerpoint slide

Metabolism of α-terpineol in rats.

In a minor pathway, epoxidation of the alkene of α-terpineol is followed by hydrolysis

to a triol metabolite, 1,2,8-trihydroxy-p-menthane, which also has been reported in
humans following ingestion of a pine oil disinfectant containing α-terpineol (Horning
and others 1976). It is expected that α-terpineol would undergo metabolism like
linalool (Chadha and Madyastha 1984), primarily by glucuronic acid conjugation and
excretion in the urine.

Bicyclic tertiary alcohols (Nr. 21, 22, 40, 42, and 43) are relatively stable in vivo, but
are eventually conjugated with glucuronic acid and excreted (Williams 1959). In
rabbits the structurally related bicyclic tertiary alcohol β–santenol (2,3,7-trimethyl
bicyclo[2.2.1]-heptan-2-ol) was conjugated with glucuronic acid (Williams 1959).

In a metabolism study using the structurally related terpenoid tertiary alcohol trans-
sobrerol in humans, dogs, and rats, 10 metabolites were isolated in urine, 8 of which
were also detected in urine from humans exposed to trans-sobrerol. In this study,
the 2 principle modes of metabolism observed were allylic oxidation of the ring
positions and alkyl substituents, and conjugation of the tertiary alcohol functions with
glucuronic acid. These are common pathways converting tertiary (Ventura and
others 1985) and secondary (Yamaguchi and others 1994) terpenoid alcohols to
polar metabolites that are easily excreted in the urine and feces (see Figure 3).

Figure 3.

 Open in figure viewer

 Download Powerpoint slide

Metabolism of trans-sobrerol in rats, dogs, and humans.

Potential pharmacokinetic outcomes in mammalian plasma were analyzed upon

intravenous administration of sclareol (Nr. 42) at 100 mg/kg bw to 2 male Wistar rats.
Plasma samples were collected at 5, 15, 30, 60, 180, 360 min, 12 and 24 h
postadministration. At 5 min post-injection, plasma levels of sclareol were 84.9
μg/mL. The sclareol plasma concentration dropped to 42.9 μg/mL after 180 min and
sclareol was not detectable at 360 min. The data indicate a rapid biphasic
disappearance of sclareol from plasma following intravenous dosing. The authors
suggest that sclareol may be distributed in fatty tissue due to its high lipophilicity
(Kouzi and others 1993).
In a 2nd study, sclareol was administered to 2 male Wistar rats by intravenous
injection at 100 mg/kg bw, and to male Wistar rats (n = unknown) by intragastric
instillation at 1000 mg/kg bw. Urine and fecal samples were collected from all rats
and bile samples were collected only from rats given intravenous injections. No
unchanged sclareol was detected in urine (with or without β-glucuronidase
treatment), or in fecal samples (only 9% of the initial dose was found in fecal samples
of rats treated orally) and very low levels (0.02% over a 3-h period) of sclareol were
found in bile samples from rats following intravenous injection. Very low levels
(0.04%) of oxidized metabolites were found in bile only after 3 h, including 3-α-
hydroxysclareol (0.24%), 3-β-hydroxysclareol (0.075%), 18-hydroxysclareol
(0.056%), and 3-ketosclareol (0.03%). The low sensitivity of the assay and possible
other metabolites were suggested as reasons for the very small percentage of
intravenously injected sclareol (<0.05%) that was accounted for (Kouzi and others
1993).

In order to determine the phase I metabolism of cedrol (Nr. 43), Bang and Ourrison
(1975) and Trifilieff and others (1975) administered cedrol to rabbits and dogs,
respectively. In rabbits, only metabolites formed by hydroxylation at C3 were
identified, and the 2 resulting epimers, α-epiisobiotol and α-isobiotol, were
subsequently dehydrated at C7/C8. The metabolic fate in dogs was far less
regioselective and metabolites that were identified were products of oxidation at
multiple locations on the ring as well as at methyl groups. Additional Phase I
functionalization must occur in humans, given that the initial glucuronidation (in
Phase II) at the hydroxyl moiety of cedrol is inefficient because of steric hindrance.
In the case of cedrol, the hydroxylation of a nonactivated saturated carbon atom is
the most likely pathway (Bang and Ourisson 1975; Trifilieff and others 1975; Ishida
2005).

Terpenoids are known to alter the activity of various drug-metabolizing hepatic

enzymes (Parke and Rahman 1969). The inhibitory effect of chamomile essential oil
and its major constituents (for example, α-bisabolol, Nr. 44) was studied on 4
selected human CYP-450 enzymes (CYP1A2, CYP2C9, CYP2D6, and CYP3A4).
Increasing concentrations of the test compounds were incubated with individual,
recombinant CYP isoforms and their effect on the conversion of surrogate
substances was measured fluorometrically; enzyme inhibition was expressed as
median inhibitory concentration (IC50) and inhibition constant (Ki) values in relation
to positive controls. α-Bisabolol (IC50 = 2.18 μM) produced a significant inhibition of
CYP2C9 and CYP2D6. As indicated by these in vitro data, chamomile preparations
contain constituents that can inhibit the activities of major human drug metabolizing
enzymes; interactions with drugs whose route of elimination is mainly through CYP
oxidation (especially CYP1A2) are therefore possible (Ganzera and others 2006).

Studies in humans, dogs, rabbits, and rats have shown that absorption of aliphatic
acyclic and alicyclic terpenoid tertiary increases with increasing lipophilicity and are
distributed primarily in adipose tissue. Oxidation to polar metabolites and/or
conjugation with glucuronic acid, followed by excretion in the urine is expected for
all of the flavoring ingredients in this group. Small amounts may be expired in
exhaled air. The esters within this group (Nr. 3 to 10, 12 to 17, 23, 33 to 37, 39 and
40) are expected to be hydrolyzed in humans to their component tertiary alcohols
and acids. The available data demonstrate that the aliphatic acyclic and alicyclic
terpenoid tertiary alcohols and structurally related substances are rapidly absorbed,
distributed, metabolized and excreted.

Toxicology

Acute toxicity

Oral median lethal dose (LD50) values have been reported for 24 of the 43
substances in this group (Jenner and others 1964; Colaianni 1967; Moreno 1971,
1973, 1975, 1976, 1977, 1982; Russell 1973; Levenstein 1975; Griffiths 1979;
Piccirillo and Hartman 1980; Yamahara and others 1985; Collinson 1989; Rhone-
Poulenc 1992; Moore 2000). LD50 values range from 1300 to greater than 36300
mg/kg bw, demonstrating that the oral acute toxicity of tertiary alcohols and related
esters is extremely low.

Short-term studies of toxicity

The results of short-term studies with representative acyclic and alicyclic tertiary
terpenoid alcohols and related substances are summarized in Table 2 and key
studies are described below.

Table 2. Short-term toxicity studies on aliphatic acyclic and alicyclic terpenoid

tertiary alcohols and structurally related substances

Nr. test NOAE

groups/ Durati L Margin
Specie Referenc
Flavor ingredient Nr. per Route on (mg/k of
s, sexa e
test (days) g safety
groupb bw/d)

1. a

M, male; F, female.

2. b

Total number of test groups does not include control animals. Total number
per test group includes both male and female animals.

3. c

Not specified.
 Nr. test NOAE
groups/ Durati L Margin
Specie Referenc
Flavor ingredient Nr. per Route on (mg/k of
s, sexa e
test (days) g safety
groupb bw/d)

4. d

Administered with citronellol as part of a 1:1 mixture.

5. e

The study was performed at a single dose level or multiple dose levels that
produced no adverse effects and, therefore, a NOAEL was not determined.
The NOAEL is probably higher than the dose level reported here which is the
highest dose level that produced no adverse effects.

6. f

Refers to maternal NOAEL derived from a developmental study.

7. g

Administered with linalyl isobutyrate and geranyl acetate as part of a mixture.

8. h

Administered with linalyl acetate and geranyl acetate as part of a mixture.

9. i

Based on a dietary concentration of 10000 ppm, which is equivalent to 1000

mg/kg bw.

10. j

Based on a dietary concentration of 1000 ppm, which is equivalent to 100

mg/kg bw.

11. k

Adverse effects seen in lowest dosage group; therefore, this is not a true
NOAEL.

12. l
 Nr. test NOAE
groups/ Durati L Margin
Specie Referenc
Flavor ingredient Nr. per Route on (mg/k of
s, sexa e
test (days) g safety
groupb bw/d)

Compared to controls, a significant increase was observed in relative liver

weights in males; however, this organ weight variation was not accompanied
by any histopathological abnormalities or any other adverse effects.
Rat; M, (Oser
1. Linalool 1/–c Dietd 84 50d, e >500
F 1967)

Gava (Lewis
1. Linalool Rat, F 3/25 11 500f >5000
ge 2006)

Rat; M, (Oser
4. Linalyl acetate 1/–c Dietg 84 24e 600
F 1967)

Rat; M, (Oser
7. Linalyl isobutyrate 1/–c Dieth 84 27e 270000
F 1967)

(Hagan
Rat; M, 250000 and
13. Terpinyl acetate 3/20 Diet 140 500e
F 0 others
1967)

24. 2,3,4-Trimethyl-3- Rat; M, 100000 (Madaras

1/10 Diet 14 10e
pentanol F 0 z 1997)

(Hagan
31. α,α-
Rat; M, 1000c 500000 and
Dimethylphenethyl 1/10 Diet 112
F ,i 0 others
alcohol
1967)

(Hagan
31. α,α-
Rat; M, and
Dimethylphenethyl 1/20 Diet 196 100c, j 500000
F others
alcohol
1967)

36. 6-
Rat; M, Gava (Griffiths
Acetoxydihydrotheaspi 1/32 91 3c 256000
F ge 1979)
rane
 Nr. test NOAE
groups/ Durati L Margin
Specie Referenc
Flavor ingredient Nr. per Route on (mg/k of
s, sexa e
test (days) g safety
groupb bw/d)

37. 6-
Rat; M, Gava 0.154 (Griffiths
Hydroxydihydrotheasp 1/32 91 220000
F ge c 1976)
irane

Rat; M, Gava (Merkel

42. (–)-Sclareol 1/20 28 8.8c 88000
F ge 2006)

Rat; M, Gava (Merkel

43. (+)-Cedrol 1/20 28 8.4c 280000
F ge 2006)

(Habersa
Rat; M, Gava <1860 310000 ng and
44. α-Bisabolol 2/40 28
F ge k 0 others
1979)

(Habersa
Dog; Gava <1860 3100,0 ng and
44. α-Bisabolol 2/6 28
M, F ge k, l 00 others
1979)

(Habersa
Gava ng and
44. α-Bisabolol Rat; F 4/–c 10 931f >10000
ge others
1979)

(Habersa
Rabbit; Gava ng and
44. α-Bisabolol 4/–c 13 931f >10000
F ge others
1979)

A mixture of linalool (Nr. 1.) and citronellol (1:1) resulting in average daily intake of
50 mg/kg bw each, and a mixture of linalyl acetate (Nr. 4), linalyl isobutyrate (Nr. 7),
and geranyl acetate at levels calculated to result in average daily intakes of 24, 27,
or 48 mg/kg bw, respectively were incorporated into the feed of male and female rats
(number and strain not specified) for 12 wk. A slight retardation of body weight gain
was observed only in males fed the linalool mixture and in females fed the linalyl
esters mixture. However, these effects were concluded by the authors to be
biologically insignificant (Oser 1967).
Four groups of 10 male and 10 female Osbourne–Mendel rats were fed the
structurally related ester of linalool, linalyl cinnamate, at dietary concentrations of up
to 10000 ppm for 17 wk, which is estimated to provide a daily intake of up to 1000
mg/kg bw/d (FDA 1993). There were no differences between treated and control
animals in the parameters evaluated (Hagan and others 1967).

Similarly, groups of 10 male and 10 female weanling Osborne–Mendel rats were fed
terpinyl acetate (Nr. 13) in the diet for 20 wk at concentrations of 0, 1000, 2500 or
10000 ppm, calculated to result in daily intakes of 0, 100, 250, and 1000 mg/kg bw/d,
respectively (FDA 1993). All animals were examined for growth, hematology, and
macroscopic and microscopic changes in the tissues. No statistically significant
adverse effects were reported at any dose level (Hagan and others 1967).

In the same study, groups of 5 weanling Osborne–Mendel rats per sex were housed
individually in wire cages and administered 0 or 10000 ppm of α,α-dimethylphenethyl
alcohol (Nr. 31) in the diet (approximately 1000 mg/kg bw/d) for 16 wk. In addition,
groups of 10 rats per sex were administered 0 or 1000 ppm of α,α-dimethylphenethyl
alcohol in the diet (approximately 100 mg/kg bw/d) for 28 wk. No effects were
observed due to the administration of the test material at either dose level (Hagan
and others 1967).

Groups of Sprague–Dawley rats (5/sex) were maintained on diets containing 2,3,4-

trimethyl-3-pentanol (Nr. 24) at a dose level of 0 or 10 mg/kg bw/d for 14 d. Test and
control groups showed no significant differences in relative or absolute kidney and
liver weights (Madarasz 1997).

During its 51st meeting, the Joint FAO/WHO Expert Committee on Food Additives
concluded that methyl 1-acetoxycyclohexylketone (Nr. 23) was not supported by a
relevant no-observed-effect level (NOEL) (JECFA 2000). Data on the structurally
related substance, 6-acetoxydihydrotheaspirane (Nr. 36) in a 90-d study in rats
(Griffiths 1979), provides a NOAEL of 3 mg/kg bw/d which is at least 100000 times
the daily per capita intake of methyl 1-acetoxycyclohexyl ketone when used as a
flavor ingredient.

In studies using groups of 16 Sprague–Dawley rats per sex per dose, 6-

acetoxydihydrotheaspirane (Nr. 36) and 6-hydroxydihydrotheaspirane (Nr. 37) were
administered by gavage at doses of 3 mg/kg bw/d (Griffiths 1979) and 0.154 mg/kg
bw/d, respectively (Griffiths 1976) for 91 d. No treatment-related clinical effects were
observed in either study. In the acetoxydihydrotheaspirane study, treated females
were reported to have significantly decreased body weights compared to controls at
week 13. In comparison to controls, a significant increase in relative spleen weight
was observed in treated males. Significant reductions in hemoglobin, hematocrit,
and red blood cell count were observed in treated females at the end of the study
relative to levels observed in controls. In the 6-hydroxydihydrotheaspirane study, 1
male death that occurred within 1 h of dosing and necropsy revealed oil in the lungs
and trachea but not in the upper part of the gastrointestinal tract, indicating a gavage
error. At week 12, treated males had significantly greater hemoglobin concentrations
relative to controls. At week 6, but not at week 12, treated males and females had
significantly greater total leukocyte counts compared to the controls (Griffiths 1976).
These hematological changes were considered variations and not adverse effects
because of the lack of reproducibility between sexes or a later time point,
respectively. No other statistically significant adverse effects were reported (Griffiths
1976, 1979).

In a 28 d oral toxicity study, Sprague–Dawley rats (10/sex/ingredient) approximately

1 mo of age, were administered 10 mg/kg bw/d of sclareol (98.3% pure, Nr. 42) or
10 mg/kg bw/d of cedrol (97% pure, Nr. 43) by intragastric instillation in
carboxymethylcellulose with the control group receiving vehicle only. It was
determined that the actual dose achieved was 8.8 mg/kg bw/d of sclareol and 8.4
mg/kg bw/d of cedrol. In rats dosed with sclareol, statistically significant increases in
mean absolute and relative liver weight in males and liver to brain weight ratio in
females were found not to correlate with evidence of macro- or microscopic
alterations or associated enzyme activity. Minor clinical pathology and
histopathological alterations were considered incidental, and therefore
toxicologically nonadverse, according to the authors. Other statistically significant
organ weight changes were isolated and not consistent between the sexes, and were
therefore considered not to be related to administration of the test substance (Merkel
2006).

Groups of 40 (strain not specified) rats (20/sex/dose) or Beagle dogs (3/sex/dose)

were administered α-bisabolol (Nr. 44) by intragastric instillation at 0, 2.0, or 3.0
mL/kg bw/d for 4 wk. These dose levels correspond to calculated daily intakes of 0,
1860 or 2790 mg/kg bw/d. In the rats, slight motor agitation and decreased body
weight gain were noted in animals of both dose groups and the mortality rate was
20% in the 2790 mg/kg bw/d group. Decreased body weight gain was noted.
Postmortem findings in the 2790 mg/kg bw/d group included inflammatory changes
in the liver, trachea, spleen, thymus, and stomach (Habersang and others 1979). In
the Beagle dog study, the high dose was increased to 3720 mg/kg bw/d after 2 wk.
Loss of appetite, reduced feed intake, and vomiting were observed in 2 of the 6 dogs
receiving 1860 mg/kg bw/d. At necropsy it was noted that the liver weight relative to
the body weight was increased. Reactions and observations were more severe in
the high dose group. No other changes were noted as compared to controls
(Habersang and others 1979).

Developmental toxicity studies

Linalool was studied for potential developmental toxicity in rats. Four groups of
presumed pregnant female Sprague–Dawley rats (25/group) were administered 0
(vehicle), 250, 500, or 1000 mg linalool/kg bw/d by intragastric instillation in corn oil
on gestational days 7 to 17. There were no clinical signs of toxicity. No deaths related
to the administration of linalool occurred. One dam in the low dose delivered early
on the day of scheduled sacrifice. This delivery was considered unrelated to
administration of linalool because the observation was not dose-dependent. There
were no macroscopic lesions at necropsy in this dam and her litter consisted of 12
live-delivered pups and 1 early in utero resorption. All pups appeared normal for this
dam and no external, soft tissue or skeletal alterations occurred in these pups. Body
weight gains were reduced by 11% in the 1000 mg/kg bw/d dose group during the
dosing period (in comparison with control) and this was accompanied by absolute
and relative feed consumption values that were reduced and significantly reduced
(by 7%), respectively, in the 1000 mg/kg bw/d dose group for the entire test period
in comparison to the control group. The lower dose groups showed no difference in
body weight gain or feed consumption when compared to controls. No Caesarean
section or litter parameters were affected by dosages of linalool as high as 1000
mg/kg bw/d. The litter averages for corpora lutea, implantations, litter sizes, live
fetuses, early and late resorptions, percent resorbed conceptuses, percent live male
fetuses and fetal body weights were similar across all test groups. No dam had a
litter that consisted of all resorbed conceptuses. All placentas appeared normal. No
macroscopic external, soft tissue or skeletal alterations appeared to be caused by
dosages of linalool as high as 1000 mg/kg bw/d. There were no dose-dependent or
significant differences in the litter or fetal incidences of any macroscopic external,
soft tissues or skeletal alterations. All ossification averages were comparable to
vehicle control group values and did not significantly differ among the groups.

Based on these data, the maternal NOAEL of linalool is 500mg/kg bw/d. The 1000
mg/kg bw/d dose caused nonsignificant reductions in body weight gain and also
reduced absolute and relative feed consumption values during the dosage period.
However, following the completion of the dosing period, these effects were reversed.
The developmental NOAEL is at least 1000 mg/kg bw/d, since no embryo–fetal
effects were observed at the highest dose tested (Lewis 2006).

A study was conducted in Sprague–Dawley and Wistar rats and in New Zealand
white rabbits to determine the potential developmental toxicity of α-bisabolol (Nr. 44).
In 2 separate experiments, groups of presumed pregnant rats or presumed pregnant
rabbits were administered 0, 0.25, 0.50 or 1.0 mL/kg in the 1st experiment, and 3.0
mL/kg bw/d of α-bisabolol in the 2nd, by intragastric instillation on days 6 to 15 of
pregnancy for rats and days 6 to 18 for rabbits. These dose levels correspond to
calculated daily intakes of 0, 233, 466, 931 and 2793 mg/kg bw/d. At the highest
dose of the test substance administered, there were reduced numbers of live fetuses
in addition to reduced fetal weights and increased numbers of resorptions in both
rats and rabbits. No adverse effects were observed in any other dose group. The
NOAEL for reproductive effects in Sprague–Dawley and Wistar rats or New Zealand
white rabbits is 931 mg/kg bw/d (Habersang and others 1979).

Genotoxicity

Genotoxicity testing has been performed on substances in this group. The results of
these tests are summarized in Table 3 and are described below.

Table 3. Studies of genotoxicity with aliphatic acyclic and alicyclic terpenoid tertiary
alcohols and structurally related substances used as flavoring ingredients
In vitro

Nr Flavoring Concentrati
End point Test object Results Reference
. agent on

1. a

With and without metabolic bioactivation system, S-9.

2. b

With metabolic bioactivation system, S-9.

3. c

Without metabolic bioactivation system, S-9.

4. d

The test material was bacteriotoxic toward all strains at greater than or equal
to 500 and 160 μg/plate with and without S-9 activation.

5. e

The test material was bacteriotoxic toward all strains at doses greater than
or equal to 500 and 150 μg/plate with and without S-9 activation.

6. f

Standard plate incorporation method.

7. g

The test material was bacteriotoxic toward all strains at greater than 200 and
100 μg/plate with and without S-9 activation.

8. h

Administered via gavage.

9. i

Calculated using molecular weight of 1-phenyl-3-methyl-3-pentanol =

178.28.
In vitro

Nr Flavoring Concentrati
End point Test object Results Reference
. agent on

0.01 to 3 μL/2
Salmonella (Eder and
Reverse mL Negative
1 Linalool typhimuriu others
mutation incubation a
m. TA100 1980)
volume

Salmonella
typhimuriu
m TA92, (Ishidate
Reverse Negative
1 Linalool TA100, 1 mg/plate and others
mutation a
TA1535, 1984)
TA1537,
TA94, TA98

Salmonella
(Rockwell
Reverse typhimuriu 0.05 to 100 Negative
1 Linalool and Raw
mutation m TA98, μL a
1979)
TA100

Salmonella
typhimuriu
m TA100, (Heck and
Reverse 10000 Negative
1 Linalool TA1535, others
mutation nL/plate a
TA1537, 1989)
TA1538,
TA98

Escherichia
Reverse 0.125 to 1
1 Linalool coli WP2 Negative (Yoo 1986)
mutation mg/plate
uvrA

3.9 to 125
Mouse
nL/mL 100 to Negative (Heck and
Forward lymphoma
1 Linalool 300 nL/mL b Weakly others
mutation L5178Y
200 nL/mL 25 Positivec 1989)
TK+/–
to 150 nL/mL

Unschedule Rat (Heck and

50 nL/mL
1 Linalool d DNA hepatocyte Negative others
43.1 μg/mL
synthesis s 1989)
In vitro

Nr Flavoring Concentrati
End point Test object Results Reference
. agent on

Chinese (Ishidate
Chromosom Negative
1 Linalool hamster 250 μg/mL and others
al aberration c
fibroblasts 1984)

Bacillus
(Oda and
subtilis H17
1 Linalool Mutation 17 μg Negative others
(rec+) and
1978)
M45 (rec–)

Bacillus
subtilis H17 10 μL/disk
1 Linalool Mutation Positive (Yoo 1986)
(rec+) and 8620 μg/disc
M45 (rec–)

Salmonella
typhimuriu
m TA1535, (Heck and
Reverse 25000 Negative
4 Linalyl acetate TA1537, others
mutation nL/plate a
TA1538, 1989)
TA98,
TA100

Fischer or
Unschedule (Heck and
SD rat
4 Linalyl acetate d DNA 300 nL/mL Negative others
hepatocyte
synthesis 1989)
s

Bacillus
(Oda and
subtilis H17
4 Linalyl acetate Mutation 18 μg Negative others
(rec+) and
1978)
M45 (rec–)

Human
peripheral
Chromosom (Bertens
4 Linalyl acetate blood 180 μg/mL Negative
al aberration 2000)
lymphocyte
s

Linalyl Reverse Salmonella 10 to 5000 Negative (Sokolowsk

5
propionate mutation typhimuriu μg/plate i 2004)
In vitro

Nr Flavoring Concentrati
End point Test object Results Reference
. agent on

m TA1535,
TA1537,
TA1538,
TA98,
TA100

Salmonella
typhimuriu
m TA1535, (Heck and
Reverse 10000 Negative
11 α-Terpineol TA1537, others
mutation μg/plate a
TA1538, 1989)
TA98,
TA100

Salmonella
typhimuriu
(Florin and
Reverse m TA1535, Up to 3 Negative
11 α-Terpineol others
mutation TA1537, μmol/plate a
1980)
TA98,
TA100

Mouse Negative
(Heck and
Forward lymphoma 250 nL/mL b
11 α-Terpineol others
mutation L5178Y 300 nL/mL Negative
1989)
TK+/– c

Bacillus
(Oda and
subtilis H17
13 Terpinyl acetate Mutation 19 μg Negative others
(rec+) and
1978)
M45 (rec–)

Salmonella
(Rockwell
p-Menth-8-en-1- Reverse typhimuriu 0.05 to 100 Negative
20 and Raw
ol (β-terpineol) mutation m TA100, μL a
1979)
TA98

(Oda and
Bacillus Negative
Terpineol Mutation 19 μg others
subtilis H17 a
1978)
In vitro

Nr Flavoring Concentrati
End point Test object Results Reference
. agent on

(rec+) and
M45 (rec–)

Salmonella
typhimuriu
1-Phenyl-3- m TA1535, (Wild and
Reverse Up to 3600 Negative
28 methyl-3- TA1537, others
mutation μg/plate a
pentanol TA1538, 1983)
TA98, and
TA100

Salmonella
typhimuriu
α, α-
Reverse m TA1535, Up to 1000 Negative (Asquith
32 Dimethylphenet
mutation TA1537, μg/plate a 1989)
hyl formate
TA98, and
TA100

Salmonella
typhimuriu 0, 1.6, 5, 16,
m TA1535, 50, 160, 500,
Reverse Negative (Scheerbau
43 (+)–Cedrol TA97a, 1600, and
mutation a, d m 2001)
TA98, 5000
TA100 and μg/plate
TA102

Salmonella
typhimuriu
0, 1.5, 5, 15,
m TA98,
Reverse 50, 100, 150, Negative
44 α-Bisabolol TA100, (King 2002)
mutation and 500 a, e, f
TA102,
μg/plate
TA1535, TA
1537

Salmonella 0, 1, 5, 10, (Gomes-

Reverse typhimuriu 25, 50, 100, Negative Carneiro
44 α-Bisabolol
mutation m TA100, 200, 300, and a, d, g and others
TA97a, 400 μg/plate 2005)
In vitro

Nr Flavoring Concentrati
End point Test object Results Reference
. agent on

TA98 and
TA1535

In vivo

Male and
0, 500, 1000,
Micronucleu female mice (Meerts
1 Linalool and 1500 Negative
s assay (Swiss CD- 2001)
mg/kg bwh
1)

Sex-linked
1-Phenyl-3- recessive Drosophila 0 or 20 mM (Wild and
28 methyl-3- lethal melanogast (3565 Negative others
pentanol mutation er μg/mL)i 1983)
(Basc test)

1-Phenyl-3- 0, 357, 624, (Wild and

Micronucleu
28 methyl-3- NMRI mice 891, or 1416 Negative others
s induction
pentanol mg/kg bw 1983)

In Vitro

No increase in reverse mutations were observed in Ames assays in Salmonella

typhimurium strains TA98, TA100, TA102, TA1535, TA1537, TA1538, TA97a, TA92,
and TA94 and Escherichia coli WP2 uvrA for linalool (Nr. 1), linalyl acetate (Nr. 4),
linalyl propionate (Nr. 5), α-terpineol (Nr. 11), p-menth-8-en-1-olβ-terpineol (Nr. 20),
1-phenyl-3-methyl-3-pentanol (Nr. 28), cedrol (Nr. 43), α,α-dimethylphenethyl
formate (Nr. 32), and α-bisabolol (Nr. 44) at concentrations up to 5000 μg/plate in
the presence and absence of bioactivation systems (Rockwell and Raw 1979; Eder
and others 1980; Florin and others 1980; Wild and others 1983; Ishidate and others
1984; Yoo 1986; Asquith 1989; Heck and others 1989; Scheerbaum 2001; King
2002; Sokolowski 2004; Gomes-Carneiro and others 2005). These results are
summarized in detail in Table 3.

When incubated with Bacillus subtilis H17 (rec+) and M45 (rec–), linalool, linalyl
acetate, and terpinyl acetate were negative in the rec assay at 17, 18, and 19 μg,
respectively (Oda and others 1978), but linalool was positive at 10 μL/disk,
corresponding to 8620 μg (56 μmol) (Yoo 1986). The positive rec assay results
reported by Yoo were most likely due to cytotoxicity to B. subtilis rather than
genotoxicity, given the high concentration.
In a mouse lymphoma assay, linalool was negative in L5178Y TK+/– cells with S-9
metabolic activation at 200 nL/mL, but weakly positive without S-9 activation at 150
nL/mL (Heck and others 1989). The authors of this study noted that the forward
mutation assay was not run under conditions controlling for changes in osmolality
and they interpreted the weak positive result not as evidence of genotoxicity but
rather related to cytotoxicity (Heck and others 1989). Negative results were obtained
in this assay when cells were treated with 250 nL/mL or 300 nL/mL of α-terpineol,
with and without S-9 metabolic activation, respectively (Heck and others 1989).

Linalool did not induce chromosomal aberrations when incubated with Chinese
hamster fibroblast cells at a maximum concentration of 250 μg/mL (Ishidate and
others 1984), nor did it induce unscheduled deoxyribonucleic acid (DNA) synthesis
(UDS) in rat hepatocytes at concentrations up to 50 nL/mL (equivalent to 43.1 μg/mL)
(Heck and others 1989). No chromosomal aberrations were observed when 0, 10,
33, 56, 100, 130, or 180 μg/mL of linalyl acetate (Nr. 4) were incubated with human
peripheral lymphocytes for 3 h with 24- and 48-h fixation times, with and without
metabolic activation (Bertens 2000). Linalyl acetate did not induce UDS in Fischer
or SD rat hepatocytes at concentrations up to 300 nL/mL (Heck and others 1989).

In Vivo

The potential of 1-phenyl-3-methyl-3-pentanol (Nr. 28) to induce sex-linked

recessive lethal mutations in adult Drosophila melanogaster was studied in the Basc
test. Mutation frequency was unaffected when flies were exposed to a 0 or 20 mM
(3565 μg/mL) solution of 1-phenyl-3-methyl-3-pentanol for 3 d (Wild and others
1983).

In a mutagenicity study, 24-h urine of Sprague–Dawley rats administered 0.5 mL of

linalool or β-terpineol was incubated with S. typhimurium strains TA98 and TA100
(Rockwell and Raw 1979). The following samples were tested: undiluted linalool and
α-terpineol (with metabolic activation), aliquots of 24-h urine from rats fed linalool
and β-terpineol (with and without metabolic activation), aliquots of the ether extract
of 24-h urine (with and without metabolic activation), and aliquots of the aqueous
phase of 24-h urine ether extract (with and without metabolic activation). A range of
volumes was tested for each type of sample. Urines were diluted to 60 mL and
incubated in the presence of chloroform and β-glucuronidase prior to ether
extraction. All tests were negative except for the ether extract of urine from α-
terpineol-treated rats.

Linalool (Nr. 1) was tested for in vivo mutagenicity in the bone marrow micronucleus
assay in Swiss CD-1 mice. Four groups (5/sex/group) received a single dose of the
test substance by intragastric instillation in corn oil; 2 of these groups were
administered 1500 mg/kg bw of linalool and 1 group each was administered 500 or
1000 mg/kg bw of linalool. Vehicle controls received corn oil and positive controls
received cyclophosphamide. Systemic toxicity signs were recorded at least once a
day. The animals were terminated at 24 or 48 h after dosing, both femurs were
removed, and bone marrow smears were prepared and analyzed for micronuclei.
The animals of the groups dosed with linalool showed no decrease in the ratio of
polychromatic to normochromatic erythrocytes, which reflects a lack of toxic effects
of this compound on erythropoiesis. No increase in the frequency of micronucleated
polychromatic erythrocytes was found in the linalool-dosed animals compared to the
vehicle controls, whereas cyclophosphamide produced an increase. Therefore,
linalool was not mutagenic in the micronucleus test under the experimental
conditions used (Meerts 2001).

In a micronucleus test, groups of 4 male and female NMRI mice (number/sex not
reported), administered single intraperitoneal doses of 0, 357, 624, 891, or 1416 mg
1-phenyl-3-methyl-3-pentanol (Nr. 27)/kg bw, demonstrated no increase in
micronucleated erythrocytes in bone marrow samples obtained 30 h
postadministration (Wild and others 1983).

Conclusion for Genotoxicity

The testing of these representative materials in vitro in bacterial test systems (Ames
assay) and in vivo in mammalian test systems (micronucleus assay) showed no
evidence of mutagenic or genotoxic potential. These results are further supported
by the lack of positive findings in the Basc test in D. melanogaster. The 2 positive
results reported in the literature are questionable at best and most likely attributable
to cytotoxic mechanisms rather than genotoxic mechanisms. The B. subtilis rec
Assay has since been superseded by the OECD-validated S. typhimurium/E. coli
reverse mutation assay in the modern genotoxicity testing battery.

Conclusion

The group of aliphatic acyclic and alicyclic terpenoid tertiary alcohols and structurally
related substances discussed here was determined to be GRAS under conditions of
intended use as flavor ingredients by the FEMA Expert Panel in 1965 and several
times thereafter (Hall and Oser 1965, 1970; Oser and Ford 1974, 1978; Oser and
others 1984; Newberne and others 1998; Smith and others 2005b, 2009; Waddell
and others 2007). In 1978, the Panel evaluated the available data and affirmed the
GRAS status of these flavor ingredients (GRASa). In 1993, the Panel initiated a
comprehensive program to reevaluate the status of all FEMA GRAS flavor
ingredients concurrent with a systematic revision of the FEMA Scientific Literature
Reviews and the Panel reaffirmed the status of this group in 1995 with GRASr status.
In 2012, this group was again reaffirmed as GRAS (GRASr2) based on knowledge
concerning their rapid absorption, metabolic conversion, and excretion in humans
and animals; their low levels of use as flavors in food; the wide margins of safety
between the conservative estimates of intake and the NOAEL or NOEL determined
from subchronic studies and the lack of significant genotoxic and mutagenic
potential. The consistency of the results obtained from subchronic studies in rodent
models support the conclusion that consumption of aliphatic acyclic and alicyclic
terpenoid tertiary alcohols and structurally related substances as part of the food
supply is not associated with any significant risk to human health.
Acknowledgments

This work was supported by the Flavor and Extract Manufacturers Assoc. The Expert
Panel of the Flavor and Extract Manufacturers Assoc. is independent but is
financially supported by the Flavor and Extract Manufacturers Assoc.

Author Contributions

Marnett, Cohen, Fukushima, Gooderham, Hecht, Rietjens, and Smith interpreted the
results and drafted the manuscript. Adams, Bastaki, Harman, McGowen, and Taylor
researched and compiled available study data and assisted in the drafting of the
manuscript.

Conflict of Interest

The authors declare that there are no conflicts of interest.

Evaluation of Ion Exchange and Sorbing Materials for Their

Adsorption/Desorption Performane towards
Anthocyanins, Total Phenolics, and Sugars from a Grape
Pomace Extract
Evangelos D. Trikas 1,2, Rigini M. Papi 2, Dimitrios A. Kyriakidis 2 and George
A. Zachariadis 1,*
1 Laboratory of Analytical Chemistry, Department of Chemistry, Aristotle University,
54124 Thessaloniki,
Greece; [email protected]
2 Laboratory of Biochemistry, Department of Chemistry, Aristotle University, 54124
Thessaloniki, Greece;
[email protected] (R.M.P.); [email protected] (D.A.K.)
* Correspondence: [email protected]; Tel.: +30-2310997707
Academic Editor: Juan F. García-Reyes
Received: 27 December 2016; Accepted: 15 March 2017; Published: 18 March 2017
Abstract: Byproducts of winery industries are treated, in many cases, as useless
wastes constituting not only a major disposal problem but also not providing any
additional profit to the industries. However, these byproducts could be utilized as a
source of various phenolic compounds, such as anthocyanins, that could be used
as nutraceuticals or natural colorants. Nine materials were tested and evaluated for
their ability to retain and elute anthocyanins, total phenolics, and sugars from a grape
pomace extract. The materials tested were the ion exchange Amberlite IRA 400 Cl􀀀,
Lewatit TP 208 and Lewatit TP 260, and the sorbing Chromosorb G-HP, Amberite
XAD 2, Zeocros CA 150, Chemviron Carbon, Oasis HLB (hydrophilic-lipophilic
balance) and Isolute C8 end-capped (EC).
The two materials with the higher anthocyanins recovery rate, Oasis HLB and Isolute
C8 (EC),
were further examined for their anthocyanin capacities which were calculated as
5.76 mg_cm􀀀3 and 3.06 mg_cm􀀀3 respectively. Furthermore, their behavior
pattern towards anthocyanins of various molecular weights was investigated using
a liquid chromatography coupled with mass spectrometry (LC-PDA-MS) system.
Keywords: grape pomace; anthocyanins; total phenolics; sugars; solid-phase
extraction; mass spectrometry
1. Introduction
The wine making industry is a huge field with yearly worldwide growth. Grapes used
for wine production contain a wide variety of valuable compounds, such as
anthocyanins and other phenolic compounds. The majority of these compounds lie
at the skin of the fruit. As a result, the byproducts remaining after the vinification
procedure (grape pomace) contain a significant percentage of the total phenolic
content of grapes. These phenolic compounds, present in wine byproducts, have
been proved to be biologically active, possessing not only antioxidant properties [1–
3], but a correlation between total phenolic content and antibacterial activity has also
been demonstrated [4,5].
Another important property of these compounds is that they are responsible for the
colors of various fruits and nuts. Anthocyanins are not only responsible for the red
color of grapes, but can also display other colorations like blue, purple, and orange.
Now that the safety of many artificial pigments is questionable [6], the industry’s
demand for attractive and edible natura l colorants has increased significantly [7,8].
The above characteristics, both biological properties and the ability to alter or
enhance a product’s color, make these compounds very valuable to the food
industry. These could Separations 2017, 4, 9; doi:10.3390/separations4010009
www.mdpi.com/journal/separations Separations 2017, 4, 9 2 of 11
be used widely as food additives such as nutraceuticals, conservatives, or colorants.
This indicates that a possible utilization of these byproducts could be of great
commercial interest.
The extraction of these compounds from grape pomace is often performed with the
use of acidified aqueous alcoholic solutions, mainly methanol or ethanol [9–13]. In
order to obtain more phenolics-rich or purified extracts, the extraction procedure
could be followed by other techniques, such as liquid-liquid extraction, filtration with
special membranes, or the absorbance of these compounds on various materials
(resins and ion exchange materials) using solid phase extraction [14–16].
Diaz-Reinoso et al. [17], used ultrafiltration and nanofiltration membranes in order to
retain catechin, epicathehin, quercetin, gallic acid, and other compounds that were
contained in their sample, distilled grape pomace pressing liquors (DGPPL). The
recovery and concentration of antioxidants from winery wastes with the use of
polymeric resins was also the objective of another study [18], while ion exchange
and adsorbent resins were also used from Kammerer et al. [19], for the bounding of
polyphenols from apple and grape crude plant extracts. The retrieval of trans-
resveratrol from grape pomace with the use of macroporous resin was the subject
of a research made by Wang [20], while Soto et al. [21], tried to adsorb polyphenols
in activated charcoal. Similar studies have also been made specifically in the field of
anthocyanins. Extensive work has been conducted by Kraemer-Schafhalter et al.
[22] that examines 16 different materials for their ability to adsorb and desorb
anthocyanins. Since the anthocyanins exist in a variety of fruits and plants, it is
expected that the research will be extended to a pleiad of matrices. So, apart from
grape pomace [23,24], the use of absorbent materials for the recovery of
anthocyanins has been investigated also in blueberries [25,26], raspberries [27],
black beans [28,29], orange pulp wash [30], and even calyces of Hibiscus sabdariffa
flower [31].
The present study consisted of two parts. The objective of the first part was to
evaluate nine different solid phase materials for their ability and easiness to retain
and desorb anthocyanins, total phenolics, and sugars from an aqueous grape
pomace extract, with a particular interest in their performance towards anthocyanins.
A wide spectrum of materials, with different chemical and physical properties, were
investigated including zeolite, anion and cation exchange resins, activated carbon,
and polystyrene resins. In the second part of the study, the materials exhibiting best
performance during anthocyanins tests were further investigated regarding their total
capacity and their anthocyanins adsorption/desorption patterns using an LC-PDA-
MS system.
2. Materials and Methods
2.1. Reagents and Solutions
All reagents were of analytical grade, and deionized water was used for extraction
procedure
and preparation of all aqueous solutions. The reagents used incuded methanol
(MeOH, Chem-Lab,
Zedelgem, Belgium), ethanol (EtOH, Chem-Lab), ortho-phosphoric acid (H3PO4
85%, Panreac,
Barcelona, Spain), formic acid (HCOOH, Panreac), sodium hydroxide (NaOH,
Sigma-Aldrich,
St. Louis, MO, USA), potassium chloride (KCl, J.T. Baker, Deventer, The
Netherlands), hydrochloric
acid (HCl 37%, Carlo Erba Reagents, Italy), acetic acid (CH3COOH, Panreac),
sodium acetate
(CH3COONa, J.T. Baker), sodium carbonate (Na2CO3, Sigma-Aldrich), Folin &
Ciocalteu’s phenol
reagent (Sigma-Aldrich), gallic acid ((HO)3C6H2CO2H, Sigma-Aldrich), potassium
sodium tartrate
tetrahydrate (KOCOCH(OH)CH(OH)COONa _ 4H2O, Sigma-Aldrich), 3,5-
dinitrosalicylic acid
((O2N)2C6H2-2-(OH)CO2H, Sigma-Aldrich), glucose (C6H12O6, Sigma-Aldrich),
phenol (C6H5OH,
Sigma-Aldrich), sodium sulfite (Na2SO3, Panreac), kuromanin chloride (cyanidin-3-
O-glucoside,
C21H21O11Cl, _96%, Extrasynthese, Lyon, France), and oenin chloride (malvidin-
3-O-glucoside,
C23H25O12Cl, _97%, Extrasynthese). Acetonitrile (ACN, LC-MS grade, Sigma-
Aldrich), methanol
(MeOH, LC-MS grade, Sigma-Aldrich), and water (H2O, Thermo Fischer Scientific,
Waltham, MA,
USA) were used as elution solvents during liquid chromatography.
Separations 2017, 4, 9 3 of 11
The nine materials investigated were the Amberite XAD 2 (Sigma-Aldrich), Amberlite
IRA 400 Cl􀀀
(Sigma-Aldrich), Lewatit MDS TP 208 (Lanxess AG, Cologne, Germany), Lewatit
MDS TP 260 (Lanxess
AG), Chromosorb G-HP (Sigma-Aldrich), Zeocros CA 150 (PQ Corporation, Rolle,
Switzerland),
Activated Carbon (Chemviron Carbon, Brussels, Belgium), Oasis HLB (Waters
Corporation, Milford,
MA, USA), and Isolute C8 (EC) (Biotage, Sweden), Table 1.
Table 1. Physical properties and structural characteristics of tested materials.
Material Material Details Pore Diameter
(Å)
Particle
Diameter (_m)
Surface Area
(m2_g􀀀1)
Chromosorb G-HP Acid washed and treated with
dimethyldichlorosilane Unknown 125–149 0.5
Amberite XAD 2 Hydrophobic crosslinked polystyrene copolymer
resin styrene-divinylbenzene (macroreticular) 90 260–840 300
Amberlite IRA
400 Cl􀀀 Styrene-divinylbenzene (gel) anion exchange column Unknown 20–25 -
Lewatit TP 208 Weakly acidic, macroporous cation exchange resin
with chelating imino-diacetate groups Unknown 390 (_30) -
Lewatit TP 260 Weakly acidic, macroporous cation exchange resin
with chelating amino methylphosphonic acid groups Unknown 420 (_50) -
Zeocros CA 150 Zeolite Unknown 3–5 -
Activated Carbon Activated carbon Unknown 150–250 -
Oasis HLB Two monomers, the hydrophilic N-vinylpyrrolidone
and the lipophilic divinylbenzene 80 30 800
Isolute C8 (EC) Medium retentive, non-polar sorbent for extraction.
Octyl (end-capped) 60 50 500
2.2. Extraction Procedure
The solid byproducts were stored at freezer (􀀀20 _C). Amount of these byproducts
was freezedried
and the resulted powder was collected and stored at fridge (5 _C). The analogy
between the
amount before the freeze drying, and the resulted powder was 3 to 1. Amount of the
powder (500 mg)
was mixed with the extraction solution (20 mL) of acidified water with 5% v/v ortho-
phosphoric
acid (H3PO4 85%). The mixture was vortexed, transferred to an ultrasound bath for
10 min, and
consequently to a water bath at 45 _C for 2 h. The next step was the centrifugation
of the mixture at
4000 rpm for 10 min and the collection of the supernatant. The procedure was
repeated enough times
in order to collect the necessary amount of extract for the scheduled studies.
The specific extraction solution was selected in order to facilitate the adsorption of
the compounds
of interest to the tested materials.
2.3. Pretreatment of Solid Phase Materials and Evaluation Tests
The resins were pretreated and activated according to the manufacturer’s
recommendation.
At first, resins were soaked in methanol overnight. After soaking they were packed
into plastic syringe
tubes with glass microfiber filters (Whatman, Little Chalfont, UK). The resulting
material volume into
the SPE columns were between 0.75 and 0.85 cm3. After filling, the SPE columns
were rinsed with
water, and equilibrated with four bed volumes (BV) of H2O + 5% v/v ortho-
phosphoric acid (H3PO4
85%), which was the solvent used during extraction procedure. The pretreatment of
ion exchange
resins was constituted of four extra steps that were performed between the steps of
rinsing with water
and before the equilibration of the column. The cation exchange resins were treated
with 2 BV of 4%
w/v NaOH solution, then rinsed with 4 BV of deionised water. Afterward, these were
treated with
2BV of a 4% v/v HCl solution, and then rinsed again with 5 BV of water. The same
procedure, but
using the NaOH and HCl solutions in reverse order, was applied during the activation
of the anion
exchange resins.
The collected byproducts’ extract was the substrate used during the evaluation tests
on the
selected resins. The pretreatment of the materials was followed by a three steps
procedure. In the
Separations 2017, 4, 9 4 of 11
first step, 1.3 BV of sample was passed through each column. The second step was
the washing of
the column with 1.3 BV of H2O, and the third step was the elution with 1.3 BV of
acidified ethanol
solution (3% v/v formic acid). After every step, the eluents of the columns were
collected and stored
for the subsequent analytical procedures.
2.4. Analytical Procedures for Total Anthocyanins, Phenolic Content, and Sugars
Determination
Total anthocyanins, phenolic content, and sugars were determined in byproducts
extract and
at the eluents collected after SPE procedure. The methods used were the pH-
differential method,
Folin-Ciocalteu (FC) method, and dinitrosalicylic acid method respectively.
pH-differential method: The total anthocyanin content was determined using the pH-
differential
method [32] using a UV-vis spectrophotometer (Jenway 6300, Staffordshire, UK).
This method is
based at the property of anthocyanins to alter their structure in different pH
conditions, resulting in a
change in their color. In brief, two dilutions of the sample were prepared using buffer
solutions, one
at pH = 1 and another at pH = 4.5, and the absorbances of these solutions at 510
nm and at 700 nm
were measured. Anthocyanin content was calculated as cyanidin-3-O-glucoside
equivalents using an
extinction coefficient of 26,900 L_mol􀀀1_cm􀀀1 and molecular weight of 449.
Folin-Ciocalteu (FC) method: The phenolic content was determined using the Folin-
Ciocalteu
(FC) method. In brief, a calibration curve was made using gallic acid solutions. In a
test tube 0.2 mL of
the sample were added, followed by 2.6 mL H2O, 2 mL Na2CO3 of a (7% w/v)
solution and 0.2 mL
of FC reagent. The mixture left in the dark for 90 min. The absorbance of the mixture
at 745 nm was
measured [18].
Dinitrosalicylic acid (DNS) method: The sugars were determined using the
Dinitrosalicylic acid
method. In brief, a calibration curve was made using glucose solutions. A solution
containing 1%
DNS, 0.2% phenol, 0.05% Na2SO3, and 1% NaOH was prepared. In a test tube, 1.5
mL of the sample
with 1.5 mL of the previous prepared solution were mixed. The mixture was heated
for 5 min at 100 _C
in a water bath. Afterwards 0.5 mL of a 40% w/v potassium sodium tartrate solution
was added and
the mixture was left at room temperature before the absorbance measurement at
575 nm [33].
2.5. LC-PDA-MS Analysis
The obtained extract along with the collected ethanolic elutions of Oasis HLB and
Isolute C8
(EC) columns were filtered through a 0.45 _m syringe filter (BGB Analytik,
Richmond, VA, USA)
and analyzed for their anthocyanin content using an LC-PDA-MS method. A
Shimadzu LCMS-2010
EV, using electrospray ionization (ESI) process and a quadrupole mass analyser,
equipped with an
SPD-M20A PDA (Shimadzu Corporation, Kyoto, Japan) detector, at positive
ionization mode was used
for the chromatographic separation and identification of the compounds while the
chromatographic
column used was a Dionex, RP-C18, Acclaim 120, 150 _ 4.6 mm _ 5 _m (Thermo
Fisher Scientific,
Waltham, MA, USA). A gradient elution program was employed, using water/formic
acid (99/1, v/v)
and acetonitrile/formic acid (99/1, v/v) as elution solvents, while the sample injection
volume was
20 _L. The flow rate was 0.5 mL_min􀀀1 with a 45 min gradient elution program as
follows: 0 min,
10% B; 0–5 min, 10%–22% B; 5–14 min, 22%–28% B; 14–35 min, 28%–42% B; 35–
40 min, 42%–100% B;
40–45 min, 100%–10% B.
3. Results and Discussion
3.1. Preliminary Experiments for All Tested Materials
Nine materials were studied in total. The reasoning for the selection of the specific
materials was
twofold. First of all, it was intended to select materials with different chemical and
physical properties.
The nine tested resins differ in chemical structure, polarity, particle size, surface area
and retain
mechanism, covering a wide range of different materials. Having this as priority it
was not possible
to test various materials from every category but we were constrained to select just
some typical
Separations 2017, 4, 9 5 of 11
examples from each one (only one anion exchange column, only one type of zeolite
and activated
carbon). Secondly, we preferred to study materials that had not been already
extensively studied in
the past, like the XAD-7 by Kraemer-Schafhalter et al. [22] considering preferable to
investigate the
effectiveness of other materials.
During the experiments in first part of the study the byproducts’ extract and the
eluents collected
after SPE procedure were analyzed for their anthocyanin, phenolic, and sugar
content with the methods
described at subsection 2.4. There are three types of collected and analyzed eluents:
the sample eluent,
the H2O eluent, and the ethanol eluent. The sample eluent is the eluent collected
after the loading of
the sample into the SPE column and its content corresponds to the compounds that
were not retained
from the column material. The H2O eluent is the eluent collected after the water
loading and its content
corresponds to the compounds that were not strongly bound to the column material.
The ethanol
eluent is the eluent collected after the ethanol loading into the column and its content
corresponds to
the compounds that were bound to the column material but were able to be
recovered relatively easy
with the use of ethanol.
There is also the sorption percentage of each material. This percentage corresponds
to the amount
of compounds that were bounded in the column and were not eluted in the sample
eluent regardless if
these were collect during second, third step, or even at all.
The results of the analyses performed to the specific eluents are expressed as the
percentage of
compounds detected in each collected eluent in comparison with the amount of the
same compounds
present at the loaded extract. Results are presented at Tables 2–4.
Table 2. Anthocyanins determined in the eluents in comparison with the
anthocyanins loaded to the
SPE column.
Material Sample Eluent
(1st Step)
H2O Eluent
(2nd Step)
EtOH Eluent
(3rd Step)
Total
Recovery
Sorption
Percentage
Chromosorb G-HP 46% 25% 20% 91% 54%
Amberite XAD 2 47% 9.2% 24% 80% 53%
Amberlite IRA 400 Cl􀀀 65% 4.6% 3.3% 73% 35%
Lewatit TP 208 36% 11% 5.2% 52% 64%
Lewatit TP 260 41% 6.8% 12% 60% 59%
Zeocros CA 150 23% 10% 53% 86% 77%
Activated Carbon 2.8% 0.6% 0.7% 4.1% 97%
Oasis HLB 0.6% 4.9% 83% 89% 99%
Isolute C8 (EC) 2.4% 2.4% 95% 100% 98%
Table 3. Total phenolics determined in the eluents in comparison with the total
phenolics loaded to the
SPE column.
Material Sample Eluent
(1st Step)
H2O Eluent
(2nd Step)
Eluent Eluent
(3rd Step)
Total
Recovery
Sorption
Percentage
Chromosorb G-HP 52% 30% 17% 99% 48%
Amberite XAD 2 48% 13% 18% 79% 52%
Amberlite IRA 400 Cl􀀀 58% 22% 8.4% 88% 42%
Lewatit TP 208 66% 31% 1.6% 99% 34%
Lewatit TP 260 59% 22% 5.3% 86% 41%
Zeocros CA 150 51% 38% 8.6% 98% 49%
Activated Carbon 1.0% 0.9% 1.4% 3.3% 99%
Oasis HLB 2.6% 3.9% 19% 26% 97%
Isolute C8 (EC) 3.5% 6.7% 87% 97% 97%
Separations 2017, 4, 9 6 of 11
Table 4. Sugars determined in the eluents in comparison with the sugars loaded to
the SPE column.
Material Sample Eluent
(1st Step)
H2O Eluent
(2nd Step)
Eluent Eluent
(3rd Step)
Total
Recovery
Sorption
Percentage
Chromosorb G-HP 59% 30% 11% 100% 41%
Amberite XAD 2 64% 21% 14% 99% 36%
Amberlite IRA 400 Cl􀀀 65% 10% 4.9% 80% 35%
Lewatit TP 208 68% 24% 4.1% 96% 32%
Lewatit TP 260 60% 14% 6.1% 80% 40%
Zeocros CA 150 61% 32% 5.4% 98% 39%
Activated Carbon 2.5% 3.6% 7.2% 13% 98%
Oasis HLB 33% 26% 3.9% 63% 67%
Isolute C8 (EC) 33% 19% 26% 78% 67%
3.1.1. Test Results Evaluation
Anthocyanin content tests results evaluation: Based on the obtained results, the nine
resins can be
divided into three categories regarding the resulting anthocyanin concentration in the
ethanolic eluent.
Oasis HLB and Isolute C8 (EC) are classified in the first category as the most
effective resins with
recovery rates above 80%. In the middle category belong the Zeocros CA 150, the
Amberite XAD 2,
and the Chromosorb G-HP with recoveries between 20% and 53%, while the rest
resins could belong
to the third group with recovery rates below 20%.
Total phenolic content tests results evaluation: Using the same methodology as for
anthocyanin
content tests results evaluation, in this case the nine resins can be divided in two
categories regarding
the resulting total phenolic concentration in the ethanolic eluent. Only the Isolute C8
(EC) can be
classified to the higher category with a recovery rate above 80%, while all other
materials exhibited
rates below 20%.
Sugar content tests results evaluation: Following the same classification criteria, the
nine resins
are divided again in two categories regarding the resulting sugar concentration in the
ethanolic eluent.
Three materials exhibit recovery rates above 20%, while the other 6 materials belong
to the lower level
with recoveries below 20%.
3.1.2. Materials Performance Remarks
The ion selective materials exhibited poor recovery for all the investigated
compounds, and
especially the anion exchange resin (Amberlite IRA 400 Cl􀀀) showed very poor
anthocyanins retention
with high color losses. The poor performance of ion selective resins comes in
agreement with the
observations of Schafhalter et al. [22]. In their research all three tested anion
exchange resins exhibited
color loss, while none of the cation exchange resins tested showed remarkable
performance.
The Chromosorb G-HP also exhibited inefficient anthocyanins retention, while a
significant
percentage of all three compound groups were eluting during water elution at the
second step of
the test procedure. This is probably due to the very small surface area of this
material, taking into
consideration that the surface area has been proved to be one of the main
parameters affecting the
anthocyanin absorbance [23,28].
The Amberlite XAD 2, a polystyrene copolymer resin, displayed relative low
anthocyanins
recovery (24%), but the total amount of anthocyanins eluted from the column during
the three steps
of the procedure was equal to 80% of the total anthocyanins load. This means that
about 20% was
still adsorbed in the resin, a quantity that could possibly be retrieved after further
successive elutions,
allowing the recovery of anthocyanins to approach or even outnumber 50% after
optimizing the total
procedure for the specific material. Furthermore, a big part of the 47% of
anthocyanins eluting from
the column during sample loading, could possibly be adsorbed after a second or a
third successive
adsorption cycle, since as Wang et al. [28], have noted the anthocyanin amount
adsorbed to the resin is
increasing rapidly during the first three cycles.
Separations 2017, 4, 9 7 of 11
The Zeocros CA150 achieved a quite satisfactory rate of anthocyanins recovery
(53%), while at
the same time did not retain total phenolics and sugars at a significant extent (<10%).
This attribute
makes it an ideal material for purposes of selective binding of anthocyanins towards
other phenolic
compounds and/or sugars. Similar behavior, but with even better performance
characteristics (83%),
was exhibited in the Oasis HLB SPE column. On the other hand, the Isolute C8 (EC)
showed similarly
impressive results on the anthocyanins recovery (95%), but without the selectivity
exhibited towards
them by Oasis HLB, since the rates for total phenolics and sugars were significantly
higher, 87% and
26% respectively.
Activated carbon showed a unique behavior compared with all other tested
materials. It adsorbed
all the studied compounds and retain them tightly, making it an ideal material choice
for specific
purposes, like the removal of these compounds from a liquid stream or aqueous
solutions. A similar
behavior of activated carbon towards phenolic compounds of distilled grape pomace
had been also
observed by Soto et al. [21].
An outline of the main remarks for each material is presented in Table 5.
Table 5. General remarks regarding the materials tested during SPE procedure.
Material Comment
Chromosorb G-HP Insufficient anthocyanins retention, compounds were eluted with
water
Amberite XAD 2 Relative low anthocyanins recovery but with 20% still absorbent in
the material meaning that
with successive elutions the recovery rate could probably approach 45%
Amberlite IRA 400 Cl􀀀 Poor retention, high color losses
Lewatit TP 208 Poor retention
Lewatit TP 260 Poor retention
Zeocros CA 150 Fair anthocyanins recovery, majority of total phenolics and sugars
were not retained
Activated Carbon Almost quantitatively absorbance of all investigated compounds,
ideal for cleaning purposes
Oasis HLB Great anthocyanins recovery, good separation from other phenolics and
sugars
Isolute C8 (EC) Great anthocyanins and total phenolics recovery, 14
of sugars recovered
The Oasis HLB and Isolute C8 (EC) columns were the materials with the higher
anthocyanins
recovery rates and since anthocyanins were the main area of interest of this
research, these materials
were selected for further studies during the second part of the study.
3.2. OASIS HLB and ISOLUTE C8 (EC) Further Studies
During the second part of our study, additional experiments were conducted in order
to calculate
the anthocyanins capacity of Oasis HLB and Isolute C8 (EC), as well as to
investigate the behavior of
these materials towards the dominant anthocyanins of the extract using an LC-PDA-
MS system.
In order to determine the sorbing capacities of the materials, small portions of the
extract (1 BV
each time) were loaded to the SPE columns until color loss started to occur. When
the columns started
to elute reddish color after sample loading, it was considered that the material was
not able to retain
any more anthocyanins without significant losses, and the procedure was
considered complete. In total
more than 30 mL of aqueous extract was passed through Isolute C8 (EC) column
and more than 50 mL
through Oasis HLB. When the sample loading procedure was completed for each
material, the elution
step came next. The elution of the adsorbed anthocyanins from the Isolute C8 (EC)
and the Oasis HLB
SPE columns was done using only 1.5 mL and 3.0 mL of acidified ethanol
respectively. That equals to
an enrichment factor of more than 14 for the first and 12 times for the second
material, also taking into
account the recovery rates. The anthocyanins amount loaded to the column, as well
as the amount
eluted, the recovery rate, and the capacity of each material are presented in Table
6.
Separations 2017, 4, 9 8 of 11
Table 6. Determination of anthocyanins capacity of Oasis HLB and Isolute C8 (EC)
materials.
Material Material
Volume
Anthocyanins
Amount Loaded
Anthocyanins
Amount Eluted
Recovery
Rate
Material Anthocyanins
Capacity
Oasis HLB 0.8 cm3 6.36 mg 4.61 mg 72.5% 5.76 mg_cm􀀀3
Isolute C8 (EC) 0.8 cm3 3.51 mg 2.45 mg 69.8% 3.06 mg_cm􀀀3
The Oasis HLB not only proved to have higher capacity compared with the Isolute
C8 (EC),
5.76 mg_cm􀀀3 to 3.06 mg_cm􀀀3, but also achieved a higher recovery rate of 72.5%
compared to 69.8%.
The higher capacity of the Oasis HLB over the Isolute C8 (EC) column could be
attributed to the larger
surface area of the first (800 m2_g􀀀1 towards 500 m2_g􀀀1), since this feature has
been characterized as
the most [23,27,28] or one of the two most [26] decisive parameter affecting the
adsorption capacity of
a resin.
The collected eluents were analyzed for their anthocyanin content with an LC-PDA-
MS
system in order to investigate how the solid phase extraction procedure using the
specific
materials affects the anthocyanins of a grape pomace extract. In order to do so, four
malvidin
based anthocyanins (the dominant anthocyanins in wine and wine byproducts) were
selected
for comparison studies. These were the Malvidin-3-O-glucoside, Vitisin B (Malvidin
derivative),
Malvidin-3-(60 0-acetylglucoside), and Malvidin-3-(60 0-p-coumaroylglucoside),
(Table 7).
Table 7. Results after LC-PDA-MS analyses.
Parameter Compound
Mv-3-O-glucoside Vitisin B (Mv
derivative) Mv-3-(60 0-acetylglucoside) Mv-3-(60 0-p-coumaroylglucoside)
Molecular weight 493 517 535 639
Anthocyanins amount
loaded to Oasis HLB 2179 _g 441.3 _g 1251 _g 301.6 _g
Anthocyanins amount
eluted from Oasis HLB 1006 _g 274.5 _g 1158 _g 290.1 _g
Recovery rate 46.2% 62.2% 92.6% 96.2%
Anthocyanins amount
loaded to Isolute C8 (EC) 1315 _g 227.9 _g 735.7 _g 161.5 _g
Anthocyanins amount
eluted from Isolute C8 (EC) 547.9 _g 152.6 _g 655.5 _g 153.9 _g
Recovery rate 41.7% 67.0% 89.1% 95.3%
As can be seen in Table 7, there is a trend observed in the behavior of both examined
materials.
In both cases, the recovery rates increased as the molecular weight of the studied
compound increased.
For the Oasis HLB it starts from 46.2% for Mv-3-O-glucoside (Molecular weight
(MW): 493) and it
reaches up to 96.2% for Mv-3-(60 0-p-coumaroylglucoside) (MW: 639), while for the
Isolute C8 (CE) the
rates are 41.7% and 95.3% respectively.
This pattern can probably be attributed to the pore diameter of the materials since
the pore
diameter plays a crucial role to the SPE procedure. If it is too small, the molecules
are not be able to
penetrate into the pores, or their diffusion is restricted. On the other hand, if the pore
diameter is too
big the molecules are not able to adsorb and are eluted faster and more easily than
intended. In our
case, the elution pattern indicated that the lower MW molecules were adsorbed less
tightly in the
column, resulting to a gradual elution of them during the subsequent sample loading
rounds.
4. Conclusions
Three ion selective and six sorbing materials with different chemical and physical
properties
were evaluated in this study for their ability to retain and desorb anthocyanins, total
phenolics, and
sugars from an aqueous solution during an SPE procedure. The sample used was
an extract from
Separations 2017, 4, 9 9 of 11
grape pomace, while the selected resins covered a wide range of materials, with
different chemical and
physical properties, from ion exchange resins, to zeolite, activated carbon, and
polystyrene resins.
The materials were divided into three categories based on the compounds
concentration in the
ethanolic eluent. The higher category included materials with recovery rates above
80%, the middle
category with recovery rates between 20% and 80%, and the lower one with
recovery rates under 20%.
Regarding anthocyanins the materials were distributed to all three categories, with
two materials in
the higher, three in the middle, and four in the lower category. For total phenolics,
only Isolute C8 (EC)
exhibited recovery rate above 80%, while for sugars there were no materials at the
top category. Though
the focus of this study was not to optimize the conditions of the SPE procedure, the
relative recovery
rates of the resins were compared to help identify the most effective materials for
further research.
The two materials with the best performance regarding anthocyanins, Oasis HLB
and Isolute
C8 (EC), were further investigated for their anthocyanin capacity and the way they
interact with
anthocyanins of different molecular weight. Results showed capacities of 5.76
mg_cm􀀀3 and
3.06 mg_cm􀀀3 for Oasis HLB and Isolute C8 (EC) respectively, while the pattern
detected showed an
increasing recovery rate for anthocyanins of higher molecular weight for both
materials, something
that could be attributed to their pore sizes.
As far as we know, it is the first time that these two commercial available materials,
Oasis HLB
and Isolute C8 (EC), have been tested for their ability to retain and elute
anthocyanins from wine
byproducts extract. Furthermore, instead of focusing our study on a specific category
of materials,
like other similar studies have done, we decided to follow a different approach by
comparing the
performance of materials with different physicochemical properties.
Acknowledgments: This work was supported by “11SYN_2_1992” action
“COOPERATION 2011” of EYDE-ETAK
funded by the Operational Program “Competitiveness and Entrepreneurship”
(EPAN-II).
Author Contributions: Evangelos D. Trikas, Rigini M. Papi, Dimitrios A. Kyriakidis,
and George A. Zachariadis
conceived and designed the experiments; Evangelos D. Trikas performed the
experiments as part of his Ph.D.
thesis, analyzed the data, and wrote the paper.
Conflicts of Interest: The authors declare no conflict of interest
Abbreviations
The following abbreviations are used in this manuscript:
Mv Malvidin
SPE solid-phase extraction
LC liquid chromatography
PDA photodiode array
MS mass spectrometry
BV bed volume
FC Folin-Ciocalteu
References
1. Alonso, A.; Guillean, D.; Barroso, C.; Puertas, B.; Garcia, A. Determination of
antioxidant activity of wine
byproducts and its correlation with polyphenolic content. J. Agric. Food. Chem. 2002,
50, 5832–5836.
[CrossRef] [PubMed]
2. Llobera, A.; Canellas, J. Antioxidant activity and dietary fibre of Prensal Blanc
white grape (Vitis vinifera)
byproducts. Int. J. Food. Sci. Technol. 2008, 43, 1953–1959. [CrossRef]
3. Trikas, E.; Melidou, M.; Rigini, P.; Zachariadis, G.; Kyriakidis, D. Extraction,
separation and identification
of anthocyanins from red wine byproduct and their biological activities. J. Funct.
Foods. 2016, 25, 548–558.
[CrossRef]
4. Sanhueza, L.; Tello, M.; Vivanco, M.; Mendoza, L.; Wilkens, M. Relation between
antibacterial activity
against food transmitted pathogens and total phenolic compounds in grape pomace
extracts from Cabernet
Sauvignon and Syrah Varieties. Adv. Microb. 2014, 4, 225–232. [CrossRef]
Separations 2017, 4, 9 10 of 11
5. Baydar, N.G.; Ozkan, G.; Sagdic, O. Total phenolic contents and antibacterial
activities of grape (Vitis vinifera L.)
extracts. Food Control 2004, 15, 335–339. [CrossRef]
6. Schab, D.W.; Trinh, N.H. Do artificial food colors promote hyperactivity in children
with hyperactive
syndromes? A meta-analysis of double-blind placebo-controlled trials. J. Dev.
Behav. Pediatr. 2004, 25,
423–434. [CrossRef] [PubMed]
7. Giusti, M.M.; Wrolstad, R.E. Acylated anthocyanins from edible sources and their
applications in food
systems. Biochem. Eng. J. 2003, 14, 217–225. [CrossRef]
8. Pazmino-Duran, A.E.; Giusti, M.M.; Wrolstad, R.E.; Gloria, M. Anthocyanins from
Oxalis triangularis as
potential food colorants. Food Chem. 2001, 75, 211–216. [CrossRef]
9. Corrales, M.; Garcia, A.; Butz, P.; Tauscher, B. Extraction of anthocyanins from
grape skins assisted by high
hydrostatic pressure. J. Food Eng. 2009, 90, 415–421. [CrossRef]
10. Acevedo, A.; Hilbert, G.; Riviere, C.; Mengin, V.; Ollat, N.; Bordenave, L.;
Decroocq, S.; Delaunay, J.; Delrot, S.;
Merillon, J.; et al. Anthocyanin identification and composition of wild Vitis spp.
accessions by using LC-MS
and LC-NMR. Anal. Chim. Acta 2012, 732, 145–152. [CrossRef] [PubMed]
11. Li, Z.; Pan, Q.; Cui, X.; Duan, C. Optimization on anthocyanins extraction from
wine grape skins using
orthogonal test design. Food Sci. Biotechnol. 2010, 19, 1047–1053. [CrossRef]
12. Corrales, M.; Toepfl, S.; Butz, P.; Knorr, D.; Tauscher, B. Extraction of
anthocyanins from grape byproducts
assisted by ultrasonics, high hydrostatic pressure or pulsed electric fields: A
comparison. Innov. Food Sci.
Emerg. Technol. 2008, 9, 58–91. [CrossRef]
13. Trikas, E.; Rigini, P.; Kyriakidis, D.; Zachariadis, G. A Sensitive LC-MS Method
for Anthocyanins and
Comparison of Byproducts and Equivalent Wine Content. Separations 2016, 3, 18.
[CrossRef]
14. Kovacevic, D.; Putnik, P.; Dragovic-Uzelac, V.; Vahcica, N.; Babojelic, M.; Levaj,
B. Influences of organically
and conventionally grown strawberry cultivars on anthocyanins content and color in
purees and low-sugar
jams. Food Chem. 2015, 181, 94–100. [CrossRef] [PubMed]
15. Cao, J.; Li, X.; Liu, Y.; Leng, F.; Li, X.; Sun, C.; Chen, K. Bioassay-Based
Isolation and Identification of
Phenolics from Sweet Cherry That Promote Active Glucose Consumption by HepG2
Cells. J. Food Sci.
2015, 80. [CrossRef] [PubMed]
16. Deineka, V.; Deineka, L.; Sidorov, N.; KostenkoI, M.; Blinova, P. Estimating the
Solubility of Anthocyanins
Using Cartridges for Solid-Phase Extraction. Russ. J. Phys. Chem. A 2016, 90, 861–
863. [CrossRef]
17. Diaz-Reinoso, B.; Gonzalez-Lopez, N.; Moure, A.; Dominguez, H.; Parajo, J.C.
Recovery of antioxidants from
industrial waste liquors using membranes and polymeric resins. J. Food Eng. 2010,
96, 127–133. [CrossRef]
18. Soto, M.L.; Conde, E.; Gonzalez-Lopez, N.; Conde, M.J.; Moure, A.; Sineiro, J.;
Falque, E.; Dominguez, H.;
Nunez, M.J.; Parajo, J.C. Recovery and Concentration of Antioxidants
fromWineryWastes. Molecules 2012,
17, 3008–3024. [CrossRef] [PubMed]
19. Kammerer, J.; Schweizer, C.; Carle, R.; Kammerer, D.R. Recovery and
fractionation of major apple and grape
polyphenolsfrom model solutions and crude plant extracts using ion exchange and
adsorbent resins. Int. J.
Food Sci. Technol. 2011, 46, 1755–1767. [CrossRef]
20. Wang, B. Orthogonal test design for optimisation of extraction of trans-resveratrol
from Pinot Noirgrape
pomace. Nat. Prod. Res. 2012, 9, 821–829. [CrossRef] [PubMed]
21. Soto, M.L.; Moure, A.; Domınguez, H.; Parajo, J.C. Charcoal adsorption of
phenolic compounds present in
distilled grape pomace. J. Food Eng. 2008, 84, 156–163. [CrossRef]
22. Kraemer-Schafhalter, A.; Fuchs, H.; Pfannhauser, W. Solid-Phase Extraction
(SPE) a Comparison of 16
Materials for the Puriücation of Anthocyanins from Aronia melanocarpa var Nero. J.
Sci. Food Agric. 1998, 78,
435–440. [CrossRef]
23. Sandhu, A.K.; Gu, L. Adsorption/Desorption Characteristics and Separation of
Anthocyanins from
Muscadine (Vitis rotundifolia) Juice Pomace by Use of Macroporous Adsorbent
Resins. J. Agric. Food Chem.
2013, 61, 1441–1448. [CrossRef] [PubMed]
24. Kammerer, D.; Kljusuric, J.G.; Carle, R.; Schieber, A. Recovery of anthocyanins
from grape pomace extracts
(Vitis vinifera L. cv. Cabernet Mitos) using a polymeric adsorber resin. Eur. Food
Res. Technol. 2005, 220,
431–437. [CrossRef]
25. Buran, T.J.; Sandhu, A.K.; Li, Z.; Rock, C.R.; Yang, W.W.; Gu, L.
Adsorption/desorption characteristics
and separation of anthocyanins and polyphenols from blueberries using
macroporous adsorbent resins.
J. Food Eng. 2014, 128, 167–173. [CrossRef]
Separations 2017, 4, 9 11 of 11
26. Liu, X.; Xiao, G.; Chen,W.; Xu, Y.;Wu, J. Quantification and Purification of
Mulberry AnthocyaninsWith
Macroporous Resins. J. Biomed. Biotechnol. 2004, 5, 326–331. [CrossRef]
[PubMed]
27. Yang, Y.; Yuan, X.; Xu, Y.; Yu, Z. Purification of Anthocyanins from Extracts of
Red Raspberry Using
Macroporous Resin. Int. J. Food Prop. 2015, 18, 1046–1058. [CrossRef]
28. Wang, X.; Hansen, C.; Allen, K. Extraction of Anthocyanins from Black Bean
Canning Wastewater with
Macroporous Resins. J. Food Sci. 2014, 79, 184–188. [CrossRef] [PubMed]
29. Wang, X.; Hansen, C.; Allen, K. Identification of Anthocyanins Isolated from
Black Bean CanningWastewater
by Macroporous Resin Using Optimized Conditions. Food Nutr. Sci. 2013, 4, 174–
181. [CrossRef]
30. Mauro, A.; Arena, E.; Fallico, B.; Passerini, A.; Maccarone, E. Recovery of
Anthocyanins from Pulp Wash
of Pigmented Oranges by Concentration on Resins. J. Agric. Food Chem. 2002, 50,
5968–5974. [CrossRef]
[PubMed]
31. Beye, C.; Tounkara, L.S.; Destain, J.; Zgoulli, S.; Ndoye, A.S.; Thonart, P. Study
of the sorption behavior of
hibiscus sabdariffa anthocyanins on a macroporous resin. J. Food Process. Eng.
2013, 36, 579–590.
32. Giusti, M.M.; Wrolstad, R.E. Characterization and measurement of anthocyanins
by UVvisible spectroscopy.
In Current Protocols in Food Analytical Chemistry; John Wiley & Son: New York, NY,
USA, 2001; p. 1000.
33. Mazauric, J.P.; Salmon, J.M. Interactions between yeast lees and wine
polyphenols during simulation of
wine aging. II. Analysis of desorbed polyphenol compounds from yeast lees. J. Agric.
Food. Chem. 2006, 54,
3876–3881. [CrossRef] [PubMed]
© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open
access
article distributed under the terms and conditions of the Creative Commons
Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Curr Anal Chem. Author manuscript; available in PMC 2009 Nov 26.
Published in final edited form as:
Curr Anal Chem. 2008 Apr 1; 4(2): 75–101.
doi: 10.2174/157341108784587795
PMCID: PMC2783603
NIHMSID: NIHMS104863

Recent Advances in Anthocyanin Analysis and Characterization

Cara R. Welch,a Qingli Wu,b,* and James E. Simonb

Author information ► Copyright and License information ►
See other articles in PMC that cite the published article.
Go to:

Abstract

Anthocyanins are a class of polyphenols responsible for the orange, red, purple and
blue colors of many fruits, vegetables, grains, flowers and other plants. Consumption
of anthocyanins has been linked as protective agents against many chronic diseases
and possesses strong antioxidant properties leading to a variety of health benefits.
In this review, we examine the advances in the chemical profiling of natural
anthocyanins in plant and biological matrices using various chromatographic
separations (HPLC and CE) coupled with different detection systems (UV, MS and
NMR). An overview of anthocyanin chemistry, prevalence in plants, biosynthesis and
metabolism, bioactivities and health properties, sample preparation and
phytochemical investigations are discussed while the major focus examines the
comparative advantages and disadvantages of each analytical technique.

Keywords: Anthocyanins, Polyphenols, Plant sources, Analytical methodology,

HPLC, CE
Go to:

1. INTRODUCTION

Anthocyanins are a class of compounds belonging to the larger flavonoids class

which comprises a subset of the poly-phenol class of compounds. Anthocyanins are
responsible for much of the red, blue, and purple colors of fruits, vegetables, grains,
flowers, and herbs, which explains their name, in Greek, anthos means flower and
kyanos means blue. Anthocyanins are predominantly found in nature as glycosides
of polyhydroxy and polymethoxy derivatives of 2-phenyl-benzopyryliurn or flavylium
salts, (Fig. (1)). Individual members are differentiated by the number of hydroxyl and
methoxyl groups of the B-ring, by the number of sugars attached to the aglycon and
the position of attachment, and by the nature and number of aliphatic or aromatic
acids attached to the sugar residues [1].

Fig. (1)
Chemical structure of some common polyphenolic compounds: phenolic acids, gallic
acid (1) and caffeic acid (2); a stilbene, resveratrol (3); a lignan, secoisolariciresinol
(4); and a flavonoid, cyanidin (5).

Flavonoids, as a class of compounds containing anthocyanins, are a member of a

larger class of compounds called polyphenols, which includes all molecules with
more than one hydroxyl group on an aromatic ring. Other polyphenol compounds
are phenolic acids, stilbenes, and lignans, (Fig. (1)). Phenolic acids consist
predominantly of two types of molecules: derivatives of benzoic acid and derivatives
of cinnamic acid. Gallic acid (1), a derivative of benzoic acid, and caffeic acid (2), a
derivative of cinnamic acid, are two widespread representatives of the phenolic
acids, (Fig. (1)). These compounds are found in various teas in relatively high
quantities and act as antioxidants by free-radical scavenging [2, 3]. Stilbenes contain
two phenyl moieties connected by a two-carbon methylene bridge and are the
smallest class of polyphenols. Stilbenes often act as antifungal phytoalexins,
compounds that are synthesized only in response to infection or injury [3]. One
example of a stilbene is Resveratrol (3), an anticarcinogenic compound produced in
grapes and peanuts that effects cell signaling pathways, cell proliferation, and
apoptosis [4–6]. Lignans are diphenolic compounds formed by the dimerization of
two cinnamic acid residues. Several lignans, such as secoisolariciresinol (4), are
considered to be phytoestrogens and are converted by intestinal bacteria into
enterolactone and enterodiol. These can mimic estrogen compounds in the body
and may reduce the effect of estrogen by displacing it from cells, leading to the
prevention of some cancers, such as breast cancer, that are estrogen dependent [2,
3, 7].

Flavonoids (5) include the group of anthocyanins which represent the largest class
of polyphenols comprising over 9000 identified compounds with more being
continually discovered [8]. These compounds share a common framework consisting
of two aromatic rings (A and B) that are bound together by three carbon atoms that
form an oxygenated heterocycle (ring C). This design can be seen in the flavonoid
structures in Fig. (2). Flavonoids are subsequently divided into several groups
differing in the oxidation state of the heterocyclic pyran ring C. The subclasses
consist of flavanols, flavanones, flavones, isoflavones, flavonols, and anthocyanins
(listed in ascending order of oxidation) and within each of these subclasses,
individual compounds are characterized by specific hydroxylation and conjugation
patterns [9]. Most flavonoids are present in nature as the glycosidic form, with the
exception of flavanols, and this contributes to their complexity and the large number
of individual molecules that have been identified [10].

Fig. (2)
The chemical structures of flavanol, (±)-catechin (6); flavanone, naringenin (7);
flavone, apigenin (8); isoflavone, genistein (9); flavonol, quercetin (10); and
anthocyanins, cyanidin (5).

Flavanols (6) exist in the monomer form as catechins and in the polymeric form as
proanthocyanidins, also known as tannins. Proanthocyanidins are found in many
types of fruit as well as red wine, but green tea and chocolate appear to be the richest
sources [2, 9]. Proanthocyanidins are responsible for the astringency of fruits and
beverages and the bitterness of chocolate and when heated in acid, yield
anthocyanidins, the aglycone form of anthocyanins [11]. Flavanones (7) are found in
high concentrations in citrus fruits and when found in nature, they exist as
glycosides, but when absorbed in humans, they are converted into aglycones [2, 9,
12]. Flavones (8), rarely found in fruits and vegetables, are found in many other
plants including parsley, celery, millet and wheat. Several prenylated flavones have
also been identified in hops and beer. These specific flavones have been
characterized as a potent phytoestrogen and have been shown to have anti-cancer
activity. Isoflavones (9) are almost exclusively found in leguminous plants, such as
soy products, in the inactive glycosidic form. However, these compounds are
metabolized in the digestive tract to their corresponding aglycones and can, only
then, be absorbed [2, 9, 13]. Flavonols (10) are the most widespread flavonoids in
foods, mostly in the form of quercetin and kaempferol. The richest sources are
onions, kale, leeks, broccoli, and blueberries as well as red wine and tea. Flavonols
are often found as glycosides in nature and tend to accumulate in the skin and leaves
of plants as their biosynthesis is stimulated by light. Not surprisingly, the
concentration of flavonols can differ between pieces of fruit on the same branch,
depending on exposure to sunlight [2,9].

Anthocyanins (5) are the most oxidized flavonoids with the C ring fully unsaturated
and a hydroxyl at position 3. The basic structure is an aglycone, or anthocyanidin,
with one or more sugars attached at most often C3, C5, or C7 and possibly
esterification on the sugars. Currently, there are 19 naturally occurring
anthocyanidins (Table 1). The six most common anthocyanidins found in edible
plants include pelargonidin, peonidin, cyanidin, malvidin, petunidin, and delphinidin
[14]. These naturally occurring anthocyanidins can all be associated with three
parent aglycone structures, pelargonidin, cyanidin, and delphinidin, due to the
substitution pattern seen in the B-ring. When referring to the six major
anthocyanidins, they can be grouped together with peonidin and cyanidin having 3′
and 4′ substitutions while petunidin, malvidin and delphinidin are trisubstituted at the
3′, 4′ and 5′ positions and pelargonidin is monosubstituted [1]. The prevalence of
sugar occurrence in natural anthocyanins is glucose, rhamnose, xylose, galactose,
arabinose, and fructose. Many anthocyanins have been found to be acylated by
aliphatic or aromatic acids, the most commonly seen acyl groups being coumaric,
caffeic, ferulic, p-hydroxy benzoic, synapic, malonic, acetic, succinic, oxalic, and
malic acids. Considering all these factors, the number of probable anthocyanin
compounds is quite large, leading to over 600 having been identified from natural
sources [15,16].

Table 1
Naturally Occurring Anthocyanidins

1.1. Plant Sources

Anthocyanins are almost exclusively found in higher plants, although a few have
been found in lower plants such as mosses and ferns [15, 17]. Generally, the types
of anthocyanins in ornamental plants, or flowers, are more complex than those found
in fruits, with the exception of grapes which consist of a vareity of anthocyanins. That
is, flower pigments can involve both polyglycosylations and polyacylations and
undergo a highly regulated series of biochemical steps that lend to a defined set of
different compounds that produce the variety of shades or hues, while fruit
anthocyanins are simpler and generally include just one or two main pigments per
plant source. Examples of such fruits containing anthocyanins includes: pome fruits,
stone fruits, berries, tropical fruits, and grapes. Anthocyanins are also found in
cereals, legumes, roots, tubers, bulbs, cole crops, grasses, and many other crops
outside of these categories. In general, the anthocyanins in most of the fruits and
vegetables are observed in concentrations from 0.1% up to 1.0% dry weight [15, 18].
Table 2 gives the concentrations of identified anthocyanins in a variety of plant
sources.
Table 2
Concentration and Plant Sources of Commonly Found Anthocyanins

Anthocyanins are found in the vacuoles of plants at the surface of the fruit and
vegetable skin, also known as the outer epidermal peel, or flower petal. Generally
speaking, as the vacuolar concentration of anthocyanins increases, the coloring of
the plant skin, flesh, or petal intensifies. Vacuolar concentration can also affect the
color hue, causing the difference between pink and deep red pigmentation [19].

1.2. Chemistry

Because anthocyanins act as pigments in a variety of fruits, vegetables, and other

plants, color intensity, hue, and stability are very important properties. These
properties are highly influenced by structure, pH, temperature, light, oxygen, and a
number of other factors. Structurally, anthocyanins undergo transformations with
changes in the pH, which has a dramatic effect on color. Studies have examined the
color and stability of anthocyanins at a range of pHs and the following scheme shown
with the cyanidin (5) structure is generally accepted, (\Fig. (3)). At a pH of
approximately 3 or lower, the anthocyanin exists as the flavylium cation and is
orange or red. As the pH is raised, kinetic and thermodynamic competition occurs
between the hydration reaction of the flavylium cation and the proton transfer
reactions related to its acidic hydroxyl groups. While the first reaction gives a
colorless carbinol pseudo-base, which can undergo ring opening to a yellow
chalcone, the latter reactions give rise to quinonoidal bases. Further deprotonation
of the quinonoidal base takes place at a pH of 6–7 with the formation of purplish blue
resonance-stabilized quinonoid anions [16, 20]. Due to the pH values typical for fresh
fruits and vegetables, each anthocyanin should accurately be represented by a
mixture of equilibrium forms [21].

Fig. (3)
Structural transformations of cyanidin (5) pigments with change in pH, modified from
[16].

The structure of each anthocyanin also has an effect on the color that is produced.
When evaluating the six common anthocyanidin compounds, the effect of hydroxyl
and methoxyl moieties on the resulting color could be elucidated. The hydroxyl group
at C-3, a position that is frequently glycosylated, is highly significant because it shifts
the color of anthocyanins from yellow-orange to red [22]. The anthocyanins can be
placed into two groups based on the colors and intensities at different pHs. Group 1,
consisting of the 3-glucosides of pelargonidin, peonidin, and malvidin, maintain
bluish colors with intensity and stability above pH 8. However, Group 2, of cyanidin,
petunidin, and delphinidin, show a shift from blue pigmentation back to reddish colors
above pH 8. Petunidin and delphinidin compounds were found to be very unstable
in the basic media, but interestingly, cyanidin was found to have similar stability as
the Group 1 compounds [52]. That is, the Group 1 aglycons with just one free
hydroxyl group on the B ring were found to give more blue colors and be more stable
in basic environments than the group 2 aglycous, with the exception of cyanidin
stability. The effects of glucosidic substitutions were also examined at different pHs.
The 3-glucoside and 5-glucoside anthocyanins also show bluer hues in alkaline
solutions as well as lower the color stability significantly. One of the explanations of
the color loss that follows glucosidic substitution is attributed to an enhanced
electrophilicity of the flavylium cation as a result of the electron-withdrawing effect
induced by introduction of the sugar moiety [53].

Along with variations in the number of sugars and position of attachment,

anthocyanins can also differ in acylating groups of the previously mentioned sugar
substitutions. Acylated anthocyanins indicate a low sensitivity to pH changes,
steadily producing a bluer color than non-acylated anthocyanins. Yet unlike the effect
of glucosidic moieties, acylated anthocyanins possess higher color stability in neutral
and alkaline solutions. This increased stability is due to the fact that acylated
anthocyanins are more resistant to hydration of the flavylium ion leading to an
equilibrium pushed towards more quinonoidal base forms [21, 54]. In effect, the
acylating groups encourage the production of the bluer pigmentation that glucosidic
moieties bring forth but also counteract the instability attributed to the sugars.

Because anthocyanins are highly reactive and therefore easily degraded, the
storage environment plays a critical role in maintaining pigmentation. Light and
temperature are both known to breakdown anthocyanins. Anthocyanins are best
kept in cool, dark environments as the presence of sunlight and high temperatures
result in the loss of pigmentation. The degradation of anthocyanins increases
proportionally with the heat of storage and exposure of pigments to temperatures
from 65°–90°C for only a short amount of time lowered the color half-life to just a few
hours. Anthocyanins have been successfully stored for several weeks at
temperatures around 2–4°C [55–58]. Relative to storage and preservation of the
anthocyanins is the bleaching effect of sulfur dioxide on these compounds. While
SO2 is often added to fruits and vegetables as an inhibitor of microbial growth, it also
acts as a nucleophile and attacks the flavylium ion of the anthocyanin, effectively
bleaching the pigment as it progresses to the colorless hemiacetal form [59–61].

Anthocyanins are also degraded by oxidative mechanisms involving the enzyme

polyphenol oxidase (PPO). PPO can be found in blueberries, strawberries, grapes,
and cherries and plays an important role in the browning of these fruit juices.
However, polyphenol oxidase cannot degrade anthocyanins on its own, but must be
in the presence of another substrate, such as caffeic acid, chlorogenic acid, or gallic
acid. All these acids are o-diphenolic compounds found in fruits and are involved in
the first step of polyphenolic oxidation. The acid is oxidized to its o-quinone form
which then oxidizes the anthocyanin to form brown polymers. The different
aglycones will result in different products with o-diphenolic anthocyanidins giving
products with part of the original acid incorporated. That is, the cyanidin degradation
products involve a coupled oxidation mechanism with partial regeneration of the
acid. The non-o-diphenolic anthocyanidins such as pelargonidin react with the
quinone to form adducts, still resulting in loss of pigmentation. Because the quinone
formation from acid is vital in polyphenol oxidation, reduction of the quinone would
effectively inhibit the oxidative degradation of anthocyanins. Addition of ascorbic acid
retards the color loss by acting as the hydrogen donor to convert the quinone back
into its acid form. As long as ascorbic acid is present in the reaction mixture, the
anthocyanins remain preserved in the presence of polyphenol oxidase [62–66].
However, ascorbic acid (vitamin C) on its own has an additional degradation effect
on anthocyanins. When ascorbic acid is present in the anthocyanin solution, the
introduction of oxygen causes the destruction of both compounds. Ascorbic acid
performs an oxidative cleavage of the flavylium ion in a manner analogous to the
attack by SO2 [67].

Another property of color hue and intensity is the presence of a copigment. A

copigment is another compound that is typically colorless, but when added to
anthocyanins, interacts with the anthocyanin and greatly enhances the color of the
solution [68]. Intermolecular copigmentation reactions are the result of associations
between anthocyanins and co-factors such as other polyphenolics, metal ions, or
organic acids to produce weak chemical bonds with increased physical and chemical
attributes. When polyphenol compounds act as cofactors towards anthocyanins, the
two compounds are held together by vertical hydrophobia stacking of the aromatic
nuclei. The flavylium cation is then stabilized by its interaction with the copigment
and avoids hydration which would form the colorless carbinol pseudo-base. Rather,
more flavylium ions are present in the solution leading to intense red colors [69, 70].
Commonly found organic copigments include catechin, epicatechin, procyanidin B2,
caffeic acid, p-coumaric acid, chlorogenic acid, myricitrin, and quercitin [71]. As
expected, the effect of copigmentation on the color and intensity of solutions varies
with anthocyanin structure and changing pH environments. The copigment effect
increases with both the number of methoxyl moieties and the number of glycosidic
groups on each anthocyanin. That is, copigmentation with pelargonidin, peonidin,
and malvidin would produce more intense colors than the other common aglycons
or a diglycosidic form would increase the intensity over a monoglycoside [68]. Also,
copigmentation increases color intensity at pH values from about 2 to neutral. This
indicates that a copigment stabilizes not only the flavylium ion but also the
quinonoidal base that is found at neutral pHs and leads to brighter red colors in acidic
media and brighter blue colors at neutrality [68]. Another form of copigmentation
found in blue flowers is the complexation of metal ions. These metalloanthocyanins
involve a supramolecular metal-complex pigment composed of simple anthocyanins,
ones that are commonly found in other flowers, flavone cofactors, and metal ions.
The metals include Fe3+, Al3+, Mg2+, and Ca2+ and are essential for the blue
pigmentation of the flower petals. All known metalloanthocyanin structures show a
chiral molecular stacking of the anthocyanins and flavone cofactors, which
distinguishes these pigments from other assemblies of compounds [72–74].

While copigmentation is used to intensify color hues by resisting hydration of the

flavylium ion, it does not necessarily increase the stability of the anthocyanins with
respect to temperature and the presence of light. Heating of the copigmented
anthocyanins causes dissociation of the complex resulting in a loss of color at the
same rate as non-copigmented anthocyanins. The influence of UV or visible light on
the stability of the anthocyanin-copigment complex is fairly similar to anthocyanins
without copigments. The pigment complex is shown to slow the degradation, but
given enough time, the complexed pigments will also become bleached [75, 76].

The resistance of anthocyanins toward hydration and nucleophilic attack is a

promising area and recently new compounds were reported in wine pigments. A type
of anthocyanin-pyruvic acid adduct has been studied recently because of an
increased stability to pH and SO2 bleaching. They are the pyranoanthocyanins
comprised of anthocyanins with the addition of pyruvic acid at C4 and a hydroxyl
group at C5 of the original anthocyanin structure. This forms a fourth ring which is
responsible for the increased stability found with this compound. The nucleophilic
attack of water and sulfur dioxide preferentially occurs at the C4 position of the
original anthocyanin structure. This converts the red flavylium ion into its colorless
hemiacetal form, causing discoloration of the solution. If this pyruvic acid adduct
adds to the C4 position, the molecule is blocked from nucleophilic attack and is
suddenly more stable at pH values up to neutral and in higher concentrations of
sulfur dioxide [61, 77].

1.3. Bioactivity and Health Properties

The ability of anthocyanins to impart color to the plants or plant products in which
they occur leads to an important role of attraction or repulsion of various animals,
birds and insects. They serve to attract animals, birds and insects for pollination in
flowers and seed dispersal in fruits [1]. The reddish color of Canadian shrub foliage
accentuates the black-colored fruits to birds leading to increased removal of the
black fruits [78]. One of the more commonly found anthocyanins, cyanidin-3-
glucoside, was reported to inhibit larval growth in tobacco worms making
anthocyanins agents of biological control as well [79]. California maple aphids flock
to and consume yellow leaves of the Japanese maple but tend to ignore the red
colored leaves [80]. Anthocyanins can serve as optical filters by protecting molecules
from being degraded by visible light. The silver beachweed, for example, holds large
amounts of thiarubrine A, a toxin acting as a defense mechanism from insects.
Thiarubrine A happens to be very photolabile and since the tree grows along the
California coast, it could easily be rendered inactive by the sunlight. However,
beachweed also contains a mix of two anthocyanins that serves to protect the
compound from sunlight by absorbing the visible light that would otherwise lead to
its destruction [81]. Anthocyanins can also protect leaves from a decrease in
photosynthesis by absorbing extra light that would otherwise be intercepted by
chlorophyll b. Anthocyanins are, in addition to optical filters, excellent scavengers of
free radicals. This is a property that has been exploited for its therapeutic purpose,
but is also evident in plant cells. Purified solutions can effectively scavenge almost
all species of reactive oxygen and nitrogen in a leaf [82].
While anthocyanins serve these important assignments in plants, they are currently
studied for their ability to protect against myriad human diseases. The natural
electron deficiency of anthocyanins makes these compounds particularly reactive
towards oxygen radicals. These antioxidative properties result from the chemical
structure of anthocyanins, particularly from the hydroxyl moieties of the C ring which
allows chelation of metal ions such as Fe or Cu. The antioxidative activity is also
increased by acylation of the sugar residues with aromatic acids. In fact, studies
show that anthocyanins can have higher antioxidative activity than vitamin E,
ascorbic acid, and β-carotene [83, 84].

The antioxidative activity of anthocyanins has been exhaustively examined [25, 26,
43, 85–91]. Anthocyanin-rich plants have historically been used to treat a number of
symptoms and diseases, such as the improvement of visual acuity. Administration
of cyanidin-3-rutinoside enhanced night and overall vision due to its effect on
rhodopsin regeneration. Studies show that the anthocyanin accelerates formation of
an intermediate to regenerate the G-protein-coupled receptor in the retina of the eye
[92]. Protection from heart attacks is also associated with administration of
anthocyanins, particularly in the form of grape juice and wine but also from other
sources [93, 94]. This role is attributed to the ability of these products to reduce
inflammation, enhance capillary strength and permeability, and inhibit platelet
formation. The mechanism of vasorelaxation that was observed is due to increased
nitric oxide release [93]. Anthocyanins can even be shown to aid in the prevention
of obesity and diabetes. Studies show that anthocyanin pigments from purple corn
inhibit both body weight and adipose tissue increases. The symptoms of
hyperglycemia that can follow a high-fat diet were also suppressed with ingestion of
this cyanidin-3-glucoside [95].

Interestingly, some antioxidants have shown anti-inflammatory activity as well by

interfering with the signaling mechanisms that regulate the cyclooxygenase (COX)
gene. Anthocyanins from tart cherries and hibiscus flowers were found to inhibit the
activities of COX-1 and COX-2 in vitro [91, 94, 96] and because the COX-2 gene
plays a role in tumorigenesis, anthocyanins can affect the spread of cancer through
a variety of mechanisms. Apoptosis is a common method of eliminating tumor cells
and apoptosis-inducing agents are expected to be excellent anticancer or antitumor
drugs. Some anthocyanidins and their glycosides, specifically delphinidin, cyanidin,
and petunidin, induce apoptosis of human promyelocytic leukemia (HL-60) cells.
These compounds differ from pelargonidin, peonidin, and malvidin by the number of
hydroxyl groups on the B ring and it appears the specific ortho position of the
hydroxyls is necessary for the apoptosis action [97, 98]. The same anthocyanidins
can also inhibit activator protein 1 (AP-1) transcriptional activity and cell
transformation in certain mouse epidermal cells. AP-1 is a transcription factor mat
promotes carcinogenesis and so the associated anthocyanins show encouraging
properties against the progression of cancer. Again, since only delphinidin,
cyaniding, and petunidin exhibited this inhibition, the researchers postulated that the
structure of the B ring produces the observed activity [99]. The chemopreventive
studies of anthocyanin pigments have mostly been limited to the aglycone form. This
does not suggest that anthocyanins lack chemopreventive activity, in fact they do
possess the same properties, but to a different degree. The biological activities of
anthocyanins tend to increase with decreasing number of glycosides and an
increasing number of hydroxyls [96]. Given the increasing number of research
studies highlighting and examining the health benefits, it is reasonable to predict that
over time, an increasing array of health benefits and applications to improve human
health and nutrition will follow as well as a greater understanding of their mode of
actions as well as their bioavailability and final form at the targeted tissue site(s) of
action.

1.4. Biosynthesis and Metabolism

The rainbow of colors and diverse functions attributed to anthocyanins has made
these pigments of great interest to biochemists and molecular biologists. The
biosynthetic pathway of anthocyanins is one of the most extensively studied
pathways of plant secondary products and has lead to the identification of the genes
and enzymes responsible for the biosynthesis of anthocyanins. A representation of
the general biosynthetic pathway is shown in Fig, (4); this pathway was mostly
formed from the investigation of three main species: maize (Zea mays), snapdragon
(Antirrhinum majus), and petunia (Petunia hybrida) [100, 101].

Fig. (4)
General representation of the biosynthetic pathway of anthocyanins, specifically
pelargonidin (18), cyanidin (5), and delphinidin (19), modified from [101]. The
intermediate compounds consist of naringenin chalcone (11), naringenin (7),
dihydrokaempferol ...

The precursors for the synthesis of all flavonoids, including anthocyanins, are
malonyl-CoA and p-coumaryl-CoA. Chalcone synthase (CHS) catalyzes the first
committed step in the biosynthetic pathway by sequential condensations of three
acetate moieties from malonyl-CoA with p-coumaryl-CoA to yield a tetraketide
intermediate that immediately cyclizes to form naringenin chalcone (11), a
tetrahydroxychalcone [100]. The accumulation of this chalcone in plant tissues is
quite rare; therefore, naringenin chalcone undergoes a Michael-type addition,
catalyzed by chalcone isomerase (CHI), to the stereospecific (2S)-naringenin (7)
with the nucleophilic attack of the deprotonated hydroxyl on the α,β-unsaturated
alkene. This isomerization can occur spontaneously, without CHI, but does so at a
much slower rate [101]. Naringenin, a flavanone, is then converted to a
dihydroflavonol (dihydrokaempferol, 12) by hydroxylation at position 3 catalyzed by
flavanone 3-hydroxylase (F3H). From this point, the dihydroflavonol can be
converted into three different molecules, leading to three different anthocyanidins.
Dihydrokaempferol can be converted to a dihydroquercetin (13) with hydroxylation
at me 3′ position, catalyzed by flavanone 3′-hydroxylase (F3′H), which will give
cyanidin pigments (5). Hydroxylation at the 3′ and 5′ positions to give
dihydromyricetin (14), catalyzed by flavanone 3′,5′-hydroxylase (F3′5′H), will
eventually lead to the blue delphinidin pigments (19). Finally, no additional
hydroxylations on the B ring will give the pelargonidin pigments (18) [101].

The dihydroflavonols are reduced by dihydroflavonol 4-reductase (DFR) to

leucoanthocyanidins (leucopelargonidin 15, leucocyanidin 16, and leucodelphinidin
17) in the course of the biosynthetic pathway. The reductase enzyme catalyzes the
stereospecific reduction of the ketone at position 4 to give (2R, 3S,4S)-
leucoanthocyanidins [100]. Anthocyanidin synthase (ANS) catalyzes the next step,
conversion of the colorless leucoanthocyanidin to 2-flaven-3,4-diol, which is then
glycosylated by flavonoid 3-O-glucosyltransferase (3-GT) and transported into the
vacuole where it is finally converted to the colored flavylium ion [100]. The ANS
belongs to a family of 2-oxoglutarate-dependent oxygenases and, therefore,
requires the presence of Fe2+, 2-oxoglutarate, molecular oxygen, and ascorbate;
however, the reaction sequence does not require any additional dehydratase,
despite the involvement of a formal dehydration step [102, 103]. In the first step, ANS
uses the ferrous ion, molecular oxygen, and 2-oxoglutarate to form an oxoferryl-
enzyme complex, along with succinate and CO2. This oxoferryl-enzyme complex
catalyzes the hydroxylation at position 2 followed by spontaneous dehydration to
give 2-flaven-3,4-diol. This product immediately isomerizes to the more stable
pseudo-base (3-flaven-2,3-diol) at neutral pH [104]. The enzyme 3-GT now
catalyzes the glucosylation at position 3 of the pseudo-base which is subsequently
transported into the vacuoles and converted to the colored flavylium ion due to die
acidic vacuolar conditions [100].

As natural anthocyanins are more complex than a simple monoglucoside, the 3-O-
glycosylation is almost always a prerequisite for further modification, such as
additional glycosylation, methylation, and acylation [101]. Glucosylation at the 5
position is catalyzed by anthocyanin 5-O-glucosyltransferase (5-GT) and uses UDP-
glucose as a co-factor. The placement of a sugar moiety at die 5 position allows for
more stable complexes in copigmentation of anthocyanins as well as modifying die
pigment color, specifically creating a purplish-blue hue [100, 104]. The glycosylation
of roses (Rosa hybrida), differs from the normal biosynthetic pathway in that it relies
on a single enzyme to achieve glucose attachment at the 3 and 5 positions of die
pseudo-base, 3-flaven-2,3-diol. Interestingly, the 3-O-glycosylation is not a
prerequisite for this species, but rather glycosylation occurs first at die 5-OH and
then at the 3-OH group. The enzyme, UDP-glucose: anthocyanidin 5,3-O-glycosyl
transferase, is referred to as RhGT1 being that it is unique to members of the R.
hybrida family. After the first glycosylation, anthocyanidin 5-O-glucoside is unstable
without the additional glycosylation at its 3-OH position and so cannot be isolated;
however, the unorthodox pathway was identified because RhGT1 cannot use
anthocyanidin 3-O-glucoside as an acceptor, only the unglycosylated or 5-O-
glucosidic anthocyanidin [105].
Methylation of the anthocyanidin hydroxyls and acylation of the sugar moieties are
common methods used to affect color hue and stability of the pigment as well as aid
in copigmentation. Methoxyl groups are found in three of the main anthocyanidins,
peonidin, petunidin, and malvidin, and lead to more stable compounds with the
methylation of the reactive hydroxyl groups. Acylated anthocyanins indicate a low
sensitivity to pH changes by increasing water solubility, steadily producing a bluer
color than non-acylated anthocyanins. Acylation also leads to higher structure
stability by protecting the suger moieties against degradation by glucosidase. The
use of aromatic acids for acylation contributes to intra- and inter-molecular stacking
seen in copigmentation, resulting in intensifying the blue pigmentation. Anthocyanin
acyltransferases (AATs) catalyze the regiospecific acyl transfer from acyl-CoA to the
appropriate sugar of the anthocyanin. The enzymes are classified on the basis of
their acyl-donor specificity into two categories: aliphatic and aromatic
acyltransferases [100, 106]. Once the biosynthesis is completed, including
modifications such as additional glycosylation, methylation, acylation, the final
product is transported into the plant vacuole for storage. The acidic environment of
this vacuole is responsible for the conversion of the pseudo-base anthocyanin to its
colored flavylium ion form. While the biosynthetic pathway of anthocyanins has been
well studied, the vacuolar transportation has several possible mechanisms. The
proposed mechanisms consist of: pH-dependent transport [107], multidrug and toxic
compound extrusion (MATE) protein-mediated system [108], MgATP-energized
transport by ATP-binding cassette (ABC) [109–111], or 24-kDa vacuolar protein
(VP24) precursor protein transport [112]. Taken together, these findings indicate that
different plant species may use distinct mechanisms to distribute flavonoids among
subcellular compartments, and multiple mechanisms may be used in individual
species [100].

The metabolism of dietary anthocyanins dictates how active the compounds are in
the human system. While extensive studies have been performed, more metabolic
studies are needed to fully ascertain the activity of anthocyanins as the knowledge
of these compounds increases. The early methods for testing anthocyanin
absorption consisted of measuring the presence of red pigments in urine after oral
administration. The evolution of testing methods has lead to observing anthocyanin
absorption in plasma and urine to determine both location of absorption and rate of
excretion. For many years, studies reported very low numbers for absorption of
anthocyanins after oral administration, on the order of 0.004% to 0.1% of the intake,
and indicated a rapid absorption and excretion with time to the maximum
concentration to be 1.5 h for plasma and 2.5 h for urine [113]. However, most of the
analyses were performed with UV-Vis detection after acidification, under the
assumption all the anthocyanins would be converted into the colored flavylium form,
and it is possible that some forms existing at neutral pH could not be colored due to
chemical reactions within the plasma or urine [113]. Also, the common technique
involved freezing and storing the urine and plasma samples before analysis but
chromatograms of samples immediately after collection showed additional peaks
that had degraded in the chromatograms of frozen samples [114]. Taking into
account the special conditions on the analysis of anthocyanins, the metabolic
products found in urine samples have included methylation, Phase II
glucuronidation, and sulfated derivatives [114, 115]. The common methylate
derivatives are cyanidin glycosides methylated to peonidin glycosides but other
glucuronyl and sulfate derivatives were not limited to any specific aglycone. Even
with the observation of these anthocyanin metabolites, the majority of ingested
anthocyanins are not recovered and therefore continued investigation is necessary
to determine the fate of the compounds in the body. As anthocyanins are rapidly
degraded by intestinal microflora, this additional metabolism could account for the
unrecovered anthocyanins [116], yet the development of a quantitative marker to
better understand the bioavailability and in situ concentration is needed.
Anthocyanidin glycosides are hydrolyzed by the microflora with cleavage of the
protective 3-glycosidic linkage. The released aglycones are very unstable molecules
under any condition, but in the neutral pH of the physiological conditions, they are
spontaneously degraded into monomeric phenolic acids and aldehydes, specifically
protocatechuic acid, syringic acid, vanillic acid and phloroglucinol aldehyde
[116,117].

Go to:

2. ANALYTICAL METHODOLOGIES

To adequately cover the analytical methodologies of anthocyanin compounds, the

complete process of analysis is reviewed from the initial extraction of anthocyanins
from plant tissues, the isolation and purification of anthocyanins from the crude plant
extract and, finally, the structure elucidation or identification of each particular
anthocyanin. The analysis of anthocyanin is complicated because of their ability to
undergo structural transformations and complexation reactions, as discussed. Also,
these compounds are difficult to measure separate of other flavonoids because they
have similar structures, see Fig. (2), and reactivity characteristics. Finally, pure
anthocyanin standards, which are integral for accurate quantification purposes, are
not readily available for purchase and are not easily purified from plant sources for
use in research in part due to the instability issues previously highlighted [18].

The specific methods of extraction from the plant tissues are best discussed after a
description of the analytical methods. There are a number of different sample
preparation techniques that are somewhat dependent on the analytical techniques
being utilized. However, the overall concept involves dried, powdered plant material
and extracted with alcohol under cold conditions, because of the susceptibility of
anthocyanins to degradation by high temperatures. The extracting solution should
be slightly acidic to maintain the flavylium cation form, which is red and stable in
highly acidic medium, but not too acidic to cause partial hydrolysis of the acyl
moieties in acylated anthocyanins [18,118,119].

The most common method for analysis of anthocyanins is high performance liquid
chromatography (HPLC). This often refers to separation methods by HPLC in
conjunction with identification methods such as UV/Vis spectrophotometry (LC/UV),
mass spectrometry (LC/MS), or nuclear magnetic resonance (LC/NMR) to elucidate
the anthocyanin structures [118]. Another common technique used for separation of
anthocyanins from crude plant extracts is capillary electrophoresis (CE) [1]. Both of
these common separation methods are demonstrated in a side-by-side comparison
in Fig. (5). While a complete description of the current abilities and recent advances
of HPLC and CE instrumentation for analyses of anthocyanins will be covered in this
review, some historical separation techniques should also be briefly described, such
as paper chromatography (PC) and thin-layer chromatography (TLC) [18, 79].

Fig. (5)
A side-by-side comparison of an LC analysis (upper) and CE analysis (lower) of a
2002 vintage Tannat red wine. The conditions are described in the original paper
and the peak identification for each follows. The LC chromatogram (A): 1,
delphinidin-3-glucoside ...

2.1. Historical Chromatographic Methods

Paper chromatography (PC) and thin layer chromatography (TLC) are two
techniques that have historically been used for the separation of anthocyanins, but
with the development of HPLC and CE methodology, there is little advancement to
describe for these two basic chromatographic techniques. Paper chromatography
was one of the first methods employed for the isolation and purification of certain
anthocyanins and depending on the specific sample and the different mobile phases,
PC did permit good resolution for some pigment mixtures. Unfortunately, this
technique did not allow large quantities of pure anthocyanins to be obtained and
generally needed long development times [118].

With the introduction of TLC, some specific advantages over PC were realized in
that TLC requires lower quantities of anthocyanin mixtures for analysis, requires
shorter times for elution, and achieves better resolution. Due to the variability in
distance the compounds travel with different solvent systems, it is necessary to use
reference compounds when using TLC to isolate anthocyanins from plant extracts.
While it is an advancement in technology, TLC does offer results directly comparable
to PC and, along with PC, does not permit pure anthocyanins to be obtained in large
quantities. In spite of their drawbacks, TLC and PC, are still being used as routine
techniques in many laboratories, due to their low cost and the constant development
of better supports and mobile phases [118].

2.2. Capillary Electrophoresis

Capillary electrophoresis (CE) separates compounds based on differences in their
electrophoretic motility and possesses excellent mass sensitivity, high resolution,
low sample consumption, and minimal generation of solvent waste. The CE
instrumentation consists of two reservoirs and a fused silica capillary tube containing
a carrier electrolyte and a high voltage source. A sample is introduced into the
capillary tube at the anode and the mobile phase will move some components of the
sample towards the cathode while others are held back at the anode by charge
attraction [1, 79]. In capillary zone electrophoresis (CZE), the most common method
used for separation of anthocyanins, the migration of a particular compound depends
on its charge-to-size ratio. That is, the total migration time for positively-charged
smaller molecules is longer than that for molecules of lesser charge and/or larger
size [121]. The commonly accepted method for separation of anthocyanins, using
basic media, involves a sodium tetraborate buffer at pH = 8.4 with 15% methanol.
The separation is carried out in positive polarity mode and a positive electroosmotic
flow with migration of the compounds from anode to cathode. Detection of the
compounds is performed by UV/Vis at 599nm, given the blue quinonoidal base form
Fig. (3) of the anthocyanins, and often include collection of the full spectrum, from
200 to 599 nm, for each peak [120, 122, 123].

The applicability of CZE in basic media is limited by the instability of anthocyanins in

basic environments; therefore, the separation of anthocyanins from plant extracts by
CZE will use acidic media and configure the system to run from cathode to anode.
The acidic media will ensure the anthocyanins remain in the protonated flavylium
form, (Fig, (3)), and the migration of anthocyanins is towards the anode, reversing
the electroosmotic flow. The speed by which they are drawn to the anode depends
on their charge-to-size ratio, as seen with normal electroosmotic flow [124]. With
acidic media and the acidic form of anthocyanin compounds, another problem arises
with the formation of ionic interactions between the flavylium cation and the anionic
silanol groups covering the capillary tube. For this reason, da Costa et al. [124]
optimized their acidic media from a phosphate buffer at pH 1.5 up to pH 1.8, after
which no interaction between the silanols and the anthocyanins could be detected.
Bicard et al. [125] shortened the analysis time by introducing a cationic surfactant,
cetyltrimethylammonium bromide (CTAB), to interact with the negatively-charged
silica on the capillary wall ensuring free-flowing anthocyanin cations in the presence
of the anionic silanols. The concentration of CTAB added to the running buffer
remains below the critical micelle concentration and effectively separates the
compounds, utilizing a reversed electroosmotic flow as they migrate from the
cathode to the anode. This experiment was performed using running buffer and
sample media at a pH 2.1 and produced excellent separation of anthocyanin peaks
as detected by UV/Vis at 520 nm [125].

There is an additional chromatographic method of CE known as micellar

electrokinetic chromatography (MEKC) that has been used to efficiently separate the
anthocyanins of elderberry pigments and grape skins. The separation principle of
MEKC is based on partitioning of the solute, consisting of non-charged anthocyanins
in a neutral environment, between an aqueous phase and a micellar pseudophase.
The anthocyanins, all cyanidin derivatives, were separated using sodium dodecyl
sulfate (SDS) in a phosphate-borate buffer at pH 7.0 and show a UV/Vis absorbance
maximum at 280 and 560 nm. While many MEKC separations might use a shorter
wavelength such as 214nm, anthocyanins are best monitored at 280 or 560 nm [124,
126].

2.2.1. CE with MS Detection

Capillary electrophoresis has grown in prominence because of its abilities to

separate complex mixtures of anthocyanins. Traditionally, the common detector
coupled to CE instrumentation has almost exclusively been UV/Vis
spectrophotometry, but the more sophisticated detection method of mass
spectrometry is growing. The use of CE/MS provides important advantages given
the combination of the great separation capabilities of CE and the power of MS as
an identification and confirmation method [127]. Regardless of the large number of
ionization techniques available for mass spectrometry, the principal interface used
for direct coupling of CE to MS has been electrospray ionization (ESI). ESI-MS
effectively couples liquid phase separation methods, such as CE, with the gas
phase-based MS. The mechanism of ESI, a soft ionization method that produces
gaseous ions from highly charged evaporating liquid droplets, causes a challenge in
coupling these two techniques because most running buffers used in CE are non-
volatile substances. This limitation must be accounted for when choosing a solvent
system for a CE-ESI-MS analysis of anthocyanins compounds [127, 128].
Additionally, a sheath liquid should be introduced into the system, running with the
system, to stabilize the spray by providing a steady stream out of the capillary.
Bednar et al. [129] examined different solvent systems for the analysis of
anthocyanin dyes, comparing an acidic running electrolyte (pH = 2) and a basic one
(pH = 9) to determine what environment provides the better results, the results of
this comparison are demonstrated in Fig. (6). Glucosylated anthocyanins were
efficiently separated with an acidic buffer of monochloroacetate-ammonium at a pH
of 2 using a sheath liquid of 80% aqueous methanol with 0.25 % acetic acid (v/v).
The migration order corresponded to increasing molecular mass as predicted in
acidic media. However, the basic running buffer, comprised of borate-ammonium at
a pH of 9, used the same sheath liquid and achieved a better overall separation of
the anthocyanins, specifically of diastereomeric pairs that could not be distinguished
in the acidic CE-ESI-MS analysis. These observations demonstrate a basic running
electrolyte should be used if possible for optimal separation of anthocyanin
compounds, but an acidic media is acceptable under certain conditions. While most
borate buffers are made with disodium tetraborate which is non-volatile and not
appropriate for ESI-MS, this borate buffer was made with boric acid having a similar
volatility to acetic acid [129].

Fig. (6)
Separation of common glucosylated anthocyanins in acidic electrolyte at pH=2 (left)
and basic electrolyte at pH=9 (right) with appropriate mass spectrometry data
included for each anthocyanins. The two figures demonstrate the difference in
migration order ...

2.3. High Performance Liquid Chromatography

As reviewed, high-performance liquid chromatography (HPLC) has been the method

of choice for the qualitative and quantitative analysis of anthocyanins. This is
because of the capability of LC in preparation of samples on the gram scale using
preparative-HPLC (prep-HPLC) and the purification of samples on the milligram
scale with semipreparative-size LC columns and on the microgram scale with
analytical-size column. HPLC is also used for identification of individual natural
compounds in the plant matrices using coupled techniques such as HPLC with
UV/Vis spectrophotometry detection (LC-UV), with mass spectrometry detection
(LC-MS), or with nuclear magnetic resonance detection (LC-NMR). The extent of
HPLC and its various detection methods will be discussed in depth as well as the
recent advances and future directions regarding the analytical methodologies of
anthocyanins.

As illustrated in Table 3, there is not a single standard procedure for the analyses of
anthocyanins using HPLC. Rather, there are a wide variety of columns and
parameters that have been used in anthocyanin characterization from the same plant
source resulting in equivalent separations. Some specific methods are described in
detail, and there are some general trends for anthocyanin analysis. A high proportion
of isolation methods for anthocyanins are generally run on reverse-phase columns,
such as octadecyl silane (ODS), polystyrene, or phenyl-bonded columns [124]. Also,
HPLC methods tend to utilize gradient solvent systems of acetonitrile-water or
methanol-water with a small amount of acid to lower the solution pH and increase
stability of the anthocyanins as mentioned with CE methodology. These solvents are
most popular due to their compatibility with gradient methods for isolation and the
various detection methods coupled to the HPLC used for identification. To obtain
reproducible results using this instrumentation, the pH of the mobile phase and
temperature of the column must be controlled due to the instability of anthocyanin
compounds in changing pH and temperature environments. For optimal results,
acidic mobile phases, lower than pH 2.0, are employed to ensure the anthocyanin
remains in the more stable flavylium form and to reduce peak tailing of the
chromatogram [1, 119, 124]. In reverse-phase chromatography, the retention time
decreases with increasing polarity, which corresponds to increasing number of
hydroxyl moieties on the flavylium ion and the elution order of delphinidin, cyanidin,
petunidin, pelargonidin, peonidin, and malvidin. Due to reverse-phase elution,
diglycosylated anthocyanins will have the lowest retention time, followed by
monoglycosylated anthocyanins, aglycones, and lastly the acylated anthocyanins
[121, 124]. In the acidic media, the flavylium cation is red-colored and gives an
absorption maximum around 520 nm, which avoids interference from other
flavonoids that may be present in the plant extract. Because of this unique
absorbance maxima, anthocyanins have been accurately identified and quantified
from very crude plant extracts [124]. Normal phase systems, or those using
unmodified silica gel, are not effective for the separation of anthocyanins due to the
polarity of these compounds [119].

Table 3
HPLC Methodology of Analysis for Selected Anthocyanins
2.3.1. HPLC with UV Detection

The most frequently used detection method for HPLC is UV/Vis spectrophotometry.
This is due to the prevalence of spectrophotometric instrumentation in laboratories
studying anthocyanin chemistry, both in the industrial and academic sectors.
Anthocyanins have a unique absorption maximum at ~520 nm which sets these
compounds apart from other flavonoids in the plant extract and simplifies the
resulting chromatograms for isolation and purification. An example of an HPLC
chromatogram with detection at 520 nm can be seen in Fig. (7). UV/Vis
spectrophotometry was not used for identification of anthocyanin compounds until
the introduction of diode-array technology, which allowed for the accumulation of
UV/Vis spectral data libraries, aiding in peak identification. Diode array detection
coupled to HPLC (LC-UV-DAD) performs a complete spectrophotometric scan on
each peak as it elutes and provides a unique chromatogram for each anthocyanin
compound that can subsequently be compared to other compound spectra and used
for identification purposes [119, 121, 124]. Generally, the spectral characteristics of
a particular anthocyanin are related to the hydroxylation pattern of the aglycone
structure from which it is derived. The anthocyanidin with just one oxygenated
position on the B-ring, pelargonidin, gives an absorption maxima at approximately
520 nm while di-oxygenated anthocyanidins, cyanidin and peonidin, give maxima at
535 nm and tri-oxygenated anthocyanidins, delphinidin, petunidin and malvidin, give
maxima at 544 run. Glycosidic substitution of the anthocyanins also affects the
spectral characteristics specifically in that 3-rnonoglycosides generate a ratio of
absorbance at 440 run to absorbance at the wavelength maxima that is twice as
large as the 3,5-diglycoside anthocyanins [176, 177]. The various sugars commonly
seen in the glycosidic substitution do not seem to have an effect on the observed
spectra but do significantly change the retention times of anthocyanins. Finally,
acylation of anthocyanins result in a fairly significant increase in the retention time
compared to nonacylated anthocyanins [177].
Fig. (7)
Base peak chromatogram from the HPLC analysis of a South African Pinotage red
wine with detection at 520 nm. The separation of anthocyanins is demonstrated and
peak identification can be found in the original paper. Modified from de Villiers et al
[130 ...
2.3.2. HPLC with MS Detection

The technique of coupling HPLC and MS instrumentation has had a significant effect
on the quantitative and qualitative analytical methodology of anthocyanins in the last
decade. This combination, LC-MS, offers the separation advantages of liquid
chromatography combined with the identification advantages of mass spectrometry.
MS is a very sensitive method of molecular analysis and due to its separation by
mass, good selectivities can be obtained and the identification of individual
compounds in a mixture of compounds is permitted [1]. However, as observed with
capillary electrophoresis, the physical coupling of these two techniques presents
some problems, such as the amount of column effluent that has to be introduced in
the MS system, the composition of the eluent, and the type of compounds to be
analyzed. The most common interface for LC and MS employs an analytical HPLC
which utilizes smaller columns and lower flow rates, therefore, facilitating the
coupling of LC to MS instrumentation. The analysis of specific types of compounds
requires an appropriate ionization technique for introducing the molecular ions into
the instrument and, to date, there is no universal interface that has been constructed.
The ionization techniques that are most suitable for anthocyanin chemistry are
continuous-flow fast-atom bombardment (CF-FAB), matrix-assisted laser desorption
ionization (MALDI), atmospheric pressure chemical ionization (APCI), and
electrospray ionization (ESI) [119, 124]. For the first few years after the introduction
of LC-MS interface systems, CF-FAB was the most common MS technique used.
However, as LC-MS technology developed, some disadvantages with fast atom
bombardment were made obvious, specifically the fact that analyses must be
preceded by purification and dissolution of the sample in a polar matrix before
injection [178]. As observed with CE-MS, ESI is the most common LC-MS interface
today because it avoids the problems seen with CF-FAB. ESI is a soft ionization
method that produces gaseous ions from highly charged evaporating liquid droplets
which makes it appropriate for small plant metabolites such as anthocyanins. Also,
ESI has the ability to introduce liquid samples to the MS with minimum manipulation,
successfully avoiding the common problems seen with other LC-MS interfaces [179].
Different types of mass analyzers are also available for various LC-MS interfaces
such as magnetic sectors, time-of-flight (TOF) analyzers, quadrupole mass filters,
quadrupole ion traps, and ion cyclotron resonance. The TOF analyzer is often
coupled with MALDI because of the precise time of ionization seen with MALDI. The
quadrupole ion trap is the most common type of analyzer for LC-ESI-MS which
allows for simplified interpretations of complex mixtures but still includes
identification of minor peaks [121].

The application of tandem MS (MS-MS or MSn) can be used to characterize

individual compounds in a mixture or to identify the structure of compounds by
separate ionization and fragmentation steps. An important mass spectrometer
structure elucidation is the triple quadrupole ion trap because it can serve as an
exceptionally high specificity detector due to a mass filter capable of transmitting
only the ion of choice. There are three quadrupoles in which the first (Q1) and third
(Q3) quadrupoles act as mass filters, and the middle (Q2) quadrupole is employed
as a collision cell, where the ionization happens [180]. This filtering process works
by mass to charge ratios (m/z) and allows for the study of fragments which are crucial
in structural elucidation. For example, the Q1 can be set to filter out an ion of a known
mass, which is fragmented in Q2, and Q3 can then be set to scan the entire m/z
range, giving information on the sizes of the fragments made as well as information
on that one specific ion fragment. Thus, the structure of the original ion can be
deduced [181]. However, as was mentioned, electrospray ionization has been an
efficient ionization technique for LC-MS but is a soft technique and does not tend to
produce many fragments, which are essential for MS/MS. Because of this, LC-
MS/MS will use two methods of ionization, ESI to introduce a parent ion into the
mass spectrometer and then collision-induced dissociation (CID) to enhance the
fragmentation of the parent ion [119]. An example of LC-ESI-MS/MS, demonstrating
precursor-ion analysis and product-ion analysis by Tian et al. can be seen in Fig. (8)
[135].

Fig. (8)
LC-ESI/MS/MS of anthocyanins in black raspberries. (A) HPLC chromatogram of
anthocyanins at 520 nm. Peak labels: (1) cyanidin 3-glucoside (cy 3-gluc); (2)
cyanidin 3-sambubioside (cy 3-sam); (3) cyanidin 3-(2-xylosylrutinoside) (cy 3-
xylosylrutin); (4) ...
2.3.3. HPLC with NMR Detection

While LC-UV-MS and LC-MS-MS are very powerful analytical methods for
anthocyanins, the use of NMR detection combined with HPLC separation (LC-NMR)
surpasses the previous methodologies in structural elucidation of unknown
compounds in crude plant extracts. When it was first introduced, LC-NMR struggled
due to a lack of sensitivity, but with progress in solvent suppression, pulse field
gradients, probe technology, and high-field magnets this technique has only grown
in popularity. The physical interfacing of these two instruments was not difficult, but
observing the analyte response in the presence of nondeuterated solvents caused
serious problems until the development of solvent suppression techniques [118].
Also, the precise separation procedures with HPLC often employ gradient solvent
systems, which cause the resonances to change frequency during analysis in the
gradient mode but with increased pulse field gradient technologies, this problem has
also been alleviated [119].

At the current time, LC-NMR methodology is mostly limited to 1H NMR spectra or,
under certain conditions, 13C NMR spectra, but only when the peak concentration of
interest is high and inverse detection experiments are acceptable. Due to the low
natural abundance of 13C isotope, the sensitivity for direct measurement of 13C-NMR
using LC-NMR instrumentation has not yet been developed. When measuring the
1H spectra of the main constituents of a crude extract, this can be done in what is

called on-flow mode. The on-flow operation requires a fairly high concentration of
the crude plant extract (at least 1 mg injection) and the spectra are acquired
continuously during the separation; what is observed with on-flow operation is a 2-D
NMR spectrum with a low sensitivity. For more precise measurements or 2-D
correlation experiments, the LC-NMR can be operated in the stopped-flow mode
which halts the flow of solvent, for a short amount of time, when the required peak
reaches the NMR flow cell. In this mode, various 2-D experiments can be performed
such as COSY, NOESY, HSQC, and HMBC, because stopping the mobile phase
flow allows for a large number of scans (or transients) to be taken for a given LC
peak [118,119].

2.4. Sample Preparation

As with the isolation and quantification methodologies, there are a numerous

methods for preparing samples for analysis and the selection of the most appropriate
method depends on the sample being analyzed and individual results desired.
Generally speaking, liquid anthocyanin-containing samples such as juices, syrups,
wines, and biological fluids require very little preparation before analysis; but
regardless of the selected preparation, it must be done under cold conditions, due
to the susceptibility of anthocyanins to degradation at high temperatures [55–58].
The majority of references analyzing red wine samples have progressed to simply
filtering the samples through a 0.45 μm filter before injection or even just a direct
injection method for anthocyanin analysis [85, 130, 167–169]. When testing
anthocyanins in biological fluids, such as urine or plasma, the samples are generally
acidified, to maintain the anthocyanin in its flavylium form, filtered and injected into
the LC [31, 114, 135]. Solid anthocyanin containing samples, however, require the
samples to be homogenized before a representative sample can be extracted. For
seeds or dried plant materials, the samples must first be crushed using a mortar and
pestle or ground using a coffee grinder, depending on the specific sample. Once the
samples have been powdered and mixed, a representative aliquot can easily be
removed for a smaller scale extraction [121]. Berry samples are different in that when
they are fresh, they need to be crushed and filtered, in order to isolate and dry the
solid residue which subsequently undergoes extraction [175]. Berries can also be
individually frozen, stored for a short amount of time, and ground in liquid nitrogen
allowing for large amounts of berries to be ground together and completely mixed
before a representative subsample is removed [182].

Once the plant tissues have been prepared, the next step involves solid-phase
extraction (SPE) or liquid-liquid extraction (LLE). Whichever method is preferred, it
must allow for recovery of the anthocyanins avoiding any chemical modification.
Anthocyanins are soluble in polar solvents, due to the hydroxyls and sugars
attached, and are therefore commonly extracted with methanol or ethanol. Because
of the instability of the compounds, extractions are performed under cold conditions,
not higher than 30°C, and with small amounts of acid to give an environment of pH
2.0 or below. The most common acids used for extraction are hydrochloric (HCl) or
formic acid; however, the use of these acids can cause hydrolysis of acylating groups
in certain anthocyanins [1, 121, 124]. Because of this, acylated anthocyanins should
be extracted with solvent containing HCl below a concentration of 0.12 mol/L or
containing organic acids (e.g., acetic or formic acid) [183]. When extracting
malonated anthocyanins, extractions are proceeded with weak acids, such as
tartaric or citric acid, to avoid hydrolysis of the dicarboxylic acid [79]. Acetone is also
used for LLE with efficient and more reproducible results than using methanol or
ethanol. This technique works well because solvent concentration can be done at
the lower temperatures needed for anthocyanins and also avoids problems with
pectins that are observed with the other LLE techniques. The pectins that are found
in the walls of some anthocyanin containing berries can cause problems when in the
acidic extraction environments, forming a turbid extract that slows the procedure;
however, acetone contains pectin-clotting properties and rapidly produces a clean
anthocyanin extract [184, 185]. For solid-phase extraction, popular methods utilize
C18 cartridges or Sephadex LH-20 for the initial separation of the anthocyanin
extracts from the crude plant sample. The polar anthocyanins bind to these
adsorbants through unsubstituted hydroxyl groups and are separated from other
plant compounds using solvents of increasing polarity. Investigations have shown
SPE to be a fast and highly efficient method for the separation of crude anthocyanin
solutions from plant material [1, 124, 186]. With the collection of the anthocyanin
extract, the next step is to remove the solvent to a known amount and continue with
the analysis. Samples which need to be stored are generally stored as dried
samples, with the solvent rotovapped and lyophilized, in cold, dark environments to
avoid decomposition of the anthocyanins.

2.5. Phytochemical Investigation Procedures

After the anthocyanin extract has been separated from the plant tissue, the extracts
are evaluated to determine whether additional preparation is needed before the
purification and quantification methods. Generally, because the results of SPE or
LLE give crude anthocyanin extracts, additional separation methods are utilized to
purify the sample before analysis. The appropriate techniques for a preliminary
separation include large-scale chromatographic techniques, such as normal-phase
column chromatography, countercurrent chromatography and preparative-HPLC
and these methods would be monitored with the previously mentioned TLC to ensure
minimum loss of the desire product. These steps are routinely employed when
working with a comprehensive phytochemical investigation of anthocyanin-
containing plant tissues.

2.5.1. Normal-Phase Column Chromatographic Separations

The first of these techniques, column chromatography, originally grew in popularity

because of the need to obtain pure anthocyanin compounds in sufficient quantities
for subsequent identification and characterization as well as to be used as reference
substances in qualitative and quantitative analyses. An effective column
chromatography technique capable of separating the impure anthocyanin
derivatization and degradation products that are inevitably present in crude plant
extracts, as well as resolving the naturally found mixture of different anthocyanins
was needed [118]. Liu et al, [187] investigated six adsorbant resins, each a cross-
linked polystyrene copolymer, and X-5 was found to be the most efficient adsorbant
of the anthocyanin as a nonpolar resin with relatively high surface area (500–600
m2/g) and pore radius (290–300 Å). The tested resins were loaded with centrifuged
mulberry juice and washed with distilled H2O until clear, which effectively removed
the sugars, acids, and other water-soluble compounds. The adsorbed anthocyanins
were then eluted with acidic ethanol (0.5% hydrochloric acid v/v) until the resins was
clean [187]. The concentrated anthocyanins were quantified using the pH differential
method with the total content calculated as cyanidin-3-glucoside which, even though
mulberry anthocyanins consist of cyanidin-3-glucoside and cyanidin-3-rutinoside,
gives acceptable calculations [188]. A more popular adsorbent resin is Amberlite
XAD-7, an acrylic ester of intermediate polarity, ~450 m 2/g surface area and ~90 Å
pore diameter. Similar to Liu’s experiment with macroporous resins, an Amberlite
XAD-7 column was loaded with the crude plant extract and washed with distilled H2O
to remove sugars, acids, and other by products after which the anthocyanin
compounds are eluted with acidic methanol (0.3–1.0 % trifluoroacetic acid v/v) [189,
190]. Following this initial separation of the anthocyanin fractions by macroporous
resins, the anthocyanins are often further purified using size exclusion
chromatography with Sephadex LH-20 or Toyopearl HW-40 and acidic methanol
elution or silica gel preparative TLC [189–191].

2.5.2. Countercurrent Chromatographic Separations

Countercurrent chromatography (CCC), in contrast to chromatography with a solid

phase, is an all liquid chromatographic technique that operates under gentle
conditions and allows non-destructive isolation of the anthocyanin compounds. Due
to the absence of any solid stationary phase, adsorption losses are minimized and
a 100% sample recovery is assured. CCC is an automated version of liquid-liquid
extraction, comparable to the repeated partitioning of an analyte between two
immiscible pases by vigorous mixing in a separatory funnel. The latest technique,
high-speed CCC (HSCCC) or multilayer coil CCC (MLCCC), creates separation by
wrapping an inert Teflon tubing around a holder in multiple layers and the coil is
subsequently rotated in a planetary fashion; that is, it rotates at ~900 rpm around its
own ‘planetary’ axis and simultaneously around a parallel ‘solar’ axis [192].
High-speed countercurrent chromatography requires the use of two immiscible
solvents, generally an organic and an aqueous phase, one to serve as the stationary
phase and the other as the mobile phase. Throughout the utilization of CCC, a
common multisolvent system composed of tert-butyl methyl ether (TBME)/n-
butanol/acetonitrile/water acidified with 0.1% (v/v) trifluoroacetic acid (TFA) has
been proven to be efficient regarding anthocyanin fractionation from a number of
different plant sources [87, 193]. A similar solvent system containing less TFA
(0.01% instead of 0.1% v/v) can be used with malonylated anthocyanins to avoid
degradation of these more labile compounds compared to other acylated
anthocyanins [192]. Winterhalter et al. [194] effectively separated the different
classes of anthocyanins in wine by using multiple solvent systems of different
polarities. This large-scale separation technique retained the solvent system
involving TBME/n-butanol/acetonitrile/water acidified with 0.1% TFA (2:2:1:5) for the
glucosidic anthocyanins, ethyl acetate/water (1:1) plus 0.1% TFA for the coumaroyl-
and caffeoyl-glucosides and ethyl acetate/n-butanol/water (4:1:5) plus 0.1% TFA for
the acetyl-glucosides [194]. Furthermore, a step gradient system was designed to
efficiently separate anthocyanin compounds of red wine marc and grape skins by
varying the polarity of the mobile phase, which is the organic phase in this case. This
protocol, meant to simplify the process, involves changing the ratio of an auxiliary
solvent, the acetonitrile concentration, while maintaining the immiscibility of the
partitioning solvents. A three step gradient was selected to shorten the
chromatographic process while keeping the level of resolution seen with an isocratic
elution and consists of the following: TBME/n-butanol/acetonitrile/water acidified with
0.02% TFA (2:2:x:5), with x = 0.1, 1.2, and 1.8 parts [195]. As described, the step
gradient method used the aqueous phase as the stationary phase, due to better
phase retention, resulting in a tail to head elution pattern while the previous
multisolvent systems all employed the lighter, organic phase as the stationary phase
leading to a head to tail elution mode [87, 192–196]. The complication most often
seen with counter-current chromatography is difficulty keeping the stationary phase
in its stationary position, whether it is aqueous or organic; because of this, stationary
chromatographic techniques are still the most popular and reported separation
methods for anthocyanins.

2.5.3. Preparative-HPLC Separations

HPLC, as discussed, is a popular method for the analysis of anthocyanin

compounds; however, when preparative separations are necessary before the
analyses, preparative-HPLC (prep-HPLC) can be used to clean up gram-scale
quantities of plant extracts. Because of the larger column size and higher flow rate
allowed with prep-HPLC, it is easy to inject as much as 100–200 mg and clean out
the crude plant extracts. The columns that are used can be the same packing
material, but a larger dimension, as is used for the analytical HPLC methods and the
same mobile phases with an isocratic or simple gradient system. TLC is used to
monitor the collected fractions and the similar TLC plates will correspond to the
fractions that can be combined, concentrated, and used to continue on with the
isolation techniques [119].
Go to:

3. CONCLUSIONS

Plant anthocyanins play an important role in the biology of the plant, and more
recently have been shown to have increasingly important applications in human
health and nutrition. From their biological activity, such as antioxidant, anti-
inflammatory, antiatherosclerotic, and anticancer properties, to their use as a natural
pigment of processed foods, the interest in anthocyanins continues to grow. This
class of compounds is found in hundreds of different plant tissues at very wide
ranges of concentration (Table 2). In this review, various chromatographic
separations and their derived analytical techniques to identify and quantitate natural
anthocyanins have been described in detail. The progression of anthocyanin
separation techniques started with traditional paper chromatography and thin-layer
chromatography, and has grown with advances in analytical instrumentations to
HPLC and CE combined with a variety of UV, MS, or NMR detectors. The range of
investigated approaches spans normal-phase column chromatography and CCC for
separation and preparation of large scale quantities to CE and HPLC for analytical
identification and quantification of complex mixtures. Though some may be more
common than others, each of these analyses has an appropriate function and
capacity for execution in the analysis of anthocyanin-containing plant tissues. The
analytical methodologies described here represent the current and foremost
methodologies for the complete analyses of anthocyanins. As improved
technologies continue to advance, so will our abilities for greater resolution at lower
concentrations. This review also detailed the importance in the preanalytical phase
of plant and sample preparation relative to maintaining the integrity of the natural
anthocyanins. We illustrated the type of chemical changes that may occur during the
processing and analysis of samples which could result in the identification of
polyphenols that are artifacts of the extraction and analytical methods employed.
Anthocyanins represent an increasingly examined class of compounds from their
role in plant biology to their use and impact in health and nutrition. As a
consequence, chemistry and analysis are the foundation upon which all other studies
involving these compounds rely whether in biochemistry, chemistry, genetics,
nutrition, foods, or pharmacognosy.

Go to:

Acknowledgments

This work was supported by the National Botanical Center for Age-Related
Diseases, NIH grant OD-00-004; and by the Partnership in Food Industry
Development for Natural Products, USAID Award No. AEG-A-00-04-00012-00.

Go to:

References
1. Kong JM, Chia LS, Goh NK, Chia TF, Brouillard R. Phytochemistry. 2003;64:923.
[PubMed]
2. Manach C, Scalbert A, Morand C, Rémésy C, Jiménez L. Am J Clin Nutr.
2004;79:727. [PubMed]
3. Kondratyuk TP, Pezzuto JM. Pharm Biol. 2004;42S:46.
4. Siemann EH, Creasy LL. Am J Enol Viti. 1992;43:49.
5. Gehm BD, McAndrew JM, Chien PY, Jameson JL. Proc Natl Acad Sci USA.
1997;9:14138. [PMC free article] [PubMed]
6. Stewart JR, Ward NE, Ioannides CG, O’Brian CA. Biochemistry. 1999;38:13244.
[PubMed]
7. Kitts DD, Yuan YV, Wijewickreme AN, Thompson LU. Mol Cell Biochem.
1999;202:91. [PubMed]
8. Williams CA, Grayer RJ. Nat Prod Rep. 2004;21:539. [PubMed]
9. Beecher GR. J Nutr. 2003;133:3248S. [PubMed]
10. Harborne JB, Williams CA. Phytochemistry. 2000;55:481. [PubMed]
11. Santos-Buelga C, Scalbert A. J Sci Food Agric. 2000;80:1094.
12. Manach C, Morand C, Gil-Izquierdo A, Bouteloup-Demange C, Remesy C. Eur
J Clin Nutr. 2003;57:235. [PubMed]
13. Cos P, De Bruyne T, Apers S, Vanden Berghe D, Pieters L, Vlietinck AJ. Planta
Med. 2003:69. [PubMed]
14. Iwashina T. J Plant Res. 2000;113:287.
15. Delgado-Vargas F, Jimenez AR, Paredes-Lopez O. Crit Rev Food Sci Nutr.
2000;40:173. [PubMed]
16. Wrolstad RE. J Food Sci. 2004;69:C419.
17. Fossen T, Slimestad R, Ovstedal DO, Andersen OM. Biochem Syst Ecol.
2002;30:855.
18. Mazza G, Miniati E. Anthocyanins in fruits, vegetables and grains. CRC Press,
Ino; Boca Raton, FL: 1993.
19. Markham KR, Gould KS, Winefield CS, Mitchell KA, Bloor SJ, Boase MR.
Phytochemistry. 2000;55:327. [PubMed]
20. Brouillard R. In: The flavonoids, advances in research since 1980. Harborne JB,
editor. Chapman and Hall; London: 1998. p. 525.
21. Torskangerpoll K, Andersen OM. Food Chem. 2005;89:427.
22. Ahmed J, Shivhare US, Raghavan GSV. Eur Food Res Tech. 2004;218:525.
23. Prior RL, Lazarus SA, Cao G, Muccitelli H, Hammerstone JF. J Agric Food Chem.
2001;49:1270. [PubMed]
24. Zheng W, Wang SY. J Agric Food Chem. 2003;51:502. [PubMed]
25. Kahkonen MP, Heinamaki J, Ollilainen V, Heinonen M. J Sci Food Agric.
2003;83:1403.
26. Wu X, Gu L, Prior RL, McKay S. J Agric Food Chem. 2004;52:7846. [PubMed]
27. Dawidowicz AL, Wianowski D, Baraniak B. LWT - Food Sci Tech. 2006;39:308.
28. Mazza G, Fukumoto L, Delaquis P, Girard B, Ewert B. J Agric Food Chem.
1999;47:4009. [PubMed]
29. Ortega-Regules A, Romero-Cascales I, Lopez-Roca JM, Ros-Garcia JM,
Gomez-Plaza E. J Sci Food Agric. 2006;86:1460.
30. Gao L, Mazza G. J Food Sci. 1994;59:1057.
31. Ichiyanagi T, Hatano Y, Matsugo S, Konishi T. Chem Pharm Bull. 2004;52:628.
[PubMed]
32. Blando F, Gerardi C, Nicoletti I. J Biomed Biotech. 2004;5:253. [PMC free article]
[PubMed]
33. Simunic V, Kovac S, Gaso-Sokac D, Pfannhauser W, Murkovic M. Eur Food Res
Tech. 2005;220:575.
34. Chandra A, Rana J, Li Y. J Agric Food Chem. 2001;49:3515. [PubMed]
35. Chaovanalikit A, Wrolstad RE. J Food Sci. 2004;69:FCT73.
36. Kim DO, Heo HJ, Kim YJ, Yang HS, Lee CY. J Agric Food Chem. 2005;53:9921.
[PubMed]
37. Gao L, Mazza G. J Agric Food Chem. 1995;43:343.
38. Torre LC, Barritt BH. J Food Sci. 1977;42:488.
39. Wada L, Ou B. J Agric Food Chem. 2002;50:3495. [PubMed]
40. Longo L, Vasapollo G. Food Chem. 2006;94:226.
41. Phippen WB, Simon JE. J Agric Food Chem. 1998;46:1734.
42. Lewis CE, Walker JRL, Lancaster JE, Sutton KH. J Sci Food Agric. 1998;77:45.
43. Chirinos R, Campos D, Betalleluz I, Giusti MM, Schwartz SJ, Tian Q, Pedreschi
R, Larondelle Y. J Agric Food Chem. 2006;54:7089. [PubMed]
44. Kammerer D, Carle R, Schieber A. Eur Food Res Tech. 2004;219:479.
45. Nielsen AH, Olsen CE, Moller BL. Phytochemistry. 2005;66:2829. [PubMed]
46. Byamukama R, Jordheim M, Kiremire B, Namukobe J, Andersen OM. Sci Hort.
2006;109:262.
47. de Ancos B, Gonzalez E, Cano MP. Z Leb Unt Forsch A. 1999;208:33.
48. Fan-Chiang HJ, Wrolstad RE. J Food Sci. 2005;70:c198.
49. Tomás-Barberán FA, Gil MI, Cremin P, Waterhouse AL, Hess-Pierce B, Kader
AA. J Agric Food Chem. 2001;49:4748. [PubMed]
50. Deineka VI, Deineka LA, Sirotin AA. Chem Nat Cpds. 2005;41:230.
51. Longo L, Vasapollo G. J Agric Food Chem. 2005;53:8063. [PubMed]
52. Cabrita L, Fossen T, Andersen OM. Food Chem. 2000;68:101.
53. Borkowski T, Szymusiak H, Gliszczynska-Swiglo A, Tyrakowska B. Food Res
Int. 2005;38:1031.
54. Fossen T, Cabrita L, Andersen OM. Food Chem. 1998;63:435.
55. Abdel-Aal ESM, Hucl P. J Agric Food Chem. 2003;51:2174. [PubMed]
56. Connor AM, Luby JJ, Hancock JF, Berkheimer S, Hanson EJ. J Agric Food
Chem. 2002;50:893. [PubMed]
57. Kiroa A, Ozkan M, Cemeroglu B. Food Chem. 2007;101:212.
58. Reyes LF, Cisneros-Zevallos L. Food Chem. 2007;100:885.
59. Goodman LP, Markakis P. J Food Sci. 1965;30:135.
60. McBean DM, Johnson AA, Pitt JI. J Food Sci. 1964;29:257.
61. Oliveira J, Fernandes V, Miranda C, Santos-Buelga C, Silva A, de Freitas V,
Mateus N. J Agric Food Chem. 2006;54:6894. [PubMed]
62. Kader F, Haluk JP, Nicolas JP, Metche M. J Agric Food Chem. 1998;46:3060.
63. Kader F, Irmouli M, Nicolas JP, Metche M. Food Chem. 2001;75:139.
64. Kader F, Irmouli M, Nicolas JP, Metche M. J Food Sci. 2002;67:910.
65. Kader F, Irmouli M, Zitouni N, Nicolas JP, Metche M. J Agric Food Chem.
1999;47:4625. [PubMed]
66. Wesche-Ebeling P, Montgomery MW. J Food Sci. 1990;55:731.
67. Garoia-Viguera C, Bridle P. Food Chemistry. 1999;64:21.
68. Mazza G, Brouillard R. Phytochemistry. 1990;29:1097.
69. Brouillard R, Mazza G, Saad Z, Albreoht-Gary AM, Cheminat A. J Am Chem
Soc. 1989;111:2604.
70. Mistry TV, Cai Y, Lilley TH, Haslam E. J Chem Soc Perkin Trans 2. 1991;8:1287.
71. Gomez-Miguez M, Gonzalez-Manzano S, Escribano-Bailon MT, Heredia FJ,
Santos-Buelga C. J Agric Food Chem. 2006;54:5422. [PubMed]
72. Ellestad GA. Chirality. 2006;18:134. [PubMed]
73. Takeda K. Proc Jap Acad Ser B. 2006;82:142. [PMC free article] [PubMed]
74. Yoshida K, Kitahara S, Ito D, Kondo T. Phytochemistry. 2006;67:992. [PubMed]
75. Bakowskia A, Kucharska AZ, Oszmianski J. Food Chem. 2003;81:349.
76. Gradinaru G, Biliaderis CG, Kallithraka S, Kefalas P, Garcia-Viguera C. Food
Chem. 2003;83:423.
77. Hakansson AE, Pardon K, Hayasaka Y, de Sa M, Herderich M. Tetrahedron Lett.
2003;44:4887.
78. Burns KC, Dalen JL. Oikos. 2002;96:463.
79. Escribano-Bailón MT, Santos-Buelga C, Rivas-Gonzalo JC. J Chromatogr A.
2004;1054:129. [PubMed]
80. Furuta K. J Appl Ent. 1986;102:93.
81. Page JE, Towers GH. Planta. 2002;215:478. [PubMed]
82. Gould KS, Vogelmann TC, Han T, Clearwater MJ. Phys Planta. 2002;116:127.
[PubMed]
83. Galvano F, La Fauci L, Lazzarino G, Fogliano V, Ritieni A, Ciappellano S,
Battistini NC, Tavazzi B, Galvano G. J Nutr Biochem. 2004;15:2. [PubMed]
84. Kowalczyk E, Krzesiński P, Kura M, Szmigiel B, Blaszczyk J. Pol J Pharmacol.
2003;55:699. [PubMed]
85. Arnous A, Makris DP, Kefalas P. J Agric Food Chem. 2001;49:5736. [PubMed]
86. Cai YZ, Xing J, Sun M, Zhan ZQ, Corke H. J Agric Food Chem. 2005;53:9940.
[PubMed]
87. Degenhardt A, Knapp H, Winterhalter P. J Agric Food Chem. 2000;48:338.
[PubMed]
88. Ichiyanagi T, Hatano Y, Matsugo S, Konishi T. Reactivity of Anthocyanin Toward
Reactive Oxygen and Reactive Nitrogen Species. CRC Press LLC; Niigata, Japan:
2006.
89. Juranic Z, Zizak Z. BioFactors. 2005;23:207. [PubMed]
90. Wang H, Cao G, Prior RL. J Agric Food Chem. 1997;45:304.
91. Wang H, Nair MG, Strasburg GM. J Nat Prod. 1999;62:294. [PubMed]
92. Matsumoto H, Nakamura Y, Tachibanaki S, Kawamura S, Hirayama M. J Agric
Food Chem. 2003;51:3560. [PubMed]
93. Folts J. Pharm Biol. 1998;36:21.
94. Christian KR, Nair MG, Jackson JC. J Food Comp Anal. 2006;19:778.
95. Tsuda T, Horio F, Uchida K, Aoki H, Osawa T. J Nutr. 2003;133:2125. [PubMed]
96. Hou DX, Fujii M, Terahara N, Yoshimoto M. J Biomed Biotech. 2004;5:321. [PMC
free article] [PubMed]
97. Hou DX, Ose T, Lin S. Int J Oncol. 2003;23:705. [PubMed]
98. Katsube N, Iwashita K, Tsushida T, Yamaki K, Kobori M. J Agric Food Chem.
2003;51:68. [PubMed]
99. Hou DX, Kai K, Li JJ. Carcinogenesis. 2004;25:29. [PubMed]
100. Springob K, Nokajima J-i, Yamazaki M, Saito K. Nat Prod Rep. 2003;20:288.
[PubMed]
101. Holton TA, Cornish EC. Plant Cell. 1995;7:1071. [PMC free article] [PubMed]
102. Nakajima JI, Tanaka Y, Yamazaki M, Saito K. J Bio Chem. 2001;276:25797.
[PubMed]
103. Nakajima JI, Sato Y, Hoshino T, Yamazaki M, Saito K. J Bio Chem.
2006;281:21387. [PubMed]
104. Saito K, Yamazaki M. New Phytol. 2002;155:9.
105. Ogata J, Kanno Y, Itoh Y, Tsugawa H, Suzuki M. Nature. 2005;435:757.
[PubMed]
106. Nakayama T, Suzuki H, Nishino T. J Mol Cat B Enz. 2003;23:117.
107. Hopp W, Seitz HU. Planta. 1987;170:74. [PubMed]
108. Debeaujon I, Peeters AJM, Leon-Kloosteriziel KM, Koornneef M. Plant Cell.
2001;13:853. [PMC free article] [PubMed]
109. Klein M, Martinoia E, Hoffmann-Thoma G, Weissenbock G. Plant J.
2000;21:289. [PubMed]
110. Marrs KA, Alfenito MR, Lloyd AM, Walbot V. Nature. 1995;375:397. [PubMed]
111. Mueller LA, Goodman CD, Silady RA, Walbot V. Plant Phys. 2000;123:1561.
[PMC free article] [PubMed]
112. Xu W, Shiori H, Kojima M, Nozue M. Plant Phys. 2001;125:447. [PMC free
article] [PubMed]
113. Manach C, Williamson G, Morand C, Scalbert A, Remesy C. Am J Clin Nutr.
2005;81:230S. [PubMed]
114. Felgines C, Talavera S, Texier O, Gil-Izquierdo A, Lamaison JL, Remesy C. J
Agric Food Chem. 2005;53:7721. [PubMed]
115. Felgines C, Talavera S, Gonthier MP, Texier O, Scalbert A, Lamaison JL,
Remesy C. J Nutr. 2003;133:1296. [PubMed]
116. Keppler K, Humpf HU. Bioorg Med Chem. 2005;13:5195. [PubMed]
117. Fleschhut J, Kratzer F, Rechkemmer G, Kulling SE. Eur J Nutr. 2006;45:7.
[PubMed]
118. Santos-Buelga C, Williamson G. Methods in Polyphenol Analysis. The Royal
Society of Chemistry; Cambridge: 2003.
119. Andersen OM, Markham KR. Flavonoids: Chemistry, Biochemistry and
Applications. CRC Press; Boca Raton: 2006.
120. Calvo D, Saenz-Lopez R, Fernandez-Zurbano P, Tena MT. Anal Chim Acta.
2004;524:207.
121. Mazza G, Cacace JE, Kay CD. Food Comp Add. 2004;87:129.
122. Saenz-Lopez R, Fernandez-Zurbano P, Tena MT. J Chromatogr A.
2003;990:247. [PubMed]
123. Saenz-Lopez R, Fernandez-Zurbano P, Tena MT. J Chromatogr A.
2004;1052:191. [PubMed]
124. da Costa CT, Horton D, Margolis SA. J Chromatogr A. 2000;881:403. [PubMed]
125. Bicard V, Fougerousse A, Brouillard R. J Liq Chromatogr Rel Technol.
1999;22:541.
126. Bednar P, Tomassi AV, Presutti C, Pavlikova M, Lemr K, Fanali S.
Chromatographia. 2003;58:283.
127. Simo C, Barbas C, Cifuentes A. Electrophoresis. 2005;26:1306. [PubMed]
128. Smyth WF. Electrophoresis. 2005;26:1334. [PubMed]
129. Bednar P, Papouskova B, Muller L, Bartak P, Stavek J, Pavlousek P, Lemr K.
J Sep Sci. 2005;28:1291. [PubMed]
130. de Villiers A, Vanhoenacker G, Majek P, Sandra P. J Chromatogr A.
2004;1054:195. [PubMed]
131. Vrhovsek U, Rigo A, Tonon D, Mattivi F. J Agric Food Chem. 2004;52:6532.
[PubMed]
132. Schutz K, Persike M, Carle R, Schieber A. Anal Bioanal Chem. 2006;384:1511.
[PubMed]
133. Cooke DN, Thomasset S, Boocock DJ, Schwarz M, Winter-halter P, Steward
WP. J Agric Food Chem. 2006;54:7009. [PubMed]
134. Nakajima JI, Tanaka I, Seo S, Yamazaki M, Saito K. J Biomed Biotech.
2004;5:241. [PMC free article] [PubMed]
135. Tian Q, Giusti MM, Stoner GD, Schwartz SJ. J Chromatogr A. 2005;1091:72.
[PubMed]
136. Anttonen MJ, Karjalainen RO. J Agric Food Chem. 2006;54:7530. [PubMed]
137. Hu C, Zawistowski J, Ling W, Kitts DD. J Agric Food Chem. 2003;51:5271.
[PubMed]
138. Cooney JM, Jensen DJ, McGhie TK. J Sci Food Agric. 2004;84:237.
139. McGhie TK, Ainge GD, Barnett LE, Cooney JM, Jensen DJ. J Agric Food Chem.
2003;51:4539. [PubMed]
140. Zanatta CF, Cuevas E, Bobbio FO, Winterhalter P, Mercadante AZ. J Agric
Food Chem. 2005;53:9531. [PubMed]
141. Seeram NP, Cichewicz RH, Chandra A, Nair MG. J Agric Food Chem.
2003;51:1948. [PubMed]
142. Seeram NP, Adams LS, Hardy ML, Heber D. J Agric Food Chem.
2004;52:2512. [PubMed]
143. Di Paola-Naranjo RD, Sanchez-Sanchez J, Gonzalez-Paramas AM, Rivas-
Gonzalo JC. J Chromatogr A. 2004;1054:205. [PubMed]
144. Kazuma K, Noda N, Suzuki M. Phytochemistry. 2003;64:1133. [PubMed]
145. Kazuma K, Noda N, Suzuki M. Phytochemistry. 2003;62:229. [PubMed]
146. Escribano-Bailon MT, Santos-Buelga C, Alonso GL, Salinas MR. Phytochem
Anal. 2002;13:354. [PubMed]
147. Es-Safi NE, Cheynier V, Moutounet M. J Agric Food Chem. 2002;50:5571.
[PubMed]
148. Ali A, Strommer J. J Agric Food Chem. 2003;51:7246. [PubMed]
149. Pomar F, Novo M, Masa A. J Chromatogr A. 2005;1094:34. [PubMed]
150. Vidal S, Meudeo E, Cheynier V, Skouroumounis G, Hayasaka Y. J Agric Food
Chem. 2004;52:7144. [PubMed]
151. Cantos E, Espin JC, Tomás-Barberán FA. J Agric Food Chem. 2002;50:5691.
[PubMed]
152. Cantos E, Garcia-Viguera C, de Pascual-Teresa S, Tomás-Barberán FA. J
Agric Food Chem. 2000;48:4606. [PubMed]
153. Lee J, Finn CE, Wrolstad RE. J Agric Food Chem. 2004;52:7039. [PubMed]
154. Alcalde-Eon C, Saavedra G, de Pascual-Teresa S, Rivas-Gonzalo JC. J
Chromatogr A. 2004;1054:211. [PubMed]
155. Choung MG, Choi BR, An YN, Chu YH, Cho YS. J Agric Food Chem.
2003;51:7040. [PubMed]
156. Montoro P, Tuberoso CIG, Perrone A, Piacenta S, Cabras P, Pizza C. J
Chromatogr A. 2006;1112:232. [PubMed]
157. Alcalde-Eon C, Saavedra G, de Pascual-Teresa S, Rivas-Gonzalo JC. Food
Chem. 2004;86:441.
158. Pissarra J, Lourenco S, Gonzalez-Paramas AM, Mateus N, Santos-Buelga C,
de Freitas V. Anal Chim Acta. 2004;513:215.
159. Pissarra J, Mateus N, Rivas-Gonzalo JC, Santos-Buelga C, de Freitas V. J
Food Sci. 2003;68:476.
160. Mateus N, de Pascual-Teresa S, Rivas-Gonzalo JC, Santos-Buelga C, de
Freitas V. Food Chem. 2002;76:335.
161. Stobiecki M, Matysiak-Kata I, Franski R, Skala J, Szopa J. Phytochemistry.
2003;62:959. [PubMed]
162. de Pascual-Teresa S, Santos-Buelga C, Rivas-Gonzalo JC. J Sci Food Agric.
2002;82:1003.
163. Fossen T, Andersen OM. Phytochem. 2003;62:1217. [PubMed]
164. Fossen T, Slimestad R, Andersen OM. Phytochemistry. 2003;64:1367.
[PubMed]
165. Vivar-Quintana AM, Santos-Buelga C, Rivas-Gonzalo JC. Anal Chim Acta.
2002;458:147.
166. Kosir IJ, Lapornik B, Andrensek S, Wondra AG, Vrhovsek U, Kidric J. Anal Chim
Acta. 2004;513:277.
167. Pati S, Losito I, Gambacorta G, La Notte E, Palmisano F, Zambonin PG. J Mass
Spectrom. 2006;41:861. [PubMed]
168. Dugo P, Favoino O, Lo Presti M, Luppino R, Dugo G, Mondello L. J Sep Sci.
2004;27:1458. [PubMed]
169. Makris DP, Kallithraka S, Mamalos A. Talanta. 2006;70:1143. [PubMed]
170. Remy S, Fulcrand H, Labarbe B, Cheynier V, Moutounet M. J Sci Food Agric.
2000;80:745.
171. Revilla I, Perez-Magarino S, Gonzalez-SanJose ML, Beltran S. J Chromatogr
A. 1999;847:83. [PubMed]
172. Pozo-Bayon MA, Monagas M, Polo MC, Gomez-Cordoves C. J Agric Food
Chem. 2004;52:1300. [PubMed]
173. Shoji T, Goda Y, Toyoda M, Yanagida A, Kanda T. Phytochemistry.
2002;59:183. [PubMed]
174. Hvattum E. Rapid Commun Mass Spectrom. 2002;16:655. [PubMed]
175. Lopes-da-Silva F, de Pascual-Teresa S, Rivas-Gonzalo JC, Santos-Buelga C.
Eur Food Res Technol. 2002;214:248.
176. Harborne JB. Biochem J. 1958;70:22. [PMC free article] [PubMed]
177. Hong V, Wrolstad RE. J Agric Food Chem. 1990;38:708.
178. Flamini R. Mass Spectrom Rev. 2003;22:213.
179. Baldi A, Romani A, Mulinacci N, Vinoieri FF, Casetta B. J Agric Food Chem.
1995;43:2104.
180. Johnstone RAW, Rose ME. Mass Spectrometry for Chemists and Biochemists.
University Press; Cambridge, UK: 1996.
181. de Rijke E, Zappey H, Ariese F, Gooijer C, Brinkman UAT. J Chromatogr A.
2003;984:45. [PubMed]
182. Mazza G, Cacace JE. J Food Sci. 2003;68:240.
183. Revilla E, Ryan JM, Martin-Ortega G. J Agric Food Chem. 1998;46:4592.
184. Garoia-Viguera C, Zafrilla P, Tomás-Barberán FA. Phytochem Anal.
1998;9:274.
185. Giusti MM, Rodriguez-Saona LE, Baggett JR, Reed GL, Durst RW, Wrolstad
RE. J Food Sci. 1998;63:219.
186. Kraemer-Schafhalter A, Fuchs H, Pfannhauser W. J Sci Food Agric.
1998;78:435.
187. Liu Z, Xiao G, Chen W, Xu Y, Wu J. J Biomed Biotechnol. 2004;5:326. [PMC
free article] [PubMed]
188. Giusti MM, Wrolstad RE. J Food Sci. 1996;61:322.
189. Byamukama R, Kiremire B, Andersen OM, Steigen A. J Food Comp Anal.
2005;18:599.
190. Pale E, Nacro M, Vanhaelen M, Vanhaelen-Fastre R. J Agric Food Chem.
1997;45:3359.
191. Sun B, Leandro MC, de Freitas V, Spranger MI. J Chromatogr A. 2006;1128:27.
[PubMed]
192. Schwarz M, Hillebrand S, Habben S, Degenhardt A, Winter-halter P. Biochem
Eng J. 2003;14:179.
193. Du Q, Jerz G, Winterhalter P. J Chromatogr A. 2004;1045:59. [PubMed]
194. Degenhardt A, Hofmann S, Knapp H, Winterhalter P. J Agric Food Chem.
2000;48:5812. [PubMed]
195. Vidal S, Hayasaka Y, Meudec E, Cheynier V, Skouroumounis G. J Agric Food
Chem. 2004;52:713. [PubMed]
196. Degenhardt A, Schwarz M, Winterhalter P, Ito Y. J Chromatogr A.
2001;922:355. [PubMed]

Anthocyanins isolated from purple corn (Zea mays L.)

Hiromitsu Aoki, Noriko Kuze, Yoshiaki Kato San-Ei Gen F. F. I. Inc.
Aoki, H., Kuze, N., Kato, Y., & Gen, S. E. (2002). Anthocyanins isolated from purple
corn (Zea mays L.). Foods and Food Ingredients Journal of Japan, 41-45.

Summary

Six anthocyanins were isolated from Peruvian purple corn seed (Zea mays L.) as
raw material for a food colorant, and their complete structures were determined by
means of spectroscopic analyses. The major anthocyanin was cyanidin 3-O-β-D-
glucoside. The other five anthocyanins were pelargonidin 3-O-β-D-glucoside,
peonidin 3-O-β-D-glucoside, cyanidin 3-O-β-D-(6-malonyl-glucoside), pelargonidin
3-O-β-D-(6-malonyl-glucoside) and peonidin 3-O-β-D-(6-malonyl-glucoside).
Pelargonidin 3-O-β-D-(6-malonyl-glucoside) was found in purple corn for the first
time. The cyanidin derivatives constitute around 70% in purple corn seed.
Seis antocianinas fueron aisladas de semillas de maíz morado peruano (Zea mays
L.) como fuente de colorante alimenticio y su completa estructura fue determinada
mediante análisis espectroscópico. La antocianina que se encontrpo en mayor
ptrroporción fue la cianidina 3-O-D-glucósido. peonidina 3-O-β-D-glucósido,
cyanidina 3-O-β-D-(6-malonyl-glucósido), pelargonidina 3-O-β-D-(6-malonyl-
glucósido) and peonidina 3-O-β-D-(6-malonyl-glucósido). La pelargonidin 3-O-β-D-
(6-malonyl-glucósido) fue encontrado en el míz morado por primera vez. La
cianidina en conjunto constituye aproximadamente el 70% del total de antocianinas
en el maíz morado.
References
1) T. Tsuda, K. Shiga, K. Ohshima, S. Kawakishi and T. Osawa,
Biochem. Pharmacol.1033-1039 (1996).
2) T. Tsuda, F. Horio and T. Osawa, Lipids, 33, 583-588 (1998).
3) J. Gabrielska, J. Oszmianski, M. Komorowska and M. Langner,
Z. Naturforsch.54, 319-324 (1999).
4) M. Yoshimoto, S. Okuno, M. Yoshinaga, O. Yamakawa, M. Yamaguchi and J.
Yamada, Biosci. Biotechnol. Biochem., 63, 537-541 (1999).
5) T. Koide, H. Kamei, Y. Hashimoto, T. Kojima and M. Hasegawa, Cancer Biother.
Radiopharm., 11, 273-277 (1996).
6) T. Koide, Y. Hashimoto, H. Kamei, T. Kojima, M. Hasegawa and K. Terabe,
Cancer Biother. Radiopharm., 12, 277-280 (1997).
7) A. Hagiwara, K. Miyashita, T. Nakanishi, M. Sano, S. Tamano, T. Kadota, T. Koda,
M. Nakamura, K. Imaida, N. Ito and T. Shirai, Cancer Lett., 171, 17-25 (2001).
8) J. Baraud, L. Genevois and J. P. Panart, J. Agric. Trop. Bot. Appl., 11, 55-59
(1964).
9) J. B. Harborne and G. Gavazzi, Phytochemistry, 8, 999-1001 (1969).
10) A. O. Lawason and B. A. Osude, Z. Pflanzenphysiol., 67, 460-463 (1972).
11) E. D. Styles and O. Ceska, Phytochemistry, 11, 3019-3021 (1972).
12) N. Nakatani, H. Fukuda and H. Fuwa, Agric. Biol. Chem., 43, 389-391 (1979).
13) S. C. Bhatla and R. C. Pant, Curr. Sci., 46, 700-702 (1977).
14) T. Fossen, R. Slimestad and O. M. Andersen, J. Agric. Food Chem., 49, 2318-
2321 (2001).
XL. Physicochemical and volatile profile alterations in pasteurized and frozen
strawberry pulp during storage

Gilma A. S. Gonçalves, Nathane S. Resende, Elisângela E. N. Carvalho,

Jaime V. de Resende, Eduardo V. de B. Vilas Boas

Abstract

This study evaluated the physicochemical characteristics and volatile profile of

strawberry pulp subjected to factors, pasteurization (unpasteurized and
pasteurized), freezing method (static air and forced air), and their interactions, during
12 months. Strawberry fruit were washed, sanitized, and pulped. The pulp was
packaged, pasteurized, and frozen (0, 2, 4, 6, and 12 months). We concluded that
pasteurization alters the strawberry pulp color. The impact of pasteurization and
freezing method on the strawberry pulp pH, titratable acidity, and soluble solids
variables is negligible. We tentatively identified 13 volatile compounds in fresh fruit
and pulp, ethyl acetate, ethyl hexanoate, and linalool being the volatiles with the
highest area percentage in the two products. The esters were predominant in both
the fruit and strawberry pulp. Time is the most determining factor in modifying the
strawberry pulp volatile profile, having an isolated effect on the increase of ethanol,
ethyl butanoate, and linalool.

Practical applications

The strawberry is very used worldwide as raw material to different products, due to
its color, flavor, and aroma. Thus, obtaining further information about the changes
caused by processing and storage is of great importance for the food industry. This
study shows the effect of simultaneous application of different processing techniques
(pasteurization and freezing methods) on physicochemical variables and volatile
profile of strawberry pulp over prolonged storage. Significant and unpublished
results involving color, pH, acidity, soluble solids, and volatile compounds of
strawberry frozen pulp over storage period are presented and may be applied for
food industry to keep the pulp quality.