J. Dairy Sci.

99:7904–7917
http://dx.doi.org/10.3168/jds.2015-10759
© American Dairy Science Association®, 2016.

Effects of diet forage source and neutral detergent fiber content on milk
production of dairy cattle and methane emissions determined
using GreenFeed and respiration chamber techniques
K. J. Hammond,1 A. K. Jones, D. J. Humphries, L. A. Crompton, and C. K. Reynolds2
Sustainable Agriculture and Food Systems Research Division, Centre for Dairy Research, School of Agriculture, Policy and Development,
University of Reading, PO Box 237, Earley Gate, Reading, RG6 6AR, UK

ABSTRACT in experiment 2 using RC). Added NDF increased (or

tended to increase) methane yield for high MS, but not
Strategies to mitigate greenhouse gas emissions from high GS diets. In the separate experiments, the GF and
dairy cattle are unlikely to be adopted if production RC methods detected similar dietary treatment effects
or profitability is reduced. The primary objective of on methane emission (expressed as g/d and g/kg of
this study was to examine the effects of high maize DMI), although the magnitude of the differences varied
silage (MS) versus high grass silage (GS) diets, without between experiments. Overall methane emission and
or with added neutral detergent fiber (NDF) on milk yield were 448 g/d and 20.9 g/kg of DMI for experi-
production and methane emission of dairy cattle, using ment 1 using GF and 458 g/d and 23.8 g/kg of DMI for
GreenFeed (GF) or respiration chamber (RC) techniques experiment 2 using RC, respectively.
for methane emission measurements. Experiment 1 was Key words: forage, fiber, milk production, methane
12 wk in duration with a randomized block continuous emission
design and 40 Holstein cows (74 d in milk) in free-stall
housing, assigned to 1 of 4 dietary treatments (n =
INTRODUCTION
10 per treatment), according to calving date, parity,
and milk yield. Milk production and dry matter in- The current United Kingdom (UK) National
take (DMI) were measured daily, and milk composition Greenhouse Gas Inventory largely estimates emissions
measured weekly, with methane yield (g/kg of DMI) from agriculture using the most simplified (Tier 1)
estimated using a GF unit (wk 10 to 12). Experiment approach to accounting (IPCC, 2007). This approach
2 was a 4 × 4 Latin square design with 5-wk periods uses generic assumptions and factors about livestock
and 4 dairy cows (114 d in milk) fed the same 4 dietary management to estimate greenhouse gas emissions,
treatments as in experiment 1. Measurements of DMI, and there is a lack of methane emission factors from
milk production, and milk composition occurred in wk livestock in different farming systems fed a variety of
4, and DMI, milk production, and methane yield were diets. Analyses of calorimetry data (Mills et al., 2001)
measured for 2 d in RC during wk 5. Dietary treat- have shown that enteric methane emission is affected
ments for both experiments were fed as total mixed by dietary concentrations of starch relative to fiber.
rations offered ad libitum and containing 500 g of si- Previous comparisons have found replacing grass silage
lage/kg of dry matter composed (DM basis) of either (GS) with maize silage (MS) increases milk produc-
75:25 MS:GS (MS) or 25:75 MS:GS (GS), without or tion from dairy cows, mostly through increased feed
with added NDF from chopped straw and soy hulls intake for MS compared with GS (Phipps et al., 1988,
(+47 g of NDF/kg of dry matter). In both experiments, 1992, 1995; O’Mara et al., 1998; Kliem et al., 2008).
compared with high GS, cows fed high MS had a higher Enteric methane emission was also found to be vari-
DMI, greater milk production, and lower methane yield ably lower with MS compared with GS diets (Reynolds
(24% lower in experiment 1 using GF and 8% lower et al., 2010; Hammond et al., 2015b), although this is
not always consistent (Hammond et al., 2015b; Living-
stone et al., 2015). An explanation for differences (and
also lack of difference) in ruminant methane emission
Received December 14, 2015. with high MS versus high GS diets may be the physi-
Accepted June 30, 2016. cal and chemical attributes of these silages, along with
1
Present address: AgResearch Grasslands, Palmerston North, New
Zealand. digestive processes associated with the quantity of feed
2
Corresponding author: [email protected] eaten. In the study of Reynolds et al. (2010), high MS

7904
 MILK PRODUCTION, METHANE, AND MEASUREMENT TECHNIQUE 7905

and high GS diets were formulated to be similar in fur hexafluoride (SF6) techniques are shown to give a
starch and NDF concentrations by manipulation of the similar estimate of daily enteric methane emission for
concentration proportion of the diet. It was concluded cattle on most occasions. However, it was concluded
that observed differences in high MS versus high GS that suitability of the GF system will be affected by
diets on methane emission was attributed to differences the experimental objectives and design. An example
in the rate and extent of degradation of carbohydrate is Hammond et al. (2015a) who used dairy cattle to
components. Intakes of fibrous diets (i.e., GS or diets compare RC, GF, and SF6 measurement techniques.
with high NDF concentration) are not expected to be Although techniques were comparable for measurement
as high as diets comprising higher proportions of read- of methane emission, it was concluded that further
ily fermentable carbohydrates (i.e., MS or diets with work was needed to determine how to best deploy the
high starch concentration) because of increased rumen GF system to detect significant changes in methane
fill and extended time required to chew and reduce the emission attributable to individual animals and treat-
particle size of fiber to enable passage from the rumen. ments, and that future studies should include a greater
Considering that MS and GS diets are applicable to number of animals per treatment than is required for
rations based on typical UK forages, further work is RC studies.
warranted to examine the effects of forage type and The primary objective of the present study was to ex-
composition on milk production and methane emission amine the effect of feeding forages differing in MS and
from ruminant livestock. GS proportions to lactating dairy cattle, with or with-
Dietary manipulation can be effective for mitigation out supplemental NDF, on feed intake, milk produc-
of methane emission from dairy cattle, and alterna- tion and composition, and methane emission. Methane
tive methods to respiration chambers (RC) are being emission was measured using RC in experiment 2 and
introduced as a less intrusive way to measure enteric GF in experiment 1, as an alternative method to RC
methane emission. Particularly lacking is the capabil- for measuring dietary effects on methane emission. It
ity to accurately measure individual methane emission was hypothesized that feed intake and milk production
from multiple animals in a production environment would be greater, and methane yield (g/kg of DMI)
over a long period of time without interference to daily lower for cows fed higher MS diets and diets without
routine. The GreenFeed (GF) system (C-Lock Inc., additional NDF, compared with higher GS diets and
Rapid City, SD) is a portable sampling unit that is diets with higher NDF concentration.
used to estimate individual daily methane emission by
integrating measurements of airflow, gas concentration, MATERIALS AND METHODS
and detection of head position during each animal’s
visit to the unit. The animal is free to move and vol- Experimental Design
untarily enters a hood where an enticement, usually in
the form of a feed supplement, is delivered, and while Two experiments using the same dietary treatments
the animal is eating, breath samples are analyzed for were undertaken simultaneously at the University of
methane emission. Depending on GF setup, animals Reading’s Centre for Dairy Research (CEDAR, Ar-
can be free to visit GF at any time of the day or access borfield, UK). All procedures were approved and
can be dictated by the investigators. Measurements of monitored under the UK Home Office Animals (Sci-
methane emission by GF are typically over short peri- entific Procedures) Act 1986. Experiment 1 was a 12-
ods (3 to 7 min) at several variable times within a day, wk randomized block continuous design experiment.
over several days, so that ultimately a 24-h individual Forty lactating Holstein dairy cows were blocked into 4
methane emission profile is estimated based on extrap- treatment groups (10 cows each) based on calving date,
olation from repeated short-term measurements. An in- parity, and milk yield determined in the 3 wk before the
depth description of the GF system for measurement of experiment commencing (wk −3 to −1, covariate pe-
enteric methane can be obtained from Zimmerman and riod) when cows were fed a common commercial TMR.
Zimmerman (2012), Hristov et al. (2015), Huhtanen et For the entire experiment, cows were loose-housed in a
al. (2015), and Hammond et al. (2016). yard with sand-bedded cubicles, weighed twice weekly,
With increasing use of GF, more studies have com- and fed using an electronic Calan Broadbent individual
pared GF estimates with methane measurements using feeding system allowing measurement of individual cow
other techniques; however, comparisons with RC are feed intake (American Calan, Northwood, NH). Dur-
difficult because measurements are not simultaneous. ing a 3-wk training period before the covariate period
In a summary of GF publications by Hammond et al. (wk −6 to −4) and from wk 9 to 12, cows had vari-
(2016), under a variety of conditions, GF, RC, and sul- able and voluntary access to a GF unit; however, GF

Journal of Dairy Science Vol. 99 No. 10, 2016

7906 HAMMOND ET AL.

measurements of methane were only considered for NDF (+47 g/kg of DM). For experiment 1, the TMR
analysis between wk 10 to 12. Measurements from cows was prepared with a Mix Max 10 Paddle Feeder (Hi
of experiment 1 included diet composition, feed intake, Spec Engineering Ltd., Bagenalstowm, Republic of
BW, and milk yield and composition during wk 1 to 8 Ireland). For experiment 2, each TMR was prepared
(production period), and diet composition, feed intake, with a Dataranger (American Calan, Northwood, NH).
milk yield, and methane emission during wk 10 to 12 For both experiments ingredients were added in the
(methane measurement period). order of straw, concentrate mix, calf pellets, limestone,
Experiment 2 used 4 lactating Holstein cows surgi- grass silage, and maize silage. Dietary treatments were
cally fitted with rumen cannulas (type #1 C, 100 mm formulated to be isonitrogenous and meet or exceed the
center diameter, Bar Diamond Inc., Parma, ID) in a recommendation for MP, minerals, and vitamins based
previous lactation. Experiment 2 was a 4 × 4 Latin on Feed into Milk (Thomas, 2007) recommendations
square design balanced for carryover effects with 5-wk (Table 1). The GF used for estimating methane emis-
treatment periods. From wk 1 to 3, animals were group- sion from individual animals in experiment 1 used calf
housed with access to cubicles bedded with rubber pellets as a form of enticement to encourage animals
mats and wood shavings, fed TMR diets ad libitum, to enter the sampling hood. Therefore, dietary treat-
and milked twice daily. During this time, animals were ments were formulated for both experiments to include
adapted to dietary treatments with feed intake mea- a commercial calf pellet that was incorporated into the
sured using a roughage intake control feeding system TMR to form 8.7% of the formulated TMR (DM basis).
(Insentec B.V., Marknesse, the Netherlands). During When the GF was used in experiment 1 (wk 9 to 12),
wk 4, animals were moved to individual tie stalls and pellets were excluded from the TMR and fed in the GF.
in wk 5 animals were staggered in pairs to 2 individual Pellets were included in the TMR throughout experi-
RC for 2 consecutive days of methane measurements. ment 2.
Measurements included diet composition, feed intake, The MS was based on a mixture of maize varieties
and milk yield and composition during wk 4, and DMI, which were combined at harvest (October 22, 2012) and
milk yield, and methane emission while in RC during stored in clamps. Grass silage was made from a third
wk 5. Cows were weighed weekly and before and after cut (August 10, 2012) Lolium perenne mixture of tetra
measurements in RC. and diploid ryegrass species. The ryegrass was wilted
for 24 h and ensiled with an additive (GENUS ULV,
Animals and Dietary Treatments Genus Breeding Ltd., Nantwich, UK; 40 mL/t). Both
forage silages remained sealed in clamps for a minimum
In experiment 1, cows averaged (±SEM) 74 ± 16.2 period of 6 wk before use.
DIM at the start of the experiment and a BW of 670
± 4.0 kg throughout the experiment. In experiment 2, Methane Emission
cows averaged (±SEM) 114 ± 3.3 DIM at the start of
the experiment and 678 ± 10.5 kg of BW throughout For experiment 1, a single GF unit was used to es-
the experiment. timate individual cow methane emission during wk 10
Cows in both experiments were fed for ad libitum to 12. Details of the GF operation and use are given
DMI (5% refusals). In experiment 1, cows were fed once by Zimmerman and Zimmerman (2012), Hammond et
daily between 0700 and 0900 h, and milked twice daily al. (2015a), Hristov et al. (2015), and Huhtanen et al.
between 0600 and 0700 h, and 1500 and 1600 h. Feed (2015). Briefly, GF operation was initiated when an
refusals were collected thrice weekly (Monday, Wednes- animal placed its head inside the GF hood. A radio fre-
day, and Friday) for estimates of individual daily DMI. quency identification reader identified the animal’s ear
In experiment 2, diets were fed twice daily at 1000 and tag and GF sampling was activated when the animal’s
1600 h from wk 1 to 3, and thereafter (wk 4 to 5) were head (located by an infrared sensor) was located close
fed 4 times daily at 0500, 1100, 1700, and 2200 h. Feed to the sampling inlet within the hood (muzzle within
refusals were collected once daily at 0800 h, and cows 30 cm of the sampling inlet as detailed by Huhtanen
were milked twice daily between 0600 and 0700 h, and et al., 2015), and it was deemed that sufficient time
1500 and 1600 h. had elapsed since the previous methane measurement
Dietary treatments fed (Table 1) in both experiments for that animal. The position of the animal’s head
were either a high MS (375 g/kg of DM) and low GS within the hood was monitored using sensors to ensure
(125 g/kg) TMR (MS), or the reverse proportions complete breath collection. The GF unit was set up
(GS), without or with additional chopped barley straw outside the end of a cubicle yard within a polytunnel
and soy hulls incorporated to increase concentration of that minimized the effect of wind on measurements.

Journal of Dairy Science Vol. 99 No. 10, 2016

 MILK PRODUCTION, METHANE, AND MEASUREMENT TECHNIQUE 7907
Table 1. Diet formulations (g/kg of DM) for total mixed rations with higher proportions of maize (MS) or
grass silage (GS), without or with added NDF (MSNDF and GSNDF) and fed to lactating cows in experiments
1 and 2 and chemical composition (DM basis, g/kg) of diets for experiment 1

Item MS MSNDF GS GSNDF

Grass silage 125 125 375 375
Maize silage 375 375 125 125
Barley straw 10 50 10 50
Cracked wheat 91 12 107 38
Maize meal 0 0 108 103
Molassed sugar beet feed 50 50 0 0
Soy hulls 12 50 0 41
Wheat feed 97 84 70 50
Soybean meal 97 104 92 105
Rapeseed meal 30 38 0 0
Molasses 8 8 8 8
Dicalcium phosphate 5 5 5 5
Salt 5 5 5 5
Mineral and vitamin mixture1 8 8 8 8
Calf pellets2 87 87 87 87
Composition, g/kg of DM
Experiment 1, wk 1 to 8
DM, g/kg of fresh matter 431 430 410 407
OM 927 923 919 907
CP 154 159 159 170
NDF 340 391 366 395
ADF 192 220 219 239
Starch 216 179 193 140
Oil 39.5 37.6 42.6 43.0
Water-soluble carbohydrate 45.8 38.7 40.6 40.1
Starch:NDF 0.64 0.46 0.53 0.35
Experiment 1, wk 10 to 12
DM, g/kg of fresh matter 431 435 383 378
OM 927 911 926 899
CP 163 178 168 161
NDF 344 411 366 401
ADF 199 242 223 243
Starch 219 144 212 141
Oil 37.7 37.7 41.7 45.1
Water-soluble carbohydrate 40.2 35.7 36.0 25.1
Starch:NDF 0.64 0.35 0.58 0.35
1
Containing (per kg): 220 g of calcium, 40 g of phosphorus, 50 g of magnesium, 80 g of sodium, 30 mg of sele-
nium, 120 mg of cobalt, 400 mg of iodine, 5,000 mg of manganese, 6,000 mg of zinc, 3,000 mg of copper, 400,000
IU of vitamin A, 75,000 IU of vitamin D, 2,600 IU of vitamin E, and 100 mg of biotin.
2
Chemical composition of calf pellets was (g/kg of DM) ash, 85.1; oil, 46.5; ADF, 174; NDF, 289; starch, 259;
water-soluble carbohydrate, 91.3; nitrogen, 27.3; CP, 171; and gross energy (MJ/kg), 18.1.

Gates were positioned to allow access to the GF by only the calibration of the analyzers, which was found to
1 animal at a time. be negligible. A gravimetrically measured amount of
The concentration of gas emitted by the animal was carbon dioxide gas was released where the animals nose
calculated using background gas concentration, the would be when feeding to check recovery of expired
differential concentration of gas during the animal’s gases at the beginning and end of the measurement
time in the GF hood, and the calibration coefficient for period. No recovery correction was required in the cur-
concentration. See Huhtanen et al. (2015) for detailed rent study. Data from GF were downloaded on a daily
calculations of GF-estimated methane emission, which basis through a web-based data management system
were used here. The calibration coefficient was based provided by C-Lock Inc.
on N, carbon dioxide, and methane gases used to calcu- Animals were adapted to GF use before the covariate
late the response of the sensors. The GF analyzers were period (wk −6 to −4) and again in wk 9, with methane
calibrated weekly using a zero baseline gas (oxygen-free measurements used for statistical analysis obtained
N) and a span gas mixture of N containing 5,000 mg/kg during wk 10 to 12. During these periods, animals were
carbon dioxide and 1,000 mg/kg methane (BOC Ltd., able to access the GF unit at any time, except dur-
Manchester, UK). This was to account for any drift in ing milking and provided it was not in use by another

Journal of Dairy Science Vol. 99 No. 10, 2016

7908 HAMMOND ET AL.

animal. However, this did not necessarily generate a intervals from each chamber, giving 15 values per
measurement of methane. A visit was defined as a chamber per hour, with automatic zero and span cali-
successful methane measurement facilitated by feed bration readings at 4-h intervals. Signal outputs from
delivery which could only occur when a specified time monitoring and gas analysis equipment were recorded
(>240 min) had elapsed since the previous visit. In this by a data logger (type DL2e, Delta-T Devices Ltd.).
case, the enticement was provided and a visit logged if The data were automatically downloaded and compiled
the animal remained correctly positioned in the unit during each 24-h period using a desktop computer
for a sufficient amount of time (>3 min) for a valid and associated software programs (7th Wave Software
methane measurement. The unit was programmed to Ltd., Pangbourne, Berkshire, UK) based on specifically
deliver feed in 50-g quantities at varying intervals over designed data logging programs. Heat production was
a 6-min period, so that up to 350 g of pellet fresh weight estimated from gaseous exchange and urinary N output
was delivered during each complete visit, with up to a using the equation of Brouwer (1965).
maximum of 6 visits per day (2 kg of DM).
For experiment 2, measurement of gaseous exchange Sample Collection and Analyses
was obtained over 2 consecutive days in wk 5 using
open-circuit respiration chambers and methods similar For experiment 1, from wk 1 to 12 samples of TMR
to those described by Cammell et al. (1981, 1986) and offered were taken 3 times per wk and frozen before
Reynolds et al. (2001, 2014). The chambers (22.3 m3 a representative monthly bulk sample was created.
capacity) were constructed from double-skin insulated This was oven-dried (model ME/850/DIG/A, Genlab
steel panels and fitted with a profiled concrete floor Ltd., Widnes, UK) at 65°C for 48 h (#930.15, AOAC
with rubber mats, tubular steel sides to the standing, International, 2005), ground, and stored for analyses
and neck yoke and food box arrangements similar to of chemical composition. An additional sample of the
those in the main experimental unit. Glazed panels bulked TMR was also oven-dried at 100°C for DM
were fitted internally and externally to the chambers, determination (#930.15, AOAC International, 2005).
so the animals had visual contact both between cham- Total mixed ration refusals and their corresponding
bers and their local surroundings. An airlock of ap- DM (oven-dried at 100°C for 24 h) were measured
proximately 5.2 m3 was provided for service access to thrice weekly and DMI calculated on a weekly basis.
the feces and urine balance equipment and for routine Milk production was determined daily throughout the
milking and animal inspection and was connected to experiment and milk samples (30 mL) were taken from
the main chamber via double doors. Each chamber 2 successive a.m. and p.m. milkings at weekly intervals
was fitted with a re-circulatory air conditioning sys- and preserved with potassium dichromate (1 mg/mL;
tem (Mueller, Caswell Refrigeration Ltd., Malmesbury, Lactabs, Thomson and Capper, Runcorn, UK) for the
Wiltshire, UK) to provide air movement of up to 20 determination of milk composition during wk 1 to 8.
times the chamber volume per hour, environmental For experiment 2, the TMR offered and refused was
control across a temperature range of 12 to 25°C ± collected daily from individual cows during wk 4 and
2°C and a relative humidity of 60 ± 10%. The present 5 for DMI determination by oven-drying at 100°C for
experiment was conducted using 6 air changes per hour 24 h. Additional daily samples were taken during wk
with environmental controls adjusted to give no more 4 and pooled for individual cows to make a composite
than ± 3°C difference from the cowshed environment. sample that was stored at −20°C for analyses of chemi-
The rate of air flow through the outlet ducting from the cal composition. Milk production was determined daily
chambers was measured using factory-calibrated tur- throughout the experiment. Milk samples (30 mL) were
bine flow-meters (AOT Systems, Andover, Hampshire, taken at every milking in wk 4 and preserved with
UK). Monitoring of temperature and relative humidity potassium dichromate for the determination of milk
in the exhaust air flows was by type RHA1 sensors composition.
(Delta-T Devices Ltd., Burwell, Cambridge, UK). The Samples of the TMR offered and refused for both ex-
concentrations of oxygen in exhaust air flow was mea- periments were analyzed by wet chemistry as detailed
sured by a dual-channel paramagnetic oxygen analyzer by Hammond et al. (2014). Samples were analyzed for N
(Servomex International Ltd., Crowborough, Sussex, (macro Kjeldahl method), NDF and ADF (procedures
UK), and carbon dioxide and methane concentrations of Robertson and Van Soest, 1981, and Mertens, 2002),
were measured by dedicated dual-channel infrared gas starch (enzymatic conversion to glucose and glucose
analyzers (ADC Manufacturing Ltd., Stanstead Ab- measured using amyloglucosidase), oil (acid extrac-
botts, Hertfordshire, UK). The gas analysis train was tion), water-soluble carbohydrates (Nelson-Somogyi
designed to allow automatic measurement at 4-min procedure), and ash (combustion) concentrations. Milk

Journal of Dairy Science Vol. 99 No. 10, 2016

 MILK PRODUCTION, METHANE, AND MEASUREMENT TECHNIQUE 7909

samples were analyzed using mid-infrared spectroscopy heterogeneous compound symmetry, autoregressive,
(Foss Electric Ltd., York, UK) to determine fat, pro- heterogeneous autoregressive or unstructured) giving
tein, casein, lactose, and MUN concentrations, and 4% the best fit based on lowest BIC value for each variable
FCM and ECM yields were calculated as detailed by of interest. Methane emission data from both experi-
Gaines (1928) and Gaillard et al. (2016), respectively. ments were analyzed for homogenous distribution and
outliers using the Univariate Procedure of SAS and
Statistical Analyses residual analysis using the Mixed Procedure. Least
squares means are reported.
For experiment 1, weekly means of variables mea-
sured for each cow were statistically analyzed from wk RESULTS
1 to 8 (production period) and wk 10 to 12 (methane
measurement period) separately. The methane emission Diet Composition
data statistically analyzed were the daily averages for
individual animals for the 3-wk measurement period. Differences in diet composition observed for bulk
Data were analyzed using the MIXED Procedure of samples taken during the production and methane
SAS Version 9.2 (2011, SAS Institute Inc., Cary, NC) measurement periods of experiment 1 (Table 1) were
and a model testing for fixed effects of forage type (1 similar to differences observed for experiment 2 (Table
df), added NDF treatment (1 df) and their interaction 2). For experiment 2, high MS diets had greater DM
(1 df), their 2-way and 3-way interactions, random ef- (P < 0.001) and OM (P = 0.002) contents, a greater
fects of cow, and repeated effects of week within cow concentration of starch (P < 0.001), and lower concen-
using the covariance structure (compound symmetry, trations of CP (P = 0.077), NDF (P = 0.025), ADF (P
heterogeneous compound symmetry, autoregressive, = 0.005), and oil (P = 0.002), compared with high GS
heterogeneous autoregressive or unstructured) giving diets (Table 2). No forage type effect was observed on
the best fit based on the lowest Bayesian information water-soluble carbohydrate. Added NDF increased con-
criterion (BIC) value for each variable of interest. In centrations of NDF (P = 0.006) and ADF (P = 0.002),
addition, averages of weekly measurements during the and decreased starch concentration (P < 0.001). Forage
3-wk covariate period were used as a covariate in the type × NDF treatment interactions were found for con-
statistical analysis. centrations of OM (P = 0.074), CP (P = 0.096), starch
For experiment 2, means of variables measured for (P = 0.094), and oil (P = 0.036). The starch:NDF ratio
each cow and period were used in the statistical analy- was higher for high MS diets (P = 0.004) and decreased
sis. Data were analyzed using the MIXED Procedure of with added NDF (P = 0.002).
SAS (as for experiment 1) and a model testing the fixed
effects of forage type (1 df), added NDF treatment (1 Animal Performance
df) and their interaction (1 df), and random effects
of cow (3 df) and the repeated effect of period (3 df) Intakes of individual dietary components are given
using the covariance structure (compound symmetry, in Table 3 for both experiments 1 and 2. Cows fed

Table 2. Chemical composition (DM basis, g/kg) of high maize (MS) or high grass silage (GS) forage diets without or with additional NDF
(MSNDF and GSNDF) for experiment 2

Dietary treatment P-value

Forage type
Item MS MSNDF GS GSNDF SEM Forage type NDF × NDF
Experiment 2, wk 4
DM, g/kg of fresh matter 425 425 401 397 6.50 <0.001 0.341 0.326
OM 930 922 911 914 2.49 0.002 0.299 0.074
CP 164 140 169 181 7.74 0.077 0.265 0.096
NDF 307 369 354 385 10.50 0.025 0.006 0.172
ADF 172 214 201 239 7.06 0.005 0.002 0.747
Starch 247 196 193 137 5.51 <0.001 <0.001 0.094
Oil 35.9 36.3 44.7 42.2 0.42 0.002 0.135 0.036
WSCHO1 48.2 43.6 41.4 38.3 5.23 0.250 0.436 0.873
Starch:NDF 0.82 0.51 0.56 0.37 0.04 0.004 0.002 0.209
1
Water-soluble carbohydrate.

Journal of Dairy Science Vol. 99 No. 10, 2016

7910 HAMMOND ET AL.

Table 3. Feed component intake (kg/d) from lactating cows fed high maize (MS) or high grass silage (GS) diets1 supplemented without or with
additional (5% DM basis) NDF (MSNDF and GSNDF) in experiments 1 and 2

Dietary treatment P-value

Forage type
Item MS MSNDF GS GSNDF SEM Forage type NDF × NDF
Experiment 1, wk 1 to 8
DM 26.4 25.9 21.8 22.0 0.35 <0.001 0.591 0.311
OM 22.7 21.4 19.3 18.9 0.70 0.001 0.292 0.333
CP 3.80 3.82 3.37 3.54 0.11 0.001 0.378 0.538
NDF 8.34 9.07 7.73 8.22 0.26 0.006 0.021 0.650
ADF 4.71 5.18 4.58 4.99 0.15 0.273 0.005 0.854
Starch 5.41 4.09 4.14 2.94 0.13 <0.001 <0.001 0.646
Oil 0.98 0.88 0.91 0.90 0.03 0.367 0.079 0.103
Experiment 1, wk 10 to 12
DM 25.2 24.1 19.5 19.0 0.67 <0.001 0.277 0.631
OM 22.9 21.5 17.5 16.6 0.59 <0.001 0.030 0.455
CP 4.04 4.20 3.18 2.97 0.10 <0.001 0.795 0.081
NDF 8.50 9.50 6.82 7.42 0.23 <0.001 0.002 0.414
ADF 4.94 5.59 4.19 4.51 0.14 <0.001 0.002 0.255
Starch 5.45 3.32 4.08 2.66 0.12 <0.001 <0.001 0.007
Oil 0.93 0.87 0.78 0.84 0.03 0.002 0.925 0.026
No. of cows 10 10 10 10
Experiment 2, wk 4
DM 21.4 21.0 18.2 18.0 1.02 0.011 0.733 0.855
OM 19.9 20.2 15.2 16.3 0.78 0.024 0.150 0.234
CP 3.29 3.28 3.03 3.36 0.13 0.152 0.047 0.033
NDF 6.50 7.69 6.48 6.75 0.50 0.383 0.210 0.429
ADF 3.65 4.60 3.76 4.10 0.31 0.545 0.089 0.382
Starch 5.52 4.16 3.55 2.40 0.20 0.001 0.002 0.183
Oil 0.81 0.77 0.75 0.75 0.06 0.558 0.671 0.757
No. of cows 4 4 4 4
1
Containing (DM basis) either 37.5 and 12.5% (MS) or 12.5 and 37.5% (GS) MS and GS, respectively.

high MS diets in experiment 1 during wk 1 to 8 had Production data (including BW, milk yield, and
greater intakes of DM (P < 0.001), OM (P = 0.001), composition) were collected from cows of experiment
CP (P = 0.001), NDF (P = 0.006), and starch (P 1 during wk 1 to 8 (Table 4). During this period, cows
< 0.001), compared with high GS, with no effect on fed high MS diets had a greater BW (P = 0.002, which
intakes of ADF and oil. Added NDF treatment in- was due to a greater average daily live weight gain;
creased intakes of NDF (P = 0.021) and ADF (P = data not shown) and milk yield (P = 0.001) than cows
0.005), and decreased intakes of starch (P < 0.001) fed high GS, with no effect of NDF treatment. There
and oil (P = 0.079), with no effects on intakes of DM, were no forage type or NDF treatment effects on FCM,
OM, or CP. During wk 10 to 12 in experiment 1, cows but cows fed high MS diets produced more ECM (P =
fed high MS diets had higher (P < 0.01) intakes of all 0.031). Yields (g/d) of milk protein (P = 0.001), lactose
individual dietary components, compared with cows (P = 0.001), and casein (P = 0.001) were greater for
fed high GS diets. Adding NDF decreased intakes of cows fed high MS than high GS diets, and added NDF
OM (P = 0.030) and starch (P < 0.001), and increased decreased milk protein (P = 0.031) and casein (P =
intakes of NDF (P = 0.002) and ADF (P = 0.002). A 0.049) yields. Forage type affected concentrations (g/
forage type × NDF treatment interaction was found kg) of milk fat (MS lower than GS; P = 0.018), lac-
for intakes of CP (P = 0.0.081), starch (P = 0.007), tose (MS higher than GS; P = 0.011), and casein (MS
and oil (P = 0.026). higher than GS; P = 0.053). Milk urea concentration
Cows fed high MS diets in experiment 2 had greater was lower (P < 0.001) for high MS compared with high
intakes of DM (P = 0.011), OM (P = 0.024), and starch GS diets. Added NDF increased milk fat concentra-
(P = 0.001), compared with high GS diets. Forage type tion (P = 0.041), with a significant forage type × NDF
had no effect on intakes of CP, NDF, ADF, or oil. treatment interaction (P = 0.049) due to a greater in-
Added NDF increased ADF intake (P = 0.089) and crease for high MS than for high GS diets. Added NDF
decreased intake of starch (P = 0.002). A forage type decreased milk protein (P = 0.021) and casein (P =
× NDF treatment interaction was found for intake of 0.066) concentrations, whereas milk urea concentration
CP (P = 0.033). increased (P = 0.001).

Journal of Dairy Science Vol. 99 No. 10, 2016

 MILK PRODUCTION, METHANE, AND MEASUREMENT TECHNIQUE 7911

In experiment 2, cows fed high MS had a greater BW NDF addition. Similarly, milk (P = 0.003) and ECM (P
(P = 0.015) and milk yield (P = 0.076), than cows fed = 0.017) yields were higher for the high MS compared
high GS diets, with no effect of NDF treatment (Table with the high GS diets, and there was no effect of NDF
4). No effect was found of forage type or NDF treat- addition. Methane production was not affected by for-
ments on FCM or ECM yields. Cows fed high MS had age type or NDF treatments (averaging 448 g/d), but
greater yields of milk protein (P = 0.043), lactose (P a forage type × NDF treatment interaction trend (P =
= 0.060), and casein (P = 0.048), compared with high 0.096) was found, with methane emission being lowest
GS diets, with no effect of NDF treatment. No effect for high MS diets without additional NDF. Methane
was found of forage type or NDF treatments on milk yield (g/kg of DMI) was 24% lower (P < 0.001) for
component concentrations, except for milk urea, which cows fed high MS compared with high GS, and added
tended to be lower (P = 0.066) for MS than GS. NDF increased (P = 0.064) methane yield for high MS
diets, but not high GS diets (forage type × NDF treat-
Methane Emission ment interaction, P = 0.093). Methane expressed per
unit of milk and ECM yields (g/kg of milk) were lower
During methane measurements in experiment 1 (P = 0.001) for cows fed high MS compared with high
(Table 5), DMI was higher (P < 0.001) for the high GS, and increased with added NDF (P = 0.016). Meth-
MS compared with high GS diets, and not affected by ane per kilogram of BW tended to increase for high

Table 4. Body weight, milk yield, and composition from lactating cows fed high maize (MS) or high grass silage (GS) diets1 supplemented
without or with additional (5% DM basis) NDF (MSNDF and GSNDF) in experiments 1 and 2

Dietary treatment P-value

Forage type
Item MS MSNDF GS GSNDF SEM Forage type NDF × NDF
Experiment 1, wk 1 to 8
BW, kg 677 677 665 661 3.87 0.002 0.686 0.673
Yield
Milk, kg/d 38.4 37.1 35.4 34.5 0.74 0.001 0.155 0.311
FCM, kg/d 37.4 37.4 38.6 37.1 0.93 0.133 0.971 0.332
ECM, kg/d 34.2 34.3 33.1 32.1 0.76 0.031 0.598 0.457
Fat, g/d 1,302 1,386 1,343 1,311 42.6 0.703 0.537 0.158
Protein, g/d 1,211 1,144 1,099 1,057 24.4 0.001 0.031 0.586
Lactose, g/d 1,723 1,673 1,576 1,532 36.7 0.001 0.204 0.925
Casein, g/d 883 838 801 769 18.6 0.001 0.049 0.718
Concentration
Fat, g/kg 34.0 37.7 38.2 38.3 0.91 0.018 0.041 0.049
Protein, g/kg 31.7 31.0 31.2 30.8 0.21 0.111 0.021 0.519
Lactose, g/kg 44.8 45.1 44.5 44.3 0.20 0.011 0.999 0.271
Casein, g/kg 23.1 22.8 22.7 22.4 0.17 0.053 0.066 0.975
Urea, mg/L 288 314 324 434 6.29 <0.001 <0.001 <0.001
No. of cows 10 10 10 10
Experiment 2, wk 4
BW, kg 693 688 664 676 21.5 0.015 0.587 0.172
Yield
Milk, kg/d 31.6 33.6 27.4 25.8 2.05 0.076 0.807 0.243
FCM, kg/d 29.6 30.8 29.6 25.5 2.39 0.256 0.583 0.296
ECM, kg/d 29.2 29.7 28.3 24.3 2.55 0.174 0.492 0.343
Fat, g/d 1,135 1,211 1,118 1,017 103 0.313 0.908 0.392
Protein, g/d 1,035 977 917 779 69.4 0.043 0.217 0.534
Lactose, g/d 1,451 1,445 1,369 1,141 6.70 0.060 0.290 0.253
Casein, g/d 765 718 667 568 54.9 0.048 0.247 0.616
Concentration
Fat, g/kg 32.0 39.6 37.8 40.8 3.37 0.467 0.410 0.640
Protein, g/kg 32.7 31.2 30.4 30.3 1.12 0.108 0.402 0.380
Lactose, g/kg 45.4 45.6 44.7 44.8 0.40 0.153 0.767 0.996
Casein, g/kg 24.2 23.0 22.2 22.1 1.01 0.109 0.499 0.438
Urea, mg/L 176 246 309 392 38.1 0.066 0.138 0.742
No. of cows 4 4 4 4
1
Containing (DM basis) either 37.5 and 12.5% (MS) or 12.5 and 37.5% (GS) MS and GS, respectively.

Journal of Dairy Science Vol. 99 No. 10, 2016

7912 HAMMOND ET AL.

Table 5. Methane emissions from lactating cows fed high maize (MS) or high grass silage (GS) TMR1 supplemented without or with additional
(5% DM basis) NDF (MSNDF and GSNDF) and obtained using a GreenFeed unit (experiment 1) or respiration chambers (experiment 2)

Dietary treatment P-value

Forage type
Item MS MSNDF GS GSNDF SEM Forage type NDF × NDF
Experiment 1, wk 10 to 12
DMI, kg/d 25.2 24.1 19.5 19.0 0.67 <0.001 0.277 0.631
Milk yield, kg/d 35.6 33.3 30.0 28.0 1.67 0.003 0.207 0.943
ECM yield, kg/d 31.7 30.6 29.1 27.9 1.06 0.017 0.287 0.904
Methane emissions
g/d 410 461 460 460 15.1 0.110 0.109 0.096
g/kg of DMI 16.5 18.9 24.0 24.1 0.68 <0.001 0.064 0.093
g/kg of milk yield 11.7 14.2 15.6 16.4 0.64 <0.001 0.016 0.200
g/kg of ECM 13.1 15.2 15.9 16.6 0.51 0.001 0.011 0.168
g/kg of BW 0.591 0.697 0.696 0.686 0.029 0.118 0.111 0.052
GreenFeed visits
Average daily per cow 2.76 2.58 3.35 3.54 0.33 0.023 0.983 0.576
Visit duration, min 4.58 5.10 4.70 4.88 0.14 0.716 0.016 0.225
No. of cows 10 10 10 10
Experiment 2, wk 5
DMI,2 kg/d 21.7 20.5 18.4 17.0 0.95 0.011 0.205 0.950
Milk yield,2 kg/d 32.9 30.7 29.5 27.1 1.83 0.004 0.024 0.820
ECM yield, kg/d 31.3 30.6 25.6 24.2 1.47 0.034 0.138 0.282
Methane emissions
g/d 495 472 462 418 26.5 0.097 0.176 0.627
g/kg of DMI 21.8 23.7 25.5 24.2 0.82 0.018 0.412 0.015
g/kg of milk yield 15.6 15.8 15.4 16.3 0.97 0.711 0.211 0.325
g/kg of ECM 16.1 16.3 16.8 17.0 0.81 0.063 0.640 0.992
g/kg of BW 0.711 0.687 0.701 0.617 0.034 0.198 0.099 0.314
No. of cows 4 4 4 4
1
Containing (DM basis) either 37.5 and 12.5% (MS) or 12.5 and 37.5% (GS) MS and GS, respectively.
2
Measurements of DMI and milk yield were taken while animals were housed in respiration chambers and so were obtained alongside measure-
ments of methane emission.

MS diets with additional NDF, but did not increase to be lower when diets included additional NDF (P =
when NDF was added to GS diets (forage type × NDF 0.099).
treatment interaction, P = 0.052).
During methane measurements in experiment 2 Methane Measurement Techniques
(Table 5), DMI was higher (P = 0.011) for high MS
compared with high GS diets, and no effect of NDF The present experiments were conducted simultane-
treatment was found. Cows fed high MS during meth- ously using lactating dairy cows of a similar BW fed the
ane measurements had a higher milk yield (P = 0.004) same dietary treatments. Although methane emission
and ECM yield (P = 0.034) than cows fed high GS, and measurements obtained using GF and RC were not
added NDF decreased milk yield (P = 0.024). Methane statistically comparable, an objective was to determine
production (g/d) tended to be higher (P = 0.097) for if dietary treatment effects on methane emission would
cows fed high MS compared with high GS, with no be detected using both techniques.
effect of NDF treatment. Methane yield (g/kg of DMI) When the GF technique was used, 2,567 visits were
was 8% lower (P = 0.018) for cows fed high MS com- made to the GF by 40 cows over 3 wk. The average time
pared with high GS diets. Although there was no effect that methane was sampled from each animal during
of NDF treatment, there was a significant forage type each GF visit was 4.8 min (Table 5), with an average of
× NDF treatment interaction (P = 0.015), with added 3.0 visits/animal per d, with 94% of cows visiting the
NDF increasing methane yield for high MS, but not for GF every day during the 3 wk of GF access. This re-
high GS diets, as observed in experiment 1. There was sulted in approximately 5 h of methane measurements
no effect of forage type when methane was expressed for each cow in experiment 1. Cows housed in RC had
per unit of milk yield (g/kg of milk), but methane per 2 consecutive days of approximately 23 h of methane
kilogram of ECM yield tended to be lower for high MS measurements, which was equivalent to 184 h of meth-
diets (P = 0.063). Methane per kilogram of BW tended ane measurements for each cow in experiment 2.

Journal of Dairy Science Vol. 99 No. 10, 2016

 MILK PRODUCTION, METHANE, AND MEASUREMENT TECHNIQUE 7913

The number of visits to the GF was affected by di- measured with GF, but not RC. Averaging (±SEM)
etary treatment, whereby cows fed MS diets visited the methane emission across dietary treatments for each
GF less frequently on a daily basis than cows fed GS technique gave similar results for both methane pro-
diets (P = 0.023; Table 5). Cows fed added NDF had a duction (GF, 448 ± 5.70 vs. RC, 458 ± 12.54 g/d) and
longer GF visit duration than diets without added NDF methane yield (GF, 20.9 ± 0.38 vs. RC, 23.8 ± 0.73 g/
(P = 0.016). The pattern of cow visitation to the GF, kg of DMI).
based on all cow visits during the 3-wk measurement For GF measurements of methane, the range in
period and cumulated over 24 h, is given in Figure 1. methane production and yield (lowest to highest value)
For methane production (g/d), a tendency for a for- was 256 to 567 g/d and 14 to 29 g/kg of DMI, respec-
age type effect was observed using RC in experiment 2 tively. The between-animal coefficient of variation for
(P = 0.097), but not with GF in experiment 1. Methane GF methane production and yield was 5.7 and 5.2%,
yield (g/kg of DMI) was measured to be lower from lac- respectively, and the within-animal coefficient of varia-
tating cows fed high MS diets compared with high GS tion for methane production and yield was 10.5 and
diets, using both GF (P < 0.001) and RC (P = 0.018) 14.4%, respectively. For RC measurements, the range
techniques, but the magnitude of the difference varied in methane production and yield was 387 to 566 g/d
between techniques (24 vs. 8% lower for high MS vs. and 19 to 29 g/kg of DMI, respectively. The between-
high GS diets for GF and RC techniques, respectively). animal coefficient of variation for methane production
For both experiments, there was (or tended to be) a and yield using RC was 8.2 and 7.3%, respectively, and
forage type × NDF treatment interaction (P = 0.093 the within-animal coefficient of variation for methane
for GF and P = 0.015 for RC) when methane emission production and yield was 6.7 and 6.4%, respectively.
was expressed per unit of DMI, with methane yield Repeatability for measurements of methane produc-
increasing with NDF addition for high MS diets, but tion and yield (calculated as described by Herskin
not high GS diets. When methane was expressed per et al., 2003) for experiment 1 was 0.772 and 0.745,
unit of milk yield (g/kg of milk), there were forage type respectively and for experiment 2 was 0.761 and 0.764,
(P < 0.001) and NDF treatment (P = 0.016) effects respectively.

Figure 1. Pattern of GreenFeed (GF) visitation, based on 3 wk of access to a single GF unit, cumulated over a 24-h period, for 40 lactating
dairy cows fed 4 dietary treatments of maize silage (MS), MS with added NDF (MSNDF), grass silage (GS), and GS with added NDF (GSNDF).
Animals had unlimited access to GF during the 3 wk, except during milking (which occurred twice daily between 0600 and 0700 h and 1500 and
1400 h) and if another animal was occupying the unit.

Journal of Dairy Science Vol. 99 No. 10, 2016

7914 HAMMOND ET AL.

DISCUSSION design. Dietary treatments were crossed over for cows

in experiment 2 and had shorter periods of adaptation
Effect of Forage Type and Added NDF on Dairy (3 wk), compared the continuous design of experiment
Cow Performance 1 where animals were maintained on the same diet for
the entire experimental duration.
Overall, in this study, high MS diets were higher In this study, adding NDF to the diet had no sig-
in starch and lower in NDF concentrations compared nificant effects on DMI or milk yield for cows of either
with high GS diets. The addition of straw and soyhulls experiments 1 or 2, except it decreased milk yield in
decreased starch and increased fiber concentrations wk 5 of experiment 2. For cows of experiment 1, added
for both MS- and GS-based diets. Differences were NDF decreased milk protein yield and concentration,
observed in dietary treatment composition between wk reflecting a decrease in diet ME concentration and rate
1 to 8 and wk 10 to 12 in experiment 1. The high MS and extent of digestible carbohydrate supply. In a study
diets had higher fiber and lower starch concentrations by Kendall et al. (2009), early lactation cows fed 28%
in wk 10 to 12, and the high GS diets had higher starch NDF and highly digestible NDF diets produced more
concentration, compared with the same respective diets milk, fat, and protein than those consuming 32% NDF
fed in wk 1 to 8. This was largely due to the influence and low digestible NDF diets. Dry matter intake was
of variable amounts of pellets dispensed by the GF unit also greater for cows consuming 28% NDF diets, but
during wk 9 to 12. The GF unit was only available this was not affected by NDF digestibility.
during wk 9 to 12, and during this period concentrate
pellets were removed from the TMR and instead pro- Effect of Forage Type and NDF Concentration
vided via the GF as enticement to generate a measure on Methane Emissions
of methane, with an allowance of up to 2 kg of DM/cow
per day. The amount of pellet animals received was de- The positive relationship between DMI (kg/d) and
pendent on actual visits to the unit, and although up to methane emission (g/d) is thoroughly documented in
6 visits/d were possible for each dietary treatment, the the literature (e.g., Mills et al., 2001) and also observed
number of visits achieved fell below this target. There in the present study, with a slope of 4.19 ± 1.53 and
were more visits to the GF unit by animals fed high GS 12.10 ± 4.3 for experiments 1 and 2, respectively (P =
diets than by animals fed high MS diets (Table 5; 3.4 0.01 and P = 0.02, respectively). Previous comparisons
vs. 2.7 visits/animal per day). have found replacing GS with MS decreased methane
As observed in previous studies (as reviewed by emission and yield to varying extents. Reynolds et al.
Kahn et al., 2015), the higher starch and lower fiber (2010), McCourt et al. (2007), Brask et al. (2013),
contents of high MS diets were likely to be responsible and van Gastelen et al. (2015) all reported higher feed
for increased DMI and milk yield for cows in both ex- intakes and lower methane yields for lactating cows
periments, compared with high GS diets. Khan et al. offered MS compared with GS, but no subsequent ef-
(2015) summarized data from 13 published studies with fect on milk production. However, Staerfl et al. (2012),
37 direct comparisons that showed inclusion of MS in Hammond et al. (2015b), and Livingstone et al. (2015)
a GS-based diets fed to dairy cows improved DMI by 2 have reported inconsistent effects of high MS versus
kg/d, milk yield by 1.9 kg/d, and milk protein concen- high GS diets on cattle methane emission. In our
tration by 1.2 g/kg, with significant increases in yields study, cows fed high MS and high GS diets had simi-
of milk protein, fat, and lactose. A similar trend was lar methane production (g/d) in experiment 1 (with a
found in this study whereby, compared with high GS significant forage type × NDF treatment interaction),
diets, high MS improved DMI by 5.4 kg/d, milk yield but greater methane production on a high MS diet in
by 5.4 kg/d, and milk protein concentration by 0.35 g/ experiment 2. For both experiments, cows fed high MS
kg for cows fed over 8 wk in experiment 1, and respec- had a lower methane yield compared with high GS di-
tive improvements of 3.4 kg/d, 6 kg/d, and 1.6 g/kg ets (24% lower in experiment 1 using GF and 8% lower
for cows fed over 4 wk in experiment 2. The high feed in experiment 2 using RC). Cows fed high MS diets had
intake of MS is the main driver of greater milk yields, greater milk yields than cows fed high GS diets; how-
with multiple mechanisms regulating DMI such as NDF ever, when expressing methane per unit of milk yield,
and starch content, rate of degradability, and rate of only in experiment 1 did cows fed high MS have a lower
rumen passage (Khan et al., 2015). The higher feed in- methane output per unit milk produced compared with
takes for lactating cows of experiment 1 compared with high GS. The lower methane yield for the high MS
cows of experiment 2 were likely due to several factors diets is likely attributed to the source of starch and
including milk yield, DIM, tie stalls, and experimental NDF affecting rates of fermentation in the rumen. High

Journal of Dairy Science Vol. 99 No. 10, 2016

 MILK PRODUCTION, METHANE, AND MEASUREMENT TECHNIQUE 7915

starch diets are known to be an effective method for Unlike experiment 2, which used RC and found
lowering enteric methane emission (Beauchemin et al., methane production (g/d) to vary between dietary
2008). Increased intake of starch enhances fermenta- treatments, methane production estimated using GF
tion pathways that decrease methane production. With was not significantly affected by dietary treatment for
increasing dietary starch concentration, rumen pH is experiment 1. This difference between experiments
lowered, which can decrease fiber digestion and cause could be due to several factors, including the animals
an inhibition of methanogen activity and therefore themselves and their gut microbes, their level of intake,
methane production (Janssen, 2010). Livingstone et al. the timing of measurements, and other environmental
(2015) found no effect on methane yield when replacing factors. The difference in the results could also be
GS with MS in a TMR for lactating dairy cows and due to differences in the timing of methane sampling
concluded higher concentrations of NDF in their high measurements relative to diurnal patterns of methane
MS diets may have counteracted negative effects of a production and feeding. Respiration chambers take a
higher starch concentration and MS composition per se continuous measurement of methane over 24 h, thus
on methane yield compared with high GS diets. This capturing varying methane emission patterns, whereas
observation is partly supported in this study where methane measurements using GF rely on animal visita-
adding NDF to the diet increased methane yield from tion, which is mostly dictated by the behavior of the
cows fed high MS, but not high GS. animal.
The reliance of a feed enticement to generate a mea-
Methane Measurement Techniques to Detect Dietary sure of methane is a limitation of the GF technique,
Treatment Effects as observed with a varying diet composition within
experiment 1 (wk 1 to 8 vs. wk 10 to 12) and com-
The GF system has capability to estimate methane pared with experiment 2. This is a concern in both
emission from greater numbers of animals, is less re- pastoral grazing systems and animal nutrition studies
strictive to animal behavior for measurement of meth- where there is the possibility of excessive or variable
ane emission, and does not require extensive laboratory contribution of attractant to the animals diet, even if
equipment or labor. In our study, although the magni- restrictions are imposed (Waghorn et al., 2013; Dorich
tude of the difference in methane yield varied between et al., 2015; Hammond et al., 2015a). Animals on a
techniques for dietary forage type (24% lower in experi- high GS diet visited the GF more regularly than ani-
ment 1 using GF and 8% lower in experiment 2 using mals on a high MS diet, and this influenced the overall
RC for MS vs. GS), overall the techniques were able composition and intakes of starch and NDF, despite the
to detect similar dietary treatment effects and overall attempt to accommodate this in the TMR formulation.
methane emission from lactating dairy cows (448 g/d A similar observation was found in Hammond et al.
and 20.9 g/kg of DMI for GF vs. 458 and 23.8 for (2015a) where more visits to the GF were made when
RC, respectively). This was similar to Hammond et al. heifers were grazing a multi-species sward compared
(2015a), who found that despite concordance analyses with ryegrass and clover.
finding no agreement between GF and RC, overall the
GF system provided an average (grand mean) estimate CONCLUSIONS
of methane emission by growing dairy cattle that was
not different from RC measurements. This study examined the effects of variations in forage
Both techniques detected a significant interaction be- proportions of MS and GS, with or without additional
tween forage type and NDF treatment, and measured a NDF concentration on feed intake, milk production and
lower methane yield from cows fed high MS compared composition, and methane emission in lactating dairy
with high GS diets. This is in contrast to Hammond et cattle, and used GF as an alternative method to RC
al. (2015a) who used 4 growing dairy cattle in a 4 × to measure dietary effects on methane emission. As
4 Latin square design and found GF unable to detect hypothesized, cows fed high MS diets had a greater
changes in methane emission due to treatment or ani- DMI, milk production, and lower methane yield (g/kg
mal effects that were detected using RC. In that study, of DMI), compared with cows fed high GS diets. Added
cattle had GF access for only 7 d of each treatment NDF to both high MS and GS diets decreased DMI and
period, and entered RC for 72 h at the end of the treat- milk yield, and increased methane yield for high MS
ment period, whereas in the present study, a greater but not high GS diets. Both the GF and RC methods
number of GF measurements was obtained daily from detected similar dietary treatment effects on methane
more animals over a longer period (3 wk) in an attempt yield, although the magnitude of the difference varied
to increase the sample size and better represent the between experiments (and techniques) for dietary treat-
daily pattern of methane emission. ments. Overall average methane production and yield
Journal of Dairy Science Vol. 99 No. 10, 2016
7916 HAMMOND ET AL.

were similar for the 2 experiments using different cows, Hammond, K. J., G. C. Waghorn, and R. S. Hegarty. 2016. The Green-
Feed system for measurement of enteric methane emissions from
experimental conditions, and measurement techniques. cattle. Anim. Prod. Sci. 56:181–189.
Herskin, M. S., R. Müller, L. Schrader, and J. Ladewig. 2003. A la-
ser-based method to measure thermal nociception in dairy cows:
ACKNOWLEDGMENTS Short-term repeatability and effects of power output and skin con-
dition. J. Anim. Sci. 81:945–954.
This study was funded by Defra, the Scottish Gov- Hristov, A. N., J. Oh, F. Giallongo, T. Frederick, H. Weeks, P. R.
ernment, DARD, and the Welsh government as part of Zimmerman, R. A. Hristova, S. R. Zimmerman, and A. F. Branco.
2015. The use of an automated system (GreenFeed) to monitor
the UK’s Agricultural GHG Research Platform project enteric methane and carbon dioxide emissions from ruminant ani-
(www.ghgplatform.org.uk). Contributions from the mals. J. Vis. Exp. 103:e52904. http://dx.doi.org/10.3791/52904.
technical staff of University of Reading’s Centre for Huhtanen, P., E. H. Cabezas-Garcia, S. Utsumi, and S. Zimmerman.
2015. Comparison of methods to determine methane emissions
Dairy Research in the daily routine of experiments and from dairy cows in farm conditions. J. Dairy Sci. 98:3394–3409.
care of cows, as well as assistance from C-Lock Inc., IPCC. 2007. Climate change 2007: The Physical Science Basis. In Con-
with regard to GreenFeed operation and use, is grate- tribution of Working Group 1 to the Fourth Assessment Report of
the Intergovernmental Panel on Climate Change. S. Solomon, D.
fully acknowledged. Qin, M. Manning, Z. Chen, M. Marquis, K. Averyt, M. Tignor,
and H. Miller, ed. Cambridge University Press, Cambridge, UK.
Janssen, P. H. 2010. Influence of hydrogen on rumen methane forma-
REFERENCES tion and fermentation balances through microbial growth kinet-
ics and fermentation thermodynamics. Anim. Feed Sci. Technol.
AOAC International. 2005. Official Method of Analysis. 18th ed.
160:1–22.
AOAC International, Gaithersburg, MD.
Kendall, C., C. Leonardi, P. C. Hoffman, and D. K. Combs. 2009. In-
Beauchemin, K. A., M. Kreuzer, F. P. O’Mara, and T. A. McAllister.
take and milk production of cows fed diets that differed in dietary
2008. Nutritional management for enteric methane abatement: A
neutral detergent fiber and neutral detergent fiber digestibility. J.
review. Aust. J. Exp. Agric. 48:21–27.
Dairy Sci. 92:313–323.
Brask, M., P. Lund, A. L. F. Hellwing, M. Poulsen, and M. R. Weis-
Khan, N. A., P. Yu, M. Ali, J. W. Cone, and W. H. Hendriks. 2015.
bjerg. 2013. Enteric methane production, digestibility and rumen
Nutritive value of maize silage in relation to dairy cow perfor-
fermentation in dairy cows fed different forages with and without
mance and milk quality. J. Sci. Food Agric. 95:238–252.
rapeseed fat supplementation. Anim. Feed Sci. Technol. 184:67–69.
Kliem, K. E., R. Morgan, D. J. Humphries, K. J. Shingfield, and
Brouwer, E. 1965. Report of sub-committee on constants and factors.
D. I. Givens. 2008. Effect of replacing grass silage with maize si-
Pages 441–443 in Energy Metabolism. K. L. Blaxter, ed. European
lage in the diet on bovine milk fatty acid composition. Animal
Association for Animal Production publication no. 11. Academic
2:1850–1858.
Press, London, UK.
Livingstone, K. M., D. J. Humphries, P. Kirton, K. E. Kliem, D.
Cammell, S. B., D. E. Beever, K. V. Skelton, and M. C. Spooner.
I. Givens, and C. K. Reynolds. 2015. Effects of forage type and
1981. The construction of open-circuit calorimeters for measuring
extruded linseed supplementation on methane production and
gaseous exchange and heat production in sheep and growing cattle.
milk fatty acid composition of lactating dairy cows. J. Dairy Sci.
Lab. Pract. 30:115–119.
98:4000–4011.
Cammell, S. B., D. J. Thomson, D. E. Beever, M. J. Haines, M. S.
Mc Court, A. R., T. Yan, and C. S. Mayne. 2007. Effect of forage
Dhanoa, and M. C. Spooner. 1986. The efficiency of energy utiliza-
type on methane production from dairy cows. Page 48 in Proc.
tion in growing cattle consuming fresh perennial ryegrass (Lolium
Br. Soc. Anim. Sci., Southport, UK. Cambridge University Press,
perenne cv. Melle) or white clover (Trifolium repens cv. Blanca).
Cambridge, UK.
Br. J. Nutr. 55:669–680.
Mertens, D. R. 2002. Gravimetric determination of amylase-treated
Dorich, C. D., R. K. Varner, A. B. D. Pereira, R. Martineau, K. J.
neutral detergent fiber in feeds with refluxing in beakers or cru-
Soder, and A. F. Brito. 2015. Short communication: Use of a por-
cibles: Collaborative study. J. AOAC Int. 85:1217–1240.
table, automated, open-circuit gas quantification system and the
Mills, J. A., J. Dijkstra, A. Bannink, S. B. Cammell, E. Kebreab, and
sulfur hexafluoride tracer technique for measuring enteric methane
J. France. 2001. A mechanistic model of whole-tract digestion and
emissions in Holstein cows fed ad libitum or restricted. J. Dairy
methanogenesis in the lactating dairy cow: Model development,
Sci. 98:2676–2681.
evaluation, and application. J. Anim. Sci. 79:1584–1597.
Gaillard, C., N. C. Friggens, M. Taghipoor, M. R. Weisbjerg, J. O.
O'Mara, F. P., J. J. Fitzgerald, J. J. Murphy, and M. Rath. 1998. The
Lehmann, and J. Sehested. 2016. Effects of an individual weight-
effect on milk production of replacing grass silage with maize si-
adjusted feeding strategy in early lactation on milk production of
lage in the diet of dairy cows. Livest. Prod. Sci. 55:79–87.
Holstein cows during extended lactation. J. Dairy Sci. 99:2221–
Phipps, R. H., J. D. Sutton, and B. A. Jones. 1995. Forage mixtures
2236.
for dairy cows: the effect on dry-matter intake and milk produc-
Gaines, W. L. 1928. The energy basis of measuring milk yield in dairy
tion of incorporating either fermented or urea-treated whole-crop
cows. University of Illinois Agric. Exp. Station Bull. 308.
wheat, brewers' grains, fodder beet or maize silage into diets based
Hammond, K. J., D. J. Humphries, L. A. Crompton, C. Green, and
on grass silage. Anim. Sci. 61:491–496.
C. K. Reynolds. 2015a. Methane emissions from cattle: Estimates
Phipps, R. H., R. F. Weller, R. J. Elliott, and J. D. Sutton. 1988. The
from short-term measurements using a GreenFeed system com-
effect of level and type of concentrate and type of conserved forage
pared with measurements obtained using respiration chambers or
on dry matter intake and milk production of lactating dairy cows.
sulphur hexafluoride tracer. Anim. Feed Sci. Technol. 203:41–52.
J. Agric. Sci. 111:179–186.
Hammond, K. J., D. J. Humphries, L. A. Crompton, P. Kirton, and C.
Phipps, R. H., R. F. Weller, and A. J. Rook. 1992. Forage mixtures
K. Reynolds. 2015b. Effects of forage source and extruded linseed
for dairy cows: The effect on dry matter intake and milk produc-
supplementation on methane emissions from growing dairy cattle
tion of incorporating different proportions of maize silage into di-
of differing body weights. J. Dairy Sci. 98:8066–8077.
ets based on grass silages of differing energy value. J. Agric. Sci.
Hammond, K. J., D. J. Humphries, D. B. Westbury, A. Thompson, L.
118:379–382.
A. Crompton, P. Kirton, C. Green, and C. K. Reynolds. 2014. The
Reynolds, C. K., S. B. Cammell, D. J. Humphries, D. E. Beever, J.
inclusion of forage mixtures in the diet of growing dairy heifers:
D. Sutton, and J. R. Newbold. 2001. Effects of postrumen starch
Impacts on digestion, energy utilisation, and methane emissions.
Agric. Ecosyst. Environ. 197:88–95.

Journal of Dairy Science Vol. 99 No. 10, 2016

 MILK PRODUCTION, METHANE, AND MEASUREMENT TECHNIQUE 7917
infusion on milk production and energy metabolism in dairy cows. compared with the IPCC default values, and the long-term meth-
J. Dairy Sci. 84:2250–2259. ane mitigation efficiency of adding acacia tannin, garlic, maca and
Reynolds, C. K., L. A. Crompton, J. A. N. Mills, D. J. Humphries, P. lupine. Agric. Ecosyst. Environ. 148:111–120.
Kirton, A. E. Relling, T. H. Misselbrook, D. R. Chadwick, and D. Thomas, C. 2007. Feed into milk: A new applied feeding system for
I. Givens. 2010. Effects of diet protein level and forage source on dairy cows. Nottingham University Press, Nottingham, UK.
energy and nitrogen balance in methane and nitrogen excretion van Gastelen, S., E. C. Antunes-Fernandes, K. A. Hettinga, G. Klop,
in lactating dairy cows. G. M. Crovetto, ed. Energy and Protein S. J. J. Alferink, W. H. Hendriks, and J. Dijkstra. 2015. Enteric
Metabolism and Nutrition Symposium, 127:736. Wageningen Aca- methane production, rumen volatile fatty acid concentrations, and
demic Publishers, the Netherlands. milk fatty acid composition in lactating Holstein-Friesian cows fed
Reynolds, C. K., D. J. Humphries, P. Kirton, M. Kindermann, S. grass silage- or corn silage-based diets. J. Dairy Sci. 98:1915–1927.
Duval, and W. Steinberg. 2014. Effects of 3-nitrooxypropanol on Waghorn, G. C., E. J. Garnett, C. S. Pinares-Patino, and S. Zim-
methane emission, digestion, and energy and nitrogen balance of merman. 2013. Implementation of GreenFeed in a dairy herd
lactating dairy cows. J. Dairy Sci. 97:3777–3789. grazing pasture. Advances in Animal Biosciences, Proceedings of
Robertson, J. B., and P. J. Van Soest. 1981. The detergent system the 5th Greenhouse Gases and Animal Agriculture Conference
of analysis and its application to human foods. Pages 123–158 in (GGAA2013), Dublin, Ireland, 4:436.
The Analysis of Dietary Fiber in Food. Vol. 3. W. James and O. Zimmerman, P. R., and R. S. Zimmerman, inventors; C-Lock Inc.,
Theander, ed. Marcel Dekker Inc., New York, NY. assignee. 2012. Method and system for monitoring and reduc-
Staerfl, S. M., J. O. Zeitz, M. Kreuzer, and C. R. Soliva. 2012. Meth- ing ruminant methane production. United States Patent number
ane conversion rate of bulls fattened on grass or maize silage as US2009028806 A1.

Journal of Dairy Science Vol. 99 No. 10, 2016