Irrig Sci (2004) 23: 11–19

DOI 10.1007/s00271-003-0088-0

O R I GI N A L P A P E R

D. A. Goldhamer Æ E. Fereres

Irrigation scheduling of almond trees with trunk diameter sensors

Received: 18 January 2003 / Accepted: 12 August 2003 / Published online: 17 October 2003

Springer-Verlag 2003

Abstract The use of plant indicators may be the ideal based his schedule primarily on SWP measurements but
method for irrigation scheduling but it is hampered by also considered the water balance, applied 900 mm.
the dynamic nature of plant water status and by the lack Estimated crop evapotranspiration was 1,030 mm. The
of suitable indicators, relative to established scheduling mean coefficients of variation for the two irrigation
methods based on atmospheric and soil observations. A treatments during the monitoring period were 0.115 and
study was conducted in an almond orchard located in 0.031 for the MDS and SWP measurements, respec-
the San Joaquin Valley of California during the 2001 tively. The stress produced by the irrigation treatments
season using trunk diameter variations as the only hastened fruit maturation, as evidenced by accelerated
indicator for determining the amount of irrigation. The hull splitting. This resulted in lower fruit hydration just
ratio of the maximum daily shrinkage (MDS) of tree prior to harvest; 17.3% and 8.0% for the two irrigation
trunks relative to a reference MDS, calculated from a schedules, respectively, compared with 27.3% for the
relationship between MDS of fully irrigated trees and grower/cooperator. Based on harvesting selected trees
atmospheric vapor pressure deficit, was used as a signal with the same nut load, fresh and dry nut weights in the
for modifying the amount of applied irrigation water. 2.75 threshold treatment were 9.0% and 10.7% less than
Applied water was increased by 10% each time the MDS those of the 1.75 threshold, which were not significantly
signal exceeded the prescribed threshold. When the different from the results for the grower cooperator. Our
MDS signal went below the threshold, applied water was results demonstrate that it is feasible to develop an
reduced by 10% in an interactive manner. Two sched- irrigation schedule for almond trees based solely on
ules were tested with signal thresholds of 1.75 and 2.75, MDS signals, which may be tailored to any desired stress
which generated mild and moderate water stress, pattern and be operated in full automation with
respectively, as indicated by their stem water potential appropriate software development.
(SWP) values. The two irrigation treatments had SWP
that varied over the season from around )0.7 to
)1.1 MPa and )0.8 to )1.7 MPa, respectively. The two Introduction
schedules resulted in seasonal water applications of
860 mm for the 1.75 and 525 mm for the 2.75 signal Water scarcity in many irrigated areas of the world is
threshold treatments. The grower/cooperator, who becoming a fact of life as demand exceeds the sustain-
able supply. For instance, water currently used in the
irrigated lands developed in California over the last five
Communicated by R. Evans decades is being considered as a possible supply for an
expanding municipal sector as well as for environmental
D. A. Goldhamer (&)
Department of Land, preservation. As its population grows and environmen-
Air, and Water Resources, tal concerns increase, California is likely to be 2.46 bil-
Kearney Agricultural Center, lion cubic meters short of water annually in the
University of California, immediate future for an average rainfall year (State of
9240 S. Riverbend Ave.,
Parlier, CA 93648, USA
California Department of Water Resources 1998). For
E-mail: [email protected] drought years, the difference between supply and
Tel.: +1-559-6466575 demand would be much greater. This gap will likely be
E. Fereres met, at least in part, by a number of measures based on
IAS-CSIC and University of Cordoba, agricultural water conservation and reuse. Growers
Apdo. 4084, 14080 Cordoba, Spain will face increasing pressure to reduce water use,
12

necessitating the adoption of improved and innovative (Klepper et al. 1971) and have been proposed as a
management practices. management tool for irrigation scheduling (Li et al.
The search for more efficient use of irrigation water 1989). In deficit-irrigated peach trees, Goldhamer et al.
has recently focused on increasing water productivity (1999a) showed that stem diameter records detected
(WP), i.e., the ratio of yield or profit relative to crop stress earlier than SWP and that the signal strength
evapotranspiration (ETc; Seckler 1996). At the irrigation (magnitude of the measurement relative to that of fully
district scale, WP varies widely, as shown in a survey of irrigated trees) of trunk maximum daily shrinkage
40 districts around the world (Molden et al. 1998). There (MDS) for detecting water deficits was greater than that
many options for increasing WP at the field scale, many of SWP (Goldhamer et al. 1999b).
of them based on improved management (Howell 2001). Establishing a threshold tree water status measure-
Horticultural crops have relatively high WP values and ment that triggers an irrigation is complicated by the
that has resulted in progressive conversion of land from fact that tree water status is affected not only by soil
low-value field crops to high-value permanent crops in moisture but also by evaporative demand. Some refer-
intensively irrigated areas, such as the San Joaquin ence or baseline number that reflects the water status
Valley of California. Orchard owners seeking improved behavior of a fully irrigated tree is required to interpret
WP beyond current levels view more precise irrigation SWP measurements for irrigation scheduling (Shackel
scheduling as an important feature of improved irriga- et al. 1997; Goldhamer and Fereres 2001a). Fereres and
tion management. Goldhamer (2003) showed that the MDS of fully irri-
Rational irrigation scheduling procedures are being gated almond trees correlated well with VPD and pro-
increasingly adopted, mostly based on carrying a water posed that this relationship could be used to develop
balance calculation to determine the timing and amount reference MDS values. Alternatively, the MDS of a
of irrigation (Leib et al. 2002). Required inputs with small number of trees specifically irrigated to be ‘‘fully’’
conventional, low application frequency irrigation sys- irrigated (applied water around 10% greater than ETc)
tems are ETc and root-zone water storage capacity. In could be used as a baseline for irrigation scheduling
intensive tree horticulture, high-frequency irrigation (Goldhamer and Fereres 2001b). The MDS signal (ac-
systems, such as drip and microsprinklers, require only tual MDS/reference MDS) embodies variation due to a
ETc information. While the accurate estimation of ETc is changing evaporative demand and should primarily re-
possible in most field crops (Allen et al. 1998), there are flect soil water availability. Thus, MDS signal threshold
uncertainties in determining orchard ETc, associated values indicate how much tree stress is desired—values
primarily with the effects of canopy architecture, degree of 1 reflect no irrigation-related stress while progres-
of canopy cover and soil surface management. Adjusting sively higher values indicate escalating stress levels.
mature orchard ETc estimates to young canopies, or to Goldhamer and Fereres (2001b) developed irrigation
situations with a cover crop or actively growing weeds, scheduling protocols for young and mature fruit trees
creates significant uncertainty in fruit tree irrigation with different irrigation systems, based on continuously
scheduling based on ETc. Application efficiency is af- recorded trunk diameter measurements. In this work, we
fected by irrigation frequency, soil texture, and system present a test of those protocols in a commercial almond
design, maintenance, and operation. The use of soil orchard using MDS measurements as the only indicator
water monitoring devices for scheduling also requires for the actual scheduling of irrigation.
some knowledge of the distribution and relative density
of roots, and the uncertainty increases when the wetted
area varies in three dimensions, as in drip irrigation and Methods
microsprinklers. Thus, assessing irrigation needs directly
from tree measurements could provide an alternative This work took place in a mature almond orchard in western Kern
County, California. The trees [Prunus dulcis (Mill.) Webb cv. Fritz]
technique for more precise irrigation management. were 6 years old and grown in a well drained, clay-loam soil (Typic
Shackel et al. (1997) have shown that midday stem Torriorthents) with a root zone extending to a depth of about 2 m.
water potential (SWP) is the most robust of the different Irrigation was done with a buried drip system having 45 cm deep
tree water status measurements (predawn and midday lateral lines located 1.5 m on either side of the tree row
leaf water potential, etc.) and thus, is best suited for use (6.40·7.32 m spacing). This resulted in 20, 3.79 l/h emitters per tree
and an application rate of 1.5 mm/h. The system was operated 2–3
in tree-based irrigation scheduling. Lampinen et al. times per day.
(1995) scheduled a regulated deficit irrigation (RDI) The orchard contained two blocks, each about 4.2 ha, and the
regime in a prune orchard, achieving not only reduced irrigation for each block could be operated independently. Data
water application but also lower fruit hydration and collected during preliminary work at the experimental site in 2000
showed that a MDS signal threshold value of approximately1.25
thus, less drying costs and potentially higher grower resulted in tree water deficits that were virtually undetectable with
profit. However, monitoring SWP requires a significant SWP measurements. Thus, a MDS signal threshold value of 1.75,
amount of labor if frequent determinations are required which we believed would result in mild stress that presumably
and, in some situations, remote and automated moni- would have little effect on production, was established for one
block while a more severe stress threshold level of 2.75 was set for
toring of tree water status may be a desirable alternative. the other block (hereafter referred to as T1.75 and T2.75, respec-
Continuous records of stem diameter have been corre- tively). Within each block, four trees were instrumented with
lated with water potential measurements for many years linear variable differential transformers (LVDTs: Model 2.5 DF;
 13

Solartron Metrology, Bognor Regis, UK) installed on the south- two experimental blocks and irrigated by the grower/cooperator
western primary scaffold. The LVDTs were mounted on holders based on a combination of SWP and water budget measurements
built of aluminum and INVAR—an alloy comprising 64% Fe and were also individually harvested for comparison (hereafter referred
35% Ni that has minimal thermal expansion (Li et al. 1989)—and to as the Ranch). A 2 kg nut sample was collected from each tree.
covered with silver foil to provide constant shade. Measurements The number of fully hull split nuts (more than 50% of the suture
were taken every 30 s and recorded on a datalogger (Model CR 10; line split) was determined. The kernels were separated from the
Campbell Scientific, Logan, Utah). Data were downloaded every shells and hulls to determine the kernel percentages on a fresh and
few days and transferred to the laboratory for MDS determination. oven dry weight basis. Nut loads were determined by multiplying
We developed a relationship between MDS and mean daily the fresh nut yields per tree by the percentage of kernels in the
atmospheric vapor pressure deficit (VPD) using MDS values from corresponding 2 kg sample. Prior to tree shaking, 50 nut samples
fully irrigated trees located in the T1.75 block collected in April and were randomly collected from the four instrumented trees in T1.75
May 2001, prior to the onset on the irrigation treatments in early and T2.75 on 14, 19, and 27 September. Four Ranch trees were also
June (Fig. 1). Mean daily vapor pressure and relative humidity sampled on these dates. These samples were bulked and analyzed
measurements were taken from a CIMIS (California Irrigation for hull split as outlined above to determine the influence of the
Management Information System) automated weather station lo- irrigation regime on this important yield component.
cated 10 km from the experimental site to calculate VPD. The
linear regression between MDS and VPD during this period
(R2=0.77) was used to determine the reference MDS value re-
quired to compute the MDS signal for the two irrigation regimes Results
evaluated.
The irrigation protocols proposed by Goldhamer and Fereres Tree monitoring
(2001b) involve adjusting the irrigation rates based on the time-
course development of the MDS signal strength. If the MDS signal
did not reach the MDS signal threshold for three consecutive days, The time courses of MDS, MDS signal, and SWP for
the irrigation rate was decreased by 10%. Similarly, if the MDS both T1.75 and T2.75 during the period under consider-
signal exceeded the threshold for three consecutive days, the irri- ation are shown in Fig. 2. Observed MDS values ranged
gation rate was raised by 10%. Thus, the goal of this experiment
was to have the MDS signals oscillate around the 1.75 and 2.75 from 0.1 to 0.9 mm depending on evaporative demand
signal strength thresholds by adjusting the irrigation rates. Applied and treatment (Fig. 2a). The T2.75 MDS was always
water was measured on each block with water meters. greater than that of T1.75, except for a few days at the
Midday shaded leaf water potential (1300–1400 h) was moni- beginning and at the end of the period. The MDS signal
tored every weekday with a pressure chamber (Model 3005; Soil
Moisture Equipment, Santa Barbara, Calif.). Two single leaves in
evolution (Fig. 2b) shows that the signals for the two
the shade and close to the trunk on each of the four trees per treatments went above their respective thresholds be-
treatment were covered with a moist cloth just prior to excision. tween 12 and 15 times during the period and, every time,
The leaf alone was placed in the chamber within 15 s of excision an adjustment of the irrigation application rate was
and precautions recommended by Hsiao (1990) were taken to made 10% upwards. Downward adjustments, which
prevent leaf water loss during measurement. Previous work
(Goldhamer and Fereres 2001a) has shown that measurements were made following three days of MDS signals below
taken in this manner in almond leaves are nearly identical to SWP. the target threshold, decreased the signal, particularly
Thus, we hereafter refer to our leaf water potential measurements for the T1.75 that approached 1 occasionally, indicating
as SWP. no water stress at that time.
On 30 September, the four instrumented and monitored trees
plus six trees of the same size randomly located within each block Measurements of SWP showed clear differences be-
were mechanically shaken and on 9 October individually harvested tween the two irrigation treatments (Fig. 2c). Values
to estimate yield. Ten trees in a third 4.2 ha block adjacent to the ranged from )0.7 to )1.4 MPa in T1.75 but oscillated in

Fig. 1 Relationship between

maximum daily trunk shrinkage
(MDS) of fully irrigated trees.
Linear regression parameters
are shown on the figure
14

Fig. 2 Seasonal evolution of

a) maximum daily trunk
shrinkage (MDS), b) MDS
signal using data from a and the
linear regression expression of
Fig. 1, and c) stem water
potential (SWP) for the two
sensor-based irrigation
scheduling regimes

a narrower range between )0.8 and )1.1 MPa for most Variability in either the MDS or SWP measurements
of the summer. In contrast, T2.75 had SWP values be- among trees of the same treatment could induce uncer-
tween )1.3 and )1.5 MPa most days and even reached tainty in determining an irrigation schedule. In this
)1.7 MPa at the end of the period (Fig. 2c). experiment, the average MDS coefficients of variation
 15

(CVs) for the period considered (103 measurements Yield and yield components
made over 106 days) were 0.124 and 0.105 for T1.75 and
T2.75, respectively (Table 1). These values are signifi- There was significant variation in the number of nuts
cantly different. The SWP CVs for the two irrigation among the ten harvested trees of the two irrigation
treatments were 0.029 and 0.032, respectively. treatments and the Ranch (up to 12%; data not shown).
Since fruit load is determined by the stress history of the
trees rather than the current year
s irrigation treatments
Applied water (Goldhamer and Viveros 2000; Esparza et al. 2001) and
we wanted to minimize the effects of fruit load on fruit
Rates and cumulative amounts of applied water in both component size (kernels, etc.), we chose five trees each in
irrigation treatments from the start of the irrigation our irrigation treatments and in the Ranch that gave us
season, together with the ETc rate of a mature almond mean fruit loads that varied by less than 1%. These
orchard calculated from published crop coefficients values and other yield data for these trees are shown in
(Goldhamer 1989) and CIMIS reference crop water use Table 2. Individual fresh and dry nut (hull, shell, and
(ETo) from the nearby CIMIS weather station, are kernel) weights for T2.75 were 9.0 and 10.7% lower,
shown in Fig. 3a, b, respectively. Applied water in both respectively, than T1.75, which was not significantly dif-
treatments met ETc through April and was then re- ferent from the Ranch (Table 2). Similarly, individual
duced in anticipation of the start of the experiment in fresh and dry kernel weights for T2.75 were lower than
early June (Fig. 3a). Subsequently, the two schedules those for T1.75 by 9.8 and 11.5%, respectively. Again,
resulted in substantial differences in applied water T1.75 and the Ranch fresh and dry kernel weights were
rates, particularly at peak demand and prior to harvest. not significantly different. Both irrigation treatments
Maximum water application rates were 7.5 mm/day for resulted in nuts with a higher fresh and dry kernel per-
T1.75 in early July and 4.6 mm/day for T2.75 in late centage than the Ranch nuts (Table 2).
July. Early in the season, applied water rates in T1.75 The water stress induced by the scheduling treatments
were well below ETc but during the experiment (early hastened fruit maturation as evidenced by the hull
June–late September), these values were similar, with splitting data. At harvest, the percentage of fully hull-
the exception of late July. At the end of the experi- split nuts was higher for T2.75 than for T1.75; 95.2 versus
ment, 860 and 525 mm of water was applied to T1.75 89.8% (Table 2). The Ranch had significantly lower hull
and T2.75, respectively; a difference of 335 mm of water. split values at harvest (82.2%) than both irrigation
Estimated ETc was 1,038 mm. treatments. The 200-nut composite sample taken from
A representative example of how applied water was four trees per block on 14 September had large differ-
managed via MDS signals is shown in Fig. 4. Following ences in hull splitting; 84.5 and 100% of the nuts were
three days when the MDS signal was below the thresh- fully split in T1.75 and T2.75, respectively, while only
old for T2.75, applied water was decreased by 10% on 25 44.5% had fully split in the Ranch. This resulted in large
July. This resulted in the MDS signal increasing to 2.91, differences between kernel hydration while the nuts were
2.86, and 2.85 from 26 to 28 July, triggering a 10% on the trees between 14 and 27 September (Table 2).
increase in the rate of applied water to 4.62 mm/day
from 4.20 mm/day (Fig. 4). Applied water was de-
creased again twice until the signal reached the threshold Discussion
again by 5 August 5. Oscillations in the MDS signal
above and below the threshold continued and adjust- Since tree productivity is closely tied to water status, most
ments were made in the amounts of applied water. irrigation researchers recognize that a plant-based stress
sensor would provide the most desirable information for
use in irrigation scheduling. This fact has driven research
Table 1 Coefficients of variation determined from mean daily and development of a variety of instruments in recent
MDS signals and SWP measurements taken from 4 June to 18 years, including those that monitor sap flow (Cohen et al.
September 1981; Eastham and Gray 1998) and stem diameter (Li
Stress MDS MDS SWP SWP CV
et al. 1989; Link et al. 1998) on a continuous basis in the
indicator signal signal CV* measurements field. The potential advantages of using stem diameter
measurements sensors for scheduling include the fact that stress indicator
parameters that can be generated from their data, such as
T1.75 103 0.124 a** 66 0.029 MDS measurements, have a high sensitivity for water
T2.75 103 0.105 b 65 0.032
NSD*** stress detection (Goldhamer et al. 1999a; Ueda and Shi-
bata 2001), a good capability for adjusting the schedule in
*CV, Coefficient of variation; standard deviation/mean very short time periods as a daily signal is generated, and
**Numbers followed by different letters are significantly different the potential for complete automation.
using Fisher
s least significant difference method (P=0.05)
***NSD indicates no statistically significant differences between Measurement of SWP indicated that both treatments
irrigation regimes using Fisher
s least significant difference method developed water stress (Fig. 2c). The SWP values in T1.75
(P=0.05) indicated mild water stress while those in T2.75 that
16

Fig. 3 Irrigation water rates

applied in the two sensor-based
scheduling regimes compared
with mature orchard
evapotranspiration (ETc) rates
with time over the season

reached values nearly 1.0 MPa lower than the reference usefulness of both SWP and MDS for irrigation sched-
SWP baseline reflected moderate to severe stress (Fereres uling and proposed using the ‘‘signal-to-noise ratio’’ as an
and Goldhamer 1990; Shackel et al. 1997). Our finding integrating parameter. Using almond baseline values
that MDS variability is high relative to SWP is consistent (Goldhamer and Fereres 2001a), the T2.75 SWP signal did
with previous reports (Ginestar and Castel 1996; Goldh- not exceed 2.0 while the MDS signal approached 3.5
amer et al. 1999a; A. Naor, personal communication). As (Fig. 2b). Thus, the higher MDS signal partially mitigates
stress increased, the MDS coefficient of variation was the higher variability and tends to equalize the signal-to-
significantly reduced (0.124 and 0.105 for T1.75 and T2.75, noise ratio for both MDS and SWP. It should also be
respectively). This did not occur with SWP. Lower MDS noted that MDS variability in this study was less than has
variability with greater stress without concomitant SWP been reported previously (Goldhamer et al. 1999a,
behavior was also reported by Goldhamer et al. (1999b) Goldhamer and Fereres 2001b).
with deficit-irrigated peach trees that transitioned The use of MDS as the only indicator for irrigation
from mild to severe stress. They suggested that both signal timing and amount has been proposed by Goldhamer
strength and noise are important in assessing the and Fereres (2001b), who suggested that baseline or
 17

Scheduling adjustments were made manually here with

irrigation duration being the only management variable.
We believe that a more detailed, mathematical analysis
of the time-course development of the MDS signals
should produce algorithms that would allow an elec-
tronic controller to better detect the trends in the MDS
signals and to react accordingly. This would likely in-
volve changing irrigation timing and durations by a
variable percentage. The development of appropriate
software should result in MDS signals that oscillate
more closely around the target threshold, leading to a
more precise management of irrigation.
The use of the two MDS signal-driven schedules re-
sulted in significant differences in applied water rates
and seasonal amounts (Fig. 3). Applied water in T2.75
was reduced almost 40% from both T1.75 and the Ranch
Fig. 4 Example of the interactive nature of the irrigation schedule (Table 2) and 50% less than mature orchard ETc
protocols where MDS signals are consistently above or below the (Fig. 3b). Treatment deviations in applied water began
target threshold triggered increases (upward arrows) or decreases in late May and continued through harvest (Fig. 3a).
(downward arrows) in the rates of applied water
While the Ranch used both SWP readings and ETc
information to develop their irrigation schedule, it is
reference MDS values are required to determine the MDS clear that some deficit irrigation occurred, since the
signal and the need to establish a threshold MDS signal Ranch seasonal amount applied was very similar to that
for adjusting the schedule every time the signal consis- of T1.75 (Table 2).
tently deviates from this threshold. In this work, we have When considering the yield and yield component re-
shown that it is feasible to schedule irrigations in the field sults, note that we report data from trees that had nearly
using only sensors that monitor trunk diameter. The identical mean fruit loads for each irrigation treatment
baseline was determined by relating MDS to VPD (Fig. 1) and the Ranch. We found that T2.75 trees had lighter nuts
and the thresholds for the two treatments were set at the and kernels on a fresh and dry weight bases than T1.75
beginning of the season based on previous experience. We and the Ranch, which were similar. Lower kernel weights
do not know if the baseline developed in this study is as the result of preharvest water stress have been well
transferable to other locations, different aged trees, documented in studies by Goldhamer and Smith (1995)
almond varieties, or other Prunus species. Our attempts to and Goldhamer and Viveros (2000), where moderate to
use the MDS of trees specifically irrigated to create non- severe water stress was imposed. In those studies, water
limiting soil moisture conditions as the reference were less stress reduced hull splitting at harvest. In contrast, our
satisfactory, primarily because of high variability and the results with mild to moderate stress showed that hull
possible effects of heavy irrigation on soil salinity and splitting was increased in T2.75 relative to T1.75 which, in
oxygen status. turn, was higher than the Ranch. The magnitude, timing,
The use of the protocols based in MDS signals re- and duration of tree water deficits clearly influence al-
sulted in MDS signal oscillations around the target mond hull split at harvest. Earlier hull splitting, which
thresholds for the two irrigation treatments, albeit with can also be viewed as accelerated crop maturation, al-
deviations both above and below the thresholds that lowed the nuts to dry more on the tree, potentially
were greater than desired (Fig. 2b). Our protocols were advancing the harvest date. This can result in price
designed to be interactive; changes in the irrigation advantages for the grower. Additionally, less ground
duration by a constant 10% (the only management drying time can reduce ant damage (Zalom and Bentley
variable) to be reflected in MDS signals (Fig. 4). 1985).

Table 2 Production components of harvested nuts, hull-splitting, and hydration prior to harvest

Irrigation Applied water Fresh nut Dry nut Fresh Dry Full hull split Kernel hydration** Full hull
regime until 18 unit unit kernel kernel nuts at 14 split nuts
September weight weight weight weight September 14 Sept 19 Sept 27 Sept at harvest
(mm) (gm) (%) (% by weight)

T1.75 860 3.57 b* 3.35 b 1.12 b 1.04 b 84.5 26.7 25.2 17.3 89.8 b
T2.75 525 3.25 a 2.99 a 1.01 a 0.92 a 100.0 24.1 14.9 8.0 95.2 c
Ranch 898 3.70 b 3.47 b 1.08 ab 1.02 b 44.5 31.2 30.0 27.3 82.2 a

*Numbers followed by different letters are significantly different using Fisher
s least significant difference method (P=0.05)
**From 200 nut composite sample; 50 nuts per monitored tree, taken prior to 30 September tree shaking
18

The impact of first year water deficits on almond Cohen Y, Fuchs M, Green GC (1981) Improvement of the heat
production may not be indicative of the long-term re- pulse method for measuring sap flow in the stems of trees and
herbaceous plants. Agronomie 9:321–325
sponse of the orchard. Indeed, Goldhamer and Smith Eastham J, Gray SA (1998) A preliminary evaluation of the suit-
(1995) found that after a single season of various deficit ability of sap flow sensors for use in scheduling vineyard irri-
irrigation treatments that all applied only 410 mm of gation. Am J Enol Vitic 49(2):171–176
water but with different application timings, in addition Esparza G, DeJong TM, Weinbaum SA, Klein I (2001) Effects of
irrigation deprivation during the harvest period on yield
to a fully irrigated control (1,020 mm), nut loads in the determinants in almond trees. Tree Physiol 21:1073–1079
deficit treatments were reduced by up to 63% the fol- Fereres E, Goldhamer DA (1990) Irrigation of deciduous fruit and
lowing season, even though the orchard was returned to nut trees. In: Stewart BA, Nielsen DR (eds) Irrigation of agri-
full irrigation. Importantly, the irrigation regime that cultural crops. (ASA monograph no. 30) American Society of
imposed the majority of the stress preharvest, leaving the Agronomy, Madison, Wis., pp 987–1017
Fereres E, Goldhamer DA (2003) Suitability of stem diameter
most water for postharvest irrigation, had no significant variations and water potential as indicators for irrigation
effect on the fruit load in the following season(s). scheduling of almond trees. J Hortic Sci Biotechnol 78:139–144
This study was designed to demonstrate that a single Ginestar C, Castel JR (1996) Utilizacion de dendrometros como
parameter gleaned from trunk diameter monitoring indicadores de estrés hidrico en mandarinos jovenes regados
por goteo. Riegos y Drenajes XXI 89:40–46
could be used to schedule irrigations in a mature almond Goldhamer DA (1989) Drought irrigation strategies for deciduous
orchard. The MDS signal thresholds were chosen to orchards. (Publication no. 21453) University of California
produce different stress levels throughout the season; not Division of Agricultural and Natural Resources, Calif.
to maximize yields. Numerous RDI studies, where stress Goldhamer DA, Fereres E (2001a) Simplified tree water status
measurements can aid almond irrigation. Calif Agric 55(3):32–37
levels are varied over the season, have shown that sea- Goldhamer DA, Fereres E (2001b) Irrigation scheduling protocols
sonal ETc can be reduced without reducing fruit yield or using continuously recorded trunk diameter measurements. Ir-
quality (Goodwin and Jerie 1992; Lampinen et al. 1995; rig Sci 20:115–125
Caspari et al. 1994) and, in some cases, actually improve Goldhamer DA, Smith T (1995) Single season drought irrigation
yield components (Mitchell et al. 1986, 1989). Teviotdale strategies influence almond production. Calif Agric 49(1):19–22
Goldhamer DA, Viveros M (2000) Effects of preharvest irrigation
et al. (2001) demonstrated that mild water stress im- cutoff durations and postharvest water deprivation on almond
posed about a month prior to harvest can significantly tree performance. Irrig Sci 19:125–131
reduce hull rot, a fungal disease that can cause shoot Goldhamer DA, Fereres E, Mata M, Girona J, Cohen M (1999a)
dieback, but pointed out that monitoring tree stress was Sensitivity of continuous and discrete plant and soil water stress
monitoring in peach trees subjected to deficit irrigation. J Am
burdensome for most growers. On the other hand, the Soc Hortic Sci 124(4):437–444
effects of long-term use of RDI on tree longevity and soil Goldhamer DA, Fereres E, Cohen M, Girona J, Mata M (1999b)
salinity have yet to be investigated. Comparison of continuous and discrete plant-based monitoring
Tree-based (as opposed to soil and atmospheric) RDI for detecting tree water deficits and barriers to grower adoption
for irrigation management. Acta Hortic 537:431–445
triggers have the major advantage of being directly Goodwin I, Jerie P (1992) Regulated deficit irrigation: from con-
linked to crop productivity. Since the MDS signal has cept to practice. Aust NZ Wine Ind J (Advances in vineyard
been shown to be a very sensitive stress indicator and irrigation) 10 July 1992:258–261
has numerous operational advantages over SWP, Howell TA (2001) Enhancing water use efficiency in irrigated
including lower labor costs and the ability to be directly agriculture. Agron J 93:281–289
Hsiao TC (1990) Measurements of plant water status. In: Stewart
incorporated into remotely operated, electronic con- BA, Nielsen DR (eds) Irrigation of agricultural crops. (ASA
trollers, it may well be a superior tree-based indicator for monograph 30) American Society of Agronomy, Madison,
use where RDI and/or precise irrigation scheduling is Wis., pp 243–279
needed. Klepper B, Browning VD, Taylor HM (1971) Stem diameter in
relation to plant water status. Plant Physiol 48:683–685
Lampinen BD, Shackel KA, Southwick SM, Olson B, Yeager JT,
Acknowledgements We sincerely appreciate the assistance of Mario Goldhamer D (1995) Sensitivity of yield and quality of French
Salinas, who made the irrigation scheduling recommendations prune to water deprivation at different fruit growth stages.
based on the MDS values, and Jesus Salinas and Miguel Marquez, J Am Soc Hortic Sci 120(2):139–147
who collected data. We also gratefully acknowledge the assistance Leib BG, Hattendorf M, Elliot T, Matthews G (2002) Adoption
of Joe Gonzales, irrigation supervisor at Paramount Farming, who and adaptation of scientific irrigation scheduling: trends from
implemented the constantly altering applied water rates. Thanks Washington, USA as of 1998. Agric Water Manage 55:105–120
also are expressed to the Paramount management and other staff, Li SH, Huguet JG, Bussi C (1989) Irrigation scheduling in mature
including Joe McIlvaine, Dennis Elam, Paulin Garcia, and Aniceto peach orchard using tensiometers and dendrometers. Irrig
Mejia. Drain Syst 3:1–12
Link SO, Thiede ME, Bavel MG van (1998) An improved strain-
gauge device for continuous field measurement of stem and fruit
diameter. J Exp Bot 49:1583–1587
References Mitchell PD, Chalmers DJ, Jerie PH, Burge G (1986) The use of
initial withholding of irrigation and tree spacing to enhance the
Allen RG, Pereira LS, Raes D, Smith M (1998) Crop evapo- effect of regulated deficit irrigation on pear trees. J Am Soc
transpiration: guidelines for computing crop water require- Hortic Sci 111(6):858–861
ments. (Irrigation and drainage paper no. 56) FAO, Rome Mitchell PD, Ende B van den, Jerie PH, Chalmers DJ (1989) Re-
Caspari HW, Behboudian MH, Chalmers DJ (1994) Water use, sponses of Barlett
pear to withholding irrigation, regulated
growth, and fruit yield of Hosui Asian pears under deficit irri- deficit irrigation, and tree spacing. J Am Soc Hortic Sci
gation. J Am Soc Hortic Sci 119(N3):383–388 114(1):15–19
 19

Molden DJ, Sakthivadivel R, Perry CJ, Fraitture C de, Kloezen State of California Department of Water Resources (1998) The
WH (1998) Indicators for comparing performance of irrigated California Water Plan Update. (Bulletin 160–98, vol. 1) Sac-
agricultural systems. (Research report 20) International Water ramento, Calif.
Management Institute, Colombo, Sri Lanka Teviotdale BL, Goldhamer DA, Viveros M (2001) Effects of deficit
Seckler D (1996) The new era of water resources management: irrigation on hull rot disease of almond trees caused by Mo-
from dry
to wet
water savings. (Research report 1) Interna- nilinia fructicola and Rhizopus stolonifer. Plant Dis 85(4):399–
tional Irrigation Management Institute (IWMI), Colombo, Sri 403
Lanka Ueda M, Shibata E (2001) Diurnal changes in branch diameter as
Shackel KA, Ahmadi H, Biasi W, Buchner R, Goldhamer D, indicator of water status of hinoki cypress Chamaecyparis ob-
Gurusinghe S, Hasey J, Kester D, Krueger B, Lampinen B, tusa. Trees 15:315–318
McGourty G, Micke W, Mitcham E, Olson B, Pelletrau K, Zalom FG, Bentley WJ (1985) Southern fire ant (Hymenoptera:
Philips H, Ramos D, Schwankl L, Sibbett S, Snyder R, Formicidae) damage to harvested almonds in California. J
Southwick S, Stevenson M, Thorpe M, Weinbaum S, Yeager J Econ Entomol 78(2):339–341
(1997) Plant water status as an index of irrigation need in
deciduous fruit trees. HortTechnology 7(1):23–29