Irrig Sci (2001) 20: 115±125
DOI 10.1007/s002710000034
O R I GI N A L P A P E R
David A. Goldhamer á Elias Fereres
Irrigation scheduling protocols using continuously recorded
trunk diameter measurements
Received: 1 October 2000 / Published online: 7 June 2001
Ó Springer-Verlag 2001
Abstract Theoretical and experimental aspects of de-
veloping irrigation scheduling strategies using continu-
ous measurements of trunk diameter are presented. The
behavior of parameters derived from trunk diameter
measurements (TDM), including maximum daily trunk
shrinkage (MDS), maximum daily trunk diameter
(MXTD), and minimum daily trunk diameter (MNTD)
is evaluated for both rapidly growing peach [Prunus
persica (L.) Batsch, cv. September Snow] and mature
almond [Prunus dulcis (Mill.) Webb cv. Price] trees sub-
jected to mild de®cit irrigation. Stem water potential
(SWP) and MDS were highly correlated in the mature
almond trees but a poor relationship was found in the
rapidly growing peach trees. Conversely, daily changes in
MXTD and MNTD correlated well with SWP in the fast
growing trees. While there was relatively high variability
(``noise'') in the MDS measurements (CV 15.8%), the
greater changes in the magnitude of the MDS (``signal'')
resulted in a signi®cantly higher signal:noise ratio than
found with the SWP measurements. In addition to soil
water and trunk growth rate, MDS patterns were in¯u-
enced by irrigation frequency and evaporative demand.
Based on the experimental results, protocols for utilizing
the trunk diameter-derived indicators for scheduling ir-
rigations are presented for three cases: (1) mature trees
under low frequency irrigation, (2) mature trees under
high frequency irrigation, and (3) young trees under high
frequency irrigation. The scheduling protocols provide
guidelines that address both under- and over-irrigation
and are predicated on the sensitivity of TDM to very
mild plant water de®cits. The necessity for and ap-
proaches to developing reference (baseline) and thresh-
D.A. Goldhamer (&)
Department of Land, Air, and Water Resources,
Kearney Agricultural Center, University of California,
9240 S. Riverbend Ave., Parlier, CA 93648, USA
E-mail: dagoldhamer@ucdavis.edu
E. Fereres
IAS-CSIC and University of Cordoba, Apdo 4084,
14080 Cordoba, Spain
old values derived from TDM are emphasized. We
conclude that protocols using TDM for precise irrigation
scheduling hold promise as an additional tool for pro-
gressive growers who want to link irrigation management
to an automated, electronic, plant-based stress indicator.
Introduction
Evidence of diurnal oscillations in the diameter of tree
trunks was identi®ed long ago (Kozlowski 1967). Short-
term changes in stem diameter have been related to
concomitant changes in plant water status as transpira-
tion loss draws water from the stem, primarily from
phloem tissues (Klepper et al. 1971; Molz and Klepper
1972). Huck and Klepper (1976) found that stem diam-
eter measurements could provide continuous records of
plant water potential in cotton plants and suggested that
plant water status measurements could be automated.
Interest in using trunk diameter measurements
(TDM) for irrigation scheduling in fruit trees was
sparked more recently by the work of Garnier and
Berger (1986) who did a comprehensive study in peach
trees and concluded that it was feasible to automate ir-
rigation based on TDM. Since then, TDM have been
taken on a number of fruit tree species, such as apple
(Huguet et al. 1992), cherry (Cabibel and Isberie 1997),
citrus (Ginestar and Castel 1996) and walnut (Cohen
et al. 1997). In all cases, good correlations were found
between the degree of trunk shrinkage and swelling and
the changes in tree water status. One advantage of TDM
is that they provide trunk growth rate records and can
characterize growth responses to the environment.
The use of TDM for irrigation scheduling requires
that some parameter be derived from the TDM that is
indicative of irrigation requirements. It would be desir-
able for TDM to provide an early warning of water
de®cits, well before the monitored stress can negatively
a€ect important physiological processes. Goldhamer
et al. (1999) evaluated the relative sensitivity of trunk
116
diameter-derived parameters and conventional indica-
tors of plant water status in peach trees, including var-
ious measurements of tree water potential, stomatal
conductance, fruit growth and photosynthesis. They
concluded that parameters derived from TDM were
more sensitive than the other indicators in that they
detected water stress earlier in a drying cycle.
Plant water status is very dynamic because it re¯ects
the balance between soil water supply and atmospheric
demand. A single observation of plant water status may
not be very informative, as a water de®cit may be caused
by either the soil water level or by changes in evapora-
tive demand. Therefore, it is not surprising that irriga-
tion scheduling is normally based on soil or atmospheric
parameters (Martin et al. 1990); not only because they
are easier to measure than plant parameters but because
they are more robust and can be readily interpreted.
Nevertheless, plant water potential and canopy tem-
perature measurements have been used successfully for
irrigation scheduling in a few crops, such as cotton, once
the measurements have been calibrated for a given en-
vironment (Grimes and Yamada 1982). In fruit trees,
research has shown that leaf water potential (LWP) can
be a reliable indicator of tree water status (Fereres and
Goldhamer 1990). More recently, the superiority of stem
water potential (SWP) over other water potential mea-
surements, including predawn LWP, has been estab-
lished for a number of fruit tree species (Shackel et al.
1997). In any case, monitoring SWP or other water
potential indicators with a pressure chamber requires
substantial manpower and cannot be automated.
Recent enhancement of the robustness of the sensors
used to measure trunk diameter and continued im-
provements in the quality and a€ordability of electronic
data collection and transfer devices suggest that irriga-
tion scheduling can be automated using a plant-based
indicator. This paper discusses the indicators that can be
Fig. 1 Parameters that can be
derived from trunk diameter
measurements (TDM), includ-
ing maximum daily trunk
shrinkage (MDS), and trunk
growth expressed as daily dif-
ferences in maximum and min-
imum daily trunk diameters
(MXTD and MNTD,
respectively)
derived from TDM and, based on theoretical consider-
ations and new experimental evidence, presents proto-
cols for using these indicators in scheduling localized
irrigations (drip or microsprinkler).
Theory
Derivation of parameters based on TDM
Trunk diameters oscillate over a 24-h cycle, reaching a
maximum value (MXTD) just before sunrise and a
minimum (MNTD) sometime in the afternoon (Fig. 1).
The di€erence between MXTD and MNTD is termed
maximum daily shrinkage (MDS); a parameter that has
been shown to increase as tree water de®cits increase in
peach (Garnier and Berger 1986), cherry (Cabibel and
Isberie 1997), walnut (Cohen et al. 1997) and other
species (Huguet et al. 1992). Goldhamer et al. (1999)
found that the MDS of ®eld-grown peach trees increased
from 0.27 to 0.63 mm when the SWP decreased from
±0.99 to ±2.1 MPa, respectively. In a few cases, MDS
has been shown to decrease as severe water stress
increases, such as in apple (Huguet et al. 1992) and
lysimeter-grown peach (Goldhamer et al. 1999).
The evolution of MXTD and MNTD also provides
useful information (Fig. 1). The di€erence between two
consecutive MXTD values gives one measure of the trunk
growth rate while the trend of MXTD establishes the
cumulative growth. Analysis of MNTD trends also gives a
measure of the trunk growth rate, as represented in Fig. 1,
and this can be one the earliest signs of water stress
(Goldhamer et al. 1999). The high sensitivity of MNTD
probably re¯ects the combined e€ects of soil water supply
and evaporative demand on maximum shrinkage while
MXTD is mostly a€ected by the rehydration process,
which depends primarily on soil water supply and may be

only indirectly a€ected by the previous day's evaporative
demand. Thus, under mild water de®cits, the trunk di-
ameter will normally recover fully overnight, while at
midday, the transpiration pull from the atmosphere and
soil water supply may amplify the combined e€ects of
limiting supply and/or high demand. This would lead to
greater shrinkage from the very beginning of the stress.
Absolute, relative and reference values
In principle, all plant-based measurements need a refer-
ence or a control. Plants are the middlemen between soils
and the atmosphere. They continually function under
water de®cits and the level of the de®cit is a function of
both soil and atmospheric conditions. Even with the soil
thoroughly wet, changes in evaporative demand induce
changes in plant water status. Thus, water potential
measurements of plants under stress, by themselves, mean
little if they are not considered relative to the measure-
ments obtained from fully irrigated plants. As with the
other plant-based indicators, it is desirable to refer pa-
rameters derived from TDM to some fully irrigated
(``control'') values. Such control values may be obtained
from trees that are irrigated speci®cally to prevent any soil
water limitation. Alternatively, TDM the day after a
heavy rain or irrigation that wets the entire root zone may
be useful as reference values but do not address oscilla-
tions due to changes in evaporative demand. In lieu of
speci®c reference trees, it may be possible to deal with
evaporative demand-related changes by developing fam-
ilies of relationships between MDS, for example, and
indicators of evaporative demand, such as reference crop
water use (ETo) or vapor pressure de®cit (VPD).
Growth e€ects on parameters derived from TDM
Since trunk growth slows as trees mature, tree age must be
taken into account when interpreting TDM. With young,
well-watered trees, trunk growth occurs throughout the
season and will be re¯ected in MXTD and MNTD
records, making these parameters potentially useful
indicators for irrigation scheduling. On the other hand,
mature tree trunk growth slows as the season progresses
and may be followed by a stoppage or a shrinking
trend during some fruit growth phases or under high
evaporative demand even without a water limitation
(Goldhamer et al. 1999). Thus, daily changes in MXTD
and MNTD have little signi®cance when the trunk is not
growing. On the other hand, rapid growth may minimize
di€erences in MDS, rendering this parameter less useful
as a water de®cit indicator in young trees.
Variability in TDM and the signal:noise ratio
There have been reports of substantial tree to tree vari-
ability in TDM regardless of the irrigation program
117
(Ginestar and Castel 1996; Naor 2000). High measure-
ment variability would require instrumentation of a large
number of trees within an orchard, increasing the mon-
itoring costs in order to decrease the uncertainty. How-
ever, variability (``noise'') must be considered relative to
the strength of the indicator (``signal''); the parameter
derived from TDM. If the signal strength is sucient, the
noise caused by the tree-to-tree variability may not be as
critical. Thus, it is important to characterize the sig-
nal:noise ratio of all plant-based indicators of water
status, including those derived from TDM, in comparing
the utility of each parameter for irrigation management.
Experimental methods
Over the past 3 years we have collected TDM in irrigation exper-
iments with peach, olive, almond, ®g and citrus trees. Here we
present two data sets for almond and peach trees that illustrate the
diversity of situations that are commonly found when monitoring
trunk diameter ¯uctuations and how the data can be used in irri-
gation management. The experiments were conducted in two lo-
cations within the San Joaquin Valley of California during the
summer of 1999. Evaporative demand, as expressed by ETo values
calculated using data from California Irrigation Management In-
formation System (CIMIS) weather stations located within 10 km
of each experiment site and a modi®ed Penman equation (Snyder
and Pruitt 1989), was similar and fairly constant throughout the
measurement period (Fig. 2).
Mature almond
The trees [Prunus dulcis (Mill.) Webb cv. Price] were 6 years old
and located in a commercial orchard near Lost Hills in western
Kern County, California. The soil was a clay loam (Typic Torri-
orthents) with a root zone extending to about 2 m depth. Micro-
sprinklers (rate 41.6 l/h) located midway between the trees in the
tree row were used for irrigation. A water budget approach was
used for irrigation scheduling with set times usually being 24 h.
Irrigation frequency was every 3±5 days. A build-up of algae in the
canal supplying this orchard and subsequent pumping problems
reduced water application rates. This de®cit irrigation presented us
with an ideal opportunity to evaluate the gradual development of
water stress in these trees.
Four trees were instrumented with linear variable displacement
transducers (LVDTs; Model 2.5 DF, Solartron Metrology, Bognor
Regis, UK) installed on the southwest primary sca€old. The
LVDTs were mounted on holders built of aluminum and invar ± an
alloy comprising 64% Fe and 35% Ni that has minimal thermal
expansion (Li et al. 1989). Measurements were taken every 30 s and
a datalogger (Model CR 10, Campbell Scienti®c, Logan, Utah) was
programmed to report 20-min means. Throughout the experimen-
tal period, the LVDTs did not have to be repositioned; the
2500 mV (2.5 mm) working range of the datalogger was not
exceeded by tree growth.
Midday SWP (1300±1400 hours) was monitored every weekday
using a pressure chamber (Model 3005, Soil Moisture Equipment,
Santa Barbara, Calif.). Single leaves close to the trunk on each of
the four trees were covered using a small black polyethylene bag
covered with silver foil for at least 2 h prior to measurement. The
leaf/bag combination was placed in the chamber within seconds of
excision and the precautions recommended by Hsiao (1990) were
taken to prevent leaf water loss during measurement.
Although four leaves were evaluated for SWP per sampling
date, the technician taking the measurements recorded only the
average of the four readings. Thus, it was not possible to determine
tree-to-tree SWP variability in the orchard where the TDM were
118
Fig. 2 Reference crop water use
(ETo) for the almond and peach
experimental sites
taken. As an alternative technique, we used a SWP data set that
was taken in a nearby orchard on single leaves from each of four
almond trees of a similar age, location, and irrigation management
approach as the trunk-sensor-equipped orchard to assess the vari-
ability in SWP. These measurements were taken on 81 days from
late June until early November.
To evaluate the relationships between MDS and parameters
related to evaporative demand, MDS data were taken during the
summer of 1999 on 4 year old almond trees in an adjacent orchard
that were speci®cally irrigated at 150% of estimated crop evapo-
transpiration (ETc), based on published crop coecients (Kcs) and
ETo. The MDS data, taken under non-limiting water conditions,
were correlated with daily ETo and mean VPD calculated from
mean daily vapor pressure and relative humidity for days 148±196.
Young peach
Two-year-old, daily drip-irrigated peach trees [Prunus persica (L.)
Batsch, cv. September Snow] grown on a loam soil (Pachic Hap-
loxeroll) in Tulare County, California, 56 km from the almond site,
were evaluated. They were part of a study assessing various soil,
plant, and atmospheric-based irrigation scheduling approaches
for practical, on-farm water management. One of the scheduling
approaches resulted in the slow development of water de®cits and
this treatment (hereafter referred to as the De®cit) and a regime
that applied 150% of estimated ETc (hereafter referred to as the
Control) are compared here.
The peach trees di€ered from the almond in their age and
irrigation frequency. The number of trees instrumented for trunk
diameter monitoring and the SWP monitoring techniques were
identical to those outlined above for the almond experiment.
Results
Mature almond
MDS and SWP for 48 days between late May and late
July 1999 are presented in Fig. 3a. The lowest MDS
occurred on the day of irrigation and increased steadily,
reaching its maximum just prior to the next irrigation.
The combination of insucient irrigation and higher
evaporative demand as the season progressed led to
higher peaks in MDS in the last three irrigations of the
period. Patterns of SWP were similar to those of MDS,
with the highest values on the irrigation days and in-
creasingly lower values as the soil dried both between
irrigations and as the season progressed. To establish a
reference, regression lines were drawn through the min-
imum MDS and the maximum SWP values, respectively
(Fig. 3a). Note that while the MDS reference line was
nearly constant for the measurement period, the SWP
reference decreased with time. Notwithstanding the mild
de®cit irrigation during this period, a decrease in plant
water potential during the season even under non-lim-
iting soil water conditions is not uncommon (Fereres
et al. 1978; McCutchan and Shackel 1992). Dividing
MDS and SWP by the baseline values generated what we
hereafter refer to as signal, as shown in Fig. 3b. The
peak MDS signal varied from 2.5 to 11.4 while the SWP
signal was much lower, varying from 1.3 to 1.4.
The greater strength of the MDS signal relative to that
of SWP has to be seen in the context of the variability of
each measurement. In other words, the signal:noise ratio.
Table 1 presents the mean MDS and SWP signals for
each irrigation cycle, their mean coecients of variation
(noise), and the signal:noise ratio. The average SWP
coecient of variation (CV) was 6.2%; a similar value of
5.2% can be calculated from the prune tree SWP data of
McCutchan and Shackel (1992). The MDS CV was
15.8% but, even with this higher value, the MDS sig-
nal:noise ratio was almost double that of SWP (Table 1).
Thus, MDS was more sensitive than SWP as a plant
water stress indicator in this particular sample of
microsprinkler-irrigated, mature almond trees.
Daily MXTD and MNTD for the experimental period
had similar general patterns although MNTD variations
were somewhat greater (Fig. 4). For instance, 1±3 days
prior to irrigation (days 155±158), MNTD decreased
 119

Fig. 3 For the mature almond

trees, a maximum daily trunk
shrinkage (MDS), stem water
potential (SWP), and baselines
for each parameter, and b indi-
cator ``signals'' (MDS and SWP
values derived by baseline val-
ues). Arrows indicates irriga-
tions

while MXTD increased (Fig. 4). While there was some Table 1 Almond ``signal'' (measured indicator/reference), ``noise''
growth (about 1 mm for the 48-day period), the growth (coecient of variation) and signal:noise ratios for almond (max-
rate was low enough to in¯uence MDS only minimally. imum daily trunk shrinkage (MDS) and stem water potential
(SWP) reported in Fig. 3
Figure 5 shows that there were moderately good
correlations between MDS of almond trees grown under Indicator Mean peak value/ Mean coecient Signal:noise
non-limiting soil water conditions and mean daily ETo referencea per of variationb ratio
irrigation cycle
(R2 0.463) and mean daily VPD (R2 0.626) for this
observation period (Figs. 5a and 5b, respectively). MDS 6.67 0.158 42.2
SWP 1.37 0.062 22.1
a
Young peach ``Signal'', b ``Noise''

MDS values for both treatments were low until day 164
when the MDS of the De®cit treatment gradually in-
creased, reaching about 200% of Control by day 170
(Fig. 6a). The greater MDS of the De®cit indicated a
lower water status; a fact that was con®rmed by the
SWP measurements (Fig. 6b). By day 178, the MDS
patterns reversed and, for a few days, MDS of the
Control was greater than that of the De®cit. After day
184, the timing of the MDS peaks and valleys for both
treatments was similar but di€ered in magnitude;
sometimes the Control was greater than the De®cit and
sometimes the reverse (Fig. 6a). This erratic behavior of
MDS during the second part of the experimental period
was not related to changes in tree water status as there
120
Fig. 4 Maximum (MXTD) and
minimum (MNTD) trunk
diameter with time for the
mature almonds
were consistent SWP treatment di€erences; the De®cit
being almost always lower than the Control (Fig. 6b).
Indeed, relative MDS (De®cit/Control) values during
the experimental period indicate that a water de®cit was
®rst detected on day 160 (Fig. 6c), well before there were
detectable SWP di€erences on day 173 (Fig. 6b). Rela-
tive MDS continued to be greater than relative SWP
until day 178, after which it declined and oscillated
around unity for the remainder of the cycle (Fig. 6c).
Since young peach MDS did not always correlate well
with SWP, as was the case in the mature almond, we
examined growth records seeking alternative tree stress
indicators derived from TDM. Daily changes in MNTD
show that following the initial period when there were
no treatment di€erences, the growth rate of the Control
was consistently higher than that of the De®cit (Fig. 7).
This behavior is similar to the SWP patterns (Fig. 6b).
Thus, for this sample of rapidly growing young trees
(trunk diameter increased 8.35 mm during the experi-
mental period), tree growth, as expressed by daily dif-
ferences in MNTD, o€ers a more consistent indicator
than MDS for scheduling irrigation. Time-course pat-
terns of the daily changes in MXTD were similar to
MNTD and, thus, this growth record could also be used
as a stress indicator (data not shown).
What caused the MDS to be inconsistent in this case?
We believe it was trunk growth rates, which we illustrate
for two situations that resulted in reversed tendencies of
relative MDS (Fig. 8a, b). During the period 14±18
June, the Control trees grew much faster than the De®cit
trees (Fig. 8a). This is illustrated by the steeper slopes of
both the MXTD and MNTD values with time and the
fact that the slopes of these parameters were similar.
This resulted in relatively low MDS values. On the other
hand, the slower-growing De®cit trees had lower MXTD
and MNTD slopes with time than in the Control
(Fig. 8a). Additionally, the Control trees had a lower
MNTD slope than MXTD, resulting in relatively high
MDS values. A di€erent situation occurred later as the
relative growth rates of the two irrigation treatments
slowed, especially in the Control. For example, on days
189±194, MDS in the Control trees was greater than that
of the De®cit trees, due primarily to higher MXTD
values. In other words, the more rapid growth rate in the
Control actually resulted in higher MDS values in this
latter case compared with the earlier case where rapid
growth resulted in lower MDS. Thus, higher growth
rates can interact with measurements of trunk shrinkage
masking the water de®cit-related di€erences in MDS
among irrigation treatments. Such growth e€ects would
be most important in the case of young trees for which
parameters derived from MNTD or MXTD seem most
valuable for irrigation scheduling.
Proposed irrigation scheduling protocols
We assume that for a particular tree species under a given
evaporative demand, TDM vary primarily due to plant
water status, tree age, and irrigation frequency. Thus, we
present irrigation scheduling protocols, in principle
adapted to Prunus species trees, based on TDM that
cover the range of the latter two factors; mature trees
under both high and low irrigation (drip or microsprin-
kler) frequencies and young trees under high frequency
irrigation, which is the likely system management to be
used. We consider trees to be mature when vegetative
growth slows in relation to previous seasons. Young trees
are considered those that are rapidly growing, normally
under 5 years for Prunus trees. Based on the experimental
results, we use MDS as the operative parameter with
mature trees and the evolution of MNTD for the young
trees. Our use of TDM for irrigation scheduling is pred-
icated on the assumption that TDM can be used to in-
dicate the presence of very mild water stress, which would
likely be undetectable by plant water potential monitor-
ing, and that fruit trees tolerate this mild stress without
negative impacts on production. We present these pro-
tocols as possible approaches to using TDM in real-world
irrigation management and recognize that, as research
 121

Fig. 5 Relationships for the

mature almond trees under
non-limiting soil moisture
between maximum daily
shrinkage (MDS) and a
reference crop water use (ETo),
and b mean daily vapor pres-
sure de®cit (VPD)

involving TDM continues, alternative protocols for
Prunus and other fruit tree species are likely to emerge.
Case 1: Mature trees under low frequency irrigation
(3 day interval or greater)
1. The irrigation start date is selected. The criteria for
starting irrigation should be based on manual soil
water measurements, the water balance using ETc, or
experience. However, irrigation should start as early
as feasible.
2. A ®xed amount is applied; the MDS of the day fol-
lowing irrigation (MDSr) is recorded to be used as a
reference.
3. The MDS of subsequent days is compared with a
threshold MDS (MDSt) established relative to MDSr
(for example, MDSt could be 150% of MDSr or
1.5´MDSr). MDSt could also be based on experience
or obtained from future research results.
4. If the MDS of three consecutive days exceeds the
MDSt before the next irrigation, the application
depth (system operating time) is raised by 10%.
122

Fig. 6 For the De®cit and

Control irrigation regimes in
the young peach trees, a maxi-
mum daily trunk shrinkage
(MDS), b stem water potential
(SWP); vertical bars are one
standard error, and c relative
(De®cit/Control) values

Similarly, if the MDS greatly exceeds MDSt by, for
example, 100% (2´MDSt), the irrigation interval
may have to be shortened.
5. The MDS of the day following the next irrigation will
become the new MDSr and will be used for the following
irrigation cycle. Alternatively, MDSr could be calcu-
lated from MDS early in the season when the pro®le is
fully wetted and corrected for changes in evaporative
demand (ETo or VPD) as the season progresses.
6. If the MDS during the next irrigation cycle never
attains the MDSt, then the application depth is
decreased by 10%.
7. Early in the season, greater changes in system oper-
ating times should be considered (20%, for example)
due to uncertainty in determining the initial appli-
cation amounts. This would allow faster convergence
of the application amounts and tree water require-
ments.
Case 2: Mature trees under high frequency irrigation
(2 day interval or less)
1. The irrigation starting date is selected and a ®xed
amount applied.
2. A MDSr must be determined so that MDSt can be
de®ned, using one of the following three approaches:
(a) recording MDS after a heavy rain or irrigation
with a fully charged pro®le, (b) maintaining a few
reference trees in the orchard that get 20±50% more

Fig. 7 Peach trunk growth ex-
pressed as daily minimum trunk
diameter (MNTD)
Fig. 8 Trunk diameter ¯uctuations (mm) in young peach trees
a from 14±18 June, and b from 8±12 July
water than the rest (by changing emitter output or
numbers) and using their MDS as MDSr, or (c) data
from future research results.
123
The value chosen for MDSr would have a threshold
level below that of case 1; for example, MDSt =
1.25´MDSr. When the MDS of three consecutive days
exceeds MDSt, the irrigation application is raised by 10%.
Similarly, if the MDS does not reach MDSt in the fol-
lowing 5 days, the irrigation depth is decreased by 10%.
Case 3: Young trees under high frequency irrigation
(2 day interval or less)
1. The irrigation starting date is selected and a ®xed
amount applied.
2. A reference daily growth rate [(dMNTD/dt)r] must be
developed by either: (a) maintaining reference trees as
outlined in case 2(b) above, or (b) data from the lit-
erature.
3. If the dMNTD/dt of three consecutive days is, for
example, 15±20% below (dMNTD/dt)r, the irrigation
depth is raised by 10%.
4. If the dMNTD/dt of three consecutive days is, say,
within 5±10% or less of (dMNTD/dt)r, the irrigation
depth is decreased by 10%. The dMXTD/dt could be
used instead of dMNTD/dt in this case, as it also
sensitive to water de®cits.
Discussion
Irrigation scheduling at the plot scale has been investi-
gated for decades and some of the ®rst instruments de-
signed for irrigation timing (tensiometers, gypsum
blocks) were commercially available more than 50 years
ago. It is not a coincidence that such instruments mon-
itored soil water levels and that subsequent methods
emphasized the use of the water budget method which
requires estimates of crop ET (Jensen 1981). Informa-
124
tion on soil water and on atmospheric demand was
considered more reliable and less subject to error than
direct measurements of the plant, even though it ap-
peared desirable to use plant-based methods that would
relate irrigation directly to crop needs.
The highly dynamic nature of plant water status
makes it dicult to interpret plant water measurements.
Thus, the development of plant-stress-sensing instru-
mentation has lagged behind that used for soil water
or ET measurements (Hsiao 1990). Nevertheless, there
have been positive developments in the characterization
of plant water status and, even though they are seldom
used for on-farm scheduling, they are very helpful as
diagnostic indicators. The primary limitation in their on-
farm use is logistical; determining water potential with
the pressure chamber requires the physical presence of a
technician, normally around midday, and a sampling
frequency of more than once a week; a labor-intensive
procedure that presents economic limitations for large
scale adoption of these techniques.
Growth is a very sensitive parameter to water de®cits
(Bradford and Hsiao 1982) and, even though the sensi-
tivity of the early dendrometers was not nearly sucient
to pick up daily oscillations in diameter, the growth
trends over weeks were indicative of the adequacy of
irrigation (Hendrickson and Veihmeyer 1941). Early in
fruit tree irrigation research, the use of dendrometers for
irrigation scheduling was proposed (Hendrickson and
Veihmeyer 1941). The current trunk diameter sensors
can generate sensitive parameters that, once the refer-
ence and threshold values for the indicator used are well
established, may be useful in irrigation decision-making.
Being able to avoid damaging water stress at all times
because of their high sensitivity to mild water de®cits,
the strong correlation with established plant water status
parameters (Goldhamer et al. 1999), and the ease of
automation are all positive features of using TDM for
irrigation management.
The scheduling protocols we suggest provide guide-
lines that address both under- and over-irrigation. Since
TDM re¯ect plant water status, they can clearly be used
as indicators of the upper boundary limit of tree stress.
On the other hand, we believe they can also be used to
sense the lower tree stress boundary. This is accom-
plished by using small reductions in applied water until
the TDM-derived parameters indicate the presence of
very mild plant water stress. Thus, the necessity of a
soil-based instrument to indicate water moving below
the root zone is removed. As with any irrigation
scheduling tool, independent veri®cation of a trunk
diameter-based program with ETc estimates or soil
sensors is desirable.
Clearly, the greatest uncertainty in the protocols
outlined above is determining MDS and dMXTD/dt
reference values. Obtaining these parameters by irrigat-
ing a small number of reference trees at 125±150% of
estimated ETc (established using a combination of dif-
ferent emitter discharge rates and/or number) to ensure
non-limiting soil water conditions can be risky, espe-
cially in poorly drained soils. Anoxia may damage root
health, which is manifested by tree water de®cits and
even tree decline and death. Additionally, excessively
irrigated reference trees may become nitrogen de®cient.
Clearly, if reference trees are used, they should be in-
spected periodically by the grower to insure that their
appearance and health matches his/her objectives for the
whole orchard.
Since MDS correlates fairly well with VPD under
non-limiting soil water regimes (Fig. 5b), it is likely that
relationships between MDS and VPD for particular tree
species and age groupings can be developed. This has
been done for SWP in fully irrigated prune trees
(McCutchan and Shackel (1992); Shackel et al. (1997). It
is probable that similar relationships exist between
dMNTD/dt and evaporative demand for particular
species of young trees under full irrigation. Research
that provides this information should improve the
precision of using TDM for irrigation scheduling.
In addition to being useful for irrigation, where the
objective is to fully meet the water needs of the trees, we
believe that TDM can strengthen regulated de®cit irriga-
tion (RDI) programs. Numerous investigators have
shown that RDI can reduce seasonal ETc without reduc-
ing fruit yield or quality (Goodwin and Jerie 1992; Caspari
et al. 1994; Lampinen et al. 1995) and in some cases, ac-
tually improve yield components (Mitchell et al. 1986,
1989; Teviotdale et al. 1995). The primary current limi-
tation for RDI is the precision of the ``triggers'' used to
implement the de®cit irrigation. While irrigating at dif-
ferent percentages of ETc for speci®c time periods is often
utilized, this approach neglects the bu€ering e€ect the soil
moisture reservoir has in delaying the development of
plant stress in response to de®cit irrigation. Since predawn
LWP (Teviotdale et al. 1994), midday LWP (Grimes and
Yamada 1982), or SWP (Shackel et al. 1997) are clearly
superior indicators for imposing plant water de®cits, it
follows that TDM can also improve RDI management.
We believe that the future holds promise for irriga-
tion scheduling of trees and vines using continuously
recorded TDM. While the current sensors are costly, it is
reasonable to forecast that cheaper, even more accurate
sensors will eventually be produced and that remote data
transmission from the sensor directly to the farmer's
computer will shortly be available and cost-e€ective.
Acknowledgements We sincerely appreciate the assistance of the
following technicians: Mario Salinas, Jesus Salinas, Miguel Mar-
quez, Merce Soler Anaya, and Dan Howes. We also gratefully
acknowledge the assistance of the management and sta€ of Para-
mount Farming, including Joe McIlvaine, Dennis Elam, Joe
Gonzales, Raul Higuera, Lou Villaruz, and Marcos Rodriguez.
References
Bradford KJ, Hsiao TC (1982) Physiological responses to moderate
water stress. In: Lange OL, Nobel PS, Osmond CB, Ziegler H
(eds) Physiological ecology. II. Encyclopedia of plant physiol-
ogy. (NS vol 12B) Springer, Berlin Heidelberg New York,
pp 264±312

Cabibel LB, Isberie C (1997) Flux de seve et alimentation hydrique de
ceresiers irrigueÂs ou non en localization. Agronomie 17:97±112
Caspari HW, Behboudian MH, Chalmers DJ (1994) Water use,
growth, and fruit yield of Hosui Asian pears under de®cit irri-
gation. J Am Soc Hortic Sci 119:383±388
Cohen M, Ameglio T, Cruisiat P, Archer P, Valancogne C, Dayau
S (1997) Yield and physiological responses of walnut trees in
semiarid conditions: application to irrigation scheduling. Acta
Hortic 449:273±280
Fereres E, Goldhamer DA (1990) Deciduous fruit and nut trees. In:
Stewart BA, Nielsen DR (eds) Irrigation of agricultural crops.
(Monograph 30) American Society of Agronomy, Madison,
Wis., pp 987±1017
Fereres E, Acevedo E, Hsiao TC, Henderson DW (1978) Seasonal
changes in water potential and turgor maintenance in sorghum
and maize under water stress. Physiol Plant 44:261±267
Garnier E, Berger A (1986) E€ect of water stress on stem diameter
changes of peach trees growing in the ®eld. J Appl Ecol 23:193±
209
Ginestar C, Castel, JR (1996) Utilizacion de dendrometros como
indicadores de estreÂs hidrico en mandarinos jovenes regados
por goteo. Riegos Drenajes XXI 89:40±46
Goldhamer DA, Fereres E, Mata M, Girona J, Cohen M (1999)
Sensitivity of continuous and discrete plant and soil water stress
monitoring in peach trees subjected to de®cit irrigation. J Am
Soc Hortic Sci 124:437±444
Goodwin I, Jerie P (1992) Regulated de®cit irrigation: from con-
cept to practice. Advances in vineyard irrigation. Aust NZ
Wine Ind J 10 July:258±261
Grimes DW, Yamada H (1982) Relation of cotton growth and
yield to minimum leaf water potential. Crop Sci 22:134±139
Hendrickson AH, Veihmeyer FJ (1941) Some factors a€ecting the
rate of growth of pears. Am Soc Hortic Sci Proc 39:1±7
Hsiao TC (1990) Measurements of plant water status. In: Stewart
BA, Nielsen DR (eds) Irrigation of agricultural crops. (Mono-
graph 30) American Society of Agronomy, Madison, Wis., pp
243±279
Huck MG, Klepper B (1976) Water relations of cotton. II. Con-
tinuous estimates of plant water potential from stem diameter
measurements. Agron J 69:593±597
Huguet JG, Li SH, Lorendeau JY, Pellous G (1992) Speci®c
micromorphometric reactions of fruit trees to water stress
and irrigation scheduling automation. J Hortic Sci 67:631±640
Li SH, Huguet JG, Bussi C (1989) Irrigation scheduling in mature
peach orchard using tensiometers and dendrometers. Irrig
Drain Syst 3:1±12
Jensen ME (1981) Summary and challenges. In: Irrigation sched-
uling for water and energy conservation in the 80's. (ASAE
special publication 23±81) pp 225±231
125
Klepper B, Browning VD, Taylor HM (1971) Stem diameter in
relation to plant water status. Plant Physiol 48: 683±685
Kozlowski TT (1967) Diurnal variations in stem diameter of small
trees. Bot Gaz 123:60±68
Lampinen BD, Shackel KA, Southwick SM, Olson B, Yeager JT,
Goldhamer D (1995) Sensitivity of yield and fruit quality of
French prune to water deprivation at di€erent fruit growth
stages. J Am Soc Hortic Sci 120:139±147
Martin DL, Stegman EC, Fereres E (1990) Irrigation scheduling
principles. In: Ho€man GJ, Howell TA, Solomon KH (eds)
Management of farm irrigation systems. American Society of
Agricultural Engineers, Madison, Wis., pp 155±203
McCutchan H, Shackel KA (1992) Stem-water potential as a sen-
sitive indicator of water stress in prune trees (Prunus domestica
L. cv. French). J Am Soc Hortic Sci 117:607±611
Mitchell PD, Chalmers DJ, Jerie PH, Burge G (1986) The use of
initial withholding of irrigation and tree spacing to enhance the
e€ect of regulated de®cit irrigation on pear trees. J Am Soc
Hortic Sci 111:858±861
Mitchell PD, Ende B van den, Jerie PH, Chalmers DJ (1989) Re-
sponses of `Barlett' pear to withholding irrigation, regulated
de®cit irrigation, and tree spacing. J Am Soc Hortic Sci
114:15±19
Molz FJ, Klepper B (1972) Radial propagation of water potential
in stems. Agron J 64:469±473
Naor A (2000) Midday stem water potential as a plant water stress
indicator for irrigation scheduling in fruit trees. Acta Hortic
537:447±454
Shackel KA, Ahmadi H, Biasi W, Buchner R, Goldhamer D,
Gurusinghe S, Hasey J, Kester D, Krueger B, Lampinen B,
McGourty G, Micke W, Mitcham E, Olson B, Pelletrau K,
Philips H, Ramos D, Schwankl L, Sibbett S, Snyder R,
Southwick S, Stevenson M, Thorpe M, Weinbaum S, Yeager J
(1997) Plant water status as an index of irrigation need in
deciduous fruit trees. HortTechnology 7:23±29
Snyder RL, Pruitt WO (1989) Reference evapotranspiration. In:
Goldhamer, DA, Snyder RL (eds) Irrigation scheduling: a guide
for ecient on-farm water management. (Publication 21454)
University of California Division of Agriculture and Natural
Resources, pp 24±27
Teviotdale BL, Michailides T, Goldhamer DA, Viveros M, Sch-
midt L, Hines V (1994) Cutting o€ irrigation early may reduce
almond hull rot. Calif Agric 48:33±36
Teviotdale BL, Michailides TJ, Goldhamer DA, Viveros M (1995)
Hull rot disease caused by Rhizopus stolonifer by early termi-
nation of preharvest irrigation. Plant Dis 79:402±405