Académique Documents
Professionnel Documents
Culture Documents
MAURITS W. VANDEGEHUCHTE
1
*
, MOHAMED BRAHAM
2
, RAOUL LEMEUR
1
AND KATHY STEPPE
1
1
Laboratory of Plant Ecology, Department of Applied Ecology and Environmental Biology, Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium
2
Institute of the Olive Tree Station of Sousse, Sousse, Tunisia
ABSTRACT
Improvement in irrigation techniques has led to an expansion of the irrigated olive area in Tunisia and consequently a rise in
agricultural water consumption. Here, the actual water use of Olea europaea L. Meski as estimated by thermal dissipation
probe measurements was compared for different irrigation regimes: three plots with irrigation doses based on PenmanMonteith
crop evapotranspiration (ETc), 0.33 ETc and the local growers experience (about 0.6 ETc), respectively. The scaling up of
sap ux densities proved to be difcult given the azimuthal and radial sap ux density variability and the conductive surface area
asymmetry. Moreover, natural thermal gradients may lead to inaccurate use of standard sap ow formulas. Depending on the
manner of upscaling, differences in water use of more than 50 l day
-1
per tree were obtained (up to 0.7 ETc irrigation dose).
Despite these difculties, sap ow calculations seem crucial to accurately determine the water needs of the orchard under
investigation, as they corresponded closely to the lowest irrigation dose which was considered sufciently high as no signicant
differences in drought stress variables were detected between the plots. Therefore, irrigation based solely on ETc leads to
wasting of water; hence, careful application of both meteorological and sap ow methods is desirable in irrigation scheduling.
Copyright 2012 John Wiley & Sons, Ltd.
key words: irrigation; PenmanMonteith; thermal dissipation probe (TDP); sap ow; radial prole; Olea europaea L. (olive) tree
Received 22 July 2011; Revised 12 April 2012; Accepted 12 April 2012
RSUM
Lamlioration des techniques dirrigation conduit lexpansion de la supercie irrigue pour la culture doliviers en Tunisie,
et par consquent laugmentation de la consommation deau par lagriculture. Dans cette tude, la consommation deau relle
de Olea europaea L. Meski telle quelle est estime par des mesures de dissipation thermique a t compare diffrents
rgimes dirrigation: trois parcelles avec des doses dirrigation sur la base de lvapotranspiration des cultures de Penman
Monteith (ETC), 33% de cette ETC et lexprience du producteur local (environ 60% de lETC), respectivement. Lintensication
des densits de ux de sve sest avre difcile en raison de la variabilit azimutale et radiale de la densit de ux de sve et
de lasymtrie de la surface conductrice. En outre, les gradients thermiques naturels peuvent conduire une utilisation
inexacte des formules standard de ux de sve. En fonction de changer dchelle, des diffrences de consommation deau
arbre de plus de 50 l arbre
-1
j
-1
ont t obtenues (jusqu 70% de la dose dirrigation ETC 100%). Malgr ces difcults,
les calculs de ux de sve semblent dune importance cruciale pour dterminer avec prcision les besoins en eau du verger
tudi car les calculs obtenus de ux de sve correspondaient troitement la dose la plus faible dirrigation qui a t juge
sufsamment leve pour ne pas gnrer de diffrences signicatives dans les variables indicatrices de stress hydrique inter
parcellaire. Par consquent, lirrigation base uniquement sur ETC mne au gaspillage deau, et une application rigoureuse
des deux mthodologies bases respectivement sur des donnes mtorologique et les mthodes de ux de sve est donc
souhaitable pour laborer le calendrier dirrigation. Copyright 2012 John Wiley & Sons, Ltd.
mots cls: irrigation; PenmanMonteith; sonde de dissipation thermique (TDP); ux de sve; prol radial; Olea europaea L. (olivier)
* Correspondence to: Maurits W. Vandegehuchte. Laboratory of Plant Ecology, Department of Applied Ecology and Environmental Biology, Faculty of
Bioscience Engineering, Ghent University, Coupure links 653, B-9000 Ghent, Belgium, Phone +32 9 264 61 12, Fax +32 9 224 44 10.
E-mail: maurits.vandegehuchte@UGent.be
Limportance des mesures de ux de sve pour estimer la consommation deau relle doliviers Meski sous diffrents rgimes dirrigation en Tunisie.
IRRIGATION AND DRAINAGE
Irrig. and Drain. 61: 645656 (2012)
Published online 15 July 2012 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/ird.1670
Copyright 2012 John Wiley & Sons, Ltd.
INTRODUCTION
Given the adaptations of olive trees (Olea europaea L.) to
long-term drought, their cultivation is widely spread in
semi-arid areas with a Mediterranean climate such as Tunisia
(Gimenez et al., 1997). While rain-fed orchards are still often
exploited, irrigated orchards are gaining ground thanks to
larger fruit yield and growth. Given the higher economic value
and to meet the growing demand for olive products, many
traditional orchards have therefore been modernized by means
of several irrigation procedures (Fernandez and Moreno,
1999), inevitably raising the consumption of water. Because
of this augmented water use, researchers are stressing the
necessity of more accurate and precise irrigation methodolo-
gies (Fereres and Evans, 2006).
Irrigation is often based on direct measurements of the
soil-water status either in terms of water content (gravimet-
ric method) or water potential (tensiometer, psychrometer).
These methods have the disadvantage that multiple sensors
are needed when dealing with a heterogeneous soil. An
alternative approach is to determine the atmospheric water
demand based on the combination of an energy and water
balance such as the PenmanMonteith method (Allen
et al., 1998). A major shortcoming of soil- and atmospheric-
based methods is that they do not take specic plantwater
interactions into account such as changes in internal water
storage. Therefore, Jones (2004) suggests that when develop-
ing new irrigation strategies, stress indicators which are
measured on the plant itself are taken into account. Possible
stress indicators are leaf water potential, stomatal resistance,
sap ux density or stem diameter variations (Nadezhdina,
1999; Moriana and Fereres, 2002; Jones, 2004; Saveyn
et al., 2007; De Pauwet al., 2008; Sevanto et al., 2008; Steppe
et al., 2008; De Swaef et al., 2009; Villez et al., 2009). Irriga-
tion methods based on direct measurements of leaf water
potential and stomatal resistance have, however, the disadvan-
tage that they are hard to automate. Moreover, they do not lead
to accurate results when dealing with isohydric species
because of the partial separation of the linkage between these
indicators and sap ux density (Buckley, 2005; Fisher et al.,
2006; Franks et al., 2007). Because of these limitations, irriga-
tion methods based on direct sap ux density measurements
are gaining in importance (Fernandez and Moreno, 1999;
Do and Rocheteau, 2002; Nadezhdina et al., 2002; Lu et al.,
2004; Steppe et al., 2008).
This study had the objective of assessing the applicability
and benets of thermal dissipation probe (TDP) measure-
ments to determine whole-tree water use of olive (Olea
europaea L. Meski) growing in a commercial orchard at
Endha, Tunisia, in comparison with classically applied
PenmanMonteith based crop evapotranspiration for irriga-
tion scheduling. To this end, water use of Meski olive trees
(Olea europaea L. Meski) irrigated based on the local
growers experience was compared with water use of trees
irrigated according to the PenmanMonteith crop evapo-
transpiration (ETc), both for 100% ETc and 33% ETc. Next
to TDP sap ux density estimates, ecophysiological mea-
surements of leaf and stem water potential and azimuthal
and radial sap ux densities were conducted.
MATERIALS AND METHODS
Plant material and experimental set-up
Measurements were carried out on 15-year-old Meski olive
trees (Olea europaea L. Meski), an important table olive
cultivar. These trees were part of an orchard located at
Endha, central Tunisia (10
22
0
E, 36
38
0
N), which belongs
to lOfce de Terres Domaniales de lEndha (OTD). This
region is characterized by a semi-arid climate with dry, hot
summers and humid, mild winters, resulting in an average
annual rainfall of 300 mm and temperature of 1825
C.
The loamy-sand soil has an average to low organic carbon
content and nitrogen content. The trees are located in grids
of 7 7 m and are pruned once a year after harvesting. In this
orchard, three plots of 10 trees were selected for the experi-
ments. In each plot, three trees were selected with a stem
diameter of 18 3.4 cm at breast height and a similar height
(3.5 0.4 m), crown shape and leaf area index (1.7 0.2).
Still, differences in stem and branch shape were detectable.
These irregular forms are typical of olive trees.
Each plot was irrigated by a localized irrigation system
with, for each tree, four drip nozzles located at a distance
of 1.0 m to the north, east, south and west of the trunk,
respectively. All plots were irrigated based on the experi-
ence of the local grower, irrigating 8 ha day
1
with a
discharge rate of approximately 1.4 l h
-1
per nozzle (approx-
imately 45 l day
1
per tree) before the experiment was
started. From April onwards, the three plots were irrigated
differently. For plot one, a water irrigation amount equiva-
lent to 100% of crop evapotranspiration (ETc) was
provided, calculated using the PenmanMonteith FAO
method (Allen et al., 1998). To obtain crop evapotranspira-
tion from reference evapotranspiration, a single estimated
crop coefcient (K
c
= 0.45) and a coverage coefcient
(K
r
= 0.67) were applied, as it has recently been shown
that the K
c
values provided by the FAO lead to overestima-
tions of crop evapotranspiration (dAndria et al., 2004;
Er-Raki et al., 2008). For plot two, one-third of this irriga-
tion dose was applied, while for plot three, the dose as deter-
mined by the local grower remained applicable.
Evapotranspiration
For evapotranspiration measurements an on-site weather
station was used to measure the relative humidity (RH)
646 M. W. VANDEGEHUCHTE ET AL.
Copyright 2012 John Wiley & Sons, Ltd. Irrig. and Drain. 61: 645656 (2012)
and the air temperature (T) using an integrated relative
humidity sensor (RHT2nl combined sensor, Delta-T
Devices Ltd, Cambridge, UK). Shortwave radiation was
measured with a photodiode 5 m above the ground surface
(ES2 Si, Delta-T Devices Ltd, Cambridge, UK) and soil heat
ux with a Heatux plate (HFP01SC, Hukseux, Delft,
Netherlands). Wind speed was measured with an anemometer
(AN1, Delta-T Devices Ltd, Cambridge, UK), installed 5 m
above the orchard oor.
Evapotranspiration calculations were carried out manu-
ally based on the FAO PenmanMonteith method developed
by Allen et al. (1998). To determine the net radiation (R
n
in
J m
-2
day
-1
), the following formula was used:
R
n
1 a R
s
R
l;in
esT
4
surf
(1)
where a is the albedo (for olive orchards with sand-loamy
soils equal to 0.2 (Moreno et al., 1996), s the Stefan-
Boltzmann constant (J K
-4
m
-2
day
-1
), e the emissivity
constant (equal to 0.98 according to Sepulcre-Canto et al.,
2006), T
surf
the surface temperature (K), R
s
the incoming
shortwave radiation (J m
-2
day
-1
) and R
l,in
the incoming
longwave radiation (J m
-2
day
-1
).
To determine R
l,in
several empirical formulas, using air
temperature (T in K) and vapour pressure (e
a
in hPa), were
evaluated. Most commonly used formulas are those of
Swinbank (1963) (Equation 2), Idso (1981) (Equation 3)
and Brunt (1932) (Equation 4):
R
l;in
0:92 10
5
T
2
sT
4
(2)
R
l;in
0:70 5:95 10
5
e
a
exp 1500=T
sT
4
(3)
R
l;in
0:51 0:066 e
a
1=2
sT
4
(4)
These formulas are adequate for use with clear skies,
which was the case in this study where they were compared
with the standard FAO method.
Water potential and stomatal resistance
Differences in drought stress between the three different
irrigation plots were assessed based on different physiological
measurements during July and August 2008 as these
months were considered to cause the highest level of drought
stress. Leaf and stem water potential were measured with a
Scholander pressure chamber (Scholander pressure chamber,
PMS Instrument Company, Oregon, USA) for nine trees
(three trees per irrigation plot). As the trees were of similar
height, crown structure and shape, stem diameter at breast
height and leaf area index, comparison of measurements was
possible. For each measurement day, one tree from each
irrigation plot was measured in four azimuthal directions from
sunset till approximately 7 pm. This was repeated for the same
trees on three comparable clear sky days. As the meteorolog-
ical conditions during the measurement period remained
approximately constant, comparison of the three measurement
repetitions per tree was possible. Hence, nine measurement
days in total were conducted: three repetitions with for every
repetition one tree per irrigation plot. For a single measure-
ment, the average of three leaves was taken.
For each tree, leaf and stem water potential and stomatal
resistance (AP4 dynamic porometer, Delta-T Devices Ltd,
Cambridge, UK) were measured in the four azimuthal
directions. To determine stem water potentials, selected
leaves were enclosed, still attached to the tree, in plastic
bags covered with aluminium foil 2 h prior to the measure-
ment (Fulton et al., 2001). Both for the stem and leaf water
potentials as for the stomatal resistance, the average of three
leaves was taken for each measurement point. Attention was
paid to choose leaves at similar height and of similar age and
exposure to sunlight.
For each measurement day, six to seven measurement
points per azimuthal direction for both stem and leaf water
potential and stomatal resistance were obtained between
sunset and 7 pm.
Sap ux density
Sap ux density was continuously monitored approximately
30 cm above soil level in four azimuthal directions using
thermal dissipation probes (TDP30, Dynamax Inc., Houston,
TX, USA) on the same trees on which stomatal resistance
and water potentials were measured. As natural temperature
gradients were considered a risk, a cyclical heating scheme
as suggested by Do and Rocheteau was applied (2002). The
thermocouple of the thermal dissipation probes was located
15 mmfromthe tip of the needle. The heater and measurement
needle were located 4 cm distant. The heater element was
discontinuously heated (10 min heating, 20 min cooling) in
order to use less energy. Using the equations of Do and
Rocheteau (2002), sap ux density was calculated as
SFD 0:00274 K
a
= 1 K
a
0:707
(5)
with SFDthe sap ux density (m m
-2
s
-1
) (Steppe et al., 2010)
and K
a
calculated as
K
a
T
a0
T
au
= T
au
(6)
with T
a0
the maximum alternating signal (T
on
T
off
with T
on
the temperature difference between the needles at
the end of the heating and T
off
the temperature difference
between the needles at the end of the cooling period) and
T
au
the measured alternating signal.
To determine whole-tree water use, sap ux density was
multiplied by the total conductive surface area. This
surface area is commonly calculated as a fraction of the total
647 WATER USE MESKI OLIVE
Copyright 2012 John Wiley & Sons, Ltd. Irrig. and Drain. 61: 645656 (2012)
cross- sectional area of the stem (considered as a perfect
disc) (Allen and Grime, 1995)
To gain more insight in the radial sap ux prole in
olive, the sensor needles were displaced at different depths
(5, 8, 15 and 20 mm) every second day with 24 h measure-
ments for each depth. Hence, on the measurement days on
which sap ux was determined at 5, 8 and 15 mm depth,
water potential and stomatal resistance were also measured
on the same trees.
For each new depth, the insertion hole was drilled a bit
deeper, taking care not to extend the diameter of the drill
hole. To be able to compare these measurements, meteoro-
logical conditions between the days should not differ too
much. To this end, average values and correlation coef-
cients of air temperature, relative humidity, shortwave
radiation, soil heat ux and wind speed for the different days
were calculated. For air temperature, shortwave radiation
and soil heat ux, correlation coefcients varied between
0.92 and 0.99 for the four depths considered. The correla-
tions for relative humidity and wind speed were less
pronounced (Table I). Table II gives the lower and upper
bounds of the signicant differences for the meteorological
variables for the different depths. Because the environ-
mental variables were statistically comparable, it was
possible to approximate the radial sap ux prole in olive
(determined by averaging the sap ow measurements of
the four azimuthal directions). This statistical analysis
also justies the comparison of the water potential and
stomatal resistance measurements for the consecutive mea-
surement days.
In addition, the actual surface area of the conductive
tissue was determined by colouring the xylem tissue.
Therefore, a waterproof receptacle was constructed around
the stem of one of the olive trees. This receptacle was lled
with an acid fuchsine solution (0.5% w/v). Several holes
were drilled underwater to prevent air bubbles from entering
the trunk. The coloured solution penetrated the conductive
tissue and was taken up by the sap stream. When colouring
of the leaves was visible, the tree was cut down. A surface
scanner (Li 3050A, Li-Cor, Lincoln, Nebraska, USA)
was used to determine the actual conductive surface area
of the trunk.
Statistical analysis
For statistical analysis, the software package S + (TIBCO
Software Inc.) was used. Normality and homoscedasticity
of the data were tested with Kolmogorov Smirnov and
modied Levene tests, respectively. Averages were then
compared based on a Tukey test (95% signicance level).
RESULTS
PenmanMonteith crop evapotranspiration
The use of different empirical formulas to determine R
l,in
led
to differences in R
n
(Figure 1a) and, subsequently, to small
deviations in reference evapotranspiration (Figure 1b).
Statistical analysis showed that only the reference evapo-
transpiration calculated according to Idso (1981) was signif-
icantly higher than the other methods.
The irrigation dose based on the reference evapotranspira-
tion, which was on average about 5.5 mm day
-1
during July
and August 2008, corresponded to approximately 75 l per tree.
For the second irrigation plot, one-third of this dose was
applied ( 25 l per tree) while, based on the experience of
the local grower, plot three was irrigated with45 l per tree.
Water potential and stomatal resistance
The average meteorological conditions for the measurement
days are presented in Figure 2. In the three different
Table I. Correlation coefcients for relative humidity and wind speed for thermal dissipation probe measurements at 5, 8, 15 and 20 mm depth
Relative humidity Wind speed
Depth 5 mm 8 mm 15 mm 20 mm 5 mm 8 mm 15 mm 20 mm
5 mm 1 0.559 0.810 0.753 1 0.775 0.632 0.706
8 mm 0.559 1 0.841 0.770 0.775 1 0.722 0.828
15 mm 0.810 0.841 1 0.821 0.632 0.722 1 0.840
20 mm 0.753 0.770 0.821 1 0.706 0.828 0.840 1
Table II. Lower and upper bound for the signicant differences
between the four depths (5, 8, 15 and 20 mm) for the
meteorological variables (Tukey, 95% signicance)
Lower
bound
Upper
bound
Temperature 8 mm 5 mm (
C) 0.694 5.75
Temperature 8 mm 20 mm (
C) 0.598 5.65
Relative humidity 15 mm 20 mm (%) 31.2 14.2
Relative humidity 8 mm 20 mm (%) 33.5 16.5
Relative humidity 5 mm 20 mm (%) 29.1 12.1
Wind speed 15 mm 5 mm (m s
-1
) 1.22 0.238
Wind speed 8 mm 5 mm (m s
-1
) 1.08 0.099
648 M. W. VANDEGEHUCHTE ET AL.
Copyright 2012 John Wiley & Sons, Ltd. Irrig. and Drain. 61: 645656 (2012)
Figure 1. Net radiation calculated according to the standard FAO method and the Idso, Brunt and Swinbank equations (a). Corresponding reference evapotrans-
piration using the net radiation calculated according to the standard FAO method and the Idso, Brunt and Swinbank equations (b)
Figure 2. Average net radiation (a) and vapour pressure decit (b) for the measurement days
Figure 3. Leaf water potentials for the different azimuthal orientations on DOY (day of the year) 225 for irrigation dose of 100% ETc (a), 33% ETc (b) and local
experience of the grower (c)
649 WATER USE MESKI OLIVE
Copyright 2012 John Wiley & Sons, Ltd. Irrig. and Drain. 61: 645656 (2012)
irrigation plots, a typical water potential pattern was found,
with values declining during the morning to reach a
minimum around midday and riseing again in the afternoon
(Figure 3). Standard deviations were relatively small
(average 0.32 MPa for leaf water potential and 0.18
MPa for stem water potential). This resulted from small
errors in the actual measurements. Besides spatial differ-
ences between the four orientations, temporal differences
occurred where relative ratios of water potentials for the four
orientations changed with time (data not shown).
Despite the use of different irrigation doses, average leaf
and stem water potential (average of four azimuthal directions
and three trees per plot) showed no signicant differences
during the measurement period (Figure 4). To conrm this,
the pre-dawn leaf water potential and leaf water potential at
13 h were statistically compared for the different irrigation
plots. This test did not yield signicant differences between
the plots (p >0.05). For the pre-dawn leaf water potential,
an overall average value of 0.5 0.11 MPa was obtained.
Larger variations were found for stomatal resistance, but
no signicant difference in average values existed between
the three irrigation plots for the measured times (p >0.05)
(Figure 5). Stomatal resistance declined strongly during
the morning to reach a minimum between 9.30 h and 10 h,
then rose to a maximum between 12 h and 14 h, subse-
quently declining again towards a second minimum in the
afternoon and nally rising again slowly in the evening.
Sap ux density and radial prole
Figure 6 shows a typical sap ux density pattern obtained by
applying the Do and Rocheteau (2002) equation. From the
temperature measurements at the end of the cooling period,
the natural thermal gradient in the wood was assessed (con-
sidering a distance of 4 cm between heater and measurement
needle) (Figure 7). This gradient was largely positive with
only short negative periods around midday.
Figure 8(a) shows sap ux density measurements for the
four azimuthal orientations for a measurement tree of the
33% ETc plot. When compared to Figures 8(b) and (c)
Figure 4. Daily pattern of average leaf water potentials (thin lines) and stem
water potentials (thick lines) for the three irrigation doses for the different
measurement days. 100% ETc: solid lines, 33% ETc: dashed lines, local
experience grower: dotted lines
Figure 5. Daily pattern of average stomatal resistances for the different measurement days for irrigation dose of 100% ETc (a), 33% ETc (b) and local
experience of the grower (c)
650 M. W. VANDEGEHUCHTE ET AL.
Copyright 2012 John Wiley & Sons, Ltd. Irrig. and Drain. 61: 645656 (2012)
showing the leaf water potential and stomatal resistance,
respectively, it is clear that no conclusive pattern for the
azimuthal variation can be detected. For the other trees,
different ratios for the sap ux densities of the azimuthal
directions were obtained. There was no statistical correlation
between differences in azimuthal sap ux density and
meteorological conditions, ecophysiological variables or
the irrigation plots.
The radial sap ux prole is summarized in Figure 9. A
similar temporal and spatial pattern was observed for the
other trees, namely a clear daily pattern of rising sap ow
from sunset till midday, falling again towards the evening,
and a radial pattern rising from the bark to the centre of
the tree till about 1214 mm depth and then falling further
towards the centre. To gain better insight into the azimuthal
inuence on the obtained radial prole, box plots were made
using the sap ux density data of the different directions and
depths (Figure 10). In the south, measurements at 20 mm
depth were relatively higher compared to the other azimuthal
directions for this measured tree. This was, however, not
always the case for the other trees measured. In general, for
all trees, it was clear that the radial sap ux density patterns
differed between azimuthal directions.
Water use based on sap ux density
Based on the sap ux density measurements at 15 mm depth,
the water use of the three irrigation plots was calculated (as an
average of the azimuthal directions and the three repetitions
per irrigation plot) and compared with the PenmanMonteith
crop evapotranspiration obtained (Figure 11a). There seems
to be good correspondence between the calculated crop
evapotranspiration and the water use based on TDP measure-
ments. Moreover, Figure 11(a) conrms the hypothesis that
no signicant drought stress differences occurred between
the three irrigation plots as there was no signicant difference
in water use. However, when the radial prole is taken into
account, the calculated crop evapotranspiration clearly overes-
timates the actual water use (Figure 11b).
DISCUSSION
More water, less stress?
Given the large differences in irrigation doses, a difference
in water potential, as a direct drought stress indicator, was
expected. When comparing leaf and stem water potential
for the three irrigation plots (Figure 3), no signicant
differences in drought stress could be detected. Moreover,
the pre-dawn leaf water potential, which is considered a
good approximation of soil water potential, showed no
signicant difference between the plots and had an overall
average value of 0.5 0.11 MPa which is considered the
upper limit for good soil water status (Michelakis et al.,
1994; Fernandez and Moreno, 1999; Tognetti et al., 2004;
Diaz-Espejo et al., 2007). This indicates that no large differ-
ences in drought stress at soil level occurred. Moreover, no
signicant difference in stomatal resistance was noted
between the irrigation plots (Figure 4). Although no signi-
cant differences in water potentials were observed between
the irrigation plots, stomatal resistance showed a maximum
shortly after midday for all plots (so-called midday closure)
(Figure 4). This can be explained by the high vapour
pressure decit during this period of the day (up to 4 kPa)
which has also been noted by Trilo et al. (2007) who
mention 45% loss in hydraulic conductivity at an applied
pressure of 3.5 MPa for twigs of Olea europaea Leccino.
From these data, it seems the lowest irrigation dose
(25 l day
-1
per tree) was sufcient for irrigation purposes,
while the irrigation dose of 75 l day
-1
per tree based on
Figure 6. Typical sap ux density pattern as obtained by the Do and
Rocheteau equation (2002)
Figure 7. Natural thermal gradient measured at the end of the cooling period
of the discontinuous heating cycle (southern sensor) from DOY 228 to 234
for one of the measured trees. Similar results were obtained for the other
trees although the gradient varied slightly between azimuthal directions
651 WATER USE MESKI OLIVE
Copyright 2012 John Wiley & Sons, Ltd. Irrig. and Drain. 61: 645656 (2012)
PenmanMonteith evapotranspiration calculations led to
water overspill.
Sap ux density
Inuence of thermal gradient: pitfall for cyclic heat-
ing. The use of crop coefcients to determine crop evapo-
transpiration from reference evapotranspiration is susceptible
to errors, because the actual physiology of the plant is not fully
taken into account (Fernandez et al., 1997; Tognetti et al.,
2004). Factors such as midday stomatal closure, pruning, crop
height, tree- to-tree distance, etc. can inuence the crop
coefcient, making it difcult to apply when no validation
measurements (for instance based on eddy covariance) have
been conducted (Teixeira et al., 2008). In order to include
the actual plantwater relations of the studied species and to
take the individual variability of trees into account, increasing
attention is currently being paid to irrigation methods that use
specic plant-based variables, including sap ow, detecting
drought stress directly at the plant level (Jones, 2004).
However, when using sap ow as a variable, attention should
be paid to natural thermal gradients. In this study, the natural
thermal gradient was mainly positive (Figure 7). This is in
contrast with the original study of Do and Rocheteau (2002).
When analysing the temperature differences between the
sensor needles at the end of the cooling period, Do and
Rocheteau (2002) found mainly positive natural gradients
(up to 1.5
C) during the night, while negative natural gradi-
ents occurred during the day (down to 2
C). This led to
an overestimation of the reference temperature difference
(for zero ow conditions) and, hence, an overestimation of
the actual sap ux density when applying the Granier formula,
Figure 8. Sap ux density (a), water potential (b) and stomatal resistance (c) for the different azimuthal directions for the irrigation dose of 33% ETc
Figure 9. Radial sap ux density prole for a tree for the irrigation dose of 33% ETc. This gure is available in colour online at wileyonlinelibrary.com/journal/ird
652 M. W. VANDEGEHUCHTE ET AL.
Copyright 2012 John Wiley & Sons, Ltd. Irrig. and Drain. 61: 645656 (2012)
because during the day the reference needle was inuenced by
the negative natural gradient. In this study, application of the
Granier method would have led to underestimation of the
sap ux density, probably due to the heat capacity of the soil.
Although the formula developed by Do and Rocheteau
(2002) has proven to be very useful when taking natural
gradients into account, it should be applied with caution.
Indeed, when the K
a
value in Equation (5) is larger than 1,
imaginary results are obtained. A K
a
value larger than 1
might occur when T
a0
>2.T
au
. This is possible for high
T
a0
or low T
au
values. Since T
a0
values are based on
zero sap ow conditions, this value always reaches a
maximum dependent on the thermal conductivity of the
tissue in which the sensor is installed and on the natural
thermal gradients that occur. For a tree, this value is more
or less constant. The reason why K
a
values can be larger
than 1 is thus mainly due to low T
au
values with
T
au
T
on;u
T
off;u
(7)
with T
on,u
and T
off,u
the temperature difference between
the needles at the end of the heating period and at the end of
the cooling period, respectively. Low values are thus caused
by either a high sap ux density (which results in a low
Figure 10. Sap ux density at the different measurement depths and for the different azimuthal orientations for one tree: (a) north, (b) east, (c) south, (d) west
Figure 11. Water use of the different irrigation plots compared to the calculated PenmanMonteith crop evapotranspiration: (a) based on the TDP measurements
at 15 mm, (b) based on the radial prole. For both (a) and (b) an outer annulus of 36 mm was considered as conductive sapwood
653 WATER USE MESKI OLIVE
Copyright 2012 John Wiley & Sons, Ltd. Irrig. and Drain. 61: 645656 (2012)
temperature difference when the needles are heated) or a
high natural thermal gradient (which means a high tempera-
ture difference between the needles when they are not
heated) or by a combination of these factors. In this study,
imaginary results were obtained for four measurement days
and mostly at midday. These data were not used in the rest
of the analysis. In total, about 4% of the data were not used
because of this problem. However, even if the results are not
imaginary, their accuracy can be questioned. Because of
these limitations, it is advisable to repeat the calibration
procedure as described by Do and Rocheteau (2002) to see
whether the used species meets the empirical relationship.
It is clear that correct application of the sensors and use
of the corresponding equations is necessary to obtain
accurate results.
Scaling up sap ow measurements: azimuthal varia-
tion. When using TDP as a basis for irrigation, it is essential
that the measurements of the probe are representative for the
whole stem. Measurements were therefore conducted in the
four azimuthal directions. It was expected that sap ux density
would be higher for the southern sensors compared to the
northern ones due to the longer and direct exposure to radia-
tion. Because the northern tree side receives no direct radia-
tion (at least in the northern hemisphere) and to minimize
thermal effects on the sensors, it has been advised to install
TDP probes in this azimuthal direction (Do and Rocheteau,
2002; Steppe et al., 2002). We found that the azimuthal
pattern for sap ux density did not match the pattern for leaf
water potential or stomatal resistance (Figure 8 as an example
of one tree). This indicates that the variation between the
azimuthal directions is probably not due to meteorological
factors. The water potential measurements conrmed this
(Figure 3). One would expect that leaf water potentials decline
more strongly during the morning and rise more quickly
during the evening in the east (reversed reasoning for the
west), but the data did not show such a pattern. Fernandez
et al. (2001) argued that spatial differences were rather due
to different orientations of the primary branches and, coupled
with this, differences in leaf area index of the olive tree and
not to differences in radiation interception. These authors
suggested that measurements should therefore be performed
at primary branch level instead of at stem level. In our study
this was not the case, as no equal ratio between the directions
could be found despite the similar structure of crown
and branches.
Scaling up sap ow measurements: radial prole.
Besides azimuthal variation, studies have indicated radial
variation in the sap ux density (Granier et al., 1994; Lu
et al., 2000; Nadezhdina et al., 2002; Saveyn et al., 2008;
Steppe et al. 2010). This radial (and temporal) variability also
existed in Meski olive trees (Figure 9): sap ux density was
lower at smaller depths, then rose towards a maximum and
then slowly descended. This radial prole was, however,
not constant for the different azimuthal orientations
(Figure 10). Azimuthal differences in the radial prole can
be attributed to the coupling between the xylem and the
crown of the tree (Tyree and Zimmerman, 2002). A locally
smaller leaf area can lead to a lower sap ux density. In
addition, it seems plausible that these differences are
inuenced by the structure of the wood as this is linked with
the development and structure of the tree. To conrm this, a
coloured section of the stem was investigated. Through
colouring of the wood, it became clear that the conductive
tissue was not uniformly distributed in stems of olive trees
and that, when different sensors are placed in the stem, the
thermocouples, although placed at the same depth, were
located in different concentric stem sections of the conductive
tissue. This might largely explain the observed azimuthal
variation in the radial prole.
Because azimuthal and radial variability appeared to
depend largely on the structure of the sapwood and not
directly on meteorological variables, it was not feasible to
develop correction factors to scale from one-point sap ux
density measurements to the whole tree. To obtain an accu-
rate estimation of the total sap ow, multiple measurements
at different depths and azimuthal directions are, hence,
necessary. Multiple point sensors such as the heat eld
deformation sensor (Nadezhdina et al., 2012) could play a
major role in the upscaling.
Scaling up sap ow measurements: which surface to
use? To determine the total sap ow or whole-tree water
use from sap ux density measurements, the sap conductive
surface area has to be taken into account. The coloured stem
section indicated that using the total surface area to scale up
from a single-point measurement to tree level led to an over-
estimation since the inner part of the wood was not coloured
and, thus, did not conduct sap. The olive tree sapwood had
an average depth of 36 mm, corresponding to 74% of the
total cut surface area. This corresponded closely with the
sapwood depth of 40 mm reported earlier for olive trees
by Nadezhdina et al. (2007). It is thus not only important
to take into account azimuthal and radial variation, but also
the shape and depth of the conductive tissue. This becomes
clear when comparing whole-tree water use calculated with
and without taking radial variation into account. When using
only the azimuthal single-point measurements (no radial
prole) and considering the whole stem cross section as con-
ductive (the surface of the section was divided in four parts
according to the four azimuthal directions), an average daily
sap ow of 113 l (2.3 mm day
-1
or 5.4 l m
leaf
-2
day
-1
) was
estimated (average of all plots as there was no signicant
654 M. W. VANDEGEHUCHTE ET AL.
Copyright 2012 John Wiley & Sons, Ltd. Irrig. and Drain. 61: 645656 (2012)
difference between them). When only using a disc section
(considering the 36 outer mm of the wood as conductive), this
amount reduced to 68 l (1.39 mm day
-1
or 3.2 l m
leaf
-2
day
-1
).
If radial variability was taken into account, the total daily sap-
ow decreased to only 34 l (0.70 mm day
-1
or 1.6 m
leaf
-2
day
-1
). If then not a perfect disc section but the actual conduc-
tive surface was used as for the cut tree, this amount was
reduced even more to 23 l (0.47 mm day
-1
or 1.1 l m
leaf
-2
day
-1
). These last values correspond well with the irrigation
dose of 33% ETc, conrming that this dose was sufcient
for this specic olive orchard. Hence, sap ux density mea-
surements that take radial and azimuthal variation into account
can lead to more accurate assessments of actual water use. It is
thus recommended to use sap ow data as validation for irri-
gation methods based on PenmanMonteith crop transpiration
calculations. Moreover, this combination can lead to a better
estimation of water use for the local grower, leading to a
decrease in water use and hence an economic advantage.
CONCLUSIONS
In this study an evaluation of the whole-tree water use of
olive trees and of frequently used irrigation methods is
presented. Based on water potential and stomatal resistance,
no difference in tree drought stress level for the three irriga-
tion doses (25, 45 and 75 l day
-1
per tree) was found, which
was a rst indication that the doses based on Penman
Monteith and the experience of the local grower led to
wasting of water. Sap ow measurements have great poten-
tial in irrigation research as they take into account the
ecophysiological response of the tree and are not primarily
based on meteorological factors. However, natural thermal
gradients played an important role when using them in Olea
europaea L. Meski, which made application of the
conventional sap ow formulas difcult. Moreover, radial
and azimuthal variability were mainly dependent on the
structure of the sapwood and not closely linked to meteoro-
logical variables such as radiation, which made scaling up a
difcult task for olive trees. Further research that could
remove the current pitfalls in sap ux density equations
could lead to even more accurate water use estimations.
ACKNOWLEDGEMENTS
The authors wish to thank the Research Foundation
Flanders (FWO) for the PhD funding granted to the
rst author and the Flemish Interuniversity Council (VLIR)
for supporting this study as part of a larger Own Initiatives-
project (ZEIN2006PR326). We are also indebted to Fethi
Ben Mariem, Olfa Boussadia and the other Tunisian
researchers and technical staff of the Olive Institute
(Sousse) and Kim Callewaert for her help with the water
potential measurements.
REFERENCES
Allen R, Pereira L, Raes D, Smith M. 1998. Crop evapotranspiration.
Guidelines for computing crop water requirements. Irrigation and Drainage
Paper 56, FAO, Rome; 229.
Allen SJ, Grime VL. 1995. Measurements of transpiration from savanna
shrubs using sap ow gauges. Agricultural and Forest Meteorology
75(13): 2341.
Brunt D. 1932. Notes on radiation in the atmosphere. Quarterly Journal of
the Royal Meteorological Society 58(247): 289420.
Buckley TN. 2005. The control of stomata by water balance. New Phytologist
168(2): 275291. doi:10.1111/j.1469-8137.2005.01543.x
dAndria R, Lavini A, Morelli G, Patumi M, Terenziani S, Calandrelli D,
Fragnito F. 2004. Effects of water regimes on ve pickling and double
aptitude olive cultivars (Olea europaea L.). Journal of Horticultural
Science and Biotechnology 79(1): 1825.
De Pauw DJW, Steppe K, De Baets B. 2008. Identiability analysis and
improvement of a tree water ow and storage model. Mathematical
Biosciences 211(2): 314332. doi:10.1016/j.mbs.2007.08.007
De Swaef T, Steppe K, Lemeur R. 2009. Determining reference values for
stem water potential and maximum daily trunk shrinkage in young apple
trees based on plant responses to water decit. Agricultural Water
Management 96(4): 541550. doi:10.1016/j.agwat.2008.09.013
Diaz-Espejo A, Nicolas E, Fernandez JE. 2007. Seasonal evolution of
diffusional limitations and photosynthetic capacity in olive under
drought. Plant Cell and Environment 30(8): 922933. doi:10.1111/
j.1365-3040.2007.001686.x
Do F, Rocheteau A. 2002. Inuence of natural temperature gradients on
measurements of xylem sap ow with thermal dissipation probes. 1. Field
observations and possible remedies. Tree Physiology 22(9): 641648.
Er-Raki S, Chehbouni A, Hoedjes J, Ezzahar J, Duchemin B, Jacob F.
2008. Improvement of FAO-56 method for olive orchards through
sequential assimilation of thermal infrared-based estimates of ET.
Agricultural Water Management 95(3): 309321. doi:10.1016/j.
agwat.2007.10.013
Fereres E, Evans RG. 2006. Irrigation of fruit trees and vines: an introduction.
Irrigation Science 24(2): 5557. doi:10.1007/s00271-005-0019-3
Fernandez JE, Moreno F. 1999. Water use by the olive tree. Journal of
Crop Production 2: 101162.
Fernandez JE, Moreno F, Giron IF, Blazquez OM. 1997. Stomatal control
of water use in olive tree leaves. Plant and Soil 190(2): 179192.
Fernandez JE, Palomo MJ, Diaz-Espejo A, Clothier BE, Green SR, Giron
IF, Moreno F. 2001. Heat-pulse measurements of sap ow in olives for
automating irrigation: tests, root ow and diagnostics of water stress.
Agricultural Water Management 51(2): 99123.
Fisher RA, Williams M, Do Vale RL, Da Costa AL, Meir P. 2006.
Evidence from Amazonian forests is consistent with isohydric control
of leaf water potential. Plant Cell and Environment 29(2): 151165.
Franks PJ, Drake PL, Froend RH. 2007. Anisohydric but isohydrodynamic:
seasonally constant plant water potential gradient explained by a stomatal
control mechanism incorporating variable plant hydraulic conductance.
Plant Cell and Environment 30(1): 1930. doi:10.1111/j.1365-
3040.01600.x
Fulton A, Buchner R, Olson B, Schwankl L, Gilles C, Bertagna N, Walton
J, Shackel K. 2001. Rapid equilibration of leaf and stem water potential
under eld conditions in almonds, walnuts, and prunes. HortTechnology
11(4): 609615.
Gimenez C, Fereres E, Ruz C, Orgaz F. 1997. Water relations and gas
exchange of olive trees: diurnal and seasonal patterns of leaf water poten-
tial, photosynthesis and stomatal conductance. Acta Horticulturae (ISHS)
449: 411416.
Granier A, Anfodillo T, Sabatti M, Cochard H, Dreyer E, Tomasi M,
Valentini R, Breda N. 1994. Axial and radial water-ow in the trunks
655 WATER USE MESKI OLIVE
Copyright 2012 John Wiley & Sons, Ltd. Irrig. and Drain. 61: 645656 (2012)
of oak trees a quantitative and qualitative analysis. Tree Physiology
14(12): 13831396.
Idso SB. 1981. A set of equations for full spectrum and 8-mu-m to 14-mu-m
and 10.5 mu-m to 12.5 mu-m thermal-radiation from cloudless skies.
Water Resources Research 17(2): 295304.
Jones HG. 2004. Irrigation scheduling: advantages and pitfalls of plant-
based methods. Journal of Experimental Botany 55(407): 24272436.
doi:10.1093/jxb/erh213
Lu P, Muller WJ, Chacko EK. 2000. Spatial variations in xylem sap ux
density in the trunk of orchard-grown, mature mango trees under changing
soil water conditions. Tree Physiology 20(10): 68369.
Lu P, Urban L, Zhao P. 2004. Graniers thermal dissipation probe (TDP)
method for measuring sap ow in trees: theory and practice. Acta Botanica
Sinica 46(6): 631646.
Michelakis NIC, Vouyoucalou E, Clapaki G. 1994. Soil moisture depletion,
evapotranspiration and crop coefcients for olive trees cv. Kalamon, for
different levels of soil water potential and methods of irrigation. Acta
Horticulturae 356: 162167.
Moreno F, Fernandez JE, Clothier BE, Green SR. 1996. Transpiration and
root water uptake by olive trees. Plant and Soil 184(1): 8596.
Moriana A, Fereres E. 2002. Plant indicators for scheduling irrigation of
young olive trees. Irrigation Science 21(2): 8390. doi:10.1007/
s00271-001-0053-8
Nadezhdina N. 1999. Sap ow index as an indicator of plant water status.
Tree Physiology 19: 885891.
Nadezhdina N, Cermak J, Ceulemans R. 2002. Radial patterns of sap ow
in woody stems of dominant and understory species: scaling errors
associated with positioning of sensors. Tree Physiology 22(13): 907918.
Nadezhdina N, Nadezhdin V, Ferreira MI, Pitacco A. 2007. Variability with
xylem depth in sap ow in trunks and branches of mature olive trees.
Tree Physiology 27(1): 105113.
Nadezhdina N, Vandegehuchte M, Steppe K. 2012. Sap ux density mea-
surements based on the heat eld deformation method. Trees Structure
and Function: 110. doi:10.1007/s00468-012-0718-3
Saveyn A, Steppe K, Lemeur R. 2007. Drought and the diurnal patterns of
stem CO2 efux and xylem CO2 concentration in young oak (Quercus
robur). Tree Physiology 27(3): 365374.
Saveyn A, Steppe K, Lemeur R. 2008. Spatial variability of xylem sap
ow in mature beech (Fagus sylvatica) and its diurnal dynamics in
relation to microclimate. Botany 86(12): 14401448. doi:10.1139/
b08-112
Sepulcre-Canto G, Zarco-Tejada PJ, Jimenez-Munoz JC, Sobrino JA, de
Miguel E, Villalobos FJ. 2006. Detection of water stress in an olive
orchard with thermal remote sensing imagery. Agricultural and Forest
Meteorology 136(12): 3144. doi:10.1016/j.agrformet.2006.01.008
Sevanto S, Nikinmaa E, Riikonen A, Daley M, Pettijohn JC, Mikkelsen TN,
Phillips N, Holbrook NM. 2008. Linking xylem diameter variations with
sap ow measurements. Plant and Soil 305(12): 7790. doi:10.1007/
s11104-008-9566-8
Steppe K, Lemeur R, Samson R. 2002. Sap ow dynamics of a beech tree
during the solar eclipse of 11 August 1999. Agricultural and Forest
Meteorology 112(34): 139149.
Steppe K, De Pauw DJW, Lemeur R. 2008. A step towards new irrigation
scheduling strategies using plant-based measurements and mathematical
modelling. Irrigation Science 26(6): 505517. doi:10.1007/s00271-008-
0111-6
Steppe K, De PauwDJW, Doody TM, Teskey RO. 2010. Acomparison of sap
ux density using thermal dissipation, heat pulse velocity and heat eld de-
formation methods. Agricultural and Forest Meteorology 150: 10461056.
Swinbank WC. 1963. Long-wave radiation from clear skies. Quarterly
Journal of the Royal Meteorological Society 89(381): 339348.
Teixeira AHD, Bastiaanssen WGM, Moura MSB, Soares JM, Ahmad MD,
Bos MG. 2008. Energy and water balance measurements for water
productivity analysis in irrigated mango trees, Northeast Brazil.
Agricultural and Forest Meteorology 148(10): 15241537.
doi:10.1016/j.agrformet.2008.05.004
Tognetti R, dAndria R, Morelli G, Calandrelli D, Fragnito F. 2004. Irriga-
tion effects on daily and seasonal variations of trunk sap ow and leaf
water relations in olive trees. Plant and Soil 263(12): 249264.
Trilo P, Lo Gullo MA, Nardini A, Pernice F, Salleo S. 2007. Rootstock
effects on xylem conduit dimensions and vulnerability to cavitation of
Olea europaea L. Trees Structure and Function 21(5): 549556.
doi:10.1007/s00468-007-0148-9
Tyree MT, Zimmerman MH. 2002. Xylem Structure and the Ascent of Sap.
Springer Series in Wood Science: New York.
Villez K, Steppe K, De Pauw DJW. 2009. Use of Unfold PCA for on-line
plant stress monitoring and sensor failure detection. Biosystems
Engineering 103(1): 2334. doi:10.1016/j.biosystemseng.2009.01.014
656 M. W. VANDEGEHUCHTE ET AL.
Copyright 2012 John Wiley & Sons, Ltd. Irrig. and Drain. 61: 645656 (2012)