Effects of regulated deficit irrigation under subsurface drip irrigationconditions on vegetative development and yield of mature almond trees
Pascual Romero1, Pablo Botia1,2 & Francisco Garcia1
1Instituto Murciano de Investigacion y Desarrollo Agrario y Alimentario. Estacion Sericıcola 30150. La Alberca,Murcia, Spain. 2Corresponding author∗
Received 6 August 2003. Accepted 31 October 2003
Key words: Prunus dulcis, regulated deficit irrigation, root distribution, subsurface drip irrigation, vegetativedevelopment, water use efficiency, yield
Abstract
The influence of several regulated deficit irrigation (RDI) strategies, applied under subsurface drip irrigation(SDI), on vegetative development and yield parameters in mature almond (Prunus dulcis (Mill.) D.A. Webb, cv.Cartagenera) trees was analysed during a 4-year field experiment. Five treatments were applied: T1 (100% cropevapotranspiration (ETc), full season); T2 (irrigated at 100% ETc except in the kernel-filling stage (20% ETc));T3 (equal to T2 but in SDI); T4 (SDI, 100% ETc, except in the kernel-filling stage (20% ETc) and post-harvest(75% ETc)); T5 (SDI, 100% ETc except in the kernel-filling stage (20% ETc) and post-harvest (50% ETc). A closecorrelation between applied water, plant water status (�pd) and tree growth parameters was observed. After fouryears, the vegetative development in T5 was reduced significantly due to a larger annual cumulative effect of waterstress on growth processes, resulting in a smaller tree size (trunk and branch growth, canopy volume and pruningweight) compared to other treatments. Moreover, water stress during kernel-filling produced a significant reductionin the leaf expansion rate and a stimulation of premature leaf abscission, resulting in a smaller tree leaf area in thistreatment. SDI produced a greater horizontal distribution of fine roots in the soil profile than surface drip system.The RDI practices applied under subsurface drip irrigation stimulated a deeper root development (40–80 cm) thansurface treatments (0–40 cm), producing also a higher root density in the subsurface treatments watered the least(T4 and T5). Water stress during the pre- and post-harvest periods had no important effect on bud development,bloom, fruit growth or fruit abscission. Moreover, there were no significant reductions in kernel dry weight orkernel percentage. Reductions in kernel yield were significant compared to T1, being of 11% in T2, 15% in T3,20% in T4 and 22% in T5. Water use efficiency (kg m−3) was increased significantly in the SDI treatments T4 andT5. A significant correlation between kernel yield and tree growth parameters was observed. We conclude that theapplication of overall reductions of water use of up to 50% during high water stress sensitivity periods (such aspost-harvest) under SDI system, is a promising alternative for water management in semiarid regions in order toimprove water use efficiency. Nevertheless prolonged water stress during kernel-filling and post-harvest can reduceexcessively the vegetative development of almond, negatively affecting the long-term yield response.
Introduction
Spain is the Mediterranean country having thegreatest production of almonds (Prunus dulcis (Mill.)D.A. Webb) and is ranked second in the world, ac-counting for 17% of world production, being the main
competitor of the USA (Alston et al., 1993). Althoughthe area of land dedicated to almond is greater in Spainthan in the USA, 629100 ha (AEA, 1999) as opposedto 165000 ha, the kernel yield is ten times higher inthe USA than in Spain, with the USA having meanvalues of between 1200 and 1800 kg ha−1, as op-posed to 150 kg ha−1 in Spain (Tous Martí, 1995).In Spain, the traditional almond plantations are gen-
170
erally located in marginal areas, on poorly drainedsoils and with no irrigation supply. This fact has res-ulted in plantations of low profitability, that have notmade use of new irrigation technologies and crop prac-tices. To improve this situation in semiarid regions ofSpain, studies are being carried out to increase theperformance of these orchards, applying new prac-tices (for example, regulated deficit irrigation, RDI)and irrigation technologies (for example, subsurfacedrip irrigation). In these areas, with limited water re-sources, high temperature and low relative humidity,RDI can be a promising alternative (Salazar and Mel-garejo, 2002) and has stimulated research to determinethe sensitivity of almond production to specific peri-ods and levels of water stress. Some studies havedetermined that almond is a drought-tolerant crop,indicating that almond yield is relatively insensitiveto mild or moderate water stress during the kernel-filling stage (dry weight accumulation in the kernel)(Girona and Marsal, 1995; Goldhamer and Shackel,1990; Goldhamer, 1996). However, the same studiesindicate that it is necessary to avoid water stress dur-ing active vegetative and fruit growth periods. Almondproductivity may be also vulnerable to water stressduring the post-harvest period because reproductivebud differentiation occurs late (early August–earlySeptember) (Goldhamer and Smith, 1995; Goldhamer,1996; Goldhamer and Viveros, 2000). It is import-ant also to consider long-term effects of water stressfor spur-bearing species such as almond, due to re-duced vegetative development and renewal of fruitingpositions (Prichard et al., 1992; Esparza et al., 2001).
The use of subsurface drip irrigation (SDI) sys-tems, applying water below the soil surface directlyto the root zone and minimising soil evaporation, hasbeen used also to save water (Camp, 1998; Lamm,1995; Phene, 1999) and improve water use efficiencyin semiarid regions (El Gindy and El Araby, 1996;Phene, 1993). In almond, the use of SDI has pro-duced a higher water application efficiency, for bothsoil and plants (Botía et al., 1998, 2000; Del Amorand Cerdá, 1997), and a better yield response thansurface drip irrigation, achieving greater kernel yieldand water use efficiency (Del Amor and Del Amor,1999). However in different soil conditions, Schwanklet al. (1999) have proposed that the utilisation of otherirrigation systems (surface drip or micro-sprinkler) isadvantageous for growth and yield. Until now, therehave been no studies in almond about the applicationof RDI under subsurface systems. Recently, Romeroet al. (2004), combining both techniques (SDI and
RDI), reported that SDI promoted rapid recovery ofsoil water status after severe water stress, maintainingadequate levels of soil water content and plant wa-ter status post-harvest and producing a higher waterapplication efficiency than surface systems.
The aim of this work was to determine the ef-fects of the application of several RDI strategies withburied drip irrigation on vegetative development andyield of mature almond trees in a four-year study. Spe-cifically, we examined the long-term effects of waterstress during the pre-harvest and post-harvest periodson leaf development, leaf abscission, shoot growth,root growth, yield determinants and the interrelation-ships between them and plant water status. Moreover,we evaluated the long–term yield response and the wa-ter use efficiency of the almond orchard under theseirrigation conditions.
Material and methods
Plant material, treatments and experimentalconditions
The experimental design has been described inRomero et al. (2004). Briefly, an irrigation experimentwas carried out from 1997 to 2000, in a commercialorchard of 13-year-old almond trees (cv. Cartagenera,grafted on almond rootstock), on drip-line irrigation.The plantation is situated in the district of Aljorra,Murcia (Southern Spain). The orchard comprised rowsof cv. Cartagenera (70%) planted alternately withpolliniser rows of cv. Ramillete (30%). The weatheris Mediterranean semi-arid, with mean annual precip-itation at the site being 280 mm, confined to autumnand spring.
The trial involved five irrigation treatments thatwere applied during four consecutive years, using twoirrigation systems, surface and subsurface drip, andthree different regulated deficit irrigation strategies:T1, 100% crop evapotranspiration (ETc) full season;T2 irrigated at 100% ETc except in the kernel-fillingstage (20% ETc); T3 equal to T2 but in SDI; T4,SDI, 100% ETc, except in the kernel-filling stage(20% ETc) and post-harvest (75% ETc)); T5, SDI,100% ETc except in the kernel-filling stage (20%ETc) and post-harvest (50% ETc). The lay-out of theexperiment has been described in detail in Romeroet al. (2004). Control (T1) was irrigated at 100%crop evapotranspiration (ETc) (ETo calculated via Panevaporation Class-A method, U.S. Weather Bureau)
over the entire crop season. Irrigation was applieddaily in short pulses, once or twice a day (high fre-quency irrigation) and was controlled and adjustedweekly according to soil matric potential and dailyclimatic data from a weather station in the vicinity(1 km) of the experiment. In both irrigation systems,a drip line was utilised with four self-compensatingdrips (type RAM, 3.5 L h−1) per tree, 1 m apart.In the subsurface drip irrigation system, the drip linewas buried at 35 cm depth. A root growth-inhibitingchemical, trifluralin, (dinitro-N, N-dipropyl-4 trifluoromethylanidine) was used in the filtration system foravoiding root intrusion in the buried drips. Soil char-acteristics and climatic parameters (annual rainfalland reference evapotranspiration (ETo)) at the experi-mental site, annual applied water for each treatment,Kc (crop coefficients) and the fertilization programapplied have been described in Romero et al. (2004).
Water stress measurements
Leaf water potential at dawn (�pd) was measured be-fore actual sunrise by using a pressure chamber (model3000; Soil Moisture Equipment. Corp., Santa Barbara,California, USA), according to the Scholander et al.(1965) and Turner (1988) technique. Two mature, butnot aged, leaves from the middle third of the tree weretaken in twelve trees per treatment (six measurementsper treatment and plot).
Vegetative development measurements
Trunk circumference was measured monthly fortwelve trees per treatment during the experimentalperiod. Six measurements per tree were taken at50 cm above the soil surface (20 cm above the graftunion). Transverse cross-sectional area was calculatedaccording to Mitchell and Chalmers (1982).
Tree canopy height and diameter (across andwithin rows (N-S, E-W)) were measured monthlyfor twelve trees per treatment. Canopy volumewas calculated according to Hutchinson (1978):Volume = (width2× height)/2.
Pruning weight was determined annually (end ofNovember) in the same trees, independently for eachtree.
During 1997 and 1998, terminal growth of fouryoung branches per tree, one from each compassdirection, was measured on four trees per treatment(16 branches per treatment) every week (from Marchto November). The length and diameter were meas-
ured with a digital calliper (Mitutoyo MTI, Corpora-tion, City of Industry, CA).
Leaf abscission was monitored in four trees pertreatment, starting with the dry weight of fallen leaves(65 ◦C, 24 h) collected periodically over the entiregrowth season in sieve boxes (mesh grade = 1 mm),quadrangular in shape (side = 2 m, height = 1 m),which occupied approximately one quarter of the totalarea spanned by the tree. Abscission over the stressperiod was expressed as a percentage of the total dryweight of accumulated fallen leaves during the year.
In 1998 and 1999, leaf area and the number ofcanopy leaves per cubic metre, at the point of greateststress, were determined via the sampling of a knownvolume of leaves (Boland et al., 1993), using cylin-ders of 20-cm diameter and 15-cm height, at variouspoints throughout the entire three dimensions of thecanopy and at different angles, at the rate of six pertree, with one tree per treatment per plot. Leaf area pertree (m2m−3) was calculated taking into account thecanopy volume.
From May to July 1999 (a representative year), leafgrowth rate (length and width) was measured weeklyin eight young (of recent appearance) leaves per treat-ment and per plot, in the middle third of the tree,at different angles. The measurements of maximalwidth and length were recorded with a digital calliper(Mitutoyo, MTI Corporation, City of Industry, CA).Relative growth rate (RGR) was calculated according
to the equation: RGR = ln M2 − ln M1
t2 − t1; M being the
value of the growth parameter measured and t the timebetween measurements.
Root distribution measurements
After three years of the experiment (1999), rootsamples were taken by soil cores (100 cm3) using anauger. The samples were taken close to the tree, per-pendicularly to an emitter at 25, 50 and 75 cm fromthe drip head, at 10-cm intervals to 100 cm depth, fora total of 4 trees per treatment (one per replicate). Theroots were washed and shaken in water and sodiumhexametaphosphate (30 g L−1) to separate the fineroots from the soil particles. Roots were filtered anddried in an oven at 80 ◦C for 24 h. The density of fineroots (diameter < 1 mm) was expressed in dry weightper soil volume (mg cm−3) (Bielorai, 1982).
In 1998 and 1999, in order to evaluate the effectof the different irrigation treatments on bud develop-ment, flowering and fruit set processes, the number ofbuds (vegetative and flower buds), flowers and fruitswere recorded on four marked branches, of 20-mmdiameter, per tree (16 branches per treatment). Thedensity of buds, flowers and fruits was expressed asnumber per metre of branch length (Johnson et al.,1992). Fruit set was calculated 75 days after full bloomas the percentage of flowers set and grown into nuts.
In 1997, 1998 and 1999, during the fruit growthphase (March–July), 40 fruits per treatment (10 perreplicate) were sampled weekly to determine diameter,length and fresh and dry weights of the entire fruit(shell + kernel) and kernel (seed).
Also in 1999 (from October 1998 to the harvest in1999), the abscission of buds, flowers and fruits wasperiodically monitored in four trees per treatment, inthe same sieve boxes (mesh grade = 1 mm) used forcollecting the leaves. The number and the fresh anddry weights of fallen buds, flowers and fruits weredetermined.
Yield parameters
Individual tree yield was measured annually in threetrees per treatment per replicate (twelve trees per treat-ment) during the experimental period 1997–2000. Theharvest of each tree was separated in two fractions,‘hull-tight’ (no evidence of hull split) and commer-cial (full hull split). Fresh and dry weights (in shell)were calculated to determine nut load. From the com-mercial fraction, a 1-kg nut sample was collected andthe kernels were separated from the shells and hulls todetermine kernel yield, single kernel weight and ker-nel percentage. Moreover the percentages of double,unfilled and faulty kernels (with visible kernel rotsymptoms) were calculated.
Regressions and statistical analysis
Relationships between parameters were fitted to linearregressions. A variance analysis (ANOVA) was usedin order to discern the main treatment effects.
Figure 1. (A), Trunk cross-sectional area growth response to irrig-ation treatments during the period 1997–2000. Each point is themean of twelve measurements per treatment (3 trees/plot). The ver-tical bars indicate the standard error of the mean. (B) Relationshipbetween annual trunk cross-sectional area growth rate and meanannual irrigation water for the period 1997–2000 (quadratic regres-sion, y = −10.33 + 0.056x − 0.0001x2, r = 0.98; P < 0.001,there was improvement in r2 for the quadratic compared the linearmodel).
Results and discussion
Shoot growth response
The growth rate in the trunk cross-sectional areashowed significant differences between treatments(Table 1), revealing different trends in the trunkgrowth, with a tendency towards greater values oftrunk cross-sectional area in the T1, T2 and T3 treescompared to T4 and T5 (Figure 1A). Treatments T2(surface) and T3 (subsurface), with the same RDIstrategy and the same amount of applied water, hadsimilar trunk growth rate (Table 1). Schwankl et al.(1999) recorded similar patterns of trunk growth inalmonds under surface and subsurface irrigation sys-
173
Table 1. Absolute growth rate (AGR) in the trunk cross-sectional area from 1997 to 2000. Mean values of branch diameter absolute andrelative growth rate (AGR and RGR, respectively) in 1998. Canopy volume and cumulative pruning weight for the experimental period(1997–2000)
n.s. not significant; ∗P < 0.05; ∗∗∗P < 0.001. For each column, mean values followed by distinct letters are significantly different;separation by Duncan’s multiple range test at the 95% confidence level.
tems in different soil conditions. A high correlation(r2 = 0.94) was observed between trunk cross-sectional area growth rate and the annual amount ofwater applied (Figure 1B). This relationship indicatesthat, in our experimental conditions, when the appliedwater was below 450 mm (threshold level) the annualtrunk cross-sectional area growth rate was reducedconsiderably.
Young branch growth, expressed as mean growthrate of branch diameter, showed significant differencesbetween RDI treatments (Table 1). During kernel-filling (in 1998), T4 and T5 had branch growth ratessignificantly lower compared to T2 and T3, but sim-ilar to T1 (Table 1). In the same way, Hutmacheret al. (1994) reported that branch growth rates declinedmore in less-watered almonds, although different re-sponses have been reported in almonds under waterstress (Girona et al., 1993). The highest rate of growthoccurred from March–May for all irrigation treatments(Figure 2A and B), as found also by Hutmacher et al.(1994). While the elongation growth period occurredprincipally in spring (March, April, May), in accord-ance with the active vegetative growth phases, thewidth growth continued during kernel-filling, althoughmore slowly (Figure 2A and B). Similar patternsof branch growth were observed in 1997 (data notshown).
Canopy volume growth was affected by irrigationtreatments during the experimental period. Canopyvolume at the end of the experimental period showed asignificant reduction for T5, around 16 and 14% com-pared to T1 and T2, respectively (Table 1). T2 and T3showed a similar growth in the canopy volume. Can-
Figure 2. Evolution of the length (A) and diameter (B) of youngbranches during 1999, for each treatment. Each point is the mean ofsixteen repetitions per treatment. Vertical bars represent the stand-ard error of the mean. Long dashed lines indicate the start and theend of the stress period in the kernel-filling stage (early June–earlyAugust).
174
Figure 3. (A), Annual canopy volume growth rate as a functionof mean annual water applied for the period 1997–2000. The lineindicates the overall linear regression (y = 0.0047 + (27 × 10−6)x,r = 0.95; P < 0.05). (B), Mean annual pruning weight as a func-tion of annual water applied (y = −7.07+0.0047x+ (3×10−6)x2,r = 0.98, P < 0.05). Points are based on treatment av-erages (12 measurements per treatment) for the different years(1997–2000).
opy volume growth rate was correlated linearly withthe annual amount of water applied (Figure 3A), ashas been described before in water-stressed almond byTorrecillas et al. (1989) and Fereres et al. (1981).
Pruning weight also was reduced by RDI, withsignificant differences between T5 and the rest of thetreatmens. There was a close relationship (r = 0.98)between amount of annual water applied and mean an-nual pruning weight (Figure 3B). Accumulated prun-ing weight, after four years, was significantly reducedby about 36% in T5 compared to T1 (Table 1). Thereduction of pruning weight under RDI may indicatea lower annual tree growth and has been describedpreviously in almond (Girona et al., 1993; Prichard,1996), pear (Pyrus communis L.) (Mitchell et al.,1989), peach (Prunus persica L. Batsch) (Johnsonet al., 1992; Boland et al., 1993) and apricot (Prunusarmeniaca L.) (Pérez, 2001). These results indicatethe potential of the more severe RDI strategy (T5)to limit tree size and vegetative development. Taking
into account that pruning costs for almond representaround 10% of total cultural costs (Klonsky and Blank,1996), a reduced pruning weight could be profitableeconomically.
Leaf growth, leaf abscission and leaf area
Water stress during kernel-filling (early June-early Au-gust) produced a significant decrease in leaf expansionrate (length and width) in all RDI treatments, with areduction compared to T1 of around 49% in T2, 39%in T3, 47% in T4 and 50% in T5 (Table 2). AmongT2, T3 and T4 there were no significant differences inthe leaf expansion rate or leaf size (Table 2). Only theSDI treatment T5 showed a leaf size (single leaf area)significantly lower than the rest of the treatments.
We observed an important effect of water stress onleaf abscission, with a premature stimulation of leaffall in the RDI treatments. There were two periodsof the year in which there was a higher abscission ofleaves, one at the end of kernel-filling (end of July,greatest stress period), mainly in the RDI treatmentsdue to severe water stress in this period, and the otherat the end of November (vegetative dormancy) duringwhich there was a pronounced physiological leaf fall,mainly in the control treatment (Figure 4A).
During kernel-filling, the percentage of leaf abscis-sion with regard to the accumulated total was about36, 30, 25 and 33%, in T2, T3, T4 and T5, respect-ively, as opposed to 15% in the control (Figure 4A).Leaf abscission was influenced by the level of waterstress (�pd) reached in this period, a correlation (r2 =0.60) being observed between these two parameters(Figure 4B). Similar patterns have been recorded instressed almond by Castel and Fereres (1982) andGoldhamer (1996).
A smaller leaf size and a higher leaf abscission, dueto water stress, resulted in a lesser development of thetree leaf area in the ‘driest’ treatment, T5. The meanreduction in the tree leaf area in T5, compared to T1,was 14% in 1999 (Table 2). In 1998, the reduction wasof 5%. The differing environmental conditions in 1998and 1999, that resulted, in general, in excessive ve-getative development and very low yields in 1998 (ananomalous year due to the environmental conditions)compared to 1999, and a more severe water stress dur-ing kernel-filling in 1999 (�pd < −2.5 MPa) withrespect to 1998 (�pd > −1.5 MPa), could explain thegreater leaf reduction observed in 1999. There was aclose linear correlation between applied water amount,tree water status and leaf area (data not shown), as has
Table 2. Relative growth rate (RGR) in leaf expansion during the kernel-filling stage in 1999. Annual meanvalues of single leaf area in 1998 and 1999. Leaf area tree−1 was estimated according to the canopy volume.Each value represents the mean of 24 measurements in three different periods of the year
Leaf expansion rate Single leaf area (cm2) Leaf area tree−1 (m2)
RGR (mm mm−1 day−1) ×10−3
Treatments Length Width 1999 1998 1999 1998
T1 15.0a 28a 5.82a 5.83a 357a 353ab
T2 8.9b 16b 6.25a 5.46a 346a 340b
T3 11.0b 17ab 5.93a 5.76a 335a 403a
T4 8.5b 17ab 5.96a 5.50a 327a 323b
T5 8.1b 13b 4.93b 5.07b 307b 338b
ANOVA ∗∗∗ ∗ ∗∗∗ ∗∗∗ ∗ ∗∗
n.s. not significant ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001. For each column, mean values followed bydistinct letters are significantly different; separation by Duncan’s multiple range test at the 95% confidencelevel.
Figure 4. (A), Seasonal patterns of leaf abscission in almond treesfor each treatment during 1999. Each point is the mean of four meas-urements per treatment. Vertical bars represent the standard errors ofthe means. (B), Leaf abscission as a function of pre-dawn leaf waterpotential (y = 4.77 − 18.45x, r = 0.77; P < 0.001). Each pointis a single measurement (one tree per treatment and plot) during theperiod of greatest stress (June–August) in 1999.
been pointed out for this species by Torrecillas et al.(1989) and for other fruit trees by Levy et al. (1978).
Root growth and distribution
The distribution of soil water content influenced theroot distribution in both irrigation systems, a moredense concentration of fine roots in the area close tothe drip (wetted soil zone) being observed (Figure 5).SDI produced a greater horizontal distribution of fineroots in the soil profile than the surface drip system(Figure 5). A deeper root distribution also was ob-served under SDI. While surface treatments (T1 andT2) had a higher density of fine roots in the first 30 cmdepth, in the SDI treatments (T3, T4 and T5) there wasa greater root development around 40–50 cm depth(Figure 5). These root patterns are similar to thosefound in other species cultivated under SDI (Pheneet al., 1991; Ayars et al., 1999).
Moreover in SDI treatments T4 and T5 there werea higher density of fine roots than in the other treat-ments (Figure 5). A stimulation of fine root develop-ment in response to a severe soil water deficit couldplay an important role in drought resistance and recov-ery (McCully, 1999). In all treatments, root densitywas practically nil below 80 cm depth. 75% of fineroots were in the upper 70 cm, as has been found fre-quently in drip-irrigated almond (Catlin, 1996; Francoand Abrisqueta, 1997). In our study, factors such asphysical properties of the soil (fine texture and a hardcalcareous layer below 80 cm) and a severe soil de-hydration (high soil strength) could have affected rootgrowth directly by restricting penetration.
The more dense distribution of absorbing roots inthe wetted volume of soil (close to the drip), observed
Figure 5. Fine roots (diameter < 1 mm) distribution as a functionof depth and lateral distance from the emitter for each treatment,in the post-harvest period in 1999. Solid lines represent iso-lines ofequal density (mg dry weight cm−3 soil) of fine roots. Black arrowsindicate the localization and depth of the emitters in both irrigationsystems.
in the driest SDI treatment, could be advantageous,allowing plants to utilise more efficiently soil water(Brown, 1996), but can be unfavourable if it leads toroot intrusion in the buried drips (Zoldoske, 1999).We observed (visually in some excavated drips) a highconcentration of roots covering the buried drips, butour study demonstrated after 4 years (at the end of theexperiment), a good SDI irrigation uniformity (97%),similar to the surface system, and no root intrusion inthe buried drips (García et al., 2000; Romero, 2002).The use of a chemical barrier (trifluralin) and pressure-compensating emitters (physical barrier) can preventroot intrusion, as demonstrated previously by Ayarset al. (1999).
The root/shoot ratio in 1999 (dry weight of fineroots (mg cm−3)/dry weight of leaves (mg cm−3))increased with water stress in the RDI treatments,mainly in treatments T4 and T5, (data not shown), asreported in plants under water stress by Kramer andBoyer (1995) and Huang and Fry (1998). This sug-gests an alteration in the transport and distribution ofcarbohydrates and nutrients between root and shoot.
Bud development, flowering and fruit growth
No significant differences were found in bud density(flower buds per branch) between treatments, althoughthere was a slight increase in the density of vegetat-ive buds in T2 compared to T1 (Table 3). Floweringdensity and fruit set were also not significantly af-fected by irrigation treatments (Table 3). Other studiescarried out with almond under moderate water defi-cit found no significant effects on flowering density(Ruiz-Sanchez et al., 1988), fruit density and fruit set(Esparza et al., 2001).
Entire (hull+shell+kernel) fruit growth (lengthand diameter and dry matter accumulation) with thetime was also similar in all treatments (data notshown). There was a tendency for the abscission ofbuds and fruits to increase in T5 compared to the othertreatments (although not significantly so) (Table 3).However, since bud and fruit abscission were highlyvariable between trees, we could not establish a cleartrend in these parameters with respect to water stress.
Yield and water use efficiency
In this study we show yield data for 1999 and 2000only, because we consider that 1997 (first year of theexperiment) and 1998 (year of very low yields, 300–400 kg ha−1, due to bad weather during bloom and
Table 3. Effect of irrigation treatments on the number (per metre of branch) of different stages of reproduction of almond treesin 1999. Fruit set and abscission of accumulated buds, flowers and fruits during 1999 (single year of the experiment)
Density (n◦ m−1) Abscission (N◦/sieve box, 4 m2)
Treatments Vegetative buds Flower buds Flowers Fruits Fruit set Total buds Flowers Fruits
T1 81a 144 124 66 49 1236 7857 127
T2 105b 164 133 59 37 1548 6739 102
T3 92ab 146 130 70 46 1046 5562 89
T4 77a 127 107 58 48 1200 6341 98
T5 78a 142 126 67 49 1442 5573 158
ANOVA ∗∗ n.s. n.s. n.s. n.s. n.s. n.s. n.s.
n.s. not significant; ∗∗P < 0.01. For each column, mean values followed by distinct letters are significantly different; separationby Duncan’s multiple range test at the 95% confidence level.
Table 4. Effect of different irrigation treatments on yield response of almond trees during two years of the experimental period(1999–2000)∗ % kernel rot.
Percent kernel, mean (%) 29.4 29.2 29.3 29.5 29.3 n.s.
Yield water use efficiency, mean (kg m−3) 0.23a 0.28b 0.26ab 0.29bc 0.33c ∗∗∗Hull-tight, mean (%) 2.33a 4.24bc 3.71abc 3.11ab 5.18c ∗Faulty kernels (%)∗ 1.58a 0.92ab 0.50b 0.96ab 0.88b ∗Annual applied water, mean (mm year−1) 603 436 436 382 330
Cumulative applied water 1997–2000 (mm) 2411 1744 1743 1528 1320
Annual reduction of applied water, mean (%) 0 28 28 37 45
n.s. not significant; ∗P < 0.05; ∗∗∗P < 0.001. For each row, mean values followed by distinct letters are significantly different;separation by Duncan’s multiple range test at the 95% confidence level.
pollinization) are not representative of the experimentfrom a production point of view.
There were significant differences in the al-mond nut yield components (shell yield, kernel yieldand fruit load) between theT1 and RDI treatments(Table 4). Mean kernel yield per tree (for the period1999–2000) was reduced by 11% in T2, 15% in T3,20% in T4 and 22% in T5 compared to T1 (Table 4).However there were no significant differences betweendeficit treatments (SDI T3, T4, T5 and T2). The ap-plication of the same RDI strategy (with the sameamount of water) under surface and subsurface sys-tems (T2 and T3) produced a similar kernel yieldresponse in both treatments (Table 4). Schwankl et al.(1999) observed a similar yield response in almondsunder surface and subsurface irrigation systems in dif-ferent soil conditions. However, other studies in the
Murcia region have shown a higher kernel yield underburied drip irrigation (Del Amor et al., 1999).
There were no significant reductions in kernel per-centage or kernel dry weight, being less than 3% inT4 and T5 (Table 4). Greater reductions in kernelweight (around 10%) have been recorded in other al-mond varieties under RDI (Goldhamer, 1996). Theseresults show that, even under severe water stress dur-ing kernel-filling (�pd < −2.3 MPa in 1999), theaccumulation of photo-assimilates in the seed was notaltered, indicating the strong sink activity of the fruitin this period. Kernel dry weight gain with time wasnearly identical for all RDI treatments and T1 throughearly June-mid to mid-July (data not shown). Onlyfrom mid-July through harvest (three weeks) did ker-nel dry matter accumulation slow slightly in all RDItreatments, as pointed out previously by Goldhamer(1996) and Girona et al. (1997).
178
We observed that yield reductions were principallydue to reductions in the fruit load (n◦ fruits tree−1).The number of fruits per tree in T4 and T5 wassignificantly reduced by 15 and 17%, respectively,compared to T1 (Table 4). Other studies in almondsunder RDI during kernel-filling (irrigated at 50% or at20% ETc) and late post-harvest showed lower reduc-tions of fruit load (between 5% and 9%), indicatingthat the principal yield component affected was kernelweight (Girona and Marsal, 1995; Goldhamer, 1996).This different yield response in almonds under waterstress could be due to different reasons: the variety ofalmond, differences in edaphoclimatic conditions andthe pattern or level of the stress.
Water use efficiency (WUE) increased signific-antly in the RDI treatments T2, T4 and T5 (between20% and 30%) compared to T1 (Table 4). Water useefficiency increased linearly with the decrease in theirrigation water amount, as found by Torrecillas et al.(1989) and Hutmacher et al. (1994). Subsurface treat-ment T5 showed also a WUE, significantly higher thanT2 and T3 (Table 4). In economic terms, these effi-ciency values can indicate that, for a fixed mean priceper kg of kernel in SE Spain (around 2.90 ¤ kg−1
for the period 1998–2002) and taking into accountthe variability in the price of water in this region,between 0.13 and 0.35 ¤ m−3, depending on the wa-ter quality, the gross profit margin estimated (incomefrom production – water price, without taking intoaccount other costs) increases for RDI treatments aswater price increases, especially in the more restrictivestrategies T4 and T5, compared to T1 (Figure 6). Theprice of water, together with the annual water savingachieved, basically decides the extent of this increasedefficiency. Normally, in the Murcia region the price ofwater is much higher than that of the water used in thisstudy (from irrigation canals), since major amounts ofirrigation water are obtained (more expensively) prin-cipally from wells used for irrigation. Currently, withthese actual prices of water, there are slight differencesfavouring deficit SDI treatments, but the foreseeableincrease in the market price of water in these zoneswill make these strategies even more interesting froman economic point of view.
In our experimental conditions, in general, weobserved under RDI a slightly earlier onset of hullsplit than in the control (between 3 and 7 days).Also, hull-tight percentage increased significantly inRDI treatments. The subsurface treatment T5 and thesurface treatment T2 had hull-tight percentages signi-ficantly higher compared to T1 (Table 4). Hull-tight
Figure 6. Gross profit margin (income from production – waterprice) as a function of water price for each treatment.
percentage was correlated closely (r2 = 0.70) withthe level of water stress reached during kernel-filling(Figure 7A), as pointed out by Goldhamer and Viveros(2000). This characteristic had no economic repercus-sion in our study, but a high percentage of hull-tightnuts (as obtained in other varieties of almond in Cali-fornia) can reduce considerably the value of kernels(Goldhamer and Smith, 1995).
There were no significant differences in the per-centage of empty and double fruits between treatments(data not shown). However, the percentage of faultyfruits (visual shell and kernel rot) was significantlyhigher in T1 compared to T5 and T3 (Table 4). Alow humidity during seed development can reducesome internal fungal diseases associated with highorchard humidity under deficit irrigation (Goldhamerand Viveros, 2000) and subsurface drip irrigation(Goldhamer, 1997b; Michailides, 1996).
In our study significant linear correlations betweenkernel yield and several growth parameters were ob-served. A simple regression model indicated that trunkcross- sectional area and canopy volume (two para-meters indicative of the overall growth of the tree)were the factors most closely correlated (positively)with the kernel yield (Table 5). Most studies of al-mond under water stress have associated yield reduc-tions with reduced canopy development and tree size(Fereres et al., 1981; Castel and Fereres, 1982; Tor-recillas et al., 1989; Hutmacher et al., 1994). Thefruit load reduction observed in our study could bedue to several factors: A cumulative effect of waterstress on shoot growth due to reduced canopy volumeand terminal shoot and branch growth. This effectwas observed principally in the subsurface treatment
Table 5. Simple regression analysis between kernel yield (1999–2000) and two growthparameters (canopy volume and trunk cross-sectional area) measured in the year 2000. ∗Astepwise multiple regression model was made between kernel yield and several growth para-meters. Trunk-cross sectional area was the only one parameter that added significantly to themodel. The other growth parameters not improved the model
Figure 7. (A), ‘Hull-tight’ percentage as a function of pre-dawnleaf water potential during the kernel-filling stage, for the period1998–2000 (y = 0.0713 − 2.4451x, r = 0.84, P < 0.001).Each point represents the mean of twelve trees per treatment. (B),Kernel yield as a function of the level of water stress reachedin the period from the end of kernel–filling to early post-harvest(y = 1438 + 237.6x, r = 0.45, P < 0.05). Each point representsthe mean of twelve trees per treatment for the period 1998–2000.
T5 (receiving the least water), resulting in a smallertree size and less fruitwood growth (fewer fruitingpositions), with more severe water stress, as foundby Goldhamer and Viveros (1991), Prichard (1996)and Esparza et al. (2001). Also, a premature leaf
defoliation during kernel-filling (severe water stressperiod) and a significant reduction of photosynthesisin this phase (Romero et al., 2004) would result in alower tree leaf area, and less carbon gain and carbo-hydrate accumulation, reducing the necessary reservesfor shoot growth in the following year. In this regard,we observed a significant linear correlation betweentree water status during this critical period (the end ofthe kernel-filling stage and early post-harvest and thekernel yield in the following year (Figure 7B).
Based on the results from Romero et al. (2004)and from these data, we conclude that the applica-tion of the same amount of water with the differentirrigation systems (T2 and T3) produced no signific-ant differences in the tree water relations, vegetativedevelopment or yield. RDI strategies T4 and T5 (witha severe decrease of irrigation water, 80% ETc, dur-ing kernel-filling and up to 50% post-harvest), undersubsurface drip irrigation conditions, did result in asignificant reduction of kernel yield compared to thewell-irrigated treatment (T1). The primary yield com-ponent affected was fruit load (number of fruits pertree), which was reduced significantly in T4 and T5.These treatments had no important effects on treewater relations parameters, bud development, flower-ing or fruit set processes and showed a significantlyhigher water use efficiency post-harvest. These irriga-tion strategies could be a good alternative in semi-aridregions, with strictly limited water supplies. In thissituation, the grower could save significant amountsof irrigation water (up to 300 mm per year), improv-ing WUE and maintaining a mean annual kernel yieldabove 1000 kg ha−1, compared with the 450 kg ha−1
currently produced in Murcia (AEA, 2001).Nevertheless, long-term application of T5 strategy
impacted shoot growth, reducing leaf area and fruit-wood, which could influence negatively the long-term yield response. The optimisation of these RDI
180
strategies under SDI would require an increase in ir-rigation two weeks before harvest in order to maintainan adequate soil and plant water status at the end ofkernel-filling and early post-harvest (�pd − 1 MPa).Other aspects, such as salts, soil and irrigation man-agement in RDI under a subsurface drip system and al-mond profitability in these irrigation conditions, needmore investigation.
Acknowledgements
The authors would like to thank Dr David Walkerfor correction of the English. This work has beensupported partially by a grant of The Institute Eur-omediterraneo de Hidrotecnica Foundation, awardedto Pascual Romero Azorín.
References
Alston J M, Christian J, Murua J R and Sexton R 1993 Restrictingflow of almonds to export markets may raise profits. Cal. Agric.47, 7–10.
Anuario de Estadística Agraria 1999 Ministerio de AgriculturaPesca y Alimentación. Madrid, 565 pp.
Anuario de Estadística Agraria 2001 Ministerio de Agricul-tura Pesca y Alimentación. Madrid. Published in Internet:http://mapa.es/estadistica.
Ayars J E, Phene C J, Hutmacher R B, Davis K R, Schoneman R A,Vail S S and Mead R M 1999 Subsurface drip irrigation of rowcrops: a review of 15 years of research at the Water ManagementResearch Laboratory. Agr. W. Manag. 42, 1–27.
Bielorai H 1982 The effect of partial wetting of the root zone onyield and water use efficiency in a drip and sprinkler-irrigatedmature grapefruit grove. Irrig. Sci. 3, 89–100.
Boland A M, Mitchell P D and Jerie R H 1993 Effect of salinewater combined with restricted irrigation on peach tree growthand water use. Aust. J. Agric. Res. 44, 799–816.
Botía O P, Romero A P, García M F, García M A, Abadía S A 1998Evaluation of RDI strategies in almonds under subsurface dripirrigation conditions in almonds. First Results. Primeros resulta-dos. Actas del IV Simposium Hispano-Portugués de RelacionesHídricas en las Plantas. 2 y 3 de Noviembre de 1998. Murcia,83–88.
Botía P, García F, Romero P and García A 2000 Yield responseof almonds to different regulated deficit irrigation strategies un-der surface and subsurface drip irrigation conditions. Actas delXVIII Congreso Nacional de Riegos. 20–22 de Junio de 2000,Huelva, 10 pp.
Brown K W, Thomas J C, Friedman S and Meiri A 1996 Wettingpatterns associated with directed subsurface irrigation. Proceed-ings of the International Conference. November, San Antonio,Texas 3-6 1996, 806–811.
Camp C R 1998 Subsurface drip irrigation: A review. Transactionsof ASAE, 41, 1353–1367.
Castel J R and Fereres E 1982 Responses of young almond trees totwo drought periods in the field. J. Hort. Sci. 57, 175–187.
Del Amor F M and Cerdá A 1997 El riego localizado subterráneoen almendro. Aspectos hidráulicos. Frut. Prof. 85, 42–47.
Del Amor F M and Del Amor F M 1999 Riego por goteo subterráneoen almendro. Aspectos de manejo y respuesta del cultivo. Frut.Prof. 104, 61–66.
El-Gindy A M and El-Arabi A M 1996 Vegetable crop response tosurface and subsurface drip under calcareous soil. In Eds. C RCamp, E J Sadler and R E Yoder Proc. Intl. Conf. on Evapotran-spiration and Irrigation scheduling, St. Joseph, Mich., ASAE,1021–1028.
Esparza G, DeJong T M, Weinbaum S A and Klein I 2001 Effects ofirrigation deprivation during the harvest period on yield determ-inants in mature almond trees. Tree Physiol. 21, 1073–1079.
Fereres E, Aldrich T M, Schulbach H and Martinich D A 1981Responses of young almond trees to late season drought. Cal.Agric. 35, 11–12.
Franco J A and Abrisqueta J M 1997 A comparison betweenminirhizotron and soil coring methods of estimating root distri-bution in young almond trees under trickle irrigation. J. Hort. Sci.72, 797–805.
García F, Botía P, Romero P and García A 2000 Evaluación de unainstalación de riego localizado subterráneo después de dos añosde funcionamiento. XVII Congreso Nacional de Riegos, Murcia11–13 de Mayo de 1999, 374–381.
Girona J, Marsal J, Cohen M, Mata M and Miravete C 1993Physiological, growth and yield responses of almond (Prunusdulcis L.) to different irrigation regimes. Acta Hort. 335, 389–398.
Girona J and Marsal J 1995 Estrategias de RDC en almendro. InZapata M and Segura P (Eds.) Riego Deficitario Controlado.Fundamentos y Aplicaciones. Mundi-Prensa España, 99–118.
Girona J, Marsal J, Mata M, Arbonés A and Miravete C 1997 Eval-uation of almond (Prunus amygdalus L.) seasonal sensitivity towater stress. Physiological and yield responses. Acta Hort. 449,489–496.
Goldhamer D A and Shackel K 1989 Irrigation cut-off and droughtirrigation strategy effects on almond. 17th Annual Almond Re-search Conference. Modesto. California, 7 pp.
Goldhamer D A and Shackel K 1990 Irrigation cut-off and droughtirrigation strategy effects on almond 18th Annual Almond Re-search Conference. Project No. 90–I2, Fresno, California, 6 pp.
Goldhamer D A and Viveros M 1991 Irrigation cut-off and droughtirrigation strategy effects on almond. Annual Report. ProjectNo. 91–I3, 19 pp.
Goldhamer D A and Smith T E 1995 Single-season drought ir-rigation strategies influence almond production. Cal. Agric. 49,19–22.
Goldhamer D A 1996 Regulated deficit irrigation of fruit and nuttrees. Proceedings of 7th International Conference on Water andIrrigation. 13–16 May. Tel Aviv, Israel, 152–167.
Goldhamer D A, Michailides T J and Morgan D P 1997 Buried dripirrigation reduces Alternaria late blight in pistachio. Acta Hort.449, 335–340.
Goldhamer D A and Viveros M 2000 Effects of pre-harvest ir-rigation cut-off durations and post-harvest water deprivation onalmond tree performance. Irrig. Sci. 19, 125–131.
Huang B and Fry J D 1998 Root anatomical, physiological and mor-phological responses to drought stress for tall Fescue cultivars.Crop Sci. 38, 1017–1022.
Hutchinson D J 1978 Influence of rootstock on the performanceof ‘Valencia’ sweet orange. Proc Int. Soc. Citriculture Cong(Orlando) 2, 523–525.
181
Hutmacher R B, Nightingale H I, Rolston D E, Biggar J W, DaleF, Vail S S and Peters D 1994 Growth and yield responses ofalmond (Prunus amygdalus) to trickle irrigation. Irrig. Sci. 14,117–126.
Johnson R S, Handley D F and DeJong T M 1992 Long-termresponse of early maturing peach trees to post-harvest waterdeficits. J. Am. Soc. Hort. Sci. 117, 881–886.
Klonsky K and Blank S C 1996 Economic considerations. In Al-mond production manual. Ed. W C Micke. UCA, Division ofAgriculture and Natural Resources. Publ. 3364, 9–19.
Kramer P J and Boyer J S 1995 Water relations of plants and soils.Academic Press, San Diego, California, 495 pp.
Lamm, F R, Manges H L, Stone L R, Khan A H and RogersD H 1995 Water requirement of subsurface drip-irrigated cornin northwest Kansas. Transactions of the ASAE 38, 441–448.
Levy Y, Bielorai H and Shalhevet J 1978 Long term effects of differ-ent irrigation regimes on grapefruit tree development and yield.J. Amer. Soc. Hort. Sci. 103, 680–683.
McCully M 1999. Roots in soil: Unearthing the complexities ofroots and their rhizospheres. Annu. Rev. Plant Physiol. PlantMol. Biol. 50, 695–718.
Mitchell P D and Chalmers D J 1982 The effect of reduced watersupply on peach tree growth and yields. J. Amer. Soc. Hort. Sci.10, 853–856.
Mitchell P D, Van Den Ende B, Jerie P H and Chalmers D J 1989 Re-sponse of Barlett pear to withholding irrigation, regulated deficitirrigation and tree spacing. J. Am. Soc. Hort. Sci. 114, 15–19.
Michailides T J, Morgan D P and Goldhamer D A 1996 Subsur-face drip irrigation reduces Alternaria Late blight of Pistachiocaused by Alternaria Alternata (Second year report). Califor-nia Pistachio Industry. Annual Report, Crop Year 1995–1996,129–136.
Pérez Pastor A 2001 Estudio agronómico y fisiológico del al-baricoquero en condiciones de infradotación hídrica. Tesis doc-toral. Universidad politécnica de Cartagena, 225 pp.
Phene C J, Davis K R, Hutmacher R B, Mead R M, Ayars J Eand Schoneman R A 1993 Maximizing water-use efficiency withsubsurface drip irrigation. Irrig J. April, 8–13.
Phene C J 1999 Subsurface drip irrigation: Why and how. Irrig. J.April, 8–10.
Phene C J, Davis K R, Hutmacher R B, Bar-Yosef B, Meek D W andMisaki J 1991. Effect of high frequency surface and subsurfacedrip irrigation on root distribution of sweet corn. Irrig. Sci. 12,135–140.
Prichard T L, Asai W, Verdegaal, Micke W and Fuson K 1992 Ef-fects of water supply and irrigation strategies on almonds. Proc.20th Annual Almond Res. Conf., Almond Board of California,Sacramento, 60–63.
Prichard T L 1996 Residual effects of water deficits and irrigationstrategies on almonds. Comprehensive Project Report, 1995–96,No. 95–M7, 8 pp.
Romero P 2002 Response of almond trees to regulated deficit ir-rigation under subsurface drip irrigation conditions. Ms Thesis.Murcia University, 288 pp.
Romero P, Botia P and García F 2004 Effects regulated deficitirrigation under subsurface drip irrigation conditions on waterrelations of mature almond trees. Plant and Soil 260, 155–168.
Ruiz-Sánchez M C, Torrecillas A, León A and del Amor F 1988Floral biology of the almond under drip irrigation. Adv. Hort.Sci. 2, 96–98.
Salazar D M and Melgarejo P 2002 Cultivos leñosos: frutales dezonas áridas. El cultivo del almendro. Mundi-Prensa, Madrid.307 pp.
Schwankl L J, Edstrom J P, Hopmans J W, Andreu L and KoumanovK S 1999 Microsprinklers wet larger soil volume; boost almondyield, tree growth. Cal. Agric. 53, 39–43.
Torrecillas A, Ruiz-Sánchez M C, León A and Del Amor F 1989The response of young almond trees to different drip irrigatedconditions. Development and yield. J. Hort. Sci. 64, 1–7.
Tous Martí J 1995 Frutales mediterráneos cultivados en California.Agricultura, 680–684.
Turner N C 1988 Measurement of plant water status by pressurechamber technique. Irrig. Sci. 9, 289–308.
Zoldoske D F 1999 Root intrusion prevention. Irrig. J. 49, 14–15.
