1. Introduction
Alentejo is the region in Portugal with the largest area devoted to olive growing, making up approximately 52% of all olive growing areas, and it is where the economic impact of olive growing is most significant [
1]. In 2010, olive groves occupied approximately 162,834 hectares in Alentejo [
2], and projections estimate that the olive oil national production will be of around 100,000 Mg by 2020 [
3]. Such an expansion requires from farmers an acute and careful irrigation water management of their orchards, as this is a region of scarce water resources, where water plays a decisive role in agricultural development. The ever-increasing water scarcity in a climate of mild winters and hot and dry summers means that olive irrigation water use (IWU) must be optimized in order to reduce the region’s water resources demand from its main water use sector. With an annual reference evapotranspiration (ET
o) of around 1200 mm, resulting in a high water demand by crops and near daily irrigation in the summer, the sustainability of olive production in the region requires improved irrigation water productivity (WP), with deficit irrigation (DI) management being advocated as a way to better yields, oil quality, and economic returns of newly commercial orchards [
4,
5]. In southern Portugal, mature drip-irrigated olive orchards with planting densities of around 300 trees ha
−1 require between 3500 and 4000 m
3 ha
−1 (350 to 400 mm) to satisfy irrigation needs for full irrigation (FI) [
6,
7]. Applying irrigation depths smaller than those, and at variable rates in distinct periods of the crop growth cycle (regulated deficit irrigation, RDI) has been advocated [
8,
9,
10] to enhance water productivity (WP). Water productivity can generally be quantified in physical or economic terms [
11,
12]. Although WP can be defined with different perspectives [
13,
14], relative to irrigation [
15] WP refers to the ratio between the yield (Y
a) and the irrigation water applied and used (IWU) by crops. Improving WP can lead to the achievement of the highest benefits from water, and hence it is viewed as a major contributor to water saving [
11]. In regions where water supply is limited and farmers are frequently forced to apply DI strategies to manage their water supply, as in southern Portugal, the increase in IWUE leads to the betterment of WP.
A recommended approach for the estimation of IWU and water stress is the dual K
c-ET
o methodology adopted by FAO56 [
16], where for DI strategies the concept of potential crop evapotranspiration (ET
c) are replaced by the ET
c (ET
c act), the result of a K
c act coefficient derived from a stress coefficient (K
s) and a soil evaporation coefficient K
e, i.e., K
c act = K
sK
cb + K
e [
17,
18,
19]. The SIMDualKc model [
7,
20,
21] provides such a computational structure [
17,
18,
22]. The implementation of DI strategies also requires knowing the plant water status, preferably determined by plant-based measurement methods [
5,
21,
23]. Aside from providing a sign indicating when to irrigate, these strategies offer valuable information on crop water stress.
The aim of this study was to evaluate the effect of two regulated deficit irrigation regimes on ETc, IWUE, Ya, WP, and Kc in “Cobrançosa” olive trees, grown in intensive orchards in southern Portugal. Deficit irrigation is compared with fully irrigated trees. Moreover, and according to the soil water conditions, the SIMDualKc model is used to obtain and adjust Kc and Kcb to plant height and density, as well as to estimate ETc.
4. Discussion
For our ‘Cobrançosa’ trees grown in orchards in southern Portugal, the main aim of this study was to evaluate the effect of two deficit irrigation (DI) regimes on tree evapotranspiration (ET
c), water use efficiency (IWUE), yield (Y
a), and water productivity (WP). Deficit irrigation was compared with fully irrigated (FI) trees. The average irrigation water supplied to FI in 2013 and 2014 was considerably different from the FI crop water uptake in the same period. Average irrigation water supplied to 70DI and 50DI were more in harmony with their crop water consumption during the years of the experiment. The irrigation water use efficiency of 50DI was the highest among treatments, closely followed by the 70DI treatment, while FI water use efficiency lagged behind in the two last seasons of the study. Significant water savings were achieved with both DI regimes, as many authors have reported [
9,
35]. As for us, water savings by DI regimes are usually viewed as major contributors to their water productivity enhancement [
11], with little impact on yield [
35,
36,
37,
38]. Also, as our data also show, many authors have reported increases in water productivity under DI regimes for olive [
39,
40,
41,
42].
Reference [
37] compared deficit irrigated to fully irrigated olive trees and reported that the DI strategy reduced ET
c, and consequently the yield, through an asymptotic yield-ET
c function [
39]. In our study, olive yields did not always increase with expanding seasonal irrigation water applied. Comparatively, and despite the larger irrigation supplies, FI used almost the same amount of water in satisfying olive seasonal consumptive use as 70DI (201 and 190 mm, respectively, in the two last seasons of the study), with no significant yield increase. Reference [
5] showed similar results for cv. ‘Cordova’ low-density olive trees grown in orchards in southern Portugal. We observed alternate bearing [
39,
43]; in the “off” years of 2011 and 2013, the mean yield difference among treatments was significant, with 45% reduction in fruit for FI as compared to 70DI, although there was no significant mean yield difference for the “on” years. FI yield reduction in the “off” year in relation to 70DI, despite the overall similarity of their Ψ
st and g
s readings, may be justified by FI stomata closure in the summer [
37,
39] and the eventual non-linear (second order) relationships mediating water use and yield [
37,
39,
44]. For the 70DI treatments, yield significantly increased with the irrigation water applied, except in 2012, when no significant differences were found between treatments. Reductions in fruit, as compared to 70DI, were 60% and 50% for 50DI in the “off” year of 2011 and 2013, and 47% in the “on” year of 2014, in agreement with the typically Y
a reported results for DI [
9]. While the irrigation water use efficiency for 50DI was the highest among treatments, meaning increases in efficiency with decreasing water application, their water productivity was lower than that of 70DI. Maximizing olive irrigation water use efficiency may not necessarily be the best option here, as growers will be likely more interested in maximizing their systems conversion of water into goods and services by consequently improving water productivity [
8]. The abovementioned increases in WP for 70DI and 50DI in relation to FI are more related to their water savings through better irrigation water use efficiency (lower irrigation water applied) than through yield, as the Y
a averages demonstrate. However, in a region short on water resources, as in Alentejo, it is prudent to look into water productivity and irrigation water use efficiency and performances for adequately selecting DI regimes. Furthermore, it is worthwhile to recall that in the abovementioned years of 2013 and 2014, T
sf for FI and 70DI were on average almost the same (201 and 190 mm, respectively) for an irrigation water application of 417 for FI and 267 mm for 70DI, respectively, which precludes considering FI regimes for cv. ‘Cobrançosa’ olive in Alentejo.
Readings of Ψ
st and g
s from 2011 through 2014 provided a good assessment of olive response to water stress induced by the distinct DI regimes [
5,
37,
40] and indirectly of their expected Y
a and WP [
25,
45]. Other authors have found similar patterns of Ψ
st and g
s versus Y
a in similar climates [
25,
46]; a result of the olive adaptation to drought, when trees under stress tend to control transpiration through stomatal closure [
25] and the lowering of their g
s readings, with subsequent decreases in net CO
2 assimilation and consequently yields [
46]. Concurrently, they increase their irrigation water use efficiency, as observed for the 50DI regime.
Our seasonal trend of T followed similar patterns to those of ET
o, with maximum T values observed in mid-summer [
25]. However, as compared to spring and autumn, ET
o values increase more in mid-summer than T values, contributing to the lower observed K
cb values in July and August than before and after mid-summer [
47]. In general, mean K
c act values in the four years of the experiment for 70DI followed the characteristic olive U-shape pattern described in Reference [
47]. They also were similar to the FAO K
c for olive crops published by Reference [
34]. Reference [
48] described a K
c for olive crops of around 0.35 during summer, and an increase thereafter. References [
49,
50] reported similar K
c values, after adjustments for ground cover, while Reference [
51] accounted for comparable K
c values for their olive experiment in northwestern Argentina (southern hemisphere). The basal mean K
cb act values in the same periods, for 70DI and 50DI, respectively, were also similar to the standard ones proposed by Reference [
34] for intensive orchards such as the one in our study (≤300 trees ha
−1). They were lower than K
c act, reflecting the partitioning of ET
c into transpiration (T) and soil evaporation (E
s). Actually, K
c act values for 70DI and 50DI, markedly influenced by soil evaporation as they integrated the seasonal evolution of ET
c act values (
Table 5), illustrated the importance and influence of E
s in the expression of olive ET
c values for olive orchards in Mediterranean climate regions [
52]; in particular, they accentuated the need for a water balance simulation model to quantify E
s [
31,
52,
53] and predict ET
c and K
c for drip-irrigated olive orchards in Alentejo. Despite the good results obtained in assessing tree transpiration and the derived K
c values from sap flow tree measurements, it is worth cautioning that a small number of uncalibrated sap flow probes per tree, as used in this study, is reported to lead to large flow variability (radial and azimuthal) within and between trees [
54].