Gas Exchanges and Stem Water Potential Deﬁne Stress Thresholds for Efﬁcient Irrigation Management in Olive ( Olea europea L.)

: With climate change and decreased water supplies, interest in irrigation scheduling based on plant water status is increasing. Stem water potential ( Ψ SWP ) thresholds for irrigation scheduling in olive have been proposed, however, a physiologically-based evaluation of their reliability is needed. A large dataset collected at variable environmental conditions, growing systems, and genotypes was used to characterize the relation between Ψ SWP and gas exchanges for olive. Based on the effect of drought stress on the ecophysiological parameters monitored, we described three levels of stress: no stress ( Ψ SWP above about − 2 MPa), where the high variability of stomatal conductance (g s ) suggests a tight stomatal control of water loss that limit Ψ SWP drop, irrigation volumes applied to overcome this threshold had no effect on assimilation but reduced intrinsic water use efﬁciency (iWUE); moderate-stress ( Ψ SWP between about − 2.0 and − 3.5 MPa), where iWUE can be increased without damage to the photosynthetic apparatus of leaves; and high-stress ( Ψ SWP below about − 3.5 MPa), where g s dropped below 150 mmol m − 2 s − 1 and the intercellular CO 2 concentration increased proportionally, suggesting non-stomatal limitation to photosynthesis was operative. This study conﬁrmed that olive Ψ SWP should be maintained between − 2 and − 3.5 MPa for optimal irrigation efﬁciency and to avoid harmful water stress levels.


Introduction
Olive (Olea europaea L.) is a drought resistant crop traditionally grown under rainfed condition in low density plantation. Recently, with the introduction of the super-high-density (SHD) system, characterized by very high planting density (1600-2000 trees per ha) and trees trained as hedgerows, olive orchard management has changed deeply. SHD orchards offer many advantages: they reach productivity very few years after planting, allow mechanization of harvesting and partial mechanization of pruning (tree topping) and a significant reduction in production costs [1][2][3]. However, increased planting density increases light interception [4], root competition [5], and plant sensitivity to drought [6]. This suggests irrigation parameters may also need to change. Irrigation is generally based on orchard water use (e.g., crop evapotranspiration, ETc), but multiple studies have demonstrated this method, because it does not consider plant water status, may overestimate the irrigation required for optimal yield [7][8][9] and reduce orchard water use efficiency. In stage II C i and g s are directly correlated showing that stomatal limitation are still predominant while in stage III C i increases as stomata close suggesting metabolic impairments to the photosynthetic system are predominant. In this last stage, iWUE decreases.
Collectively, the results demonstrate that a physiological characterization of stress levels is needed for a more sensitive plant based deficit irrigation scheduling, with the objective of increasing WUE while maintaining yield.
In this study, we aim to characterize the relationship between midday stem water potential and gas exchange for olive at variable crop conditions. We used a large dataset from two different experiments carried out in recent years [7,17].
The variability of environmental conditions (two different locations in two different years), genotypes ('Arbequina', 'Nocellara del Belice', and 'Olivo di Mandanici') and planting design and tree density (super high density, high density, and low density) provide sufficient wide dataset to smooth out variability in physiological behavior generally associated to difference in experimental conditions [31].
The different mechanisms involved in stomatal and non-stomatal control of photosynthesis at the different stress levels are investigated and used to define Ψ SWP thresholds for irrigation of olive orchards that satisfy good physiologically-based criteria for a sustainable and efficient irrigation management.
Overall aim was to provide practical recommendation to improve irrigation management and avoid harmful stress while maintaining orchard productivity.

Materials and Methods
Data were collected during two different experiments. The first experiment was carried out in 2009 season in the same SHD orchard (cultivar 'Arbequina', 1905 trees/ha) where Marra et al. [7] made their experiment.
The orchard was planted in 2004 on a private farm located in Southern Italy (37 • 48 0 N, 12 • 26 0 E, 12 m altitude). The climate of the area and the experimental soil have been described in Marra et al. [7].
Five different irrigation treatments were imposed, corresponding to 16%, 21%, 41%, 62% and 83% of Crop Water Requirement (CWR). The FAO CROPWAT model for deficit irrigation scheduling was used to calculate CWR starting from the ETc and rain data.
ETc was established a priori using FAO procedure published as FAO Irrigation and Drainage Paper No. 56 [32]. In particular, ETc was calculated as ETc = ETo × Kc × Kr, where ETo (30-year average reference evapotranspiration) was estimated using the Penman-Monteith equation and environmental data provided by a public weather station (SIAS, Servizio Informativo Agrometeorologico Siciliano), located in the proximity of the orchard; Kc is crop coefficient obtained from literature [32][33][34][35][36] and varied from 0.50 to 0.75, depending on the phenology of the trees; Kr is a coefficient related to the percentage of ground covered by the crop and resulted 0.58 based on measurements of canopy volume and subsequent calculations of the percentage of ground area coverage (42%).
Details on the irrigation system and orchard practices were also given in Marra et al. [7]. A randomized block design was used with 5 blocks of 45 trees each (9 trees per treatment), distributed between three adjacent rows; within each block one tree per irrigation treatment was randomly selected in the central row and subsequently monitored. Stem water potential (Ψ SWP ) and gas exchanges were measured on the 11 August, when the trees are more resistant to drought, and 1 September, when the trees are sensitive to drought [37].
The second experiment was carried out in 2014, in two adjoining olive orchards located near Sciacca (37 32 N, 150 m above sea level) in southwest Sicily (Italy), the same area where Marino et al. [15] conducted their experiment. A more detailed description of the climate and the soil is reported in Marino et al. [6].
Out of the two adjoining orchards, one was a traditional 10-year-old widely spaced (200 trees/ha) orchard (cv. 'Nocellara del Belice'). Trees were trained as vase shape and rain-fed all over the season.
The adjoining orchard was a hedgerow trained high density (HD, 1000 trees/ha) 3-year-old orchard. Two genotypes ('Nocellara del Belice' and 'Olivo di Mandanici'), characterized by different vigor and productive potential [38] were monitored for their response to short term water deficit. The irrigation was supplied from June, every 7-10 days. At the beginning of July, irrigation was stopped for a number of days until the midday Ψ SWP decreased below −2.5 MPa [7,26]. A re-watering period followed. Detail about orchard characteristics, experimental design and irrigation management are reported in Marino et al. [6]. Stem water potential and gas exchange were monitored every 10 days, from June until October.
Stem water potential was measured at midday using a pressure chamber (PMS Instrument Co., Corovallis, OR, USA). Fully-exposed shoots of the current growth season with five or six expanded leaves were covered with plastic envelopes and aluminum reflective foil at least 1 h before measurement in order to reduce leaf transpiration [39] and equilibrate stem water potential with branch water potential. On the same days when the Ψ SWP was measured, we conducted a series of leaf gas exchange measurements on fully expanded leaves, using a CIRAS-2 (PP system ® ) portable gas exchange system (CO 2 and H 2 O) connected to a gas exchange chamber (Parkinson Leaf Cuvette). The system measured principal eco-physiological parameters, such as light-saturated net CO 2 assimilation (A n , µmol m −2 s −1 ), stomatal conductance (g s , mmol m −2 s −1 ), atmospheric pressure, air and leaf temperature, and air CO 2 and intercellular CO 2 concentration (C i ) [40,41]. Intrinsic water use efficiency (iWUE, mmol mol −1 ) was calculated as the relationship between A n and g s .
A commercial software package (TableCurve 2D; SYSTAT Software Inc., Chicago, IL, USA) was used to find the best-fit and the function parameters of the relationship between the principal eco-physiological variables.
Statistical analysis of the data (GLM) was carried out using the Systat (SYSTAT Software Inc., Chicago, IL, USA); significance was set at p ≤ 0.05.

Results
As shown in Figure 1, A n and g s were well correlated with midday Ψ SWP (r 2 of 0.61 and 0.51, respectively).   Data were fitted against a double exponential function. The high variability of the response of g s and A n relative to Ψ SWP at different degree of stress supports this approach of curve fitting and distinguished three regions of Ψ SWP .
For Ψ SWP under the threshold of approximately −3.5 MPa, gas exchange maintained relatively constant and low rates, and Ψ SWP slightly affected gas exchange. For instance, each unit increase in Ψ SWP increased g s by 25 mmol m −2 s −1 (8% of total g s increase) and A n of 1.7 µmol m −2 s −2 (12% of total A n increase).
When Ψ SWP increased, its relative influence on gas exchange was higher. One unit increase in Ψ SWP within the range of −3.5 to −2 MPa increased g s by 88 mmol m −2 s −1 (25% of total g s increase) and A n by 2.6 µmol m −2 s −2 (18% total A n increase).
Above the threshold of −2 MPa, for each unit increase in Ψ SWP , g s increased by 137 mmol m −2 s −1 (40% of total g s increase) while A n increased by 5 (35% of total A n increase). Higher dispersion of data points around the curve was observed at increasing Ψ SWP , suggesting the superimposition of other factor at low stress levels. This was particularly clear at Ψ SWP above −2 MPa where, for the same Ψ SWP value, g s varied from 98 to 547 mmol m −2 s −1 .
Sub-stomatal CO 2 concentration (C i ) curvilinearly decreased, from approx. 250 µmol mol −1 to minimum values of 160 µmol mol −1 , as g s decreased from 600 to 100 mmol m −2 s −1 , a ( Figure 2). For g s under the threshold of 150 mmol m −2 s −1 a deviation from the curve was observed and C i started to increase for a group of datapoints. These datapoints, characterized by g s lower than 150 mmol m −2 s −1 and C i higher than 220 µmol mol −1 (average of all the data), were plotted separately and were all characterized by Ψ SWP lower than −3.5 MPa. Figure 1. Relationship between (A) stomatal conductance (gs, mmol m −2 s −1 ) and (B) net photosynthesis (An, μmol m −2 s −1 ) relative to the stem water potential (ΨSWP, MPa) measured at midday. The best fit relationship for the entire gs data pool was performed using a double exponential function: y = 39 × e (−0.08x) + 744 × e (0.63x) ; r 2 = 0.51: p < 0.0001; the best fit relationship for the entire An data pool was performed using a double exponential function: y = 3.1 × e (0.29x) + 18.9 × e (0.30x) ; r 2 = 0.61, p < 0.0001.
The relation between g s and A n was positive and exponential ( Figure 3). To better visualize the effect of the stress on stomatal control of photosynthesis, we separately plotted data in three groups based on the analysis of the previous curves (Figures 1 and 2): data characterized by Ψ SWP lower than −3.5 MPa, data characterized by Ψ SWP between −3.5 and −2 MPa and data characterized by Ψ SWP higher than −2 MPa.  For Ψ SWP below −3.5 MPa, g s values were below approx. 150 mmol m −2 s −1 and the A n increased sharply and linearly from 0 to 8 µmol m −2 s −1 as stomata progressively opened. For Ψ SWP ranging from −2 to −3.5 MPa, the relation between g s and A n was curvilinear. A n increased from 8 to 15 µmol m −2 s −1 and g s from 150 to 350 mmol m −2 s −1 . The points characterized by a Ψ SWP above −2 MPa were in the asymptotic part of the curve highlighting a wide variability in g s (ranging from 350 to 600 mmol m −2 s −1 ) corresponding to a low variability in A n (from 15 to 17 µmol m −2 s −1 ).

Discussion
The non-linear relation observed by analyzing the pooled data of ΨSWP and gs clearly suggests that the mechanisms involved in stomatal control vary according to the stress level [42].
Multiple studies have reported on the threshold for this relationship [23,26,43]. Generally, the best fit, in accordance to our study, is positive and exponential [44][45][46] but also linear [23] and logistic regressions were observed [26]. The main reason for this discrepancy can be the limited variability of ΨSWP values. ΨSWP data set of this experiment widely ranged, from −1 to −7 MPa, and gs increased from 20 to more than 600 mmol m −2 s −1 . If only the data points characterized by a ΨSWP above −3.5 MPa are used, a linear and weaker regression is observed also in this work (data not shown); if only data above −2 MPa are analyzed no relationship between ΨSWP and gs is observed.
The marked variability in gs (ranging from 600 to 250 mmol m −2 s −1 ) detected in this work for ΨSWP above −2 MPa, suggest that a ΨSWP-gs model at very low level of stress is not reliable.
Such high variability of gs is probably due to the superimposition of environmental factors affecting stomatal behavior in well-irrigated trees [42]. This has been previously demonstrated for olive by Moriana et al. [12], reporting vapor pressure deficit to affect the relationship between ΨSWP andgs only at low to moderate stress while no effect was observed at high stress. Oscillation of stomatal conductance that occur in olive, mainly in non-stressed plants [47,48], may contribute to increased variability in measured gs values at low to no-stress conditions.

Discussion
The non-linear relation observed by analyzing the pooled data of Ψ SWP and g s clearly suggests that the mechanisms involved in stomatal control vary according to the stress level [42].
Multiple studies have reported on the threshold for this relationship [23,26,43]. Generally, the best fit, in accordance to our study, is positive and exponential [44][45][46] but also linear [23] and logistic regressions were observed [26]. The main reason for this discrepancy can be the limited variability of Ψ SWP values. Ψ SWP data set of this experiment widely ranged, from −1 to −7 MPa, and g s increased from 20 to more than 600 mmol m −2 s −1 . If only the data points characterized by a Ψ SWP above −3.5 MPa are used, a linear and weaker regression is observed also in this work (data not shown); if only data above −2 MPa are analyzed no relationship between Ψ SWP and g s is observed.
The marked variability in g s (ranging from 600 to 250 mmol m −2 s −1 ) detected in this work for Ψ SWP above −2 MPa, suggest that a Ψ SWP -g s model at very low level of stress is not reliable.
Such high variability of g s is probably due to the superimposition of environmental factors affecting stomatal behavior in well-irrigated trees [42]. This has been previously demonstrated for olive by Moriana et al. [12], reporting vapor pressure deficit to affect the relationship between Ψ SWP andg s only at low to moderate stress while no effect was observed at high stress. Oscillation of stomatal conductance that occur in olive, mainly in non-stressed plants [47,48], may contribute to increased variability in measured g s values at low to no-stress conditions.
As the stress intensified (−3.5 MPa < Ψ SWP < −2 MPa in this work), stomata progressively closed suggesting hydraulic feedback largely controls the mechanism [42]. Mildly stressed olive trees restrict excessive water loss and prevent an excessive drop in Ψ SWP by modulating stomatal closure [49], which is the earliest response to drought, and the major limitation to photosynthesis at mild to moderate drought [27].
At very high levels of stress (Ψ SWP < −3.5 MPa) the small variation of both g s and A n in response to Ψ SWP confirm that water deficit override the effect of other parameters on olive physiology and on stomatal control of gas exchange [50].
A double exponential equation was used because allowed to better distinguish the different mechanisms involved in stomatal response to drought [42] with respect to a single exponential curve. Variation in g s affected the assimilation rate differently depending on the drought level as demonstrated by the analysis of the curve in Figure 3. This allows definition of different regions in this relationship, characterized by the different effects of drought on photosynthesis.
The first region is represented by the curvilinear and slight decrease in A n for g s decreasing from maximum values to 250 mmol m −2 s −1 . Very high maximum values of g s were observed reaching up to 600 mmol m −2 s −1 . These values are higher then what reported by different authors [31,51,52]. However, similarly to what reported by Fernandez et al. [32], the increase in transpiration did not correspond to an increase in A n that showed maximum value of 20 µmol m −2 s −1 .
In this range of values, a decrease in g s corresponded to a parallel decline in the sub-stomatal CO 2 concentration (Figure 2), which directly affected the photosynthetic rate.
Considering that plants were characterized by Ψ SWP values above −3.5 MPa, this suggested that at low to mild stress levels, stomatal limitations to photosynthesis were dominant. At this level of stress, consistent with earlier reports [49,53,54], progressive drought increased iWUE ( Figures 2B and 4).
Non-stomatal factors were predominant at higher levels of stress. For example, as Ψ SWP dropped to −3.5 MPa and g s dropped below 150 mmol m −2 s −1 , C i increased proportionally (Figure 3), reflecting the impaired photosynthetic metabolism in these plants, and, consequentially, iWUE decreased (Figures 2A and 4). For instance, a steeper slope of the function A n for g s , observed for values lower than 150 mmol m −2 s −1 , reflects the superimposition of non-stomatal factors affecting photosynthesis ( Figure 2). Other authors found a similar pattern with C i , initially declining with increasing stress and then increasing as drought stress became more severe [55]. Other studies confirm the possibility of using C i as an indicator of the stomatal or non-stomatal limitations to photosynthesis [27,56,57]. In an experiment conducted on nine different conifer species, Brodribb [58] observed that a rapid increase in C i values at certain levels of water stress was accompanied by a rapid loss of fluorescence from PSII, suggesting damage to the photosynthetic apparatus of plants under drought beyond the minimum C i , also called the C i inflexion point by Flexas et al. [20]. Finally, analysis with data from literature, using g s as a reference [19,25], demonstrated that the point at which C i starts to increase is accompanied by a steep reduction in the net photosynthetic rate (A n , reduced by 70%), in the CO 2 -saturated rate of photosynthesis (A sat , reduced by 50%), in the apparent carboxylation efficiency (ε, reduced by 50%) and in the rate of light-saturated electron transport (ETR, reduced by 40%).

Conclusions
The results of this experiment provide physiological data to support the plant based irrigation schedule of an olive orchard designed to increase water use efficiency.
Based on the effects of drought stress on the principal ecophysiological parameters, we were able to define three different levels of stress represented by thresholds in Ψ SWP values. The first, Ψ SWP above about −2.0 MPa, is characterized by the absence of stress. In this state, a wide variation in stomatal conductance is observed unrelated to Ψ SWP variation.
Superimposition of environmental and plant or site specific factors affects stomatal behavior in well-irrigated trees.
Increased g s in this state was not clearly improving leave assimilation rates and, as a consequence, resulted in reduced iWUE.
Moderate stress was described by a Ψ SWP between about −2 and −3.5 MPa, and was characterized by stomatal regulation of gas exchange. A reduction in water volume remaining in this range of potential improved iWUE, determining water savings. On the basis of the study conducted by Marra et al. [7], this stress level benefits oil quality without affecting system productivity. Finally, Water 2018, 10, 342 8 of 10 below about −3.5 MPa, non-stomatal limits to photosynthesis are predominant. This represents a threshold value for high stress levels, which should be avoided because it is dangerous to the plant and has a marked and negative effect on both gas exchange and productivity.