Introduction

An evaluation of several studies of modern olive orchards in Mediterranean climatic conditions has demonstrated that irrigation benefits fruit and oil yield (Gucci and Fereres 2012). Nevertheless, the water scarcity and competition for water with other sectors of society that occurs worldwide has generated pressure to reduce the water used in agriculture (Fereres et al. 2003; Fernández and Torrecillas 2012). For this reason, several researchers have indicated that regulated deficit irrigation (RDI) is a viable management tool for improving leaf-level water-use efficiency (WUE, µmol CO2/mol H2O) and overall crop water productivity (WP, kg/m3) in fruit trees (Lampinen et al. 2001; Ortega-Farias et al. 2012; Fereres et al. 2014). For RDI to be a practicable solution, adequate water stress indicators are required to properly schedule the timing of water application. Both stomatal conductance (gs) and plant water potential (Ψ) have been suggested as potential indicators for olive orchards (Moriana et al. 2002, 2012; Tognetti et al. 2009; Agüero Alcaras et al. 2016).

Stomatal conductance is controlled by soil water availability, osmotic adjustment, xylem hydraulic conductivity, and environmental factors such as vapor pressure deficit and involves complex interactions between internal and external leaf factors (Flexas and Medrano 2002; Medrano et al. 2002; Fernández 2014). According to Cifre et al. (2005), gs has been identified as a very good indicator of water stress in vineyards, because a decrease in gs is an early response to water deficits in grape leaves. In olive trees, leaves have been shown to close their stomata progressively when soil water availability decreases to reduce water loss by transpiration (Tl) (Fernández et al. 1997; Hernandez-Santana et al. 2016). Stomatal closure in olives also reduces CO2 diffusion into the leaf, thereby affecting net assimilation (An) (Giorio et al. 1999; Fernandes-Silva et al. 2016).

Several studies have found significant linear correlations between gs and An for olive trees (Angelopoulos et al. 1996; Moriana et al. 2002; Tognetti et al. 2007). However, some other studies have reported nonlinear correlations between gs and An for olive trees (Sofo et al. 2009; Marino et al. 2018) and vineyards (Medrano et al. 2002; Cifre et al. 2005). In young, potted olive plants, Sofo et al. (2009) found that gs decreased from an upper limit of approximately 0.20 mol m2 s−1 in well-watered plants as water stress increased and that An was more sensitive to water stress in plants exposed to full sunlight than in shaded plants. This suggests that both stomatal limitations and photoinhibition were likely to be of importance in the response to water stress. In vineyards, Cifre et al. (2005) observed a nonlinear correlation between gs vs. An. In this study, the threshold values of gs > 0.15, between 0.15 and 0.05 and < 0.05 mol m2 s−1 indicated null-mild, moderate, and severe water stress, respectively. Under mild water stress conditions, water-use efficiency (WUE) increased, because gs and transpiration (Tl) had a greater rate of decrease than that of An (Cifre et al. 2005). This occurred, because An remains relatively high under partial stomatal closure due to limited decreases in the substomatal CO2 concentration (Ci) (Angelopoulos et al. 1996; Medrano et al. 2002; Cifre et al. 2005). However, under high levels of water stress, Ci further decreases and both stomatal and nonstomatal limitations, such as decreased electron transport rate and carboxylation efficiency, begin to be more important (Medrano et al. 2002). Despite some advances, there is little information about gs thresholds that define different levels of water stress in olive orchards under field conditions. López-Bernal et al. (2017) recently cautioned against extrapolating gs values from young potted olive plants to older field-grown plants due to significant stomatal oscillations in young plants under water stress conditions.

Using a pressure chamber to determine water potential (Ψ) has been suggested as a good method for monitoring irrigation scheduling in vineyards and orchards, because it integrates the effects of soil water content, atmospheric conditions, and cultivar on leaf water status (Scholander et al. 1965; Meyer and Reicosky 1985; Williams and Araujo 2002; Naor 2006; Ortega-Farías and López-Olivari 2012). The measurement of water potential can be carried out using three different measurements: (1) the predawn leaf water potential (Ψpd); (2) midday leaf water potential (Ψleaf) in leaves exposed to the full sun; and (3) midday stem water potential (Ψstem) measured in leaves previously enclosed in plastic and aluminum foil to obtain equilibrium with the plant xylem (Ahumada-Orellana et al. 2017; Tognetti et al. 2005; Fulton et al. 2001). Several researchers have suggested that Ψstem can be used to evaluate water deficit, because it is very sensitive to water deficit, has low variability between measurements in a given tree, and is highly correlated with other physiological variables such as gs and An (Williams and Trout 2005; Naor 2006; Tognetti et al. 2007; Ennajeh et al. 2008; Ben-Gal et al. 2010).

Recent studies have indicated that Ψstem values between − 2.5 and − 3.5 MPa are appropriate to maintain adequate olive oil yield and quality (cv. Arbequina) (Naor et al. 2013; Trentacoste et al. 2015; Marra et al. 2016; Ahumada-Orellana et al. 2017, 2018). Such values likely represent mild-to-moderate water stress when applied over extended portions of the growing season. Moriana et al. (2002) also proposed that Ψstem values < − 4.0 MPa represent severe water stress in cv. Picual. Nevertheless, more information is still needed regarding Ψstem thresholds in order for effective application of RDI in olive trees (Marino et al. 2018). As one step in defining such Ψstem thresholds, it is necessary to understand the effects of different water deficit levels on leaf gas exchange. Defining the relationships among leaf gas exchange variables (An, gs and Tl) and Ψstem can help to establish threshold values that affect oil quality and yield. Therefore, the objective of this study was to determine gs and Ψstem threshold values for water stress in a super-intensive, drip-irrigated olive orchard (cv. Arbequina) in Mediterranean climatic conditions. Gas exchange relationships were developed for olive trees under different irrigation treatments, which generated different degrees of water stress.

Materials and methods

Site description and experimental design

The experiment was conducted during three consecutive growing seasons (2011–2012 to 2013–2014) in a 6-year-old drip-irrigated olive orchard (Olea europaea L. cv. Arbequina) established in 2005 and located in the Pencahue Valley, Maule Region, Chile (35°, 232′L.S; 71°442′W; 96 m altitude). The olive trees were trained under a hedgerow system with a planting framework of 1.5 × 5.0 m (1333 tree ha−1) and irrigated using two 2.0 L h−1 drippers per tree. The climate is Mediterranean with an annual rainfall of 620 mm that is concentrated in the winter period (Ortega-Farías and López-Olivari 2012).

The experimental design was described in detail by Ahumada-Orellana et al. (2017). Briefly, the control treatment (T1) was maintained by irrigation at Ψstem values between − 1.4 and − 2.2 MPa during the growing season, with the most negative values tending to occur in the summer when crop demand was greatest. Irrigation was cutoff in the other treatments from fruit set (20 days after full bloom) until reaching Ψstem thresholds of approximately − 3.5 MPa for T2, − 5.0 MPa for T3, and − 6.0 MPa for T4. Upon reaching the specific thresholds, irrigation was reinitiated, so that Ψstem recovered to values close to those of T1. The seasonal irrigation amount for T1 averaged approximately 245 mm over the three growing seasons, and was 200, 180, and 160 mm for the T2, T3, and T4 treatments, respectively. There were four replicate plots per treatment with each plot consisting of five trees.

Plant water status measurements

To evaluate olive water status between 12:30 and 14:00 h, the midday stem water potential (Ψstem) was measured weekly during each growing season using a Scholander-type pressure chamber (Model 1000, PMS Instrument Company., Albany, Oregon, USA) (Moriana and Fereres 2002). Two apical stems per plot were used for this measurement. The stems were located in the center of the hedgerow with at least five leaf pairs per stem (Secchi et al. 2007; Rousseaux et al. 2008), and they were covered with plastic bags and aluminum foil for 1–2 h before measurement (Meyer and Reicosky 1985).

Leaf gas exchange measurements

Gas exchange was measured during each growing season, between 12:00 and 14:00 h every 7–14 days, from December to February. Measurements were performed on two mature, sun-exposed leaves per plot, located at chest height on the hedgerow exterior (Tognetti et al. 2007).

An infrared gas analyzer (Model LI-6400, Li-Cor, Inc., Lincoln, NE, USA) was used to estimate values of stomatal conductance (gs), transpiration (Tl), substomatal CO2 concentration (Ci), and net photosynthesis (An). Measurements were made on sunny days with PAR and CO2 concentration ranging between 1100–1700 µmol m−2 s−1 and 380–400 µmol mol−1, respectively. Moreover, the leaf chamber temperature was maintained between 25 and 35 °C, molar air flow rate was 400 µmol s−1, and relative humidity was between 40 and 50%.

Data analysis

Relationships between gs and the other gas exchange variables, including Ψstem, were evaluated. In addition, gs threshold values were obtained by plotting the relationship between An and gs using a piecewise linear regression, which is an effective technique for modeling changes in slope (Toms and Lesperance 2003). In this analysis, gs values were segmented and regression analyses were done separately for each segment (Malash and El-Khaiary 2010). Finally, Ψstem vs gs relationship was used to obtain Ψstem and gs thresholds that defined water stress.

Results

The overall relationship between An and gs was curvilinear with a r2 value = 0.88 (Fig. 1). Piecewise linear regression analysis indicated that there were three lines segments of different slopes over the range of gs values explored, which coincided with: gs > 0.18 (Zone I), gs between 0.18 and 0.09 (Zone II), and gs < 0.09 mol m−2 s−1 (Zone III). A change in slope at a gs value of 0.18 mol m−2 s−1 indicated an upper threshold above which water stress was mild or absent. This threshold value was consistent among seasons (Table 1). However, the point of inflection between the lower two line segments (Zones II, III) could not be determined the first season (2011–2012) and was different among seasons for 2012–2013 and 2013–2014. In Zone I, An was not significantly affected when gs > 0.18 mol m−2 s−1, presenting a clear plateau at 17.3 µmol m−2 s−1. Values of An for Zone II decreased linearly from 17.2 to 10.5 µmol m−2 s−1 as gs values declined from 0.18 to 0.09 mol m−2 s−1, while An in Zone III decreased linearly from 10.5 to 1.00 µmol m−2 s−1 as gs decreased from 0.09 to nearly 0.01 mol m−2 s−1. For Zones II and III, the ratios of An to gs were 78.2 and 107.2 µmol m−2 s−1/mol m−2 s−1 with a r2 of 0.55 and 0.63, respectively. Therefore, Zones I, II, and III were used to interpret the behavior of the leaf gas exchange relationships (gs vs Ci and Tl vs gs).

Fig. 1
figure 1

Relationship between stomatal conductance (gs) and net assimilation (An). Vertical dashed lines separate three line segments into zones (I, II, III) using piecewise regression

Full size image
Table 1 Estimation of stomatal conductance (gs) thresholds using piecewise linear regression
Full size table

There were linear relationships between Ci and gs for Zones I and II, with r2 values equal to 0.79 and 0.48, respectively (Fig. 2). Ci values for Zone I increased linearly from 170 to 250 µmol m−2 s−1 for gs > 0.18 mol m−2 s−1, while those for Zone II decreased linearly from 170 to 150 µmol m−2 s−1 as gs diminished from 0.18 to 0.09 mol m−2 s−1. For Zone III, there was not a significant relationship between Ci and gs, since the data were highly scattered. Finally, the relationship between Tl and gs was curvilinear (r2 = 0.82) with highly scattered data for Zone I and low scattering for Zone III (Fig. 3). The r2 values for Zones II and III were 0.25 and 0.81, respectively.

Fig. 2
figure 2

Relationship between stomatal conductance (gs) and internal CO2 concentration (Ci)

Full size image
Fig. 3
figure 3

Relationship between stomatal conductance (gs) and leaf transpiration (Tl)

Full size image

Significant nonlinear relationships were observed between An and Ψstem and gs and Ψstem, with r2 values of 0.65 and 0.67, respectively (Fig. 4). Values of gs and An decreased rapidly at first and then more gradually as water status became more negative. The relationship between Ψstem and gs was used to establish threshold values of Ψstem. This relationship was developed using values of Ψstem and gs ranging from − 0.9 to − 6.5 MPa and 0.05 to nearly 0.395 mol m−2 s−1, respectively. A piecewise linear regression indicated that a break point in the gs–Ψstem relationship (i.e., a significant change in slope) occurred when gs was 0.12 mol m−2 s−1 and Ψstem was − 2.3 MPa (Fig. 4b), and two linear equations were obtained (Ψstem = − 2.77 + 4.36gs for gs > 0.12 mol m−2 s−1 and Ψstem = − 5.29 + 25.54gs for gs < 0.12 mol m−2 s−1). A more detailed analysis of the gs–Ψstem relationship indicated that crop load did not significantly affect the relationship for the range of crop load (6–10 kg tree−1) evaluated (Fig. 5).

Fig. 4
figure 4

Relationships between a net assimilation (An) and midday stem water potential (Ψstem) and b stomatal conductance (gs) and midday stem water potential (Ψstem). The intersection of the black solid lines indicates a Ψstem threshold value (− 2.3 MPa) at gs = 0.12. The red dashed line indicates the Ψstem value (− 2.0) that separates different zones in the gs vs An relationship in Fig. 1

Full size image
Fig. 5
figure 5

Relationship between stomatal conductance (gs) and midday stem water potential (Ψstem) for different growing seasons. Crop load was 8.3, 6.5, and 10.3 kg tree−1 in 2011–2012, 2012–2013, and 2013–2014, respectively. The range of crop load evaluated did not affect the gs vs. Ψstem relationship

Full size image

Discussion

In comparison with other agronomic species such as grapevines, there has been little emphasis on establishing water stress thresholds in olive trees using gas exchange measurements (Cifre et al. 2005; Medrano et al. 2002; Marino et al. 2018). Stomatal conductance values have been used successfully as an integrative parameter for determining the degree of water stress in vineyards (Zsófi et al. 2009). The water stress thresholds of 0.015 and 0.05 mol m−2 s−1 for gs, observed by Medrano et al. (2002) and Cifre et al. (2005) in grapevines, have also been proposed for other C3 plants

In our study, as water deficit progressed, values of gs, An, Ci, and Tl decreased, but at different rates. In addition, the gs–An relationship has been described in olives and other species, because the decrease in photosynthesis is greatly controlled by stomatal closure (Naor and Wample 1994; Angelopoulos et al. 1996; Medrano et al. 2002; Cifre et al. 2005; Galmés et al. 2007; Tognetti et al. 2007). Moriana et al. (2002) observed that the relationship between gs and An was linear in olive trees. However, the relationship was nonlinear in our study (Fig. 1). This difference can likely be explained by the range of gs values evaluated in the two studies. Moriana et al. (2002) evaluated the gs vs An relationship up to maximum gs values of about 0.2 mol m−2 s−1. In contrast, gs reached values of 0.4 mol m−2 s−1 under our field conditions, but An did not increase when gs was greater than 0.18 mol m−2 s−1.

Based on the gs–An relationship, two zones of water stress were identified with water stress being mild or absent in Zone I (gs > 0.18 mol m−2 s−1) and water stress increasing from moderate to severe as gs decreased significantly below 0.18 mol m−2 s−1 in Zone II. In Zone I, An presented a constant value of approximately 17.0 µmol m−2 s−1, which is similar to the maximum value obtained by Marino et al. (2018), although a higher gs (> 0.25 mol m−2 s−1) was needed to reach the highest An rates. Fernández (2014) has reported that An values of C3 plants usually do not exceed 25 µmol m−2 s−1. In contrast to An, Ci did not appear to reach maximum values within the measured range (i.e., up to gs of 0.4 mol m−2 s−1). In Zone II, An decreased linearly to 10.5 µmol m−2 s−1 at gs of 0.09 mol m−2 s−1. In this zone, stomatal closure is likely to limit photosynthesis due to diminishing Ci in the leaf mesophyll (Flexas and Medrano 2002). Finally, with gs under 0.09 mol m−2 s−1, An decreased steeply, but Ci did not show any pattern as a result of highly scattered data. Nonstomatal limitations to photosynthesis become dominant in this range, as has been reported by several studies (Medrano et al. 2002; Cifre et al. 2005; Zsófi et al. 2009).

Given the above discussion, the value of 0.18 mol m−2 s−1 could be used as a gs threshold for determining water stress in olive trees (cv. Arbequina). With the decrease in Ψstem, the An and gs values decreased rapidly at first; later, as Ψstem became more negative, the decreases were more gradual (Fig. 4a, b). Such a nonlinear response coincides with the results reported by other studies with olive trees (Angelopoulos et al. 1996; Ennajeh et al. 2008; Marino et al. 2018). In addition, our results showed that gs values were highly scattered when Ψstem was greater than − 2 MPa. Similar results were observed by Marino et al. (2018), who suggested that a Ψstem–gs model for low-level stress (Ψstem > − 2 MPa) is not reliable. This data scattering at high Ψstem values is caused by the sensitivity of gs to other factors such as temperature, vapor pressure deficit, and radiation influx under well-watered conditions (Ennajeh et al. 2008; Marino et al. 2018). Later, as stress intensified (Ψstem < − 2 MPa), gs decreased more gradually. Based on these physiological responses, the − 2.0 MPa value of Ψstem could be the break point between well-watered and water stress conditions. Therefore, − 2.0 MPa of Ψstem is proposed as the threshold value for programming irrigation in super-intensive ‘Arbequina’ olive orchards.

These results complement the previous studies in the same orchard in Chile under Mediterranean climate conditions (Ahumada-Orellana et al. 2017, 2018). Fruit and oil yield were not affected in this orchard under 4 years of water deficit treatment when irrigation was cutoff after fruit set each year until reaching a Ψstem value of − 3.5 MPa (Ahumada-Orellana et al. 2017). Moreover, the quality of the olive oil was not affected negatively by this level of water stress (Ahumada-Orellana et al. 2018). Similar results were observed by Marra et al. (2016) who suggested maintaining a Ψstem between − 3.5 and − 2.5 MPa for moderate, sustainable yields with good oil quality.

When interpreting Ψstem and gs values in field situations, it should be recognized that crop load in a given year can influence such variables with Ψstem tending to decrease late in the season as crop load increases (Naor et al. 2013). This indicates that agronomic factors, as well as climatic conditions may influence target Ψstem and gs values to some degree. In our study, crop load did not influence the gs–Ψstem relationship (Fig. 5) likely because the range of crop load was fairly narrow (6.5–10.3 kg tree−1) in the three seasons evaluated. Ψstem of − 2.0 MPa and gs of 0.18 mol m−2 s−1 found in this study are proposed to provide a guideline as to when water stress occurs under moderate yields in Mediterranean climates. The empirical relationships obtained between these variables and photosynthetic rate could be of use in validating simulation models addressing different aspects of irrigation and global change such as OliveCan (López-Bernal et al. 2018). The obtained Ψstem and gs thresholds also give valuable information for maintaining trees well-watered in critical phenological stages such as flowering, and could be used in baseline, reference plots when RDI is implemented in super-intensive ‘Arbequina’ olive orchards.

Conclusion

Results obtained in the present study over three growing seasons showed that there were significant nonlinear correlations between An vs gs,An vs Ψstem, and gs vs Ψstem with r2 values ranging between 0.65 and 0.88. Water stress was considered to be mild or absent when gs and Ψstem values were above 0.18 mol m−2 s−1 and above − 2.0 MPa, respectively. These threshold values should provide valuable information for maintaining trees well-watered in critical phenological stages such as flowering, and could be used in establishing baseline or reference plots when regulated deficit irrigation is implemented in super-intensive ‘Arbequina’ olive orchards. Further research is necessary to establish threshold values of gs and Ψstem for olive trees under severe water stress.