Introduction

Salinity causes conditions of stress for plants via osmotic imbalances, reductions in water uptake, and eventual reduction in growth and production. Plants exposed to salinity suffer from the cost of diverting energy from growth and production into mechanisms for stress avoidance or tolerance (Bernstein 1975; Munns 2002). Health and production of crops irrigated with water containing salts involves complex, interrelated biological and physical factors determined not only by osmotic stress and specific-ion toxicity, but also by changes in water availability via feedback mechanisms with soil water content and hydraulic conductivity. For this reason, and due to challenges in integrating and up-scaling subcellular mechanisms involving specific ions, contemporary quantitative treatment of crop response to salinity is mostly based on knowledge regarding whole-plant and field-scale responses (Shani et al. 2005, 2007).

Management of water containing salts for irrigation commonly utilizes application of excess water to maintain minimum root zone salinity and thus avoid or minimize salinity-induced yield reduction (Ayers and Westcot 1985). Maximization of crop yields when irrigation water contains dissolved salts depends on provision of plant water needs (transpiration, T) and evaporative losses as well as minimizing soil solution salinity through leaching. Evapotranspiration (ET) requirements can be estimated by measuring or calculating potential or reference ET (ET0), which is a function of climate, and through the use of crop factors that consider plant size and physiological stage (Allen et al. 1998). Salinity is not commonly a factor in the calculation of ET from ET0, but this omission is likely to lead to overestimation of ET, due to salinity-caused reductions in T (Ben-Gal et al. 2008; Dudley et al. 2008; Meiri et al. 1977). Leaching requirements for maximized production, dependent on climate and soil type (Dudley et al. 2008), are largely determined by the salinity of the irrigation water and by the relative tolerance of the crop being irrigated (Rhoades 1989).

The leaching fraction (LF) is the relative volume of applied water transporting salts out of the root zone and is most often characterized as drainage (D) divided by irrigation (I). Since, in controlled systems over long enough time periods, ET = I–D, when ET is known, such as in lysimeters, characterization of irrigation requirement including leaching can be alternatively given as irrigation relative to actual evapotranspiration or I ETa−1.

Increased demand for olive oil production and understanding that olive (Olea europaea) is relatively tolerant to many environmental stress-causing factors has led to increased irrigation in areas where the cultivation was previously rain-fed and increased planting of olives in arid regions where soil salinity is common and where large quantities of irrigation water are required. Available water for irrigation in relatively dry areas is commonly of marginal quality: recycled wastewater and brackish groundwater, both characterized by relatively high concentrations of dissolved salts.

Olive trees are customarily reported to be relatively salt tolerant, with ECe of up to between 3 and 6 dS m−1 causing no effect to growth or yields (Ayers and Westcot 1985). Salinity tolerance mechanisms of olive trees apparently include a strong ability to exclude potentially toxic ions from above-ground tissues but appear to be particularly cultivar specific (Ben-Gal 2011; Chartzoulakis 2005; Gucci and Tattini 1997; Soda et al. 2016). Evidence regarding whole-tree response to salinity is contradictory, with the bulk of the literature reporting short-term controlled studies on young trees and showing decreases in photosynthetic activity, vegetative growth, and fruit and oil production as a function of increasing root zone salinity (Aragues et al. 2005; Ben Ahmed et al. 2008; Gucci and Tattini 1997). A number of longer-term studies on mature producing trees in field conditions reported, however, that drip irrigation with water salinity as high as EC = 4.5 dS m−1 and including at least some winter leaching, did not negatively affect yields (Klein et al. 1994; Weissbein et al. 2008; Wiesman et al. 2004). Salinity additionally has been shown to be beneficial to olive oil quality attributes, particularly by increasing polyphenol content (Ben Ahmed et al. 2009; Stefanoudaki 2004; Wiesman et al. 2004).

We hypothesized that root zone salinity increased alternatively by increasing irrigation water salinity or decreasing leaching level would lead to reduced uptake and consumption of water, and growth and production of olives. Our main objective was to measure tree-scale evapotranspiration, growth, and yields of olives exposed to a large range of root zone salinity levels. A further objective was to determine the salinity response reduction function for olive and apply it in ANSWER (Shani et al. 2007), a water and salt balance-driven simulation model considering water quality and quantity and consequential plant water uptake.

Materials and methods

Twenty seven Barnea variety olive trees were grown from August 2012 until November 2015 in 2.5 m3 weighing-drainage lysimeters located at the Gilat Research Center in Israel (31o20′N, 34o40′E). The trees were planted as 60-cm-tall 1-cm-diameter single two-year-old trunks at the beginning of the experiment. Two variables, irrigation water salinity (ECi) and LF, were evaluated independently, each at 5 levels in 3 replicated lysimeters. Leaching fraction is given in terms of amount of irrigation (I) relative to ETa where treatments are summarized in Table 1 with one treatment (I ETa−1 = 140 %, ECi = 5 dS m−1) common to the two sets of variables. Until treatment initiation in April 2013, all trees were provided 0.5 dS m−1 ECi at 140 % I ETa−1. Rainfall and data to calculate daily reference ET (ET0) according to a modified Penman–Monteith equation (Allen et al. 1998) were measured at a standard meteorological station, part of the Israeli agricultural meteorological network (http://www.meteo.co.il/), located 200 meters from the lysimeter site. The station is located on a bare soil without grass. The lack of grass, as is the norm in Israel and other arid regions, is expected to well represent the local area, its agriculture, and the lysimeters while having occasionally led to higher reference values than expected in other and grass covered locations.

Table 1 Irrigation salinity and relative irrigation quantities given to lysimeter-grown olive trees
Full size table

Lysimeters and water balance

The lysimeters consisted of polyethylene containers (1.4 m high × 1.5 m diameter) filled uniformly with sandy loam soil, a bottom layer of highly conductive porous rockwool media in contact with the soil, and drainage piping filled with rockwool extending downward from the lysimeter bottom. The rockwool drainage extension (Ben-Gal and Shani 2002) disallowed saturation at the lower soil boundary while permitting movement of water out of the soil and leachate collection. Trees in lysimeters were automatically provided with water and fertilizer, and drainage water was automatically collected (Tripler et al. 2007). The soil surface of each lysimeter was covered by water-permeable silver-woven polypropylene cloth (Palrig Plastic Industries, Neot Mordechai, Israel) to minimize evaporation losses. The lysimeters were placed every 2.5 m, seven to a row in four rows with 4 m spacing, and were surrounded by border trees. Each individual lysimeter was positioned on a square weighing platform with load cells situated in each corner. By distributing load cell output current only over the relevant range of interest (4–5 tons), a resulting resolution of ±15.5 g was reached. Actual evapotranspiration (ETa) was calculated daily according to:

$${\text{ETa}} = I - P - D - \Delta W$$
(1)

where I is irrigation (predetermined), P is precipitation (measured), D is drainage (measured) and ΔW is change in soil water (derived from the change in lysimeter weight). The trees were irrigated daily.

Treatments were initiated on April 21, 2013, and continued until trees were removed in November of 2015. In addition to the preordained salinity levels maintained by adding NaCl, a constant concentration of fertilizer was supplied in the irrigation solutions. Nutrient concentrations in irrigation water were: N-NH4: 2.8 mg L−1, N–NO3: 62.0 mg L−1, P: 11.7 mg L−1, K: 71.0 mg L−1, Ca: 125.8 mg L−1, and Mg: 16.7 mg L−1, achieved by dissolving NH4H2PO4, K2SO4, and MgSO4·7H2O, and CaNO3 into solution.

Soil and plant analyses

Drainage and irrigation solutions were sampled weekly from each lysimeter and analyzed for EC. Soil was sampled in April 2014 and in September 2015, using an auger, from 0 to 20, 20 to 40, 40 to 60, 60 to 80, and 80 to 100 cm depths from the soil surface in each lysimeter. Saturated paste extract from the samples was evaluated for EC (ECe).

Trunk cross-sectional area (TCSA) was calculated using monthly measurements of trunk circumference of each tree at a marked point 40 cm above the soil surface. Trunk growth was shown, in an earlier study in the lysimeters, to be an accurate proxy for vegetative growth and leaf area in ‘Barnea’ olives trees (Bustan et al. 2016). Biomass was measured from annual pruning in December of 2013 and 2014, fruit harvests in September of 2014 and 2015 and the final removal of the trees in November 2015 when they were dissected into trunk, branches, stems, and leaves. Each of the components was weighed for fresh mass and samples were taken, weighed, dried at 70 °C, and weighed again to calculate dry biomass. Fruit was manually harvested from each tree upon reaching ripening according to 50 % black coloring (ripeness index between 2 and 3). Tree-scale total fresh fruit weight and total above-ground dry biomass production were determined. A near-infrared (NIR) spectrometer (OliveScanTM, Foss, Denmark) was used to determine oil content in fruit paste.

Data and statistics

Best-fit linear and nonlinear regressions were determined and portrayed using curve-fitting and analysis tools of SigmaPlot 13.0 (Systat Software, San Jose, CA).

Salinity reduction functions

Both total ETa and above-ground dry matter production for the experimental period were used to determine a salinity response function. We fitted normalized (actual as a function of maximum in the database) data to the van Genuchten response function (Van Genuchten and Gupta 1993)

$$f_{\text{EC}} = \frac{1}{{1 + \left( {\frac{ECe}{{ECe_{50} }}} \right)^{p} }}$$
(2)

where f EC is a reduction function due to salinity, ECe 50 represents the EC of the soil saturated paste where relative yield = 0.5, and p is a plant parameter determining the steepness of the decreasing sigmoidal curve.

Plant response simulation model

The analytical salt water (ANSWER) model (Shani et al. 2007) is a one-dimensional mechanistic-based tool governed by the soil–plant–atmosphere continuum that considers multiple environmental variables and stressors and their combined effect on plant response, expressed as water use or biomass production. The ANSWER model captures essential factors of the soil–plant–atmospheric system as a closed-form analytical solution. The analytical approach is based on four assumptions: environmental conditions in the root zone affecting plant growth and uptake can be represented by effective values of relevant parameters including water content and soil salinity (expressed as EC); conditions are steady state with respect to water and salt status (thus there are no time dimensions); a proportional relationship exists between the ratio of yield to potential yield which relates to biomass production; and finally that the climate conditions are static. The ANSWER model consists of water balance, salt balance, a hydraulic water content/movement model (Brooks and Corey 1966), a root water uptake response model (Nimah and Hanks 1973), and the van Genuchten (van Genuchten and Gupta 1993) uptake reduction due to salinity function. These governing factors essentially relate management (irrigation water quantity and quality), biological (plant response), and physical variables (soil and water) in a closed-form equation that can be used to make management decisions as it can be run for given situations with easily-accessible input data concerning soil, climate, crop, and water management (Shani et al. 2007). We ran the ANSWER model for water application rates ranging from 0 to 1.9 times reference ET for each of the ECi levels evaluated in the lysimeter experiment. Input values for environmental and crop parameters are shown in Table 2.

Table 2 Soil and plant parameters for lysimeters and model input parameters
Full size table

Results

ET and rain

Weather conditions at the experimental site followed typical Mediterranean patterns (Fig. 1). Summers were hot and dry with daily reference ET reaching values of 11 mm in 2014 and 9 mm in 2015. Winter ET0 fell as low as 0.5 mm day−1. Rainfall was limited to October through May with a total of 280 mm prior to the 2014 season and 250 mm before the 2015 season (Fig. 1).

Fig. 1
figure 1

Daily reference evapotranspiration (ET0) and rainfall over the course of the experiment. ET0 calculated based on modified Penman–Monteith (Allen et al. 1998) equation from meteorological station data at the Gilat Research Center, adjacent to the experimental site

Full size image

Drainage EC

The EC of the water draining out the lysimeters (ECd) increased quickly with time following initiation of treatments in April 2013 (Fig. 2a, b), reaching plateau values by September 2013. While ECd fluctuated seasonally thereafter, the non-equilibrium most particularly was due to winter rainfall events. Average values of ECd were consistently differentiated by treatment with higher ECd as a function of increased input irrigation water EC (Fig. 2a) and lower ECd as a function of increasing leaching from 105 to 140 % treatments with no changes as leaching further increased to 160 and 180 % ETa (Fig. 2b).

Fig. 2
figure 2

Time courses of drainage water electrical conductivity (ECd; a, b), tree-scale evapotranspiration (ETa; c, d), and trunk cross-sectional area (TCSA; e, f) as a function of irrigation water solution EC (ECi; a, c, e), and leaching level as given by irrigation as a function of actual ET (I/ETa; b, d, f). Data are average of 3 replicate lysimeters for each treatment

Full size image

Actual ET

Tree-scale water consumption followed patterns reflecting both the growth of the trees and seasons. Soon following treatment initiation differentiation of ETa values was apparent between the treatments with daily ETa increasing as a function of both decreasing ECi (Fig. 2c) and increasing LF (Fig. 2d). Maximum daily ETa for trees receiving the lowest salinity water reached over 250 L tree−1 during the summer of 2015, while maximum ETa values for trees with both the highest irrigation water salinity or with ECi of 5 dS m−1 and low leaching fraction were only around one fifth or 50 L tree−1 day−1. Similarly to the trends in ECd, ET of leaching treatments at 140, 160 and 180 % I ETa−1 was comparable.

Trunk cross-sectional area

Monthly calculations of TCSA based on circumference measurements (Fig. 2e, f) showed consistent growth over time with clear seasonal effects where growth rate increased in summers and decreased in winters. Similar to the previous parameters, TCSA strongly responded to ECi, decreasing with increased salinity and less strongly responding to leaching. By the end of 2013, the TCSA of the 140–180 % treatments was not differentiated, while that of  the 125 % leaching treatment was slightly lower, and the TCSA of the 105 % treatment was significantly lower.

Lysimeter and treatment average root zone salinity

Soil profile ECe values for the lysimeters (Soda et al. 2016) showed that average ECe increased as a function of irrigation water treatments. Increased ECi continuously increased ECe, while increasing leaching level decreased ECe up until the 140 % treatment with no further effect of the two higher leaching levels. There was no overall effect of depth on ECe for either the irrigation water salinity or leaching treatments. A linear correlation was found between average ECd for each lysimeter during April 2014 and linearly profile-averaged ECe measured from soil sampled during the same period (Fig. 3). This relationship, ECe = 0.323*ECd, was used to calculate a time-averaged soil profile salinity for the entire experiment. The relationship was reconfirmed with additional soil sampling taken in 2015 (Soda et al. 2016). These time- and profile (root zone)-averaged ECe values were significantly correlated with treatments (Fig. 4a, b) with increased ECi causing a linear increase in ECe ranging from 1.2 dS m−1 for ECi = 0.5 to 7.5 dS m−1 for ECi = 11.0. Increasing leaching of water having ECi of 5.0 dS m−1 decreased ECe as well with the best-fit regression represented as a polynomial quadratic function with relative decreases in ECe declining as leaching level increased (Fig. 4b). The lowest leaching treatment had average ECe of 7.5 dS m−1, 140 % had 4.0 dS m−1, while 180 % was characterized by 3.5 dS m−1.

Fig. 3
figure 3

Correlation of drainage water electrical conductivity (ECd) and profile-averaged saturated paste extract electrical conductivity (ECe) during April 2014. Data are from individual lysimeters. Linear regression is best fit with intercept = 0

Full size image
Fig. 4
figure 4

Time- and profile-averaged saturated paste soil extract electrical conductivity (ECe) as a function of experimental treatment variables; a irrigation water EC (ECi) and b leaching level characterized as irrigation amount (I) relative to actual evapotranspiration (ETa). Symbols are averages of 3 replicate lysimeters, error bars are standard deviations, and lines are best-fit regression equations

Full size image

Final ETa, TCSA, and above-ground dry weight

Tree-level ETa accumulated over the entire experimental period (Fig. 5a), final TCSA (Fig. 5b) and total above-ground dry weight produced by the trees (Fig. 5c) all showed similar response curves to time- and root zone-averaged ECe. The significant best-fit regression lines were defined as exponential decay with diminishing relative effects as ECe increased. The data generated from both variables causing differences in ECe, ECi and LF, were well characterized by single-response curves. Slight discrimination from the curves was apparent for the two highest leaching levels, I ETa−1 = 160 % (ECe = 3.6 dS m−1) and I ETa−1 = 180 % (ECe = 3.5 dS m−1). In these cases, the ETa, TCSA, and above-ground dry biomass (AGDW) measurements were low compared to the predicted statistic based on all the data.

Fig. 5
figure 5

a Final accumulated actual evapotranspiration (ETa), b trunk cross-sectional area (TCSA) and c above-ground dry biomass (DW) as a function of time- and profile-averaged saturated paste soil extract electrical conductivity (ECe). Symbols are averages of 3 replicate lysimeters, error bars are standard deviations, and lines are best-fit exponential decline regression equations for entire data set. Open symbols are treatments with varied irrigation water salinity (ECi), closed symbols varied leaching fractions (LF) and the gray symbols common to both (treatment 5, 140)

Full size image

Fruit yield and oil content

Basic harvest data for each of 2014 and 2015 is summarized in Table 3 and Fig. 6. Fruit yield was similar between the two years with an average of 25 kg tree−1 in 2014 and 26.5 kg tree−1 in 2015. The treatments leading to highest ECe values (ECi of 11 ds m−1 with 140 % I ETa−1 and ECi of 5 ds m−1 with 105 % I ETa−1) consistently had the lowest yields. The biggest difference between the years was for the lowest salinity treatment (ECi = 0.5 dS m−1, I ETa−1 = 140 %) which had a low yield (18.1 kg tree−1) in the first year and the highest (37.3 kg tree−1) in the second. Oil in fruit was higher in 2014 with an average of 14.7 % compared to 2015 with 13.2 %. The lowest oil percentage was found in both years for the lowest salinity treatments and was particularly low (8–11 %) in the ECi = 0.5 dS m−1 treatment. The two-year total fruit yield was weakly linearly decreased with increasing ECe (Fig. 6a). Regression indicated a decrease of some 6 % per unit EC (dS m−1). Fruit yield ranged from a maximum of ~66 kg tree−1 at ECe = 2.8 dS m−1 (treatment ECi = 2 dS m−1, I ETa−1 = 140 %) to 31–32 kg tree−1 at ECe = 7.5 dS m−1 (treatments ECi = 11.0 dS m−1, I ETa−1 = 140 % and ECi = 5.0 dS m−1, I ETa−1 = 105 %). Percent oil in fruit increased exponentially from ~9 to ~16 % as root zone ECe increased, with the biggest increase found between the two lowest salinity levels and increases in % oil tapering off with further increased salinity over ECe = 4.0 dS m−1 (Fig. 6b).

Table 3 Tree-scale fruit yield and oil percentage in fruit for 2014 and 2015 harvests
Full size table
Fig. 6
figure 6

a Fruit fresh weight, total of 2 years, 2014 and 2015, and b percent oil in fruit, average of the 2 years, each as a function of time- and profile-averaged saturated paste soil extract electrical conductivity (ECe). Symbols are averages of 3 replicate lysimeters and error bars are standard deviations. Lines are best-fit linear regression in a and exponential increase to maximum in b calculated for the entire data set. Open symbols are treatments with varied irrigation water salinity, closed symbols varied leaching fractions, and the gray symbols common to both (treatment 5, 140)

Full size image

Discussion

Analysis and application

We found that increases in root zone salinity decreased olive tree growth and water use without differentiation between causal agents, whether irrigation water salinity or leaching level (Figs. 2, 5). The fact that time- and depth-averaged ECe was successful in characterizing response curves for all the treatments in the experiment is far from trivial as while the average ECe stemming from the two variables may well be similar, roots may have been expected to experience different salinity regimes and therefore to develop and grow differently. For example, a root zone of high salinity caused by high ECi with high leaching fraction in which the entire root zone is expected to witness a more or less uniform salinity would not necessarily be equivalent to that in a case of moderate ECi and low leaching, where a similar average ECe could possibly be characterized by relatively low ECe in the shallow soil and increasing ECe with depth. Leaching decreased time- and depth-averaged ECe only until the intermediate treatment level of 1.4 times ETa for ECi of 5.0 dS m−1 (Fig. 4b). While minimal leaching caused ECe of 7.5, increasing to 1.25 and 1.40 times ETa decreased the ECe to 5.97 and 4.03, respectively, and 1.60 and 1.80 times leaching resulted only in minor additional decrease in ECe of 3.63 and 3.53 dS m−1, respectively. Simple calculation, assuming a factor of around 1.4 connecting EC of the solution in the soil to saturated paste solution, shows that maximal leaching, creating soil water solution EC equal to ECi, with ECi = 5.0 dS m−1, can reduce ECe only to around 3.5. Therefore, at not much over the leaching treatment of 140 %, additional water was inefficient in further reducing soil salinity. The underperforming of the treatments receiving the highest leaching levels may even suggest that the excess water caused wetter-than-optimal conditions, especially in the final summer of the experiment, when the irrigation and drainage fluxes through the lysimeters, due to relatively large trees and high ET demand, were particularly large. We suspect that the 5 ds m−1 ECi, 160 and 180 % I ETa−1 trees were occasionally exposed to less-than-optimal root zone oxygen conditions introducing an additional stress not considered in the experimental design. It has been previously documented that deficient oxygen caused by excess water is harmful to olives, particularly when combined with exposure to salinity and that oxygen may be important in root-scale mechanisms for salt tolerance (Aragues et al. 2004; Isidoro and Aragues 2006). The topic of root distribution, root growth dynamics, and uptake of salt ions was published in Soda et al. (2016). In that study, the increased exposure to salinity was found to cause reduction in number and length of roots and increased root turnover, with the most drastic effects occurring at the first level of the salt gradient. Soda et al. (2016) credited restricted ion transport from roots to aerial tissues as the chief mechanism for response to salts, coming at a high cost as apparent toxicity occurred in young roots, typified by their decreased growth and increased mortality.

We found high correlation between the effect of salinity on tree size and effect of salinity on water consumption (Fig. 5). Comparing final AGDW or TCSA with accumulated ETa of thee individual lysimeters gives linear correlation with slopes not different from 1 and intercepts not different from 0 (p < 0.01) and coefficients of determination, R 2, of 0.85 and 0.84, respectively. This result joins rich evidence from the literature that, while failing to indicate cause or effect, show linear relationships between (evapo)transpiration and biomass production in crops including grapevines (Shani and Ben-Gal 2005), date palms (Tripler et al. 2007), vegetables (Ben-Gal et al. 2003) and field crops (Hanks 1983; Steduto et al. 2007) affected by abiotic stressors including water and salinity.

Fruit yield response was smaller and less significant than responses of the water use and growth parameters. This is not surprising since the trees were young, in their first two years of bearing, and had vigorous vegetative growth throughout the experiment. It is well known that olive fruit production comes from the previous year’s vegetative growth, and therefore, treatment-related effects for fruit will always be delivered at least a year after vegetative effects. The relatively low significance of fruit response to salinity in the experiment was also strongly due to the fact that in the first year, 2014, the trees receiving the lowest exposure to salt, while having greater vegetative growth than the other treatments (Figs. 2, 5), had fruit yields close to the lowest (Table 3). In the second year, the yields of the 0.5 dS m−1 ECi, 140 % I ETa−1 treatment were the highest measured (Table 3). We are not that surprised that young trees living under luxurious conditions of excess water and no stress, like those of the lowest ECi treatment, would initially invest in vegetative growth and postpone reproductive efforts to future seasons. We suspect that had the experiment and exposure to salinity continued, stronger linear decrease in yield, better following the trends for water use, tree growth, and tree size, would have become apparent. Oil yield will be a function of the fruit yield as well as the percent of oil in the fruit, and while the faster growing, larger trees are anticipated to have more fruit (Ben-Gal et al. 2011), this will be at least somewhat counteracted by the lower oil content. Increasing soil water salinity from ECe 1.8 to around 4.0 dS m−1 led to some 40 % greater oil percentage (Fig. 6b). A more in depth evaluation of the effects of the salinity on fruit and oil characteristics and quality parameters will be the subject of an additional manuscript.

Salinity response functions

Conversion of the results from the experiment to normalized reduction functions due to salinity allows comparison of the findings with those from other studies, locations, varieties, and species, and application in plant response models to be used for predictions, planning, and understanding of soil, and water–crop interactions. A best-fit (using the solver tool in Microsoft Excel) curve according to Eq. (2) for combined total ETa and AGDW was characterized by ECe50 = 4.23 and p = 2.08 (Fig. 7). This generalized reduction-due-to-salinity curve gave no indication of a threshold soil salinity under which there was no negative effect on growth or yields. The responses of water use and tree size (Figs. 5, 7) suggest that a threshold, if existing at all, falls below the lowest ECe level reached in the experiment, 1.2 dS m−1 and a subsequent response of ~15 % per unit dS m−1 increase in ECe. While the slope is similar, due to the lack of threshold this suggests greater sensitivity compared to the data from the few studies of salinity leading to response functions of bearing olive trees found in the literature. Aragues et al. (2005) conducted field observations of densely spaced (1400 trees ha−1) 3–5-year-old ‘Arbequina’ trees over three years where ECe was measured. Their results indicated that the relative effect of salinity increased over time and presented an average three-year response curve characterized by a threshold value of 4.0 dS m−1 and subsequent response slope of ~14 % reduction in yield per 1 dS m−1 increase in ECe based on trunk growth. In a separate publication (Aragues et al. 2004), a threshold of 4.0 dS m−1 ECe and slope of 12 % for every 1 dS m−1 was found as ECe was determined for a single year of monitoring three-year-old ‘Arbequina’ olives. The response curves from the Aragues et al. papers are compatible with Maas and Hoffman’s (Maas and Hoffman 1977) ‘moderately tolerant’ category commonly assumed for olive. Studies in Israel’s Negev Highlands on ‘Barnea’ and other varieties have indicated that irrigation with EC 4.2 dS m−1 water had no influence on vegetative growth, measured as increased trunk circumference (Klein et al. 1994; Klein et al. 1992; Weissbein et al. 2008; Wiesman et al. 2004). In the Wiesman et al. (2004) and Weissbein et al. (2008) studies, vegetative growth was reduced by more than 20 % after at least three years of irrigation with EC 7.5 dS m−1 water. Our data show ECe50 values between 4 and 5 for biomass production or ETa, respectively (Fig. 7). This is much lower, suggesting greater sensitivity, than what would be suggested for ‘Barnea’ and ‘Arbequina’ varieties based on the previously published Spanish and Israeli studies. Comparisons with the prior Israeli studies are particularly problematic due to the challenge of accurately characterizing root zone salinity in the face of the three dimensionality of drip irrigation in the field experiments, the dynamics of changing salinity due to apparent high winter leaching regimes, and their not reporting ECe values.

Fig. 7
figure 7

Relative above-ground dry biomass (treatment average biomass divided by maximum data set biomass; closed squares) and total experimental period evapotranspiration (ETa, open circles) as a function of time- and profile-averaged saturated paste soil extract electrical conductivity (ECe). Symbols are averages of 3 replicate lysimeters, error bars are standard deviations, and the line is best fit to the van Genuchten salinity response function (Eq. 2) equation; for the entire data set including both variables

Full size image

We are hesitant to translate our fruit data into response-to-salinity functions, since the fruit harvest (Table 3; Fig. 6) was likely not yet affected completely by the salinity treatments and is expected with time to better reflect the ETa and biomass curves. There is strong evidence that the most dominant driver for fruit and oil yield on an olive tree is fruit number and that fruit load is strongly related to the tree’s canopy size and previous year’s vegetative growth (Ben-Gal et al. 2011; Bustan et al. 2016). That said, the 2015 and total harvest data also suggest no threshold over 1.2 dS m−1 but a decrease of only around 6 % per unit EC (Fig. 6), leading to a ECe50 for fruit of ~8 dS m−1 for this initial period.

Modeling

Crop water production curves resulting from simulations in the Shani et al. (2007) ANSWER model (Fig. 8) are shown with measured total data for ETa, TCSA, and AGDW. The data are for the most part well represented by the model results. Correlation between predicted and measured values was linear with slopes not significantly different from 1 and intercept not different from 0 and with R2 values of 0.81 for ETa, 0.91 for TCSA, and 0.68 for AGDW. Exceptions of notable model and data incongruence occurred for all three parameters for the 5 ds m−1 ECi, 1.05 leaching treatment, where the model failed to predict the extent of reduction. As has been shown previously in both numeric and analytical modeling, results are particularly sensitive to the empirical reduction -due-to-salinity functions (Oster et al. 2012; Skaggs et al. 2006). Parameterization of the functions will be unique to the choice of the data; in our case whether ETa or TCSA or biomass or some combination was chosen. The choice to use combined ETa and AGDW data to determine the function parameters explains, at least partially, some apparent overestimation of AGDW and underestimation of ETa by ANSWER (Fig. 8). Interestingly, the best-fit results were for TCSA, the parameter not used in calibrating the van Genuchten curve and a popular measure of tree size, changes in which have been shown to often be equivalent to canopy growth (Bustan et al. 2016). The overall success of the model in mimicking the experimental results for the two variables causing root zone salinity is encouraging as it allows evaluation of any other combination of irrigation water EC and leaching strategy and consideration of other locations, conditions, and varieties with different anticipated salinity response functions. That said, the success may partially be due to the matching of the model’s assumption of steady-state conditions and the pseudo-steady state created in the lysimeters by definition of the treatments. The results should be evaluated against actual field conditions where the assumption of steady state may have greater repercussions.

Fig. 8
figure 8

Water production functions calculated by the Shani et al. (2007) ANSWER model for the irrigation water salinity levels evaluated experimentally. Symbols are measured values of trunk cross-sectional area (TCSA, squares), total evapotranspiration (ETa, triangles), and total accumulated above-ground dry weight of biomass (AGDW, circles). Error bars are standard deviations. Lines are simulated results using input values found in Table 2. Line style and color of lines and symbols correspond to given irrigation water salinity levels (color figure online)

Full size image

The current study brings novel information on and benefits of understanding of irrigation of perennial crops with low-quality water by quantifying responses of tree growth, yield, and water consumption using the unique lysimeter-based growing platform. Of course, roots and root growth, ion uptake and accumulation, physiological tree- and leaf-based behavior, and fruit and oil quality parameters are all likely to be influenced by the exposure to salinity. Each of these topics should therefore be addressed in additional studies and manuscripts.

Conclusions

In a lysimeter study evaluating root zone salinity and its effect on olive tree water use, growth, and production, average root zone salinity was found to reduce all of the measured parameters. Tree-scale response to increased salinity did not demonstrate any sign of a threshold value and was not differentiated by the cause of salinity, be it changes in input irrigation water salt concentrations or changes in leaching fraction. Soil salinity, measured as ECe and maintained at stable levels over time, decreased tree water consumption and tree size measured as trunk area or above-ground biomass by 40–60 % as it increased from 1.2 to around 3.5–4.0 dS m−1. Further increases in ECe brought the ETa and biomass production parameters to 20–30 % of the treatment with low salinity. Fruit yield also decreased with increasing salinity, albeit less drastically, with the differences likely due to the relative young age of the trees and complex interactions between vegetative and reproductive growth in olives. Fitted parameters for the van Genuchten salinity reduction curve (van Genuchten and Gupta 1993), ECe50 = 4.23 and p = 2.08, were used to successfully simulate the results using the analytical model ANSWER (Shani et al. 2007).