You are on page 1of 12

Agricultural Water Management 245 (2021) 106577

Contents lists available at ScienceDirect

Agricultural Water Management


journal homepage: www.elsevier.com/locate/agwat

Evaluation of regulated deficit irrigation strategies for oil olives: A case


study for two modern Israeli cultivars
Alon Ben-Gal a, *, Yonatan Ron a, Uri Yermiyahu b, Isaac Zipori c, Sireen Naoum d, Arnon Dag c
a
Environmental Physics and Irrigation, Institute of Soil, Water and Environmental Sciences, Agricultural Research Organization – Volcani, Gilat Research Center, Israel
b
Soil Chemistry, Plant Nutrition and Microbiology, Institute of Soil, Water and Environmental Sciences, Agricultural Research Organization – Volcani, Gilat Research
Center, Israel
c
Fruit Tree Sciences, Institute of Plant Sciences, Agricultural Research Organization – Volcani, Gilat Research Center, Israel
d
National Agricultural Research Center, Jordan

A R T I C L E I N F O A B S T R A C T

Handling Editor - J.E. Fernández Profits from olive oil production are expected to benefit from regulated deficit irrigation (RDI) regimes designed
to optimize yield and oil quality with minimal inputs of water. We evaluated a number of potential strategies for
Keywords: implementing RDI including choosing phenological periods for reduced irrigation, use of midday stem water
RDI potential (SWP) for scheduling, considering fruit load, and irrigation to reach optimal fruit water content at
Olea europaea
harvest. We base our discussion on the potentials and challenges of practical commercial RDI regimes on results
Stem water potential
from a 6-year orchard experiment on two olive oil cultivars used in intensive orchards (’Barnea’ and ’Askal’) in
Pit hardening
Fruit load Israel, focusing on attempts for long-tern optimization of yield and oil quality. Results showed a general trend of
Irrigation scheduling increased water productivity as a function of reduced irrigation but since yields tended to be reduced when
irrigation was decreased, there was not a clear benefit of RDI over sustained deficit irrigation. The exception to
this was a treatment based on SWP where water application was reduced by some 40% compared to irrigation
regimes bringing highest yields, and a non-significant accumulated yield reduction of less than 15% after 6 years
in both cultivars. Oil quality, measured in terms of free fatty acids and polyphenol content, was often improved
when irrigation was decreased, with no dependence on the seasonal timing of deficit amounts and increased tree
water stress.

1. Introduction In previous work it was found that a sustained 70–80% supply of po­
tential crop evapotranspiration (ETp), throughout the whole irrigation
Irrigated olives (Olea europaea L.) often are found to yield 3–4 times season, optimized yield and quality over time in a mature olive (cv
more than rain fed orchards in typical traditional growing areas (Lavee Barnea and Souri) orchard in Israel (Ben-Gal et al., 2011a, 2011b).
et al., 1990). Irrigation by itself obviously reduces water stress and Olives are biennial bearers, with typical 2-year cycles with a high
subsequently increases tree and orchard productivity (Ben-Gal et al., fruit load following a season with low fruit load (Lavee, 2007). In Israel,
2011b), and increases the tree’s capacity for mineral nutrient uptake such cycles are particularly drastic and, depending on actual conditions
(Zipori et al., 2015). Irrigation additionally allows movement of olive and cultivar, yields in "On" years can be 2–3 times those of "Off" years
production into dry areas, and intensification of orchard management (Dag et al., 2009). Response to irrigation in such cases was found to be
including nutrient application via the irrigation system, introduction of sensitive to fruit load, with high yielding years effected more strongly by
vigorous high yielding varieties bred to respond to water, densely the same relative water stress (Ben-Gal et al., 2011b; Naor et al., 2012).
spaced trees, and mechanized harvesting. At least part of the explanation to such lies in the high need for assim­
Because of the importance of oil quality and a tendency for quality ilates from photosynthesis when fruit is a prominent sink and a tree’s
parameters to benefit from conditions causing water stress to the trees failure to trigger and process water saving mechanisms, like stomatal
(Ben-Gal et al., 2011a; Dag et al., 2008, 2015), optimization of irrigation closure, which otherwise would be activated under conditions of
of oil olives depends on maintaining at least some level of water deficit. limiting water (Bustan et al., 2011). In a controlled lysimeter study, trees

* Corresponding author.
E-mail address: bengal@volcani.agri.gov.il (A. Ben-Gal).

https://doi.org/10.1016/j.agwat.2020.106577
Received 11 April 2020; Received in revised form 23 August 2020; Accepted 3 October 2020
Available online 26 October 2020
0378-3774/© 2020 Elsevier B.V. All rights reserved.
A. Ben-Gal et al. Agricultural Water Management 245 (2021) 106577

with medium to high fruit yield were shown to consume 20–30% more Table 1
water on a per canopy area basis compared to trees with little to no fruit Annual fruit load status, winter rainfall, dates of irrigation season and dates of
(Bustan et al., 2016). harvest.
The stress, measured as decreasing stem water potential (SWP) was Bearing Raina Irrigation Harvest
found to be consistently less in the first months of irrigation each year, Year ’Askal’ ’Barnea’ (mm) Start Stop ’Askal’ ’Barnea
following winter rains, and increased (SWP becoming more negative)
2012 Off On 1-Julb 7-Nov 24-Oct 10-Dec
with time over the season until olive fruits were harvested. SWP in "On"
2013 On On 616.5 31-Mar 14-Nov 28-Oct 6-Nov
years was typically 20–30% lower than in "Off" years. SWP in May in the 2014 On On 241.8 6-Apr 12-Nov 30-Sep 8-Dec
Israeli studies was around − 0.8 MPa in "Off" years and − 1.1 MPa in 2015 On Off 594.8 3-May 7-Oct 20-Sep 14-Oct
"On" years. In July the values decreased to − 1.7 MPa and − 2.3 MPa and 2016 Off On 548.8 4-May 29-Nov 15-Sep 10-Nov
in September the SWP readings were around − 2.3 and − 3.0 in "Off" and 2017 On Off 271.4 23-Apr 15-Nov 17-Sep 1-Nov
2018 Off On 303.9 15-Apr 4-Nov 03-Sep 12-Nov
"On" years, respectively (Ben-Gal et al., 2011b).
a
Regulated deficit irrigation (RDI) exploits phenological stages in September previous year until April current year.
b
annual tree physiology allowing much lower water application than Beginning of irrigation treatments.
supplying full crop evapotranspiration demand during times when the
stress has no or minimal effect on yields (Fereres and Soriano, 2006). In foothills of the Judean Hills in Israel. The soil of the orchard was a clay
olive, fruit pit hardening, occurring over a 4–5 week period in the vertisol (sand 20%, clay 65%). The experimental site has a typical
summer, is believed to be such a stage where water stress has little or no Mediterranean climate with hot dry summers and mild winters with
influence on production over time (Gucci et al., 2009). annual precipitation averaging ~500 mm falling exclusively from
Olive irrigation with RDI strategies has been reported previously, October until April and concentrated over November–March. Irrigation
particularly in Spain and Italy. Fernández et al. (2013) and Gómez-del- began each season 3–4 weeks following the last significant (more than
Campo (2013) each evaluated RDI strategies on super high density 30 mm) rainfall event in the spring and ended following rainfall of at
’Arbequina’ olives in Spain and both concluded that moderate DI during least 40 mm within a week in the autumn. Annual rainfall, irrigation
July and August (end of pit hardening) were best for water savings. In seasonal start and finish dates and seasonal amounts are given in
this way, Fernández et al. (2013) showed water savings of 72% and yield Table 1. Patterns of rainfall and reference ET (ET0) are illustrated in
decreases of 26% while Gómez-del-Campo (2013) saved 27% in water Fig. 1.
with a 15% decrease in oil production. Gucci et al. (2019) evaluated RDI Actual daily irrigation (I) was computed by:
before or during harvesting on ’Frantoio’ olives in Italy and did not find
IrrLev
advantages to either over "full" irrigation as both reduced water use and I = ETp ∗ ; ETp = ET0 ∗ fc (1)
100
yield equivalently. From the literature it seems therefore that deficit
irrigation (DI) strategies, while succeeding to reduce water applied as where ETp is potential evapotranspiration, IrrLev is the level of irriga­
irrigation, are generally accompanied by yield reductions, often are tion for the different treatments (%), and fc is a cover factor estimated by
beneficial to oil quality, and only sometimes increase water productiv­ midday shaded area. The IrrLev fraction, found in Fig. 2 for the various
ity. Similar results were found in attempts to schedule RDI based on treatments, is equivalent to a crop factor (Kc) as found in FAO crop water
midday stem water potential (SWP) in experiments conducted in two consumption methods (Allen et al., 1998). The cover factor, fc, was
locations in Spain on ’Cornicabra’ and ’Morisca’ olives (Moriana et al., estimated as 40% in 2012 and 60% from 2013 to 2018. Frequency of
2012). Evaluation of these studies must consider that the "full" irrigation irrigation events was twice weekly on Sundays and Wednesdays. Irri­
regimes, used to compare the RDI strategies vary from location to gation was via a single line of pressure compensated drippers (UniRam,
location. None of these studies considered or adjusted for fruit load. Netafim, Hatzerim, Israel) with flow rate of 1.6 L/h spaced every 50 cm.
We hypothesized that deficit irrigation, inducing stress in olives Statistical design was a randomized block with five replicates per
during fruit pit hardening, would increase water productivity (yield per treatment. The two cultivars were in separate, adjacent fields. Each of
applied irrigation volume) without significantly decreasing oil yield or the 25 experimental units consisted of three adjacent tree rows with at
quality. We additionally hypothesized that shifting timing and extent of least 5 olive trees per row. Two of the three central trees of each cultivar
deficit irrigation according to expected needs as a function of fruit load were monitored while those surrounding them served as border-guard
would add to the benefit. Based on these hypotheses, our objective was trees. The orchard was irrigated with effluent that, originating as
to significantly reduce water application and increase water produc­ municipal wastewater in the city of Jerusalem, underwent secondary,
tivity and profit without corresponding yield reductions and without
compromising oil quality. This was evaluated by treatments including
pre-scheduled RDI regimes and a regime where irrigation was applied
according to set points based on stem water potential measurements.
We present results of a 6-year field experiment conducted in Israel
and discuss them in terms of: Choice of periods for RDI; Success of RDI over
sustained DI; Use of SWP for scheduling; Choice and use of thresholds; fruit
load considerations; Irrigation to reach optimal fruit water content at harvest
and; Optimizing yield and oil quality.

2. Materials and methods

2.1. Orchard and general management description

The research was conducted on ’Barnea’ and ’Askal’ olive cultivars


in a two hectare section of a 4 year old (at onset of treatments) com­
mercial olive orchard with general upkeep and horticultural activities in
accordance with accepted local commercial practice. Tree density was 7 Fig. 1. Reference evapotranspiration ET0, shown as daily, 7-day running
m between rows and 4 m between trees in the rows (350 trees/ha). The averaged values and daily rainfall over the experimental period. X-axis tick
orchard was located (31◦ 78′ N, 34◦ 82′ E) adjacent to Revadim in the marks indicate 1 January for each year.

2
A. Ben-Gal et al. Agricultural Water Management 245 (2021) 106577

MARCH APRIL MAY JUNE JULY AUGUST SEPTEMBER OCTOBER NOVEMBER DECEMBER

Beginning Beginning Bloom and Pit Start of colour End of Start of colour
of shoot of fruit set hardening change in OFF fruit Harvest
change in ON
growth irrigaon years growth years

T1 80% of ETp throughout the whole irrigation season

T2 80% of ETp 50% of ETp 80% of ETp Adapt irrigation to FWC

ON 80% of ETp 50% of ETp 80% of ETp Adapt irrigation to FWC


T3
OFF 80% of ETp 50% of ETp Adapt irrigation to FW C

ON 40% of ETp 80% of ETp Adapt irrigation to FWC


T4
OFF 80% of ETp 40% of ETp Adapt irrigation to FW C

ON Minimum SWP = -1.5 Minimum SWP = -3.0 Minimum SWP = -2.5 Adapt irrigation to FWC
T5
OFF Minimum SWP = -2.5 Minimum SWP = -3.5 Adapt irrigation to FWC

Fig. 2. Schematic of olive irrigation annual cycle and treatments. ETp = potential crop ET = reference ET X cover factor. FWC = fruit water content.
SWP = midday stem water potential (MPa).

activated sludge treatment and was stored in a nearby reservoir. Ni­ water content is thought to be optimum for commercial mill operation
trogen and potassium fertilizer was given with irrigation water as rec­ (Zipori et al., 2016). Following harvest, irrigation continued at a rate of
ommended by the local extension service equally to all treatments 40% ETp until the first significant (> 40 mm) rainfall event.
according to area-based calculations including consideration of nutri­
ents originating in the recycled wastewater (Erel et al., 2019). Season­
2.3. Measurements, sampling and analyses
ally, a total of 200 kg nitrogen, 22 kg phosphorous, and 205 kg
potassium, including that originating in the water and that added as
Plant water status was assessed by SWP using a Scholander-type
fertilizer were applied per hectare. The effluent was low in salts and had
pressure chamber (MRC, Israel) according to Shackel et al. (1997). In
no other characteristic expected to influence plant performance. Typical
treatments T1–T4, SWP was measured three times a year, early (May),
characteristics of the irrigation water have been previously published
mid-season during pit hardening (July) and late (August–September).
(Erel et al., 2019; Zipori et al., 2015).
Actual dates of SWP measurements were 28 May, 23 July and 01 Sept
2013; 13 May, 22 July and 26 August 2014; 26 May, 07 July and 11 Aug
2.2. Irrigation treatments 2015; 31 May, 02 Aug and 30 Aug 2016; 16 May, 11 July and 07 Sept
2017; and 06 June, 11 July and 29 Aug 2018. Vegetative growth rate
We applied a number of configurations of RDI treatments on two was measured by annual measurement of trunk circumference at a
olive cultivars (’Barnea’ and ’Askal’) over 6 and 1/2 years (initial season marked point 40 cm above the ground on each tree each March.
plus 3 potential biennial bearing cycles). Both are relatively new Israeli Beginning each season in August, fruit water content was determined
cultivars bred for intensive, irrigated cultivation (Lavee et al., 2003). weekly before harvest by sampling 10 representative fruits from around
The treatments, outlined in Fig. 2, included: T1, a control with sustained the circumference of each measured tree at 2–2.5 m height and weigh­
regulated irrigation of 80% calculated crop ET; T2, RDI with irrigation ing before and after drying in an oven at 70 ºC.
reduced to 50% ETp during the period leading up to and into pit hard­ Yield was determined for each monitored tree. A representative
ening (June–July), with no consideration of fruit yield; T3, RDI with sample of 2 kg of fruit from each tree was analyzed. Cold-pressed virgin
treatment identical to T2 in "On" years and the deficit period shifted to oil was obtained with an Abencor system (MC2, Ingenieria y Sistemas,
pit hardening and after in "Off" years (July–August); T4 RDI treatment Seville, Spain) as described by Ben-David et al. (2010). Oil and water
with 40% supply of ETp in the first half of the irrigation season percentage of harvested fruit was measured in olive paste by near
(beginning of annual irrigation through July) in "On" years and 40% infrared (NIR) measurement (Naor et al., 2012; Zipori et al., 2016) using
supply during pit hardening (July–August) in "Off" years; and T5, a RDI OliveScan (Foss, Denmark) equipment calibrated with ~3000 samples
treatment based on stem water potential thresholds. Each treatment for with Soxhlet (Zipori et al., 2016). Water productivity was calculated as
each cultivar was irrigated via an independently automated valve and kg oil per cubic meter of water irrigated. Determination of free fatty
main line including water meter and fertilizer pump and meter. Pit acids (FFA) was carried out following analytical methods described in
hardening beginning and end were determined by manual observations ISO (International Organization for Standardization) 660. FFA was
following first order estimations based on historical average data expressed as percent oleic acid. Phenolic compounds were isolated from
(Fig. 2). The target SWP levels for T5 were adapted from results from a a solution of oil in hexane by double extraction with methanol/water
previous experiment (Ben-Gal et al., 2011b). In "On" years target SWP (60:40 vol/vol). Total polyphenols (as Tyrosol equivalent) were deter­
was − 1.5 MPa from initiation of irrigation through May, − 3.0 MPa mined with a UV–visible spectrophotometer (Beckman Coulter, Inc.
from May through July, and − 2.5 MPa in August–September. In "Off" Fullerton, CA, USA) at 735 nm using Folin-Ciocalteu reagent (Swain and
years the SWP target was − 2.5 MPa until the end of June and Hillis, 1959).
− 3.5 MPa during July–August. SWP in T5 was measured twice a week, Data were analyzed using JMP 14 software, (SAS Institute, Cary,
once the day before and once the day after an irrigation event and target NC). SigmaPlot 13 (Systat Software, San Jose, CA) was used to prepare
values were set to be averages of these measurements. Determination of figures. Fruit yield, fruit oil and water content, and free fatty acids and
whether each cultivar was in an "On" or "Off" year was made following polyphenols in oil were analyzed using two-way (treatment and year)
fruit set each year based on visual observations (Table 1). All treatments ANOVA (supplemental Table S1). Annual results of fruit yield, fruit oil
other than the control were irrigated at the end of the season, from the and water content, oil quality parameters, trunk circumference and total
first week of September until harvest, according to fruit water content, change in circumference, total fruit and oil yield, midday stem water
aiming for 50–55% water in fruit arriving to the olive mill. This fruit potential for three periods each year, for "On" and "Off" years separately

3
A. Ben-Gal et al. Agricultural Water Management 245 (2021) 106577

and for all years, and annual and total water productivity, were analyzed Table 2
with one-way ANOVA to determine significance of treatments. Accumulated trunk growth measured as the difference in circumference from
Tukey–Kramer was used for means separation. All statistical analyzes 2013 until 2019 and accumulated fruit oil yield as a sum of yields in years from
were conducted at P < 0.05. 2013 until 2018 per cultivar and treatment. Oil yield is total oil determined by
multiplying fruit yield with percent oil measured by NIR, commercial oil mill
yields are expected to be around 0.85 of this. Shown are average (n = 10) re­
3. Results
sults. No significant differences were found for combinations of cultivar and
treatment from analysis of variation.
Annual status of bearing, winter precipitation, irrigation season
Cultivar Treatment Total Change in trunk Total Total
dates, and harvest dates are shown in Table 1. For both cultivars, ob­
irrigation circumference fruit oil
servations of fruit load status indicated more "On" compared to "Off" applied (cm) yield yield
years. ’Barnea’ trees were determined to be "On" in three consecutive (mm) (kg/ (t/ha)
years from 2012 to 2014 and ’Askal’ to be "On" from 2013 to 2015. In tree)
each of the last 4 years of the experiment, the cultivars had opposite Barnea 1 3243 35.8 305.7 20.3
bearing level status (Table 1). The beginning and end of irrigation sea­ 2 2765 35.4 283.6 20.2
sons were determined by winter rainfall, particularly first and last 3 2689 37.2 267.4 18.8
4 2454 33.6 244.2 17.8
effective (more than 30 mm) rain events (Fig. 1) with irrigation starting
5 1732 32.7 265.0 19.9
between the end of March and beginning of May and ending between the Askal 1 2955 34.3 205.6 16.6
beginning of October and end of November (Table 1). Annual irrigation 2 2548 34.5 204.5 17.0
amounts for the control treatments T1 ranged from 446 to 557 mm for 3 2501 31.8 191.8 15.0
’Barnea’ and 417–493 for ’Askal’ (Fig. 3). Since ’Askal’ trees are har­ 4 2212 33.5 187.0 15.4
5 1735 30.9 190.9 16.3
vested early in order to avoid problematic levels of free fatty acids in the
oil, their irrigation continued for quite some time following harvest.
Harvest of ’Barnea’ trees more typically coincided with first substantial
winter rains and the end of seasonal irrigation. 80
Deficit treatments applied less water. The least amounts of water (A) Barnea
were consistently given to T5, the SWP triggered treatment, with annual
amounts ranging from 96 to 336 mm (’Barnea’) and 199–331 mm
(’Askal’). Total, 7 year, irrigation application amounts (Table 2) for T1
60
reached 3243 mm for ’Barnea’ and 2955 mm for ’Askal’. The deficit
treatments in ’Barnea’ reduced this total water applied by 15% (T2),
17% (T3), 24% (T4) and 47% (T5). For ’Askal’ the reductions were 14% 40
Trunk circumference (cm)

(T2), 15% (T3), 25% (T4) and 41% (T5).

600 20
(A) Barnea
500

80
0
400
(B) Askal
T1
T2
300 T3
60 T4
Annual irrigation applied (mm)

T5
200

40
100

0
20
500 (B) Askal
T1
T2
400 T3
T4 0
1 2 0 1 3 0 14 0 1 5 0 1 6 0 1 7 0 1 8 0 1 9
T5

300 20 2 2 2 2 2 2 2
Fig. 4. Trunk circumference measured in March of each year as a function of
200 deficit irrigation treatment. Average for treatment, n = 10. (A) Barnea, (B)
Askal. According to one-way analysis of variation, there were no significant
differences in any year between treatments.
100
3.1. Vegetative growth
0
12 13 14 1 5 16 17 18 Trunk size increased each year with larger increases in "Off" years
20 20 20 20 20 20 20 and smaller increases in "On" years (Fig. 4). Average trunk circumfer­
ence increased in ’Barnea’ trees by 3.3 cm in "Off" years and 6.3 cm in
Fig. 3. Annual total applied irrigation per treatment and cultivar. (A) Barnea,
(B) Askal. In 2012 irrigation treatments began in July.
"On" years. For ’Askal’ trees the increase was 3.3 and 6.1 cm in "Off" and

4
A. Ben-Gal et al. Agricultural Water Management 245 (2021) 106577

Table 3
Midday stem water potential (MPa) measured in early (May) midseason during pit hardening (July) and late (August–September) each year and averages of "On" and
"Off" years 2013–2018 for n = 10 trees per treatment. Different small case letters represent significance for each row, reflecting combinations of date and cultivar.
Barnea

Treatment T1 T2 T3 T4 T5

2013 ON May − 1.39 ab − 1.28 bc − 1.23 c − 1.46 a − 1.33 bc


July − 2.24 c − 2.43 abc − 2.3 bc − 2.52 ab − 2.59 a
September − 2.73 b − 2.45 b − 2.41 b − 2.48 b − 3.16 a
May − 1.61 a − 1.72 a − 1.61 a − 1.87 a − 1.77 a
2014 ON July − 3.04 ab − 3.12 ab − 2.81 b − 3.30 a − 3.47 a
September − 2.88 bc − 3.17 ab − 2.93 bc − 3.28 a − 2.69 c
May − 1.76 b − 1.75 b − 1.77 b − 1.82 b − 2.29 a
2015 OFF July − 1.94 c − 2.34 b − 2.08 bc − 2.38 b − 3.09 a
September − 2.40 d − 2.61 cd − 2.82 bc − 3.26 a − 3.07 ab
May − 1.90 b − 1.88 b − 1.85 b − 1.77 b − 2.85 a
2016 ON July − 2.54 a − 3.22 a − 2.57 a − 3.30 a − 3.42 a
September − 2.42 a − 2.66 a − 2.54 a − 2.57 a − 2.64 a
May − 2.12 b − 2.09 b − 2.05 b − 2.20 b − 2.90 a
2017 OFF July − 2.76 c − 3.15 b − 2.98 bc − 3.19 b − 3.69 a
September − 2.53 b − 2.39 b − 3.13 a − 2.92 a − 2.86 a
May − 2.48 c − 2.68 bc − 2.71 b − 3.18 a − 2.58 bc
2018 ON July − 3.01 c − 3.67 b − 3.54 b − 4.01 a − 4.26 a
September − 3.07 b − 3.29 ab − 3.18 b − 3.20 b − 3.46 a
May − 1.85 b − 1.89 b − 1.86 b − 2.07 a − 2.13 a
Average ON July − 2.71 c − 3.12 ab − 2.80 bc − 3.28 a − 3.43 a
September − 2.78 b − 2.89 ab − 2.77 b − 2.89 ab − 2.99 a
May − 1.92 b − 1.92 b − 1.91 b − 2.01 b − 2.56 a
Average OFF July − 2.35 c − 2.75 b − 2.58 bc − 2.78 b − 3.39 a
September − 2.46 b − 2.5 b − 3.00 a − 3.09 a − 2.97 a
May − 1.87 c − 1.90 c − 1.88 c − 2.05 b − 2.27 a
Average all July − 2.59 d − 2.99 bc − 2.73 cd − 3.12 b − 3.42 a
September − 2.67 b − 2.76 b − 2.85 ab − 2.96 a − 2.98 a

Askal

Treatment T1 T2 T3 T4 T5

2013 ON May − 1.28 b − 1.22 b − 1.26 b − 1.44 a − 1.35 ab


July − 2.97 c − 3.53 ab − 3.30 bc − 3.56 ab − 3.72 a
September − 3.67 a − 3.86 a − 3.68 a − 3.65 a − 3.79 a
May − 1.68 a − 1.51 b − 1.63 ab − 1.70 a − 1.63 ab
2014 ON July − 3.33 ab − 3.18 b − 3.07 b − 3.31 ab − 3.91 a
September − 2.76 a − 3.11 a − 2.99 a − 3.17 a − 2.85 a
May − 1.77 b − 1.72 b − 1.67 b − 2.06 a − 2.06 a
2015 ON July − 2.41 b − 2.74 b − 2.54 b − 2.77 b − 3.53 a
September − 3.09 b − 3.27 ab − 2.90 b − 3.06 b − 3.55 a
May − 1.68 b − 1.74 b − 1.67 b − 1.73 b − 2.73 a
2016 OFF July − 2.83 b − 3.82 a − 3.10 ab − 3.24 ab − 3.81 a
September − 2.85 b − 2.98 b − 3.15 b − 3.31 ab − 4.04 a
May − 2.02 b − 2.18 b − 2.08 b − 2.50 a − 2.00 b
2017 ON July − 2.89 b − 3.40 a − 3.32 a − 3.50 a − 3.68 a
September − 3.19 a − 2.98 ab − 2.92 ab − 2.70 b − 2.78 ab
May − 2.22 bc − 2.58 ab − 2.30 bc − 2.05 c − 2.82 a
2018 OFF July − 3.03 d − 3.58 b − 3.47 bc − 3.15 cd − 4.28 a
September − 2.97 c − 2.87 c − 3.54 b − 3.59 b − 4.28 a
May − 1.69 b − 1.66 b − 1.66 b − 1.93 a − 1.75 b
Average ON July − 3.00 c − 3.21 bc − 3.06 bc − 3.26 b − 3.71 a
September − 3.18 a − 3.29 a − 3.12 a − 3.14 a − 3.24 a
May − 1.95 b − 2.16 b − 1.99 b − 1.89 b − 2.78 a
Average OFF July − 2.93 c − 3.70 ab − 3.32 bc − 3.22 c − 4.07 a
September − 2.91 c − 2.89 c − 3.35 bc − 3.46 b − 4.16 a
May − 1.73 b − 1.76 b − 1.71 b − 1.80 b − 2.12 a
Average all July − 2.91 c − 3.37 b − 3.13 bc − 3.25 b − 3.83 a
September − 3.09 b − 3.17 b − 3.20 b − 3.25 b − 3.54 a

"On" years, respectively. No effect of deficit irrigation treatment was 3.2. Stem water potential
found on trunk size for the first 4 years of the experiment. A non-
significant tendency for reduced growth as a function of reduced irri­ Stem water potential was highest (less negative) early in the season
gation water application appeared thereafter. This tendency was stron­ and lower during pit hardening and before harvest in the late summer
ger in ’Barnea’ than ’Askal’ trees. Between 2012 and the end of the (Table 3). Average (for all treatments) SWP measured for ’Barnea’ trees
experiment in the beginning of 2019 trunk circumference increased by in May was − 2 MPa, in July − 3 MPa and in August–September
up to 37.2 cm in ’Barnea’ and 34.5 cm in ’Askal’ (Table 2). While the − 2.8 MPa. For ’Askal’ trees the average SWP was − 1.82 MPa in May,
differences to other treatments were not significant, the lowest rates of − 3.29 MPa in July and − 3.25 MPa in August–September. Autumn and
trunk growth were found for the T5 treatments with 32.7 cm for ’Bar­ post-harvest found increase in SWP (data not shown). Treatments of RDI
nea’ and 30.9 cm for ’Askal’. were less evident in the early season and became more significant in
July. SWP measured during pit hardening periods each year was

5
A. Ben-Gal et al. Agricultural Water Management 245 (2021) 106577

consistently highest for T1, intermediate for T2 and T3, and T4, and higher for T1 compared to T2, T4 and T5 while T2 was not significantly
lowest for T5 that sometimes reached values as low as ~− 4 MPa. Dif­ different than any of the other treatments. The ’Askal’ fruit was often
ferences between T1 and T5 for ’Barnea’ averaged 1 MPa with T1 at below the target level of 50–55% water at harvest. Evaluation of each
− 2.6 MPa and T5 at − 3.4 MPa (Table 3). Equivalent data from ’Askal’ year reveals that fruit water at harvest was consistently highest in T1
trees was − 2.9 MPa for T1 and − 3.8 MPa for T5. No significant dif­ and lowest in T5 with the other treatments falling in between, but this
ferences were found between "On" and "Off" years regarding the trends was not always significant (Table 4). ’Askal’ fruit had significantly lower
of measured SWP and no significant effect of T3, which considered water content in T5 compared to T1 in 2012 and in T5 compared to T1,
bearing level in its RDI strategy was found compared to T2 which did T2 and T3 in 2016. The other treatments were not significantly different
not. Treatment T4, with a more drastic deficit irrigation level, often had than either T1 or T5 in those years. For ’Barnea’ fruit the trend of lower
significantly lower SWP in July (’Barnea’) and July and September water content at harvest as a function of lower irrigation application
(’Askal’). level was significant in 2015, 2016 and 2017.

3.3. Treatment T5 3.5. Yields

Stem water potential patterns and target values are shown for the Fruit yields of ’Barnea’ trees (Fig. 6) ranged from ~60 kg/tree in
three final seasons (2016–2018) of the experiment (Fig. 5) Data from all "On" years 2014, 2016 and 2018 to ~30 kg/tree in "Off" years 2013,
6 year of the experiment can be found in supplementary material 2015 and 2017. Two-way ANOVA for ’Barnea’ fresh fruit yield indicated
(Fig. S1). Actual values followed the target values (Fig. 2) with varia­ significant (P < 0.0001) differences between years but not for treat­
tions of between 1 and 5 MPa above the thresholds on days after irri­ ments and with no interaction (supplementary Table S1). Fruit yields of
gation and below the threshold on days prior to irrigation. Early in the ’Askal’ trees (Fig. 6) were also significantly influenced by only year
season each year, trees in "Off" years were not irrigated as they took time (P < 0.0001) with no interaction between year and treatment. Yields for
to reach their target SWP values. Also in interim periods when SWP ’Askal’ were greater (~40 kg/tree) in the "On" years 2013, 2015 and
targets were lowered, the trees had periods when they were not 2017 and lower (~20–25 kg/tree) in the "Off" years 2014, 2016 and
irrigated. 2018. Note that the bearing status was miss-diagnosed at the beginning
of the season in 2013 for ’Barnea’ and in 2014 for ’Askal’ when they
3.4. Fruit water content were characterized as "On" in spite of ultimately having relatively low
yields.
Two-way ANOVA (supplemental material, Table S1) of all the fruit Over the course of the experiment, non-significant, non-consistent
water content at harvest data (Table 4) indicated significance of both trends of reduced yield as a function of reduced water application due to
year and treatment (P < 0.0001) for ’Barnea’ and of year but not RDI strategies were found (Fig. 6). Average ’Barnea’ fruit yield ranged
treatment for ’Askal’. There was significant interaction between year from 20 to 40 kg/tree (7–14 t/ha) in "Off" years to 40–75 kg/tree
and treatment for both cultivars (’Barnea’, P < 0.001, ’Askal’ (14–26 t/ha) in "On" years. Yield of ’Askal’ tended to be lower;
P = 0.0035). ’Barnea’ fruit, which was harvested later in each season 10–30 kg/tree (3.5–10.5 t/ha) in "Off" years and 30–45 kg/tree
than the ’Askal’ fruit, reached the oil mill with relatively high per­ (10.5–16 t/ha) in "On" years (Fig. 6). Cumulative fruit yield (Table 2),
centage of water, often higher than the targeted 50–55% (Table 4). summed for the last 6 years of the experiment, from 2013 until 2018,
Average fruit water content at harvest for ’Barnea’ was significantly ranged from 306 kg/tree for T1 in ’Barnea’ and 205–206 kg/tree for T1

Fig. 5. Midday stem water potential (SWP, target and average measured) dynamics for 3 years (2016–2018) for treatment T5. In these years, the ’Barnea’ trees were
determined to be in "On" years in 2016 and 2018 and "Off" in 2017 while the ’Askal’ trees were in opposite bearing stages. Dotted lines are target values while
connected symbols show actual measured values on two days weekly, one prior to and the second following irrigation events.

Table 4
Average water content (%) of olive fruit at harvest each season. Determined by NIR for paste in olive mill. Average of N = 10 trees. Different letters indicate significant
differences among treatments for each cultivar and year.
Treatment 2012 2013 2014 2015 2016 2017 2018

Barnea T1 60.0 53.3 58.6 56.8 a 57.3 a 54.6 a 58.0


T2 60.6 55.3 57.9 55.8 ab 50.6 c 50.5 b 56.2
T3 61.7 55.6 59.2 54.3 ab 52.5 bc 49.8 b 56.7
T4 61.1 54.9 58.0 53.7 ab 54.8 ab 47.2 b 54.9
T5 60.1 51.8 59.1 53.7 b 54.7 ab 47.3 b 55.2
Askal T1 51.7 b 48.0 50.3 45.9 44.4 a 48.0 44.6
T2 52.6 ab 50.3 50.4 45.5 44.3 a 46.4 48.4
T3 52.3 ab 49.8 49.7 47.7 44.1 a 47.4 48.0
T4 53.0 ab 48.6 49.9 47.2 42.9 ab 48.2 45.3
T5 54.1 a 50.3 49.5 46.3 40.3 b 45.1 51.8

6
A. Ben-Gal et al. Agricultural Water Management 245 (2021) 106577

35
(A) Barnea

30
T1
T2
T3
T4
T5
25

20

Fruit oil content (% fresh weight)


15

10

(B) Askal
30

25

20

15

10
12 13 14 15 16 17 18 ag
e
20 20 20 20 20 20 20 er
av
Fig. 6. Annual fruit yield as a function of irrigation treatment. (A) Barnea, (B)
Fig. 7. Fruit oil content (% of fresh weight) by NIR for paste in olive mill at
Askal. According to analysis of variance, no significant differences were found
harvest for each year per treatment and as average of all years. (A) Barnea, (B)
between treatments in any of the years of the experiment.
Askal. Different letters indicate significant differences between treatments in
each year.
and T2 in ’Askal’ to 244 kg/tree for ’Barnea’ T4 and 265 kg/tree for T5
and 192, 187 and 191 kg/tree for T3, T4 and T5, respectively for ’Askal’.
3.6. Oil quality

3.5.1. Fruit oil content


3.6.1. Free fatty acids
The general two-way ANOVA model indicated significance between
Two-way ANOVA indicated significance of year (P < 0.0001) but not
years (P < 0.0001) but not treatments for oil content in fruit for both
treatment and no interaction between them for free fatty acid in oil for
cultivars and no interaction between treatment and year (Fig. 7; Sup­
both cultivars (Supplementary Table S1). No significant differences were
plementary Table S1). Oil content in fruit as percentage of fresh weight
found in oil free fatty acid percentage as a function of treatment in any of
tended to increase as a function of decreased annual irrigation/increase
the individual years. In spite of that, after the first 3 years, free fatty acid
water stress (Fig. 7). No effect of bearing level was found on fruit oil
percentage in oil from ’Barnea’ fruit was consistently higher in T1 and
percentage measured in paste in the mill. Fruit from ’Askal’ typically
lower in treatments with greater deficit irrigation regimes (Fig. 8). From
reached between 24% and 29% oil (based on NIR) and from ’Barnea’
2015 onwards, treatment T5 tended to have the lowest free fatty acid
between 17% and 25%. For ’Barnea’ in some years differences between
level. A similar trend was found for ’Askal’ in the "Off" years 2016 and
percent oil content in T1 and T5 were strongly significant, with T5
2018. Noticeable are years (2015 and 2018 for ’Barnea’, 2018 for
having as much as 5% more oil in fruit. In ’Askal’ such differences were
’Askal’) when T1 fatty acid levels were above the 0.8% standard for
less drastic, with T5 typically having 1–2% more oil per fresh weight
extra virgin olive oil while RDI treatments, and especially T5, were
than T1 and with the exception of the 2017 season when, while fruit oil
under the threshold.
content in all treatments was low but T1 (and T3) had only 18% oil while
T5 had greater than 22%.
3.6.2. Polyphenols
Reflecting the contrasting trends of greater fruit yield and lower fruit
Both treatment and year were found to be significant factors
oil percentage, differences in cumulative oil yield (Table 2), summed for
(P < 0.0001) influencing oil polyphenol content (Fig. 9) for both ’Bar­
the last 6 years of the experiment, from 2013 until 2018 were not sta­
nea’ and ’Askal’ (Supplementary Table S1). No interaction was found
tistically significant between the treatments. Total oil yield for the 6
between year and treatment for either cultivar. Average polyphenols for
years ranged from ~20 (T1, T2) to ~18 (T4) t/ha for ’Barnea’ and ~17
’Askal’ were significantly highest in T5 oil (490 mg/kg oil), second
(T1, T2) to 15 t/ha (T3) for ’Askal’.
highest in T4 oil (372 mg/kg oil) and lower in T1 (279 mg/kg) T2
(327 mg/kg) and T3 (307 mg/kg) which were not significantly different
one from another. Polyphenol content of the produced oil tended to
increase as a function of decreased applied irrigation water starting in

7
A. Ben-Gal et al. Agricultural Water Management 245 (2021) 106577

2.0
(A) Barnea

1.5

1.0

.5
Free fatty acids (%)

0.0

T1 (B) Askal
T2
T3
T4
1.5 T5

1.0

.5

0.0
12

13

14

15

16

17

18
20

20

20

20

20

20

20

Fig. 9. Polyphenol content of oil produced each season per treatment. (A)
Fig. 8. Fatty acid content of oil each season per treatment. (A) Barnea, (B) Barnea, (B) Askal. Different letters indicate significant differences between
Askal. Error bars are standard deviations. There are no significant differences treatments in each year.
between treatments within each year and cultivar.
4. Discussion
2013, the first full season of treatments. Polyphenol content varied from
season to season and ranged from 7% lower (2014) to 150% higher 4.1. Absolute values of yields
(2015) in T5 compared to T1 for ’Barnea’. Similar trends were found for
’Askal’ with polyphenols 6% lower in T5 than T1 in 2012 but 30–385% The values measured for yields in the current experiment (Table 2,
higher in all subsequent years. Average polyphenol content was Fig. 6) were high. Average commercial yields of both fruit and oil are
~250 mg/kg for T1, T2 and T3, 337 for T4 and 372 mg/kg for T5 for expected to be some 20–30% lower than those presented here. We
’Barnea’ and between 260 and 320 for T1-T3, 342 for T4 and 487 for T5 believe that there are several explanations for this. First, yield was
for ’Askal’. In 2016 and 2018, both "Off" years, polyphenols in oil from measured only on pre-determined trees. In choosing the measured trees,
the ’Askal’ trees were significantly higher in T5 than in all of the other exceptional individuals, particular small or low bearing trees were
treatments. avoided. The yield from measured trees was harvested and weighed
individually (per tree) with little expected losses or missed fruit. Oil
3.7. Water productivity yield was calculated by multiplying fruit yield by percent oil in fruit. Oil
in fruit was determined using NIR, a method that measures total oil,
Productivity of irrigation water calculated as kg oil per cubic meter while commercial mills typically recover only around 85% of the total.
of water irrigated (Fig. 10) was obviously greater for treatments where The yields for fruit and oil therefore should be taken as high end po­
decreases in irrigation were greater than consequential decreases in tential rather than actual commercial quantities and best referred to
yields. With an overall trend of increased productivity as a function of relatively per treatment.
decreased irrigation amount, in some of the years, productivity of T5
was significantly greater than most of the other treatments. In any given
4.2. Choice of periods for RDI
year, water productivity ranged from under 0.5 to more than 1.5 kg oil/
m3 with obvious higher productivity in "On" compared to "Off" years.
A number of phenological stages have been identified as potential for
Total 6-year water productivity (Fig. 10) ranged from 0.68 to 0.78 kg
strategies for reduced irrigation with no or minimal negative effect on
oil/m3 in ’Barnea’ for treatments 1–4 with no significant difference
olive oil yield or quality. We chose to evaluate regulated deficit irriga­
while T5 was 1.28 kg oil/m3 and significantly greater. Similarly, ’Askal’
tion in the early spring, as water availability in soil, following winter
had a non-significant range of water productivity of 0.6–0.76 kg oil/m3
rains, is relatively high, and during the mid-summer pit hardening
water for treatments 1–4 and significantly higher 1.05 kg oil/m3 in T5
period (considered less sensitive to water stress). Similar RDI strategies
(Fig. 10).
with combinations of stress levels given in the spring with onset of
irrigation and during pit hardening in the summer were found to also not
effect yield significantly and to therefore increase water productivity in

8
A. Ben-Gal et al. Agricultural Water Management 245 (2021) 106577

Fig. 11. Water production functions comparing prior study from sustained
levels of deficit irrigation (DI) from Ben-Gal et al. (2011a) and Moriana et al.
(2003) with results from the current study’s regulated deficit irrigation (RDI)
strategies. I/ETp is seasonal irrigation amount divided by seasonal potential
evapotranspiration. Oil yield is normalized by maximum measured in
each experiment.

6.4 for the 60% treatment.

4.3. Success (or not) of RDI over DI

We ask the question: does RDI actually have advantage over or dif­
ference from sustained DI regimes? There have been few studies
comparing DI to RDI directly and these are inconsistent in their findings.
Iniesta et al. (2009) in a study in Cordoba, southern Spain on ’Arbe­
quina’ olives, compared sustained DI to RDI with midsummer (July 1 to
September 10–15) without irrigation. Both treatments were around 25%
Fig. 10. Water productivity (kg oil per cubic meter of irrigation water) per of calculated full ET needs and received the same total annual irrigation.
season and accumulate total for the experiment per treatment. (A) Barnea, (B) During stress period, SWP fell to below − 2 MPa for DI and − 3 MPa for
Askal. Different letters indicate significant differences between treatments in RDI. Oil yield was reduced for both treatments compared to full control
each year. irrigation by around 20% but no differences were found between them.
Moriana et al. (2003) had an RDI treatment where SWP decreased to
a sister experiment in Jordan on ’Nabali Baladi’ and ’Nabali Muhasan’ − 7 MPa in summer during the deficit while sustained DI of same annual
olive trees (Naoum et al., 2016). Under typical Mediterranean and deficit decreased only to − 3.8 MPA.
Middle-Eastern conditions, the soil in olive orchards tends to have In order to evaluate if RDI had added benefit compared to full season,
substantial tree-available water in the springtime following winter rains. sustained DI, we compare results from the current experiment with
This was particularly true for the case study reported here where the previously published data from DI studies. In Fig. 11 water production
local soil was heavy and could hold large amounts of water. This, on one curves are shown based on a DI trial on ’Barnea’ olives located in the
hand, supports the claim that little if any irrigation is needed during this same area as the current study (Ben-Gal et al., 2011a) and on the 2003
period while, on the other hand, limits the potential for significant water experiment on ’Picual’ olives by Moriana et al. (2003) in Spain where
savings. Certainly, there is major importance in determination of when both DI and RDI treatments were studied. Ben-Gal et al. (2011a) found a
irrigation should be initiated each spring. In the current study, the T5 curve with reduced yields when irrigation water was less than fully
treatment, based on SWP trigger values (Fig. 5), indicated that common satisfying potential ET requirements. Moriana et al. (2003) suggested a
practice tends to begin irrigation earlier than necessary and to over- single response curve for both DI and RDI treatments, suggesting no
irrigate olives in the first month or more of the season, depending of added benefit for RDI as both treatments reduced yields by 15–20%. The
course on the extent and pattern of the previous winter’s rainfall. curve from Moriana et al. (2003) was similar to Ben-Gal et al. (2011a)
Trentacoste et al. (2019) found that RDI in spring was effective in saving but steeper (larger response to relative decreases in applied water). They
water and reducing undesirable vegetative growth in super-intensive published a best-fit curve for their oil yield data where
hedgerow cv Arbosana olives in Argentina. Yield = − 2.78 + 0.11 ET − 0.0006 × 10− 3 ET2. The data from the
Lavee et al. (2007) tested seven RDI strategies in Israel on ’Muhasan’ current study for the most part follow the previously published response
olives and found the most efficient irrigation schedule was based on curves, with decreased seasonal water application leading to decreased
withholding water application until the end of endocarp (pit) hardening yields, with the exception of T5, which had clearly greater water
applying all the annual water from that stage until 2 weeks prior to productivity.
harvest. They found 4-year reduction of yield relative to full irrigation of
12.5% (oil) with 50% savings in water. Fernández et al. (2013) practiced 4.4. Use of SWP for scheduling
RDI from late June to late August. In that study midday SWP fell to
between − 2 and − 3.5 MPa when irrigation was reduced to 60% and In the current study the SWP guided irrigation treatment, based on
− 3 and − 4.5 MPa when reduced further to 30% of full irrigation. Total dynamic thresholds ranging from − 1.5 to − 2.5 MPa in the spring in
yield after three experimental years was reduced 25–30% by the RDI "On" and "Off" years, respectively, reduced to − 3 to − 3.5 MPa (on and
treatments with no differences between them. Water productivity was off years) during pit hardening and increasing back to − 2.5 MPa later in
2.5 kg oil ha− 1 per mm for full irrigation, 2.8 for the 30% treatment and the season, during the oil accumulation stage (Fig. 2). These target set

9
A. Ben-Gal et al. Agricultural Water Management 245 (2021) 106577

points were based on measurements in a previous experiment (Ben-Gal situations, in many of the years the actual load levels fell somewhere in
et al., 2011a, 2011b) where seasonal SWP measurements in well irri­ between (Fig. 6). Furthermore, flowering level is not always correlated
gated treatments followed a pattern ranging from − 1.5 MPa in the to final yield, especially when unfavorable climatic conditions occur
spring to − 2.5 MPa pre-harvest in September each year and where DI during flowering and harm the fruit set. Our results ended up showing
treatments had SWP values of − 3 to − 3.5 MPa during it hardening. As little or no benefit to changing irrigation practice according to fruit load,
mentioned already, this treatment consistently led to the least amount of likely due to this issue. That said, compared to the preliminary study
applied irrigation water and the greatest water productivity and was the under the same environment as in the current experiment (Ben-Gal
only treatment suggesting true advantage of RDI over DI (Fig. 11). et al., 2011a), water status measured as minimum SWP were much more
Midday stem water potential used for evaluating water status and similar, even when comparing high and low yield trees and years
scheduling irrigation was reported previously (Moriana et al., 2012). (Table 3). This may be due to the differential RDI treatments but may
Moriana et al. (2012) evaluated SWP in two locations in Spain based on also be due to less extreme fruit load conditions.
thresholds of − 2 MPa for DI while full irrigation maintained from
− 1.2 MPa to − 1.4 MPa from pit hardening. The DI treatment led to 4.6. Irrigation to reach optimal fruit water content at harvest
70–80% reduction in irrigation, reduced vegetative growth and around
15% reduction in fruit yield. Contrary to our current study, there was In the current study, we attempted to use measurements of fruit
significant rainfall during the irrigation seasons in the Moriana et al. water content in order to direct end of season irrigation such as to reach
(2012) trial. Ahumada-Orellana et al. (2017) evaluated SWP thresholds the olive mill at optimum status in order to promote mill efficiency and
for irrigation in a super high density ’Arbequina’ orchard in Chile. They oil quality and to avoid early ripening due to water stress (Table 4). The
compared thresholds of − 1.4—2.2 MPa considered full irrigation to approach has additional logic and potential value based on strong
treatments initiating irrigation at SWP thresholds of − 3.4, − 4.5 and established relationships between tree water status and fruit water
− 6 MPa. After irrigation was triggered all treatments were irrigated at content (Fernandes et al., 2018). The strategy was practically challenged
100% level. A threshold for irrigation initiation of − 3.4 MPa did not by a number of issues. First, the method for determining water content
affect yield while saving 20–25% water. Threshold of − 4.5 MPa saved was time consuming. Sampling, weighing, drying, and re-weighing took
~24–40% water with decrease of 30% in fruit yield while − 6 MPa between 4 and 7 days to supply results. Second, response of fruit water
threshold saved ~30–45% water at a cost of 40% decreased yield. content to changes in irrigation were slow, taking sometimes up to 2
It is not at all clear in the current study if the chosen threshold values weeks to be quantified. This combination meant that the technique had
were optimal or if the differences in thresholds between "On" and "Off" low sensitivity and limited effectiveness. What more, there appeared to
years contributed to their success. While the treatment and its results are be disconnect between water content values measured from the sam­
of great interest, the large cost, particularly in labor, to conduct them is pling process and those determined on samples from harvest. This could
expected to be prohibitive for many commercial operations. Certainly be due to lack of representation of the chosen representative fruit (taken
more work is needed to fine-tune the ultimate threshold values for best from one certain height from the ground around the tree’s circumfer­
results, including evaluation of other cultivars and environments. We ence) compared to actual averaged samples of all fruit from the entire
suggest that better alternatives could be developed either to build crop tree. Commercial measurements of water in fruit based on large volume
factor-reference ET RDI protocols based on the results of this and other samples and water content evaluation using NIR on olive paste after
studies or to find a proxy for the manual pressure bomb based mea­ milling would expedite the method. The approach could become even
surements, possibly using continuous measurements of SWP itself more practical in the future following successful development of sensors
(Dixon and Tyree, 1984; Meron et al., 2015) or some other water status for accurate and rapid determination of fruit water in the field (Trapani
sensitive plant based sensor method (Ben-Gal et al., 2010). Possible et al., 2017).
methods are sap flow sensors or dendrometers, although they have not
proven effective in monitoring olive trees to date (Carr, 2013), and 4.7. Optimizing yield and oil quality
thermal based crops water stress indices from remote imaging or prox­
imal sensing (Agam et al., 2013). Total polyphenols (and oxidative stability) decrease with increased
irrigation, while free acidity increases. We found profound improvement
4.5. Identifying and managing fruit load in oil quality as a function of lowered seasonal irrigation (Figs. 8 and 9).
This is not surprising, as prior studies have indicated the same, including
Fruit load has been successfully identified as a driver for olive tree for ’Barnea’, ’Koroneiki’ and ’Souri’ (Ben-Gal et al., 2011a; Dag et al.,
water status and transpiration. Ben-Gal et al. (2011b) showed minimum 2008, 2015). The connection between water stress and higher poly­
seasonal stem water in "On" compared to "Off" Barnea and Souri trees for phenols is well established. Artajo et al. (2006) suggested that the water
a large range of deficit irrigation levels as well as for well-watered trees status of trees affects phenol production in olive fruit and consequently
with SWP values as much as 0.5 MPa lower in "On" trees in May and as the phenol content of the olive paste and oil. They and others (Tovar
much as 1 MPa lower in September. Bustan et al. (2016) quantified as et al., 2002) indicate that water stress triggers augmented phenol syn­
much as 30% greater tree scale transpiration of Barnea trees as a func­ thesis. The phenomenon where acidity is decreased under increased
tion of fruit load when comparing loaded to low load or fruitless trees. water stress is less evident in the literature and may be dependent on
Naor et al. (2012) studied interactions between irrigation during oil climate and cultivar since studies reporting the response tend to be
accumulation and fruit yield of ’Koroneiki’ olives by irrigating accord­ based in regions outside of the traditional center of olive production in
ing to varied SWP thresholds. Oil yield was practically unaffected by southwestern Europe (Berenguer et al., 2006; Dag et al., 2008, 2015;
irrigation rate at low crop loads but significantly increased with Ghrab et al., 2014; Toplu et al., 2009).
increased irrigation at high crop loads. Moriana et al. (2003) considered Of interest in the current study’s results is that the timing of stress did
fruit load by applying full irrigation during bearing "On" years and no not seem to have effect on its influence on oil quality. Even treatments
irrigation at all (rainfed) in "Off" years. This strategy was found unde­ where stress was mostly felt by trees early in the season and when the
sirable as the alternate year irrigation decreased yields by 25% in their last several months prior to harvest were irrigated at minimal deficit
3-year experiment. levels, the positive effects on oil quality measured by increased poly­
In the current study, while we were confident in the logic behind the phenols and decreased free fatty acid content were evident.
treatments with differential RDI strategies between high and low load
years, we were often challenged in correctly defining "On" vs "Off" years
in the springtime. While treatments were based on binary "On" or "Off"

10
A. Ben-Gal et al. Agricultural Water Management 245 (2021) 106577

5. Conclusions was further supported technically by Yulia Subbotin, Amir Elkrinawi,


and Talaal Huashala of The Gilat Research Center.
Regulated deficit irrigation of olives, in which commercial irrigation
application levels were reduced 30–40% alternatively during springtime Appendix A. Supporting information
(April–June) or mid-summer pit hardening, led to reductions in accu­
mulated multi-seasonal water application by 15–25%. The RDI treat­ Supplementary data associated with this article can be found in the
ments had a non-significant trend of reduced yield as a function of online version at doi:10.1016/j.agwat.2020.106577.
reduced applied irrigation water with yields decreased up to 12%
compared to the control treatment. Furthermore, quality parameters
References
were consistently improved under the RDI treatments, in both cultivars.
An attempt to consider alternate bearing and shift RDI treatment Agam, N., Cohen, Y., Berni, J., Alchanatis, V., Kool, D., Dag, A., Yermiyahu, U., Ben-
periods accordingly did not appear to be beneficial, as long-term water Gal, A., 2013. An insight to the performance of crop water stress index for olive trees.
savings, yields and water productivity were unchanged. We believe that Agric. Water Manag. 118, 79–86.
Ahumada-Orellana, L.E., Ortega-Farías, S., Searles, P.S., Retamales, J.B., 2017. Yield and
while fruit load is obviously important to water requirements and water productivity responses to irrigation cut-off strategies after fruit set using stem
response to water stress, identification of actual bearing level and water potential thresholds in a super-high density olive orchard. Front. Plant Sci. 8,
matching irrigation levels prove to be hard to achieve or, at least, need 1280.
Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration-Guidelines
to be determined only after fruit set, around the end of May. for computing crop water requirements-FAO Irrigation and drainage paper 56. Fao.
Attempts to irrigate according to pre-set target values of SWP did Rome 300 (9), D05109.
significantly increase water productivity as accumulated water applied Artajo, L.S., Romero, M.P., Tovar, M.J., Motilva, M.J., 2006. Effect of irrigation applied
to olive trees (Olea europaea L.) on phenolic compound transfer during olive oil
was reduced by ~40% and yield decreased by less than 15% (non- extraction. Eur. J. Lipid Sci. Technol. 108, 19–27.
significantly) over the 6 years of the experiment. Choice of optimal SWP Ben-Gal, A., Dag, A., Basheer, L., Yermiyahu, U., Zipori, I., Kerem, Z., 2011a. The
threshold needs further study, including its universality across envi­ influence of bearing cycles on olive oil quality response to irrigation. J. Agric. Food
Chem. 59, 11667–11675.
ronments and cultivars. While SWP based irrigation scheduling seems to
Ben-Gal, A., Kool, D., Agam, N., van Halsema, G.E., Yermiyahu, U., Yafe, A., Presnov, E.,
be effective in driving RDI, its high labor demand and cost make its Erel, R., Majdop, A., Zipori, I., Segal, E., Ruger, S., Zimmermann, U., Cohen, Y.,
practicality in commercial olive orchards questionable. We therefore Alchanatis, V., Dag, A., 2010. Whole-tree water balance and indicators for short-term
drought stress in non-bearing ‘Barnea’ olives. Agric. Water Manag. 98, 124–133.
hope for and await reliable proxies based on continuous or remote
Ben-Gal, A., Yermiyahu, U., Zipori, I., Presnov, E., Hanoch, E., Dag, A., 2011b. The
sensing of traits related to tree water status. influence of bearing cycles on olive oil production response to irrigation. Irrig. Sci.
End of season irrigation designed to bring fruit to oil mills with 29, 253–263.
optimal water concentration was also found to be less than successful. Ben-David, E., Kerem, Z., Zipori, I., Weissbein, S., Basheer, L., Bustan, A., Dag, A., 2010.
Optimization of the Abencor system to extract olive oil from irrigated orchards. Eur.
This due to an overly long response time, both for measuring fruit water J. Lipid Sci. Technol. 112, 1158–1165.
content and for the measure to respond to changes in irrigation. In situ Berenguer, M.J., Vossen, P.M., Grattan, S.R., Connell, J.H., Polito, V.S., 2006. Tree
sensing of fruit water could improve this method. irrigation levels for optimum chemical and sensory properties of olive oil.
HortScience 41, 427–432.
Oil quality, measured in terms of decreased fatty acids and increased Bustan, A., Avni, A., Lavee, S., Zipori, I., Yeselson, Y., Schaffer, A.A., Riov, J., Dag, A.,
polyphenols, was improved by reductions in applied irrigation water, 2011. Role of carbohydrate reserves in yield production of intensively cultivated oil
with no dependence on when the reductions occurred over the irrigation olive (Olea europaea L.) trees. Tree Physiol. 31, 519–530.
Bustan, A., Dag, A., Yermiyahu, U., Erel, R., Presnov, E., Agam, N., Kool, D., Iwema, J.,
season. Zipori, I., Ben-Gal, A., 2016. Fruit load governs transpiration of olive trees. Tree
We understand that we were looking for marginal benefits of Physiol. 36, 380–391.
sometime nuanced differences in water application strategies. We still Carr, M., 2013. The water relations and irrigation requirements of olive (Olea europaea
L.): a review. Exp. Agric. 49, 597–639.
believe that such investigation for subtle increases in water productivity
Dag, A., Ben-Gal, A., Yermiyahu, U., Basheer, L., Nir, Y., Kerem, Z., 2008. The effect of
can be important to profitability of irrigated olives, particularly as oil irrigation level and harvest mechanization on virgin olive oil quality in a traditional
prices decrease and water becomes more valuable. While much of our rain-fed ‘Souri’ olive orchard converted to irrigation. J. Sci. Food Agric. 88,
1524–1528.
results regarding superiority of RDI were not significant, we believe that
Dag, A., Bustan, A., Avni, A., Lavee, S., Riov, J., 2009. Fruit thinning using NAA shows
the discussion of them and the potential and challenges in applying them potential for reducing biennial bearing of ‘Barnea’and ‘Picual’oil olive trees. Crop
have value. It appears that RDI can indeed be more beneficial than Pasture Sci. 60, 1124–1130.
sustained DI as seen in our experiment’s results regarding the treatment Dag, A., Naor, A., Ben-Gal, A., Harlev, G., Zipori, I., Schneider, D., Birger, R., Peres, M.,
Gal, Y., Kerem, Z., 2015. The effect of water stress on super-high-density
T5 where SWP was used to guide irrigation. Those results indicate that ‘Koroneiki’olive oil quality. J. Sci. Food Agric. 95, 2016–2020.
irrigation levels during deficit periods in spring and during pit hard­ Dixon, M., Tyree, M., 1984. A new stem hygrometer, corrected for temperature gradients
ening could be significantly lower than those chosen in the study, and calibrated against the pressure bomb. Plant Cell Environ. 7, 693–697.
Erel, R., Eppel, A., Yermiyahu, U., Ben-Gal, A., Levy, G., Zipori, I., Schaumann, G.E.,
regardless of whether they are a product of SWP measurements or more Mayer, O., Dag, A., 2019. Long-term irrigation with reclaimed wastewater:
simple adaptation as crop factors for irrigation according to reference implications on nutrient management, soil chemistry and olive (Olea europaea L.)
ET. performance. Agric. Water Manag. 213, 324–335.
Fereres, E., Soriano, M.A., 2006. Deficit irrigation for reducing agricultural water use.
J. Exp. Bot. 58, 147–159.
Declaration of Competing Interest Fernandes, R.D.M., Cuevas, M.V., Diaz-Espejo, A., Hernandez-Santana, V., 2018. Effects
of water stress on fruit growth and water relations between fruits and leaves in a
hedgerow olive orchard. Agric. Water Manag. 210, 32–40.
The authors declare that they have no known competing financial Fernández, J.E., Perez-Martin, A., Torres-Ruiz, J.M., Cuevas, M.V., Rodriguez-
interests or personal relationships that could have appeared to influence Dominguez, C.M., Elsayed-Farag, S., Morales-Sillero, A., García, J.M., Hernandez-
the work reported in this paper. Santana, V., Diaz-Espejo, A., 2013. A regulated deficit irrigation strategy for
hedgerow olive orchards with high plant density. Plant Soil 372, 279–295.
Ghrab, M., Ayadi, M., Gargouri, K., Chartzoulakis, K., Gharsallaoui, M., Bentaher, H.,
Acknowledgements Psarras, G., Mimoun, M.B., Masmoudi, M.M., Mechlia, N.B., 2014. Long-term effects
of partial root-zone drying (PRD) on yield, oil composition and quality of olive tree
(cv. Chemlali) irrigated with saline water in arid land. J. Food Compos. Anal. 36,
This research was generously supported by USAID/MERC, Middle
90–97.
East Regional Cooperation fund (Grant TA-MAU-M36-001) and The Gómez-del-Campo, M., 2013. Summer deficit-irrigation strategies in a hedgerow olive
Chief Scientist of Israel’s Ministry of Agriculture and Rural Development orchard cv.‘Arbequina’: effect on fruit characteristics and yield. Irrig. Sci. 31,
(project #20-10-0064). Gratitude to TZ”ABAR K”AMA for providing the 259–269.
Gucci, R., Caruso, G., Gennai, C., Esposto, S., Urbani, S., Servili, M., 2019. Fruit growth,
orchard and orchard management support from Nir Kilman and to yield and oil quality changes induced by deficit irrigation at different stages of olive
Menachem Elia of Megadlai Darom for field measurements. The work fruit development. Agric. Water Manag. 212, 88–98.

11
A. Ben-Gal et al. Agricultural Water Management 245 (2021) 106577

Gucci, R., Lodolini, E.M., Rapoport, H.F., 2009. Water deficit-induced changes in Shackel, K.A., Ahmadi, H., Biasi, W., Buchner, R., Goldhamer, D., Gurusinghe, S.,
mesocarp cellular processes and the relationship between mesocarp and endocarp Hasey, J., Kester, D., Krueger, B., Lampinen, B., 1997. Plant water status as an index
during olive fruit development. Tree Physiol. 29, 1575–1585. of irrigation need in deciduous fruit trees. HortTechnology 7, 23–29.
Iniesta, F., Testi, L., Orgaz, F., Villalobos, F., 2009. The effects of regulated and Swain, T., Hillis, W., 1959. The phenolic constituents of Prunus domestica. I.—the
continuous deficit irrigation on the water use, growth and yield of olive trees. Eur. J. quantitative analysis of phenolic constituents. J. Sci. Food Agric. 10, 63–68.
Agron. 30, 258–265. Toplu, C., Önder, D., Önder, S., Yıldız, E., 2009. Determination of fruit and oil
Lavee, S., 2007. Biennial bearing in olive (Olea europaea L.). Ann. Ser. Hist. Nat. 17, characteristics of olive (Olea europaea L. cv.‘Gemlik’) in different irrigation and
101–112. fertilization regimes. Afr. J. Agric. Res. 4, 649–658.
Lavee, S., Avidan, B., Meni, Y., 2003. Askal. a new high-performing oil variety for intensive Tovar, M.J., Romero, M.P., Girona, J., Motilva, M.J., 2002. l-Phenylalanine ammonia-
and super-intensive olive orchards. Olivae 97. lyase activity and concentration of phenolics in developing olive (Olea europaea L cv
Lavee, S., Hanoch, E., Wodner, M., Abramowitch, H., 2007. The effect of predetermined Arbequina) fruit grown under different irrigation regimes. J. Sci. Food Agric. 82,
deficit irrigation on the performance of cv. Muhasan olives (Olea europaea L.) in the 892–898.
eastern coastal plain of Israel. Sci. Hortic. 112, 156–163. Trapani, S., Migliorini, M., Cecchi, L., Giovenzana, V., Beghi, R., Canuti, V., Fia, G.,
Lavee, S., Nashef, M., Wodner, M., Harshemesh, H., 1990. The effect of complementary Zanoni, B., 2017. Feasibility of filter-based NIR spectroscopy for the routine
irrigation added to old olive trees (Olea europaea L.) cv. Souri on fruit characteristics, measurement of olive oil fruit ripening indices. Eur. J. Lipid Sci. Technol. 119,
yield and oil production. Adv. Hortic. Sci. 135–138. 1600239.
Meron, M., Goldberg, S., Solomon-Halgoa, A., Ramon, G., 2015. Ramon Embedded stem Trentacoste, E.R., Calderon, F.J., Contreras-Zanessi, O., Galarza, W., Banco, A.P.,
water potential sensor. In: Stafford, John V. (Ed.), Precision Agriculture’15. Puertas, C.M., 2019. Effect of regulated deficit irrigation during the vegetative
Wageningen Academic Publishers, p. 2. growth period on shoot elongation and oil yield components in olive hedgerows (cv.
Moriana, A., Orgaz, F., Pastor, M., Fereres, E., 2003. Yield responses of a mature olive Arbosana) pruned annually on alternate sides in San Juan, Argentina. Irrig. Sci. 37,
orchard to water deficits. J. Am. Soc. Hortic. Sci. 128, 425–431. 1–14..
Moriana, A., Pérez-López, D., Prieto, M., Ramírez-Santa-Pau, M., Pérez-Rodriguez, J., Zipori, I., Bustan, A., Kerem, Z., Dag, A., 2016. Olive paste oil content on a dry weight
2012. Midday stem water potential as a useful tool for estimating irrigation basis (OPDW): an indicator for optimal harvesting time in modern olive orchards.
requirements in olive trees. Agric. Water Manag. 112, 43–54. Grasas y Aceites 67, e137.
Naor, A., Schneider, D., Ben-Gal, A., Zipori, I., Dag, A., Kerem, Z., Birger, R., Peres, M., Zipori, I., Yermiyahu, U., Erel, R., Presnov, E., Faingold, I., Ben-Gal, A., Dag, A., 2015.
Gal, Y., 2012. The effects of crop load and irrigation rate in the oil accumulation The influence of irrigation level on olive tree nutritional status. Irrig. Sci. 33,
stage on oil yield and water relations of ‘Koroneiki’ olives. Irrig. Sci. 1–11. 277–287.
Naoum, S., Albalawneh, A., Ayoub, S., Diab, M., Amayreh, I., Ammoush, M., Kawaleet,
B., Daoud, L.2016. Productivity of water, growth and yield of olive trees under
deficit irrigation. VIII International Olive Symposium 1199, 261–266.

12

You might also like