Vegetative

The olive is a xerophytic evergreen tree that can attain a height of 8–15 m but under cultivation it is kept at 4–5 m by pruning. The small, silvery-green leaves are thick and leathery with a waxy upper surface. The lower surface is protected by a mass of peltate hairs with numerous, sunken stomata (at densities of 420–540 stomata mm⁻²) with small openings (c. 11 × 5 μm). Cultivars differ in the morphological and structural leaf adaptations to protect against water loss. This does not mean that the olive tree will not respond to irrigation, only that it is adapted to withstand drought (Bacelar et al., Reference Bacelar, Correia, Moutinho, Gonçalves, Lopes and Pereira2004; Beede and Goldhamer, Reference Beede, Goldhamer, Ferguson, Sibbett and Martin1994; Hendrickson and Veihmeyer, Reference Hendrickson and Veihmeyer1949). A detailed review of the photosynthetic processes (C3 pathway), at the leaf and canopy levels, has been prepared by Connor and Fereres (Reference Connor and Fereres2005). The optimum temperature for net photosynthesis is about 28 °C.

The rate of expansion of the tree trunk is particularly sensitive to water stress. This sensitivity increases as the tree ages. Branch length is also sensitive to water stress in both young and mature trees. This means that the number of leaf nodes (which are where the flower buds form) is reduced. The amount of fruit on a tree also influences vegetative growth. For example, in the early years after field planting especially, a large fruit load will reduce both trunk and canopy expansion rates (Martín-Vertedor et al., Reference Martín-Vertedor, Rodríguez, Losada and Castiel2011a). The length of time it takes for a tree canopy to develop depends in part on how the soil surface (whether it is cultivated or not) and any understorey vegetation (weeds, pasture or intercrops) are managed. There is always a risk of soil erosion on sloping land when the area between trees is cultivated. But in a rain-fed orchard the risk of water stress increases with the size of the crop canopy (Connor, Reference Connor2005; Fereres, Reference Fereres, Steduto, Hsiao, Fereres and Raes2012; Martín-Vertedor et al., Reference Martín-Vertedor, Rodríguez, Losada and Castiel2011a; Villalobos et al., Reference Villalobos, Orgaz, Testi and Fereres2000).

Flowering

Flowers are born on inflorescences that, following a period of winter dormancy, develop in the spring from buds in the axils of leaves formed on wood produced during the previous year. The terminal bud on a shoot is nearly always vegetative. The flowers are either ‘perfect’ hermaphrodite, containing both male and female parts, or ‘imperfect’ staminate, capable of only producing pollen. In a controlled water deficit experiment, Rapoport et al. (Reference Rapoport, Pérez-Priego, Orgaz and Martins2011) found that water stress during winter dormancy had no lasting effect on the flowering process, recovery being rapid on re-watering. Water stress during inflorescence development reduced the number of inflorescences, the number of flowers and the number of imperfect flowers, and inhibited ovule development. Water stress during flowering and initial fruit set (spring and early summer) caused flowers to dry and abscise as a unit, exposing senescent stigmas, which were no longer receptive to pollination. The flowers are largely wind pollinated, but less than 2% of the flowers set fruit.

Fruiting

The fruit is a small drupe. It contains a single seed, known as a pit (USA) or stone (UK), which develops from at least one of four ovules present in the ovary. Fruit numbers are reduced when there are fewer flowers and/or fertilization is inhibited. Increased size of individual fruits may not always compensate for any reduction in fruit number (Rapoport et al., Reference Rapoport, Pérez-Priego, Orgaz and Martins2011). Connor (Reference Connor2005) proposed a three-part explanation for successful oil production under severe water stress in Mediterranean environments. First, during the early stages of olive fruit formation (lasting about eight weeks) the tree can tolerate moderate water stress (pre-dawn leaf water potentials between –2 and –3 MPa) without much reduction in its potential to accumulate oil. Second, during the later stages of fruit development, carbon fixation by the fruit itself, together with the inherent capacity of the olive tree to use water efficiently, sustains limited growth, enabling the fruit to survive. Third, following rain in the autumn, the tree water status recovers rapidly, and substantial quantities of new assimilate are produced (when even previously ‘wrinkled’ fruit will recover). It takes about 15–18 months from flower bud initiation to fruit ripening, including four to five months for fruit filling. The pulp to pit ratio is an important quality criterion for both table and olive oil fruit (Connor, Reference Connor2005).

The components of yield for olive oil can be summarised as follows:

\begin{eqnarray*} {\rm Oil}\,{\rm yield}\,({\rm kg}\,{\rm tree}^{ - 1} ) &=& {\rm Number}\,{\rm of}\,{\rm fruits}\,({\rm tree}^{ - 1} )\; \times \;{\rm Mean}\,{\rm fresh}\,{\rm fruit}\,{\rm weight}\,({\rm g}) \\ &&\times\, {\rm Oil}\,{\rm content}\,(\% \;{\rm fresh}\,{\rm weight})\; \div \;1000 . \end{eqnarray*}

Irrigation may increase the fresh weight of an individual fruit, but may reduce the oil concentration of the mesocarp, which depending in part on the cultivar may result in a reduction in the oil yield. Oils accumulate during the late summer and autumn. The final oil content also has implications on the costs of processing. For table olives, fruit size is important. A reasonable commercial yield of fresh fruit from a rain-fed crop is 2–5 t ha⁻¹ and from an irrigated crop 12–15 t ha⁻¹ (averaged over two years, to allow for on–off syndrome, see below). Oils from irrigated orchards are usually less bitter and pungent than those from rain-fed areas (Fereres, Reference Fereres, Steduto, Hsiao, Fereres and Raes2012).

Fruits are distributed irregularly in the canopy. They are formed preferentially on the illuminated sides of the canopy, namely the top and southern sides in the northern hemisphere. Pruning practices in an olive orchard are designed to allow light to penetrate into the canopy and to promote fruiting sites (Barranco, et al., Reference Barranco, Fernández-Escobar and Rallo2010). The design of hedgerow systems to facilitate mechanisation, while optimising the interception and distribution of incident solar radiation over the canopy, has been the subject of detailed research by Connor (Reference Connor2006) and Connor et al. (Reference Connor, Gómez-del-Campo and Cornas2012).

Olive has an alternate bearing pattern whereby fruit production fluctuates between large and small yields in alternate years. These are often called ‘on’ and ‘off’ years. Cultivars differ in the degree to which biennial bearing occurs. As fruits are formed on last year's wood, excess vegetative growth in one year can lead to a high fruit yield the next, but less vegetative growth (because of competition for assimilates). This, in turn, means a low yield in the following year. In this way biennial bearing is initiated. The issue is further complicated when mature trees are pruned to reduce vegetative growth. Without management interventions the ‘yield ratio’ (ratio yield in ‘on’ years to that in ‘off’ years) can be as much as 10:1 (Martín-Vertedor et al., Reference Martín-Vertedor, Rodríguez, Losada and Castiel2011a).

Roots

Little research into the root distribution and root activity of olive trees has been reported. In general terms, Fereres (Reference Fereres, Goldhamer, Sadras, Steduto, Hsiao, Fereres and Raes2012) described the root system of olive as being ‘extensive and vigorous’, with most roots found in the top 1 m of soil, although they can extend to depths of 2–3 m in deep alluvial soils. In California (USA), Hendrickson and Veihmeyer (Reference Hendrickson and Veihmeyer1949) recorded water extraction by roots of unirrigated olive trees (cv. Manzanillo; 25-year old, spaced 9.1 × 9.1 m or 120 trees ha⁻¹), growing in a sandy loam soil, from depths of at least 1.8 m.

Apart from observations such as these, very little systematic research on the actual distribution and dynamics of the root system of the olive tree appears to have been published. Exceptions include a detailed investigation by Fernández et al. (Reference Fernández, Moreno, Cabrera, Arrue and Martín-Aranda1991) in Spain. They traced roots of previously unirrigated, 20-year-old trees (cv. Manzanillo) to depths of 2 m and at distances of more than 2.5 m away from the trunk (trees spaced 7 × 7 m, sandy loam soil). Root densities with drip irrigation were recorded, but the data, as presented, are not easy to interpret or summarise. The greatest root density occurred in the wetted areas, down to a depth of 0.6 m, the most abundant being roots with a diameter <0.5 mm. Root length densities were also monitored in three commercial orchards in north-west Argentina in relation to different drip irrigation emitter arrangements. Approximately 70% of the root system (cv. Manzanillo) was concentrated in the 0–0.50 m (sandy) soil layer, and most of the roots were within 0.50 m of the drip line. The total root length per unit leaf area averaged 1.8–3.5 km m⁻² across the three sites (Searles et al., Reference Searles, Saravia and Rousseaux2009). This research was done in an area with very low rainfall so that the soil away from the influence of the emitters was dry, which explains why there were so few roots outside the wetted zone.

A recent paper by Polverigiani et al. (Reference Polverigiani, Lodolini and Neri2012) describes the results of observations made in a rhizotron (a glass window 0.8 m deep and 1.3 m wide, situated 0.5 m away from the trunks of four drip-irrigated trees) over a 10-month period (November 2007 to August 2008) in central Italy (43°29′N 13°07′E). Root growth (extension) continued throughout the winter (November to mid-March). The growth of very fine (<0.2 mm) roots peaked in February, and fine roots (>0.2 mm), to a lesser extent, in April/May. (For the same mass, very fine roots have 1.55 times the surface area of fine roots). During the summer, from mid-June to late August, root growth virtually ceased. At this time, the average daytime temperature was 25.4 °C, shoot extension was rapid and the fruits were expanding and ripening. Both shoot and root processes were competing with the roots for assimilates. Root mortality was first observed in late April, peaking in May. Mortality declined with increase in soil depth.

The proportion of the total biomass partitioned to the roots has been estimated to be about 0.26–0.30 (Connor and Fereres, Reference Connor and Fereres2005).

Cultivars

There are at least 2000 recognised olive tree cultivars, but only a few are grown on a large scale (Barranco et al., Reference Barranco, Fernández-Escobar and Rallo2010). The ones listed in Box 1 have been the subject of research summarised in this paper (see FAO, 2005; Oliveaustralia, 2012 for details).

Summary: crop development

1. 1. The olive is an evergreen, xerophytic tree, well adapted to survive dry conditions. Stomata are only found on the lower surface of leaves.

2. 2. Commercial production of olive is possible throughout the warm temperate regions and especially in the dry subtropical or Mediterranean regions. The olive is responsive to irrigation.

3. 3. For successful cultivation the minimum average annual rainfall needed is about 600 mm. Olives can survive with 200–250 mm. Rainwater harvesting is an alternative to formal irrigation.

4. 4. A period of low temperatures (c. 10 weeks at less than 12 °C) is required to induce flowering.

5. 5. Flowers are born in the spring on inflorescences that develop from buds in the axils of leaves formed on wood produced the previous year. Less than 2% of flowers set fruit.

6. 6. It takes from 15–18 months from flower bud induction to fruit ripening.

7. 7. The olive tree is prone to biennial bearing, which complicates experimentation.

8. 8. There has been little research reported on root systems of the olive tree, but roots can extend to depths of at least 2 m: root growth virtually ceases during the summer.

Plant water status and gas exchange

The leaf (or stem) water potential is commonly measured with a pressure bomb, with readings taken immediately before dawn and/or at midday. Leaf water potential is measured on exposed leaves, while for stem water potential the leaves are shaded. The stem water potential is a much more stable indicator of the water status of mature trees than leaf water potential (Fereres, Reference Fereres, Goldhamer, Sadras, Steduto, Hsiao, Fereres and Raes2012). Sap flow sensors are used to monitor changes in sap speed, which is an indirect measure of transpiration in individual trees. Displacement sensors (dendrometers) are used to monitor trunk diameter fluctuations, which are considered to be early indicators of water stress. All three of these techniques (pressure bomb, sap flow sensors and dendrometers, together with leaf diffusion porometers) have been used in research projects with the olive tree. Typical reference values of stem water potentials are listed in Table 1.

Table 1.

Typical values of stem water potential (MPa) at different levels of water stress on sunny, summer days (ETo = 5–6 mm d⁻¹) (based on Fereres, Reference Fereres, Goldhamer, Sadras, Steduto, Hsiao, Fereres and Raes2012).

Spain: Diurnal measurements of leaf water potential and stomatal conductance at different times during the season in south-west Spain (37°17′N 06°03′W; alt. 30 m) showed how olive trees (26-year-old; cv. Manzanillo) conserved water under conditions of high evaporative demand (saturation deficit up to 3.5 kPa) by closing their stomata (Fernández et al., Reference Fernández, Moreno, Girón and Blázquez1997). The stomata on young leaves exhibited better water use control than those on leaves that had been on the tree for a year. Maximum stomatal conductances occurred at relatively low photon flux densities (c. 500 μmol m⁻² s⁻¹). At similar radiation levels and saturation deficits, the stomata were wider open in the morning than in the afternoon. In well-irrigated trees, pre-dawn leaf water potentials never fell below –0.50 MPa, but reached –2.47 MPa at midday late in the season. The corresponding values for unirrigated trees were less than –1.60 MPa and –3.63 MPa respectively. After watering, both these variables matched those of irrigated trees within two days.

In order to characterize with greater precision the gas exchange processes of olive trees at various levels of water stress, Moriana et al. (Reference Moriana, Villalobos and Fereres2002) took measurements within a deficit irrigation experiment in Cordoba (38°N 4.8°W; alt. 110 m) during 1998 (trees spaced 6 × 6 m = 280 trees ha⁻¹). Both leaf photosynthesis and stomatal conductance responded diurnally and seasonally to changes in the tree water status and to the evaporative demand in similar ways (there was a close linear correlation between these two variables, r² = 0.92). Maximum values again occurred in the morning, followed by a continuous decline throughout the afternoon. As water stress increased, both the daily maximum and minimum values fell. In the absence of water stress, midday photosynthesis rates and stomatal conductances were high during the autumn and low on days when the air was dry (large saturation deficits). Transpiration efficiencies (mmol CO₂ mol H₂O⁻¹) followed a diurnal course similar to those for photosynthesis and stomatal conductance. At stem water potentials greater than –4 MPa, an inverse hyperbola described the relationship with the saturation deficit of the air (R ² = 0.91). Stem water potentials fell to –8 MPa in the rain-fed treatment, which is a much lower value than those observed in most crop plants subjected to drought (Moriana et al., Reference Moriana, Villalobos and Fereres2002). This capacity of the olive tree to reduce its stem water potential so low explains why it was able to extract water from the 2.4 m deep soil profile (192 mm) in the rain-fed plots at water contents below the conventionally accepted definition of permanent wilting point (soil matric potential = –1.5 MPa), also known as the lower limit of extraction (available water holding capacity = 150 mm).

As part of a modelling study to predict photosynthesis under drought conditions, the photosynthesis capacity was observed to decline over the growing season in both irrigated and unirrigated olive trees (cv. Manzanilla). The seasonal reduction in leaf nitrogen content (on a leaf area basis) was used in the model to predict photosynthesis under drought conditions (Díaz-Espejo et al., Reference Díaz-Espejo, Walcroft, Fernández, Hafidi, Palomo and Girón2006).

Previously, Jorba et al. (Reference Jorba, Tapia and Sant1985) had described the results of a container experiment at the University of Barcelona, Spain. In well-watered plants, leaf water potentials fell from –0.7 MPa at sunrise to –2.5 MPa at midday. Under dry conditions, values as low as –4.0 MPa were recorded. There was a linear relationship between the net photosynthesis rate and leaf water potential over the range of –1.0 to –3.5 MPa. The three cultivars studied (Arbequina, Manzanillo and Sevillana) all responded to dry conditions in similar ways. In another container experiment leaf conductance was found to be more sensitive to water stress than vegetative growth (Gómez-del-Campo, Reference Gómez-del-Campo2007).

At the same site in south-west Spain referred to above, Fernández et al. (Reference Fernández, Díaz-Espejo, Infante, Durán, Palomo, Chamorro, Girón and Villagarcía2006) subsequently compared the effects of ‘partial root zone drying’ and ‘regulated deficit irrigation’ on various water status indicators, including leaf water potential, stomatal conductance and net CO₂ assimilation rates. No evidence could be found to show that partial root zone drying, which involved keeping half the root zone dry for two to three weeks prior to switching the irrigation from one side of the tree to the other, had a positive effect on any of these variables compared with regulated deficit irrigation. In many species, for example the vine, allowing part of the root zone to dry in this way initiates a signalling mechanism. This triggers partial stomatal closure, which in turn reduces water loss by transpiration, without a proportional loss in yield. For olive, sap flow measurements confirmed that daily water consumption was similar in both deficit irrigation treatments on most days throughout the irrigation period. There was no evidence to show that alternating the water application (by drip) from one side of a tree to the other had any advantages compared with conventional regulated deficit irrigation.

As a result of small reductions in stomatal conductance, trees with part of the root system in a drying soil maintained midday leaf water potentials at levels similar (c. –1.5 MPa) to those in trees in which the whole soil profile was kept close to field capacity (Torres-Ruiz et al., Reference Torres-Ruiz, Fernández, Diaz-Espejo, Martin-Polemo, Morales-Sillero, Muriel and Romero2011). However, in rain-fed trees, large reductions in stomatal conductances did not prevent midday leaf water potentials (between –3.5 and –5 MPa) from falling below those of the irrigated trees (–1 to –2 MPa).

In a complex study, conducted at two contrasting sites, Moriana et al. (Reference Moriana, Pérez-López, Prieto, Ramírez-Santa-Pau and Pérez-Rodriguez2012) attempted to evaluate the possibility of using stem water potential as a guide for scheduling irrigation. The results are not easy to summarise, since the responses varied with site and season. This was an ambitious project, but the practical outcomes are not immediately obvious.

The presence of fruit on an olive tree influences its water status. For example, in an experiment in south-west Spain, mid-morning, stomatal conductances (cv. Morisca) were about 17% higher during the late summer and early autumn, that is from a few weeks after fruit set until harvest, in trees with fruits than in those without fruits. This response was observed at all levels of water supply (Martín-Vertedor et al., Reference Martín-Vertedor, Rodríguez, Losada and Castiel2011a). The lowest midday leaf water potentials recorded were about –4.0 MPa in young trees, and –3.0 MPa in mature trees, both recorded in the early autumn. Leaf water potential values were less in trees with a medium/heavy fruit load than in those without fruit.

In recent years, the so-called ‘magnetic leaf patch clamp pressure probe’ has been successfully used to measure the pressure potential (turgor pressure) within a leaf. A series of papers demonstrate how this new sensor (known commercially as a ZIM probe) has the potential to monitor on a continuous basis the water status of olive trees in an orchard (Fernández et al., Reference Fernández, Rodriguez-Dominguez, Perez-Martin, Zimmerman, Rüger, Martin-Palomo, Torres-Ruiz, Cuevas, Sann, Ehrenberger and Diaz-Espejo2011b; Rodriguez-Dominguez et al., Reference Rodriguez-Dominguez, Ehrenberger, Sann, Rüger, Sukhorukov, Martin-Palomo, Diaz-Espejo, Cuevas, Torres-Ruiz, Perez-Martin, Zimmerman and Fernández2012).

Italy: In southern Italy (41°06′N 14°43′E; alt. 250 m), Tognetti et al. (Reference Tognetti, d'Andria, Morelli, Calandrelli and Fragnito2004) made a sequence of measurements in a field experiment. Diurnal changes in sap flow (compensation heat pulse technique), leaf water potential and stomatal conductances were recorded between May and October 2002 in an 11-year-old orchard (cv. Kalamata grafted on DA121 rootstock; tree spacing 3 × 6 m = 555 trees ha⁻¹) on irrigated and rain-fed trees. Pre-dawn leaf water potentials never fell below –0.5 MPa in irrigated trees, or below –2.5 MPa at midday. Both these values were very close to those reported in Spain. Sap flow rates for irrigated trees peaked at midday, reaching maximum values of 4–5 kg h⁻¹ in June before progressively declining until October. Mean daily stomatal conductance also declined as the summer progressed. There was a positive linear relationship (r² = 0.83, irrigated trees) between hourly sap flow rates and the saturation deficit of the air (range 0 to 3.5 kPa). Similarly, there was a linear correlation (r² = 0.72) between daily sap flow and daily estimates (Penman–Monteith equation) of reference crop ETo (range 0 to 6 mm d⁻¹). The hydraulic conductivity of a tree (represented by the slope of the line relating xylem water potential and sap flow) declined from June onwards, more in the rain-fed trees than in those that were irrigated. A polynomial quadratic curve represented the close relationship between the photosynthetic rate and stomatal conductance (R ² = 0.87).

Morocco: In a ‘partial root-zone drying’ field experiment in Morocco (31°38′N 08°04′W) with mature olive trees (spaced 6 × 6 m = 280 trees ha⁻¹), Centritto et al. (Reference Centritto, Wahbi, Serraj and Chaves2005) found, when similar depths of water were applied, that the leaf water status in trees (irrigated on alternate sides) was similar to that recorded in the control trees (irrigated on both sides). By contrast, applying half that quantity of water on one side of the tree only and switching sides every two weeks, reduced the leaf water potential, but only marginally relative to the 50% water saving. The relative water contents of the leaves were similar, and their photosynthetic capacity was maintained. Wahbi et al. (Reference Wahbi, Wakrim, Aganchich, Tahi and Serraj2005) have described the yield responses in this experiment (see below).

New Zealand: In New Zealand, Greven et al. (Reference Greven, Neal, Green, Dichio and Clothier2009) investigated the impact of short-term water stress on some physiological processes in five-year-old trees (cv. Verdale) growing in a stony silt loam soil with a water-holding capacity of only about 70 mm m⁻¹. Roots did not extend below a depth of 1 m. Using the compensation heat-pulse technique, sap flow rates were monitored regularly over a two-month period. On warm, sunny days, these peaked in irrigated trees at about 3 L h⁻¹, equating to 30 L d⁻¹ tree⁻¹, while the corresponding flow rates in adjacent trees without irrigation, or significant rainfall, for two months declined to 0.4 L h⁻¹. Leaf water potentials in irrigated trees varied between –0.4 MPa at dawn and –2.1 MPa at noon, while stem water potentials did not fall below –1.4 MPa. In unirrigated trees, pre-dawn leaf water potentials declined to less than –4.0 MPa, a very low value that was below the limit of measurement. Upon re-watering, the recovery in sap flow rates was rapid (within seven days), while, because of variability between leaves on the same tree, it took up to 10 days for leaf water potentials to return to pre-drought levels. This is unusual for tree crops, as normally it is the opposite way round, with recovery in water status preceding the recovery in transpiration rates.

Tunisia: In southern Tunisia (34°N 10°E), Masmoudi et al. (Reference Masmoudi, Ayachi, Gouia, Laabidi, Reguaya, Amor and Bousnina2010) recorded midday leaf water potentials and stomatal conductances in five cultivars of six-year-old olive trees over a summer season. There were three deficit irrigation treatments (20, 50 and 100% ETc replacement, where ETc is the potential water use by olives). Due to variability in the data, it is not easy to interpret the results presented. Although there were seasonal and cultivar differences, the responses to the irrigation treatments were inconsistent. Early in the summer, cv. Pincholine had the highest leaf water potentials under both wet (–0.97 MPa) and dry (–1.5 MPa) soil conditions. The corresponding values for the other cultivars were lower (range –1.63 to –2.13 MPa). Later in the season, the leaf water potentials were least for cv. Coratina (–4 MPa) compared with –3 MPa for the other four cultivars. Masmoudi et al. (Reference Masmoudi, Ayachi, Gouia, Laabidi, Reguaya, Amor and Bousnina2010) stated that the recommended threshold stem water potentials at midday for initiating irrigation of olives in central Spain were –1.2 MPa during ‘on’ years and –1.4 MPa during ‘off’ years. It is not clear on what evidence this assertion was based.

Surprisingly, diurnal changes in leaf conductance were judged to be more consistent than leaf water potentials (since sometimes it was difficult to judge when the end point occurred in the pressure bomb). In all cases (cultivars and watering levels), there was evidence of (partial) midday stomatal closure. Conductance fell during the morning, reaching minimum values around midday, before increasing again during the afternoon. There were significant differences in stomatal conductance between cultivars and also between the irrigation deficit treatments (Masmoudi et al., Reference Masmoudi, Ayachi, Gouia, Laabidi, Reguaya, Amor and Bousnina2010). This was not an easy paper to follow.

The main observation that came out of another deficit irrigation experiment in an arid region of Tunisia (34°43′N 10°41′E) was the adverse effects of high temperatures during the summer on photosynthetic rates for both irrigated and rain-fed trees (cv. Chemlali). Mean monthly maximum daily temperatures were 34.5 °C (June), 37 °C (July) and 39 °C (August). It was not until the weather cooled in September and October (31 °C) that the gas exchange processes (stomatal conductance, photosynthesis and transpiration) in the irrigated trees returned to values recorded in the spring. The more severe the water stress experienced the slower the rate of recovery. On a diurnal basis in June, the stomata were partly open early in the morning, but from 0800 h onwards there was progressive stomatal closure for the remainder of the day. A detailed analysis showed that for an irrigated tree the optimum temperature for photosynthesis was 30 °C (corresponding to a photosynthetically active radiation intensity of 1250 μmol m⁻² s⁻¹), declining rapidly to almost zero at 40 °C. For a severely water-stressed tree the temperature response curve was largely flat, at a low photosynthetic rate, before peaking at about 28 °C. Ahmed et al. (Reference Ahmed, Rouina and Boukhris2007) refer to the response of the olive to conditions in the summer period as ‘a rest phase’.

Trunk diameter variability (TDV)

Reports of olive TDV use a number of terms that are defined in Box 2 in an attempt to minimise confusion.

Spain: During 2009, the usefulness of maximum trunk diameter measurements to detect water stress in mature, 12-year-old olive trees (cv. Arbequina), heavy with fruit, was studied in a commercial orchard in south-west Spain (37°30′N 05°44′W; alt. 60 m) by Fernández et al. (Reference Fernández, Torres-Ruiz, Diaz-Espejo, Montero, Alvarez, Jiménez, Cuerva and Cuevas2011c). There were two treatments: One in which the soil was irrigated to field capacity twice a week, and another in which the soil was allowed to dry until the fruits were observed to shrivel, whereupon the same cumulative total quantity of water was applied as in the first treatment. In addition, there were two control trees that were supplied with excess water (1.5 ETc) to ensure that they were never short of water. The potential ETc was determined using the following relationship:

\begin{equation*} {\rm ETc} = {\rm Kc}\; \times \;{\rm Kr}\; \times \;{\rm ETo},\end{equation*}

The ‘trunk diameter variations’ were analysed by Fernández et al. (Reference Fernández, Torres-Ruiz, Diaz-Espejo, Montero, Alvarez, Jiménez, Cuerva and Cuevas2011c) in a number of different ways with the overall aim of identifying an index that could usefully and reliably identify water stress in an orchard in south-west Spain. They found that the most useful approach was to calculate, on a daily basis, the difference between the maximum trunk diameter in a control well-watered tree and that of the tree under comparison. This is referred to as the ‘maximum trunk diameter difference’. Its value mimicked closely changes in the water status of the soil and plant, in contrast to the so-called signal intensity for ‘maximum daily shrinkage’ (=actual value/reference value) with which it was compared. For each comparison, four instrumented trees were enough to provide reliable data from which to detect the onset and severity of water stress for olive trees with a heavy fruit load (cv. Arbequina).

In a similar comparison, again in Spain, of the different ways in which ‘trunk diameter variations’ can be expressed (Box 2), Cuevas et al. (Reference Cuevas, Torres-Ruiz, Álvarez, Jiménez, Cuerva and Fernández2010) came to the conclusion that, for trees with large root zones (low/medium planting density, old trees), the various derived indices were of little value where the aim was to keep the soil profile close to field capacity. However, the ‘maximum daily shrinkage’ and ‘daily recovery’ indices could be useful indicators to avoid fruit shrivelling, since this can adversely affect oil quality. In the same study, good relationships (linear) were obtained, for well-watered trees, between ‘maximum daily shrinkage’ and maximum daily air temperature (r² = 079–0.81), and also with the maximum daily saturation deficit of the air (r² = 0.74–0.80).

According to Moriana et al. (Reference Moriana, Fereres, Orgaz, Castro, Humanes and Pastor2000), the relationship between the maximum daily shrinkage and stem water potential had two components. Under conditions of mild water stress (stem water potential at midday above –1.5 MPa) the response was linear, i.e. the maximum daily shrinkage increased as the stress level increased. But at stress levels greater than –1.5 MPa, the maximum daily shrinkage got less. In other words, there was not a straightforward relation between these two parameters. However, maximum daily shrinkage and weather variables such as temperature (in particular) and the saturation deficit of the air were correlated. Moriana et al. (Reference Moriana, Moreno, Girón, Conejero, Ortuño, Morales, Corell and Torrecillas2011) analysed daily weather and trunk shrinkage data, collected over four successive years (2005–2008), from fully irrigated, 37-year-old trees (cv. Manzanillo). Data from the fifth year (2009) were used to validate the relationships. The best correlations (all were linear) were between the maximum daily shrinkage of the trunk and the air temperature at midday (very highly significant, r² = 0.85; range c. 16 to 40 °C)) and between the maximum daily shrinkage and the saturation deficit of the air (range 0 to 6 kPa), also at midday (r² = 0.81). There were seasonal differences in the relationships; the slope and/or intercept varied. But for years when the fruit loads (‘on’ years and ‘off’ years) were similar, the correlations were also similar. It was expected that benchmarks of this nature would help facilitate the use of plant-based indicators as tools to improve irrigation water management.

In a further attempt to develop plant-based indices for detecting the onset and severity of water stress in mature olive trees with a heavy fruit load, Fernández et al. (Reference Fernández, Moreno, Martin-Palomo, Cuevas, Torres-Ruiz and Moriana2011a) analysed the relations between sap flow, trunk diameter variation, midday stem water potential, relative soil water content and atmospheric demand. The measurements were made in a deficit irrigation experiment during an ‘on’ year (2006) in an orchard planted with large, 38-year-old trees (cv. Manzanillo de Sevilla, spaced 7 × 5 m = 285 trees ha⁻¹). The main conclusion from this detailed analysis was that the water needs of an orchard like the one described could be best assessed through the simultaneous use of the daily difference in the maximum trunk diameter of a stressed tree and a control well-watered tree, together with the corresponding difference in the daily water use (sap flow, L tree⁻¹ d⁻¹). Using other derivatives such as signal intensity offered no advantages. The number of instrumented trees still needed to be assessed.

Cuevas et al. (Reference Cuevas, Martín-Palomo, Díaz-Espejo, Torres-Ruiz, Rodriguwez-Dominguez, Perez-Martin, Pino-Mejías and Fernández2012) extended the search for a suitable indicator to a high-density hedgerow olive orchard (1667 trees ha⁻¹; cv. Arbequina). Based on measurements made in a regulated deficit irrigation experiment with four-year-old trees with a heavy fruit load in south-west Spain (37°15′N 5°48′W), they concluded that the daily difference in water use was a better indicator of when to irrigate than the daily difference in maximum trunk diameter. This could either be used alone (although care was needed in interpreting the data), or in association with the daily difference in the maximum trunk diameter. The maximum daily shrinkage was not a reliable water stress indicator. The search goes on! But, since even relatively simple irrigation scheduling techniques are not being widely used by irrigators in commercial orchards, it can be questioned whether any of these sophisticated techniques is likely to be taken up in practice.

Osmotic regulation

Greece: Previously, Chartzoulakis et al. (Reference Chartzoulakis, Bosabalidis, Patakas and Vemmos2000) had demonstrated in a pot (volume 50 L) experiment lasting two years in Crete (Greece) the occurrence of osmotic regulation in severely stressed olive plants (cv. Koroneiki; pre-dawn leaf water potential down to –4.8 MPa). This was a result of solute accumulation (mainly mannitol and to a lesser extent glucose). There were also increases in the cell wall elasticity and reductions in the volume of intercellular spaces. All of these attributes contributed to drought tolerance by maintaining cell turgor.

Tunisia: In a well-written paper, Ennajeh et al. (Reference Ennajeh, Tounekti, Vadel, Khemira and Cochard2008) reported the results of an experiment in southern Tunisia (33°50′N 10°05′E) in which the responses to drought of two contrasting cultivars, Meski and Chemlali, were compared. In a shaded, greenhouse study, the relative contributions to drought tolerance of osmotic adjustment and resistance to xylem cavitation (embolism) were assessed over a 60-day period. Several studies (cited by Ennajeh et al., Reference Ennajeh, Tounekti, Vadel, Khemira and Cochard2008) had shown that osmoregulation (by means of the synthesis and accumulation of solutes in cells) occurs in olive. By maintaining a positive pressure potential in the cell, the water potential gradient between the soil and the leaf is maintained, thereby facilitating water uptake from the soil. Another drought adaptation of olive trees is the capacity of the leaves to release about 60% of the water stored in their tissues at full turgor to transpiration.

The value of osmotic adjustment can however be negated during a drought by cavitation in the xylem vessels, which reduces the hydraulic conductivity of the xylem. Ennajeh et al. (Reference Ennajeh, Tounekti, Vadel, Khemira and Cochard2008) found that cv. Chemlali had a greater capacity than cv. Meski for osmotic adjustment and that both cultivars exhibited embolism. The xylem potential inducing a 50% loss of stem conductivity approached –7 MPa in cv. Meski but was slightly less (more negative) than this in cv. Chemlali, suggesting that cv. Meski was more vulnerable to embolism than cv. Chemlali. In addition, cv. Meski maintained a slightly higher stomatal conductance and transpiration rate than cv. Chemlali throughout the period of severe drought. There was also a close (linear) relationship between pre-dawn leaf water potential and soil water potential (at a depth of 0.20 m) for both cultivars, with a small divergence from linearity for cv. Chemlali under very dry conditions. Although the differences were all small, they confirmed the superior drought resistance of cv. Chamlali compared with cv. Meski as previously reported by Ennajeh et al. (Reference Ennajeh, Vadel, Khemira, Ben Mimoun and Helali2006). Measurement of resistance to embolism may provide an effective way for screening cultivars suited to dry areas (Ennajeh et al., Reference Ennajeh, Tounekti, Vadel, Khemira and Cochard2008).

Summary: plant water relations

1. 1. Olive trees conserve water under conditions of high evaporative demand by closing their stomata.

2. 2. Stomatal conductances (generally) peak in the morning and then progressively decline during the remainder of the day.

3. 3. Rates of photosynthesis and transpiration (and instantaneous transpiration efficiencies) follow a similar diurnal trend, except under conditions of high evaporative demand when transpiration peaks later in the day than photosynthesis.

4. 4. Mean daily stomatal conductances decline as the summer progresses.

5. 5. In well-watered trees, rates of photosynthesis decline rapidly at temperatures greater than 28–30 °C, reaching zero at 40 °C.

6. 6. In well-irrigated trees, pre-dawn leaf water potentials remain above –0.5 MPa, and at midday above –2.5 MPa.

7. 7. Under dry conditions, midday stem water potentials can fall as low as –8 MPa, which is much lower than those experienced by most crop plants subjected to drought. The presence of fruit on a tree reduces the midday leaf water potential.

8. 8. There is a positive linear relation between rates of sap flow and the saturation deficit of the air.

9. 9. The hydraulic conductivity of a tree declines as the summer progresses, more so in rain-fed trees than in those that are irrigated.

10. 10. There is no evidence that partial root zone drying offers any advantages over regulated deficit irrigation in terms of tree water status.

11. 11. There have been many attempts to develop ways of exploiting the sensitivity of trunk expansion to water availability as a means of scheduling irrigation. No clear consensus has yet emerged.

12. 12. Maximum daily trunk shrinkage is correlated with maximum/midday daily temperature and maximum/midday daily saturation deficit.

13. 13. Among all the tree indicators related to water status, the most promising for detecting water stress in an olive orchard are the difference in the maximum trunk diameter of a stressed tree and a control well-watered tree, and the corresponding difference in sap flow rates, after calibration of the sensors (both calculated on a daily basis).

14. 14. Osmotic regulation (due to solute accumulation) and resistance to cavitation (due to narrow xylem vessels) contribute to the capacity of the olive to withstand drought. Cultivars differ in these attributes.

Transpiration

Spain: In southern Spain (37.85°N 4.80°W; alt. 110 m), Villalobos et al. (Reference Villalobos, Perez-Priego, Testi, Morales and Orgaz2012), using large closed chambers in an olive orchard, monitored carbon exchange and transpiration of individual trees (cv. Arbequina, spaced 3.5 × 7 m = 408 trees ha⁻¹) during 2006 and 2007. Canopy conductance peaked in the early morning (0800–0900 h GMT) and then declined until sunset in both well-watered and the deficit-irrigated trees. By contrast, transpiration by the well-watered trees reached a maximum in early afternoon (1300–1400 h GMT) and declined from 1600 h onwards. Daily transpiration rates equated to 0.6 to 3.4 mm d⁻¹. For deficit-irrigated trees, transpiration was virtually constant at a lower level (0.4 to 2.9 mm d⁻¹) from 0900 to 1600 h. For comparison, ETo varied between 1.1 and 7.1 mm d⁻¹. Net assimilation peaked early in the morning (0800–0900 h), before declining steeply from about 1500 h. Daily total assimilation varied from 10.8 to 21.3 g CO₂ d⁻¹. Instantaneous water-use efficiency, which was inversely related to the saturation deficit of the air (range 0 to 5.8 kPa), declined throughout the day, from about 30 g CO₂ L⁻¹ at sunrise down to 4–7 g CO₂ L⁻¹ at sunset.

Argentina: The olive industry is developing rapidly in Argentina, particularly in the arid areas in the north-west of the country. In the past 15–20 years, large (>100 ha) commercial olive orchards have become common (Rousseaux et al., Reference Rousseaux, Benedetti and Searles2008). By contrast to the Mediterranean region, there is no winter rainfall in this part of Argentina. During the summer, the total rainfall received is 100–400 mm. In Rioja province, irrigation is practised throughout the year with water obtained from subsurface aquifers 100–200 m below the surface. It is not known whether recharge of the aquifers is sufficient to replace the water being used for irrigation. Meanwhile, global climate change models predict a reduction in precipitation in this region (Rousseaux et al., Reference Rousseaux, Figuerola, Correa-Tedesco and Searles2009).

This is the context in which Rousseaux et al. (Reference Rousseaux, Benedetti and Searles2008) began to investigate the leaf level responses of olive (cv. Manzanillo) to irrigation over a 40-day period in the winter (La Rioja province, 28°33′S 66°49′W; alt. 800 m). During this time period, there were only small reductions in transpiration rates and leaf conductances in the unirrigated trees compared with those that were well irrigated. Reference crop ETo (FAO, Penman) during the first four weeks averaged 2–3 mm d⁻¹, increasing to 4–5 mm d⁻¹ towards the end of the experiment when the soil profile to a depth of 0.90 m was close to permanent wilting point. Irrespective of the irrigation regime there was a strong curvilinear relationship between leaf conductance and the saturation deficit of the air, with conductance declining rapidly as the saturation deficit (recorded at midday) increased from 2 to 3 kPa, and declining less rapidly from 3–6 kPa. The stomata of both well-irrigated and dry trees behaved in identical ways.

Evapotranspiration

Spain: Recognising the need for a methodology that would allow reasonable estimates of Kc to be determined for a range of practical situations, Testi et al. (Reference Testi, Villalobos, Orgaz and Fereres2006) developed a model for determining monthly Kc values that took into account four components: tree transpiration, direct evaporation of water intercepted by the crop canopy, evaporation from the soil surface and evaporation from the areas wetted by drip irrigation emitters. The model was validated against actual measurements of ET made over three years (1998–2000) in a young olive orchard (4 ha) in Cordoba, Spain, using both the eddy correlation technique and the water balance method (Testi et al., Reference Testi, Villalobos and Orgaz2004). The model was then used to simulate the water use (and its principal components) by an olive orchard at two sites, Cordoba and Fresno, in California, and for two hypothetical case studies, a traditional low-density (100 trees ha⁻¹) orchard converted to irrigation, and an intensive (300 trees ha⁻¹) modern orchard. The annual average values for Kc were 0.57 (low density) and 0.63 (high density), and monthly values for July (mid-summer) were 0.49 and 0.53. These are comparable with those values obtained by direct measurement in a young orchard (trees were from one to four years old): for a ‘typical’ summer day, without irrigation, Kc = 0.15, when the crop cover was about 5%, and 0.3 at 25% ground cover. When the wet soil surface patches associated with drip irrigation were taken into account the corresponding values for Kc were 0.22 and 0.35 respectively (Testi et al., Reference Testi, Villalobos and Orgaz2004). In the winter, when the soil surface was frequently wetted by rain, Kc sometimes reached 0.8.

In addition to the standard, representative weather data needed to calculate ETo, to run the model it is necessary to know the tree density and canopy volume, the ground area wetted by the emitters and the frequency of irrigation/rainfall events. Using 20 years of daily data from Andalusia, Spain, the same researchers (Orgaz et al., Reference Orgaz, Testi, Villalobos and Fereres2006) presented examples of the model's outputs for five different scenarios: a traditional olive orchard at immature and mature stages of development, a semi-intensive orchard (spacing 7 × 7 m = 204 trees ha⁻¹) also at two stages of development and a modern, highly intensive orchard (spacing 7 × 3.5 m = 408 trees ha⁻¹). The range of values for the four components of the annual water requirements covering these five simulations were as follows: tree transpiration 156–708 mm (32–65% of total ETc), soil surface evaporation 223–283 mm (25–53% of total ETc), evaporation from wet soil 23–63 mm (5–7% of total ETc) and intercepted rainfall evaporation 19–72 mm (4–10% of ETc). For irrigation scheduling purposes, the model could be used to predict water use by using long-term average weather data, adjusted as necessary at the end of the month.

Over a six-year period (2002–2007), the impact that the number of fruits on an olive tree (the crop load) had on its crop water status, ETc and productivity was studied in south-west Spain (38°51′N 6°40′W; alt. 200m) by Martín-Vertedor et al. (Reference Martín-Vertedor, Rodríguez, Losada and Castiel2011a, Reference Martín-Vertedor, Rodríguez, Losada and Castielb). There were two experiments, one with young trees and another with more mature, annually pruned trees (both cv. Morisca), in which deficit irrigation (including excess irrigation) was imposed (experiment 1: 75, 100 and 125% of ETc; experiment 2: 60, 100 and 115% of ETc). Crop loads were added as a variable varied, either artificially by thinning the fruit, or by comparing ‘on’ and ‘off’ years. The following observations were made:

• • The FAO method (Doorenbos and Pruitt, Reference Doorenbos and Pruitt1974), with Kc proposed by Pastor et al. (Reference Pastor, Orgaz, Vega, Hidalgo and Castro1998) and Kr taken from Fereres et al. (Reference Fereres, Martinich, Aldrich, Castel and Schulbach1982), underestimated ETc in the control (100% replacement) treatment.

• • A better estimate of ETc was obtained by using the method proposed by Orgaz et al. (Reference Orgaz, Testi, Villalobos and Fereres2006).

• • Over the period from flowering to fruit colour change, ‘on’ year trees depleted more water from the soil profile (to a depth of nearly 1.0 m) than ‘off’ year trees.

• • The absence of fruits reduced ETc compared with medium and high crop loads, but canopy size was a much more important determinant of orchard water requirements than crop load.

Run-off and run-on: As Castro et al. (Reference Castro, Romera, Gómez and Fereres2006) showed, rainfall redistribution in an orchard could be substantial because of localised run-off and run-on induced by the spatial variability of infiltration rate, rainfall interception by the trees and surface roughness. A physically based model, calibrated locally (Cordoba, southern Spain, average annual rainfall of about 550 mm), predicted annual cumulative infiltration differences of 40 to 100 mm between the area beneath the trees (olive) and the adjacent bare soil inter-row area. Such differences could represent fruit yields equivalent to 40 mm × 25 kg ha⁻¹ mm⁻¹ (=1000 kg ha⁻¹) and 100 mm × 25 kg ha⁻¹ mm⁻¹ (=2500 kg ha⁻¹) (see below). These yield gains are sufficiently large to justify developing rainwater run-off structures for rain-fed olive orchards.

Argentina: Following the previous work by Rousseaux et al. (Reference Rousseaux, Benedetti and Searles2008), summarised above, the same team undertook a detailed study of the actual water use of drip-irrigated olives (cv. Manzanillo fina, spaced 4 × 8 m = 312 trees ha⁻¹; crop cover 23%) over a 12-month period with the aim of specifying appropriate values for Kc for use in north-west Argentina (28°33′S 66°49′W; alt. 800 m) (Rousseaux et al., Reference Rousseaux, Figuerola, Correa-Tedesco and Searles2009). Transpiration was monitored using the sap flow (heat balance) method. Values obtained ranged from highs of 2.0 mm d⁻¹ in the mid-spring and summer to 0.23 mm d⁻¹ in the winter. Direct evaporation from the soil surface (based on data from micro-lysimeters) varied between 0.2 and 0.6 mm d⁻¹, depending on the area of soil surface that was wetted and the frequency and duration of wetting. For a well-irrigated orchard, transpiration represented 0.45–0.60 × ETo (except in the winter when values were close to 0.15), and evaporation from both the tree zone and the inter-row areas averaged about 0.20 × ETo. These values were based on Penman–Monteith estimates of ETo using weather data collected over a bare soil (i.e. not grass). On average, transpiration represented about 70–80% of potential crop evapotranspiration (ETc), except when the soil surface was kept exceptionally wet. Daily transpiration per unit of leaf area was linearly related to daily ETo values (range 2 to 10 mm d⁻¹; r² = 0.84) and solar radiation (range 10 to 30 MJ m⁻² d⁻¹; r² = 0.80), but exponentially related to average daily saturation deficits of the air (range 0.5 to 3.5 kPa; R ² = 0.85). There was a suggestion that the relationship with daily mean temperature was two-stage linear, with no transpiration at temperatures below 13 °C but linear from 13 to 30 °C (r² = 0.93). There was good agreement between estimated transpiration rates based on sap flow measurements and values calculated from the model proposed by Orgaz et al. (Reference Orgaz, Villalobos, Testi and Fereres2007).

Australia: In Australia, the olive industry has to compete with established irrigated crops such as grape vines for the scarce water resource. Prudent management of water is essential for a viable olive industry. This was the context in which the water use by four, carefully selected commercial olive orchards (groves) was monitored over two growing seasons (1999/2001) in South Australia (c. 34°S 140°E; alt. 300–500 m) (Nuberg and Yunusa, Reference Nuberg and Yunusa2003). The trees were >10 years old and densities varied between 133 and 204 trees ha⁻¹. The soil textures varied between deep sandy, sandy loam over clay, clay loam over clay and clay loam over silty clay. Changes in soil water content (measured by using a neutron probe), from which ET was determined by using a simple water balance equation, light interception (leaf area index and ceptometer) and sap flow (heat pulse sensors, on one site over one season only), were monitored. Transpiration was estimated using an energy balance, micro-meteorological approach together with calculated values for stomatal conductances. The Priestley–Taylor equation was used to estimate evaporation from an open water surface (Eo) for the whole district using local weather data. The validity of the sap flow data was confirmed by a linear regression between sap flow and daily estimates of transpiration. In mid-summer, when daily average maximum temperatures and saturation deficits of the air reached 30 °C and 2 kPa respectively, Eo peaked at close to 10 mm d⁻¹. Across the three irrigated orchards, total water use over a season (ET) was in the range 490–670 mm. By comparison on the rain-fed orchard, ET averaged only 316 mm. It was judged that not enough water was applied during mid-summer by the irrigators to match potential transpiration rates and that the orchard had been under-irrigated by about 15%. The full water requirement was probably 900 mm at one orchard where there was a permanent cover crop, and 700 mm at the two orchards where drip irrigation was used.

On average, where microsprinklers were used, transpiration represented 0.63 × ET, but 0.76 × ET where drip irrigation was practised and there was less surface evaporation. Peak rates of transpiration reached 120–147 L tree⁻¹ day⁻¹. Crop coefficients based on the Priestley–Taylor estimate of Eo averaged 0.38 on the irrigated estates and 0.23 on the rain-fed estate. (Note that in this case Kc = ET/Eo, not ETc/ETo as used elsewhere). Soil water extraction did not occur below 2 m depth. There was evidence that the stomata closed from midday onwards on hot, dry days. Estimates were provided of the monthly water requirements, and corresponding Kc values, for a drip-irrigated orchard at different stages of development (10, 20, 40 and 60% ground cover) (Nuberg and Yunusa, Reference Nuberg and Yunusa2003). This was an interesting approach to obtaining useful information in situations where funds and facilities did not exist for conventional field experiments.

Morocco: In a semi-arid area region of Morocco (31.601°N 7.974°W), Er-Raki et al. (Reference Er-Raki, Chebouni, Boulet and Williams2010) evaluated the FAO-56 dual approach (Allen et al., Reference Allen, Pereira, Raes and Smith1998) for estimating separately the soil evaporation and crop transpiration components of actual ET in a large (275 ha), flood-irrigated olive orchard (240-year-old trees, density 225 trees ha⁻¹). In a detailed comparison over two years (2003 and 2004), model simulations were compared with measured values of ET (eddy correlation technique), and transpiration (scaled-up from sap flow measurements). The agreement between the two approaches was considered to be good. The value so derived for the basal crop coefficient (K_cb, transpiration component only) was 0.54 × ETo (where ETo was calculated from the Penman–Monteith equation). Using this dual approach, Er-Raki et al. (Reference Er-Raki, Chebouni, Boulet and Williams2010) evaluated the irrigation scheduling practices used by one farmer on the project. This suggested that although twice as much water was being applied in a growing season (800 mm) than was needed (411 mm), the olive trees still experienced water stress during the summer when ETo values were at their peak.

Portugal: In Portugal, the olive is traditionally a rain-fed crop and plays an important role in the rural economy. Existing orchards are now being irrigated and new ones created. The questions to be answered include how best to schedule irrigation in an orchard that for many years previously had been rain-fed only. This was the justification for a deficit irrigation experiment in southern Portugal described initially by Santos et al. (Reference Santos, Valverde, Ramos, Reis and Castanheira2007). Sap flow was measured at 30-minute intervals for four months in 2005 and for eight months from March to the end of August in 2006 in a low-density (12 × 12 m = 70 trees ha⁻¹), previously rain-fed orchard planted with mature (cv. Cordovil, >80-year-old) trees. The results for individual trees were scaled up in order to estimate transpiration from an equivalent orchard by monitoring the sapwood and canopy areas of randomly chosen trees and comparing these values with those of the sampled trees. In 2006, transpiration in the fully irrigated treatment totalled 700 mm, in the 60% sustained deficit replacement treatment the total was 750 mm, in the regulated deficit treatment the total was 650 mm and in the rainfall-only control the total was 400 mm. Rainfall over the eight months totalled 240 mm, the ETo total was 1000 mm, while the depths of irrigation water applied were 880, 450 and 70 mm respectively (Ramos and Santos, Reference Ramos and Santos2009). In the well-irrigated trees the T/ETo ratios declined over the summer months from 0.75–0.8 in March/April to 0.6 in June/July, before increasing to 0.9 in September. This reflected the effect of the high-potential ETo (=7–8 mm d⁻¹) experienced in mid-summer, which exceeded the capacity of the trees to transfer water from the wet soil to the sites of evaporation in the leaf at the needed rate (Santos et al., Reference Santos, Valverde, Reis, Ramos and Castanheira2012). The implications of the results of this experiment in terms of yield are considered below.

Summary: crop water requirements

The principal source for each summary statement is given at the end of the sentence.

1. (a) Transpiration

   1. 1. For a well-watered crop, the canopy conductance is low early in the morning, reaches a peak in mid-morning and then progressively falls for the remainder of the day (Spain).

   2. 2. Because of strong coupling between the olive tree and the bulk atmosphere, the aerodynamic conductance is many times greater than the canopy conductance.

   3. 3. Transpiration reaches a maximum in the early afternoon and then declines from 1600 h onwards; net assimilation of CO₂ peaks early in the morning, and declines steeply from mid-afternoon onwards (Spain).

   4. 4. Instantaneous water-use efficiency declines throughout the day; it is inversely related to the saturation deficit of the air (Spain).

   5. 5. There is a close relationship (linear, Spain, Moriana et al., Reference Moriana, Villalobos and Fereres2002; curvilinear, Argentina) between leaf conductance and the saturation deficit of the air irrespective of the irrigation regime.

   6. 6. After re-watering droughted trees, stomatal conductance and leaf water potential usually return to pre-drought levels within two days (Spain), but it has taken longer elsewhere (New Zealand). Sap flow rates can take up to seven days.

1. (b) Evapotranspiration

   1. 7. It is not appropriate or indeed possible to specify a single value of Kc for a widely spaced, evergreen tree crop like the olive.

   2. 8. When estimating ET, it is necessary to take into account tree transpiration, direct evaporation of water intercepted by the leaf canopy, evaporation from the soil surface and evaporation from the areas wetted by the drip or sprinkler systems.

   3. 9. The seasonal values of Kc to be used as a first approximation for the computation of ETc for mature olive orchards with a crop cover of 50–60% in semi-arid areas with Mediterranean-type climatesFootnote ² and arid locations are listed in Table 2 (Fereres et al., Reference Fereres, Villalobos, Orgaz and Testi2011).

   4. 10. Trees with a heavy fruit crop deplete more water from the soil profile than low yielding trees (Spain).

   5. 11. Because of localised run-off and run-on, rainfall redistribution in an orchard can be substantial (Spain).

   6. 12. In a well-irrigated orchard, transpiration represents 0.70–0.80 × ETc, or 0.45—0.60 × ETo (except in the winter) (Argentina, Morocco).

   7. 13. The T/ETo ratio can change over the growing season (in the northern hemisphere) from as low as 0.4–0.5 up to 0.75–0.80 in March/April to 0.60 in June/July before increasing to 0.90 in September. This reflects the effect of high evaporative demand on stomatal conductance (Portugal and others).

   8. 14. Total water use (ET) over a season (depending on how this is defined) is in the range 490–750 mm; of which transpiration is 0.76 × ET with drip irrigation, and 0.63 × ET with micro-sprinklers (Australia, Portugal).

Table 2.

Values of Kc for olive orchards in semi-arid and arid locations (figures from Fereres et al., Reference Fereres, Villalobos, Orgaz and Testi2011).

Summary: water productivity

1. 1. Numerous experiments have been done to compare the effect of sustained deficit irrigation and regulated deficit irrigation with rainfall only and/or fully irrigated control treatments.

2. 2. This is in response to the international need for evidence to justify using scarce water supplies to irrigate olives, traditionally a rain-fed crop.

3. 3. A confounding factor in this work is the wide disparity in the tree densities (from <100 to >1000 trees ha⁻¹) represented.

4. 4. No one appears to have looked systematically at the interaction between soil water availability and tree density in olive orchards.

5. 5. Other complications include biennial bearing, and initial variability in tree size, canopy volume and fruit load. For comparative purposes, trees with similar properties need to be selected.

6. 6. With deficit irrigation experiments it is always necessary to specify the control (100% replacement of ETc) treatment with precision, as this defines the benchmark against which other treatments are judged.

7. 7. Deficit irrigation (when supply is less than ETc) limits vegetative growth; this is particularly important for tree management (Spain, Argentina and Italy).

8. 8. Deficit irrigation also slows the rate of fruit expansion. Fruit yields peak with less than full replacement of water (when Kc = 0.7–0.85). Compensatory fruit growth occurs when the rains begin or when irrigation is resumed. Fruit number is reduced when Kc < 0.7. (Argentina and Spain).

9. 9. Oil yield is reduced less than fruit yield for a given irrigation deficit. The quantity of oil produced per unit of intercepted solar radiation is the same regardless of treatment (Spain).

10. 10. There is no convincing evidence that the timing of regulated deficit irrigation is important; fruit yields are proportionate to the amount of water applied (Israel).

11. 11. Similarly, there is no evidence (from one experiment only) that there are any benefits from partial root-zone drying (Morocco).

12. 12. Fruit yields under sustained deficit irrigation are similar to those from regulated deficit irrigation (mid-summer): deficit irrigation delays ripening compared with full irrigation.

13. 13. There is no evidence of an adverse effect of irrigation on oil quality (Portugal) and there can be a net benefit (Tunisia). Reduction in oil concentration in well-irrigated fruits is more than compensated for by increase in fresh weight (Spain, Israel and USA). Sometimes, the oil concentration does not change (Italy and Morocco).

14. 14. Water productivities are summarised in Table 3. Curvilinear relations were developed between yields of both fresh fruit and oil (Y, t ha⁻¹) over two years and ET (mm) (Spain).

Table 3.

Summary table showing yield (oil and fresh fruit) responses to water (kg ha⁻¹ mm⁻¹) in terms of total depth of irrigation water applied (I), or the equivalent totals for evapotranspiration (ET) and transpiration (T).

Water productivity: FAO reconciliation

In the FAO Irrigation Drainage Paper 33, Doorenbos and Kassam (Reference Doorenbos and Kassam1979) gave a value for water productivity of fresh olives (30% water content) of 15–20 kg ha⁻¹ mm⁻¹. The corresponding figures for oil were 0.3–0.5 kg ha⁻¹ mm⁻¹. In the recent FAO Paper 66, the (relative) fruit yield response curve to (relative) ET developed in Spain by Moriana et al. (Reference Moriana, Orgaz, Pastor and Fereres2003) was found, within the margins of error, to fit data for other cultivars and locations (Fereres, Reference Fereres, Goldhamer, Sadras, Steduto, Hsiao, Fereres and Raes2012). For relatively small applications of water (100–150 mm), the water productivity was about 30 kg ha⁻¹ mm⁻¹, for example at replacements between 0.5 × ETc and 0.8 × ETc, but the yield response then tapered off rapidly between 0.8 × ETc and 1.0 × ETc.

Water quality

The olive tree is traditionally considered to be moderately tolerant of salinity (e.g. Ayers and Westcot, Reference Ayers and Westcot1985; Maas and Hoffman, Reference Maas and Hoffman1977), a view that has been largely confirmed by the results of a long-term (1998–2006) field experiment in Spain (37.28°N 4.26°W). This was designed to assess the effect of low-quality irrigation water on the growth and yield of the olive tree. Irrigation water with three levels of salt concentration (0.5, 5 or 10 dS m⁻¹, mixtures of sodium chloride and calcium chloride) was applied to mature olive trees (cv. Picual; spacing 7 × 7 m = 204 trees ha⁻¹) by daily drip irrigation during the summer months for nine years (Melgar et al., Reference Melgar, Mohamed, Serrano, García-Galavís, Navarro, Parra, Benlloch and Fernández-Escobar2009). In addition, there were two irrigation regimes, one in which the soil was maintained close to field capacity and another in which 20% extra water was applied (a leaching fraction) to wash out any salts from the soil.

Numerous growth parameters were monitored throughout that extended period but no differences between treatments were identified. The only exception was that the oil content (on a fresh weight basis) increased (significantly) by two to three percentage points with the increase in salt content of the irrigation water in four of the nine years. It was considered that in areas with similar rainfall amounts (long-term average c. 700 mm) and distribution (mainly during the winter, only c. 30 mm from June to September, inclusive) it is possible to use saline water (up to 10 dS m⁻¹) to irrigate olive trees. The wisdom of this advice is constrained by the view that the cultivar used in this experiment (cv. Picual) is relatively salt-tolerant compared with the many other genotypes that are grown commercially (Melgar et al., Reference Melgar, Mohamed, Serrano, García-Galavís, Navarro, Parra, Benlloch and Fernández-Escobar2009).

A great deal of research has been undertaken on the mechanisms of salt tolerance in olive, and why cultivars differ in their responses. Much of the work has been done on seedlings under controlled conditions (e.g. see Cimato et al., Reference Cimato, Castelli, Tattini and Traversi2010; Gucci and Tattini, Reference Gucci and Tattini1997) and its relevance to long-term field conditions is uncertain. A recent paper described the responses of two contrasting mature cultivars (Barnea, salt tolerant, and Proline, salt sensitive) to the application of saline water (drip irrigation) over a period of 11 years. The tolerance of cv. Barnea to salinity was believed to be associated with the relatively large mass of fine roots that developed around the emitters (compared with cv. Proline). This enabled water uptake to occur under high evaporation rates despite the saline conditions (Rewald et al., Reference Rewald, Rachmilevich, McHue and Ephrath2011).

Another aspect of water quality as it affects olives involves its role in disease transmission. A recent study in southern Spain has confirmed that irrigation water plays a part in the dispersal of the Verticillum wilt disease of olive (García-Cabello et al., Reference García-Cabello, Pérez-Rodriguez, Blanco-López and López-Escudero2012). This threatening disease (Verticillium dahliae Kleb) is particularly prevalent in the Guadalquivir Valley where there are 1.5 million ha of olive orchards, of which about 240,000 ha are irrigated. Verticillium is a threat to the olive industry worldwide as a result of the conversion of rain-fed orchards to irrigation. The pathogen survives in the sediment and is distributed in soil and plant particles suspended in the irrigation water. It has been found in all the components of the irrigation water distribution system, from the pumping station, through the main canal to the soil surrounding the emitters on a drip irrigation lateral pipe. Other crops within the irrigation schemes (e.g. cotton) act as hosts to the disease.

Irrigation strategies

There are three ‘deficit irrigation’ strategies that have been used successfully with olives (Fereres, Reference Fereres, Goldhamer, Sadras, Steduto, Hsiao, Fereres and Raes2012): First, ‘sustained deficit irrigation’, in which the planned deficit is spread evenly over the season; second, ‘regulated deficit irrigation’, when the deficit is concentrated in the summer months (e.g. from pit hardening to the end of the summer); third, alternating short cycles of water stress and relief depending on the growth stage. As Fereres (Reference Fereres, Goldhamer, Sadras, Steduto, Hsiao, Fereres and Raes2012) stated, and as this review confirms, there is no evidence of the superiority of one strategy over another, it all depends on local circumstances and the need to limit the amount of water used. Where the irrigation system is the limiting constraint, any one of these strategies could be implemented. For soils with low water availability, regulated deficit irrigation is likely to work better than sustained or continuous deficit irrigation. This is also probably the most appropriate strategy in years with well-distributed late summer rainfall, or when there is a shortage of water delivered to the farm. In general, and particularly in years with no summer or early autumn rainfall (as can often occur in southern Portugal, for example) the sustained deficit irrigation regime may be the better option for scheduling irrigation (Ramos and Santos, Reference Ramos and Santos2009). A contributing factor to decision-making is the capacity of olive trees to extract substantial quantities of water early in the season, where there is winter rainfall, from inter-row areas away from the strips of land wetted by drip irrigation.

Irrigation scheduling

Fereres et al. (Reference Fereres, Villalobos, Orgaz and Testi2011) expressed their surprise that olive tree growers had not used the water budget method more widely for irrigation scheduling despite its simplicity. One difficulty is in quantifying the volume of available water within the extensive root zone and, with that knowledge, specifying the allowable soil water deficit. This challenge is due in part to the spatial variability in soil physical properties, and in particular to the largely unknown depth and distribution of the effective roots of olive trees. Fereres et al. (Reference Fereres, Villalobos, Orgaz and Testi2011) also made the point that the olive has the capacity to extract water from soils at water potentials below the generally accepted lower limit of available water (permanent wilting point). This could be significant in fine textured soils but not in sands. The allowable soil water deficit probably also varies with the stage of crop development, with the flowering and fruit set stages being particularly sensitive to water stress.

Monitoring the changes in the maximum daily shrinkage of the trunk of a tree with a dendrometer has been proposed as a possible way of automating the irrigation scheduling of trees, including olive. For this method to work, baseline reference values for trees well supplied with water are a prerequisite so that the influence of other climatic variables can be taken into account. Moreno et al. (Reference Moreno, Canejero, Martin-Palomo, Girón and Torrecillas2006) undertook such an analysis over 104 summer days during 2005 in southern Spain. For mature well-irrigated trees (cv. Manzanillo), the maximum daily shrinkage was closely correlated with the reference crop ETo (range maximum daily shrinkage 0.2–0.9 mm d⁻¹, ETo 3–9 mm d⁻¹; r² = 0.55, n = c. 90), with the saturation deficit of the air and temperature recorded at midday being the most important individual weather variables (range saturation deficit 1–6 kPa, r² = 0.83; temperature 24–38 °C, r² = 0.79).

In an interesting comparison of how sap flow measurements and models of plant transpiration could be best used to schedule irrigation in four fruit crops, including olive, Fernández et al. (Reference Fernández, Green, Caspari, Diaz-Espejo and Cuevas2008) concluded that the transpiration ratio method offered the best approach. This is defined as the ratio of the actual daily water use divided by the potential daily water use of similar plants under non-limiting soil water conditions. Actual water use could be measured using sap flow techniques, while potential water use was best estimated using a modelling approach, provided that it could be simplified so that it became a realistic practical option in commercial orchards. They dismissed the use of the ‘sap velocity profile’ method, proposed by Fernández et al. (Reference Fernández, Palomo, Diaz-Espejo, Clothier, Green, Girón and Moreno2001) and Nadezhdina et al. (Reference Nadezhdina, Nadezhdin, Ferreia and Pitacco2007), as it was too insensitive to be a reliable indicator of when to irrigate.

In order to determine the potential for variations in stem diameter to be used for scheduling irrigation in commercial orchards, Fernández and Cuevas (Reference Fernández and Cuevas2010) undertook a comprehensive review of the research literature. They urged caution for the following reasons: stem diameter variability is affected by many other factors other than water stress, including seasonal growth patterns, weather, crop load and plant age and size, which all require expert interpretation; the relationship between stem diameter variability and stem water potential is not always straightforward; there is considerable plant-to-plant variation, which may not always be compensated for by sensitivity analysis. The installation and maintenance of sensors must also be of the highest standards to avoid erroneous readings. Although the reliability of this method can be improved if it is combined with other plant stress indicators, such as sap flow in the case of olive, variation in stem diameter is not yet robust enough to be used as the primary indicator of when to irrigate a commercial orchard.

Irrigation method

Given the scarcity of water in most olive growing areas, and the variable tree densities that exist, the choice of irrigation method is limited to those methods that can apply water (and nutrients) uniformly, close to the tree with minimum waste. Both drip (trickle) and micro-sprinklers have the potential for precision irrigation, provided they are well managed. In nearly all the experiments reported in this paper, one of these two methods, usually drip, has been used with apparent success.

Practical issues associated with the planning and selection of irrigation systems suitable for olive orchards are well described in a report by the International Olive Council (IOC, 2007).