Water use and root zone water dynamics of drip-irrigated olive (Olea europaea L.) under different soil water regimes

Abstract

A field experiment was conducted over 2 years in a semiarid region of Australia to determine the root water uptake–soil water regime relationship and water use of drip-irrigated olive (Olea europaea L.) (cv. Corregiolo). A soil water balance approach was used to estimate the olive trees' water use. Olive evapotranspiration during the first (‘on’) year was 620, 685 and 723 mm for rainfed (during pit-hardening period), deficit (during pit-hardening period) and irrigated treatments, respectively. These values were 555, 610 and 673 mm during the second (‘off’) year. Soil water monitoring in three dimensions was used to infer vertical and lateral root distribution and root activity. The soil water depletion pattern indicated that most of the roots were in the top 60 cm soil depth and aligned along the tree rows. This information is important for fertilizer application and for the placement of drip laterals and soil water measurement equipment used for irrigation scheduling. It also underscores the importance of using three-dimensional models to understand root zone hydrology of drip-irrigated trees.

Keywords:

• Australia
• evapotranspiration
• olive
• relative extractable water
• root zone water balance

Introduction

Crop evapotranspiration is an important parameter in irrigation scheduling and design. Evapotranspiration of a given crop depends on atmospheric evaporative demand and soil water availability. Since evapotranspiration of a given crop varies from region to region (Palomo et al. 2002), it is essential to determine its quantity under different climatic and water availability conditions. Although olive (Olea europaea L.) is traditionally grown under rainfed conditions, its intensive modern production system requires irrigation. Its water requirement has been determined in regions where olive is traditionally grown (Fernandez & Moreno 1999; Palomo et al. 2002; Moriana & Orgaz 2003; Tognetti et al. 2006). However, this information is of limited use in regions such as Australia where olive production is an emerging industry. The most common approach to calculate evapotranspiration (ET) of a given crop is the Food and Agriculture Organization's crop coefficient method (Allen et al. 1998). However, the variability of the crop coefficient measured at different locations makes it difficult to apply this method to locations where no experimental information exists. For fruit trees such as olive, the in-field root zone water balance of individual trees can be used (e.g. Mickelakis et al. 1993). Using water balance, actual crop evapotranspiration is estimated from the measurement of other components of the root zone water balance: rainfall, irrigation, deep drainage, runoff and change in soil water storage. However, it is not possible to directly measure some of these parameters. Deep drainage and runoff can be estimated using the Darcy equation and curve number method, respectively. Change in soil water storage can be measured using a neutron probe and time domain reflectometer (TDR). The use of the water balance approach in localized irrigation systems requires monitoring of soil water content in three dimensions; along and across the tree lines and at different depths (Michelakis et al. 1993; Palomo et al. 2002). This is important in order to understand how a reduced wetted soil volume under drip-irrigation system influences root distribution and water extraction.

For efficient use of fertilizer and higher crop yield, it is important to know the rate and pattern of water absorption by olive trees. Most of the studies about the response of olive trees to water availability focused only on the above-ground part of the plant (Melgar et al. 2008; Rousseaux et al. 2009). The limited studies of the below-ground part of olive trees were mainly on root distribution and root density. However, root distribution tells only a little about root activity; root activity might not always be higher in zones of higher root density. In fact, the opposite might be true as a result of a compensation mechanism (Fernandez et al. 1991). Dichio et al. (2002) observed that olive root growth responds positively to irrigation with irrigated olives having 34% more soil volume explored by roots and 22% more root length than non-irrigated trees. In Argentina, Searles et al. (2009) observed a high correlation between root distribution and soil water content in a drip-irrigated olive grove. They found that 70% of the root length density was within 0.5 m of the drip line and more than 90% was within 1 m; more than 70% of the roots were located in the top 0.5 m soil depth. Michelakis (1986) and Michelakis & Vougioucalou (1988) showed cv. Kalamon olive trees had three to seven times more roots at the 0–120 cm horizontal distance from the drippers compared to a distance of 120–300 cm.

Understanding root distribution and root activity is important: (1) to differentiate soil zones from which water extraction occurs; (2) to establish an appropriate model (one-, two- or three-dimensional) that enables the estimation of water and solute transport in the root zone; and (3) to select representative sites for the location of soil water sensors. This study was performed to determine olive evapotranspiration using the soil water balance method, and to assess root water extraction at different depths and distances from the tree trunk under three irrigation regimes: (1) rainfed, where the trees were not irrigated during the pit-hardening period; (2) deficit, where the trees were irrigated at 50% of the full irrigation requirement; and (3) irrigated, where the trees were irrigated to fulfil their full water requirement.

Materials and methods

Experimental site and design

The experiment was conducted during the 2010–11 and 2011–12 seasons on nine-year-old cv. Corregiolo olive trees planted at 5 × 7 m spacing at the Wagga Wagga campus of Charles Sturt University (35°03′S; 147°21′E; 235 m asl), New South Wales, Australia. A strip of about 2 m on either side of a tree row was sprayed with herbicide, while a central strip of about 1.5 m between the rows was regularly mowed short. The first year (2010–11) was an ‘on’ year with high orchard yield, whereas the second year (2011–12) was an ‘off’ year with very low yield. The soil physical characteristics are given in Table 1. The historical average annual rainfall in the area is 560 mm, fairly evenly distributed throughout the year, while the minimum and maximum evapotranspiration occurs in the months of July and January, respectively (Fig. 1). The mean minimum and mean maximum temperatures are in July and January, respectively. The long-term average minimum and maximum daily temperatures during the olive production period (flowering to harvest) are 11 and 25 °C, respectively.

Figure 1 A, Mean monthly minimum and maximum temperatures, and B, mean monthly rain and pan evaporation at Wagga Wagga, New South Wales, Australia.

Table 1 Physical characteristics of the soil at the experimental site.

	Particle size distribution (%)
Depth (cm)	Sand	Silt	Clay	Field capacity (% vol)	Permanent wilting point (% vol)	Bulk density (g/cm^3)
0–30	59	11	30	24	15	1.58
30–60	42	10	48	31	22	1.51
60–90	40	9	51	36	25	1.60
90–120	38	9	53	37	27	1.66

Each tree row was irrigated using two drip laterals, laid 50 cm from each side of the rows. The drippers were fitted with 60 cm spaced pressure compensating drippers discharging 2.3 L/h. The olive fruit pit-hardening period is a relatively less sensitive stage to water stress (Lavee & Wonder 1991; Goldhamer 1999; Moriana et al. 2003). Therefore, three irrigation treatments were selected based on the amount of water supplied during the pit-hardening period: rainfed, deficit and irrigated. The rainfed treatment was not irrigated during the pit-hardening period as water stress at this stage is considered to not affect olive yield significantly. Irrigated trees were irrigated to fully meet crop water demand. Deficit treatment trees were irrigated with about 50% of the amount of water applied to irrigated trees. Soil water content was monitored under a total of nine trees (i.e. three trees per treatment) to a depth of 120 cm using a neutron probe soil moisture meter calibrated at the site (Gardner & Kirkham 1952). A separate calibration equation was used for the top 30 cm soil layer. Under each replicated tree, there were four neutron probe access tubes. One probe was located at a distance of 75 cm from the tree row (referred to as X75) and one at a distance of 200 cm from the row (referred to as X200). The other two tubes were located across the tree rows at 75 cm (referred to as Y75) and 200 cm (referred to as Y200) from the tree trunks.

The depth of irrigation water applied during the experimental seasons is presented in Table 2. For the 2010–11 experiment, as a result of exceptionally high rainfall during the spring and early summer months, irrigation started only at the beginning of January 2011. The total amount of irrigation applied for the fully irrigated and deficit irrigated treatments was 267 and 130 mm, respectively; rainfed treatment trees were not irrigated. In the 2011–12 hydrological year, fully irrigated, deficit irrigated and rainfed treatments received 299, 211 and 123 mm, respectively.

Table 2 Amount of irrigation water applied to the three water treatments during the two experimental seasons (2010–11 and 2011–12).

Amount of irrigation water applied (mm)
	2010–11			2011–12
Month	Rainfed	Deficit	Irrigated	Rainfed	Deficit	Irrigated
September	0	0	0	15	15	15
October	0	0	0	42	42	42
November	0	0	0	58	58	58
December	0	0	0	8	22	42
January	0	38	78	0	31	69
February	0	46	96	0	21	39
March	0	24	48	0	4	4
April	0	22	45	0	18	30
Total	0	130	267	123	211	299

Soil water balance

Water balance has been one of the most widely used methods to determine evapotranspiration of tree crops (Petillo & Castel 2007). Although the method has a low temporal resolution, is time consuming and laborious, it is widely used for research purposes. The value of crop evapotranspiration, ET_c, (between two soil water measurements) is estimated using the following equation:

(1)

where R is the rainfall (mm), I is the water applied by irrigation (mm), ΔS is the change in soil water storage in the soil profile assumed as exploited by the roots (mm), D is the water lost by drainage (mm) and R_o is the surface water runoff (mm). Runoff values for different rainfall events were estimated using the SCS curve number method (Soil Conservation Service 1985) as:

(2)

where S is the initial abstraction (mm) and R is the rainfall (mm).

The potential maximum initial abstraction, S, was calculated from the curve number as:

(3)

The curve number, CN, is a dimensionless parameter indicating the runoff response characteristic of a surface. It is related to land use, land treatment, hydrological condition, hydrological soil group and antecedent soil water condition. In this study, hydrological soil group B (loam), average antecedent moisture condition and poor hydrologic condition (no runoff protection) were considered in selecting the curve number.

The change in soil water storage, ΔS, over a time interval, t₂–t₁, (days) was calculated as:

(4)

where z₁ is the initial depth (mm), z₂ is the final depth (mm) and θ is the volumetric water content of soil at a given depth and time (cm³ cm⁻³).

The Darcy equation was used to estimate the amount of deep percolation, DP, below the 90 cm depth as:

(5)

where K(θ) is the unsaturated hydraulic conductivity at the water content, θ, of the soil layer below the rooting zone (cm d⁻¹); Δh is the matric potential difference between two points between which the flux is calculated (90 cm and 120 cm depths in this study) (cm); Δz is the distance between the bottom of the root zone and the immediate next depth's soil moisture monitoring point (120 – 90 cm = 30 cm in this case) (cm).

The water retention curve of the soil in the study site was used to determine K(θ) using the van Genuchten closed form equation (van Genuchten 1980):

(6)

(7)

For each θ at a given depth, the matric head, h, at that location was estimated as:

(8)

where K_s is saturated hydraulic conductivity (cm d⁻¹), θ_s is the water content at saturation (cm³ cm⁻³), θ_r is the residual water content (cm³ cm⁻³), θ is the water content at which K(θ) was estimated (cm³ cm⁻³) and m, n, α are fitting parameters of the soil water retention curve. The soil water retention curves for the four layers of the root zone soil profile are given in Fig. 2. The RETC model (van Genuchten et al. 1991) was used to determine these parameters from the soil water retention curve of the 90–120 cm depth soil layer as: K_s = 6.24 cm d⁻¹; θ_s = 0.41 (cm³ cm⁻³); θ_r = 0.095 (cm³ cm⁻³); m = 0.237; n = 1.31; α = 0.019.

Figure 2 Soil water retention curve for the four depth layers of the Red Kandosol soil in the study area.

Relative extractable water

Relative extractable water (REW) is a depth equivalent of water defined as a fraction of maximum extractable water content (Granier 1987) expressed as:

(9)

where R and R_min are the actual soil water content and the minimum soil water content observed during the experiment, respectively. R_max is the soil water content at field capacity, i.e. 72, 93, 108 and 111 mm in 0–30, 30–60, 60–90 and 90–120 cm depth layers, respectively.

Data analysis

Analysis of variance of the data set was performed for two representative dates during which olive water use was considered to be high (18 January 2011 and 27 January 2012). Error bars were also determined and indicated on the figures in such a way that the clarity of the figures is not compromised.

Results

Evapotranspiration

Cumulative values of the water balance components of the three irrigation treatments are presented in Table 3 and Fig. 3. The water balance components were calculated at 2–4 week intervals and summarized seasonwise. It can be observed that there was soil water depletion during the first hydrological year (2010–11) and soil water addition during the second hydrological year (2011–12). This is also evident from the soil water profiles in Fig. 4. The soil water contents, both at the end of the first year (20 September 2011) and at the end of the second year (20 September 2012), were lower than the values at the beginning of the experimental period (20 September 2010). However, the soil water content at the end of the second year was higher than that at the end of the first year, indicating refilling of the root zone during the second year. The difference in evapotranspiration between the treatments was mainly during the regulated deficit irrigation period of the summer months. During this period, lower ET_c was observed in the rainfed treatment than in the deficit and fully irrigated treatments. The results in Table 3 indicate that annual evapotranspiration of fully irrigated olives was 723 and 673 mm for the 2010–11 and 2011–12 seasons, respectively. For the rainfed treatments, these values were 620 and 529 mm, respectively.

Figure 3 Cumulative values of soil water balance components in the 2010–11 and 2011–12 experimental period for rainfed, deficit and irrigated treatments. Each data point is an average of three values. End-of-hydrological year values are indicated on respective data plots. DP, deep percolation; ET_c, crop evapotranspiration.

Figure 4 Average soil water content at the beginning of the first year (20 September 2010), end of first year or beginning of second year (20 September 2011) and end of the second year (20 September 2012) for the three treatments: rainfed, deficit and irrigated.

Table 3 Seasonal and annual water balance components of rainfed, deficit irrigated and fully irrigated olives during the two experimental years (2010–11 and 2011–12).

	Rainfall (mm)	Irrigation (mm)	ΔS (mm)	Runoff (mm)	DP (mm)	ETc (mm)
2010–11 Rainfed
Spring	214	0	−5	31	39	150
Summer	352	0	−31	101	36	246
Autumn	104	0	−38	1	4	138
Winter	99	0	3	5	4	88
Total	769	0	−72	137	83	621
2010–11 Deficit
Spring	214	0	−1	33	42	139
Summer	352	87	−16	107	82	266
Autumn	104	43	−56	1	9	193
Winter	99	0	3	5	4	88
Total	769	130	−70	147	137	685
2010–11 Irrigated
Spring	214	0	1	31	62	121
Summer	352	174	−13	117	119	303
Autumn	104	93	−56	7	44	203
Winter	99	0	0	2	0	97
Total	769	267	−68	157	224	723
2011–12 Rainfed
Spring	203	47	1	0	5	244
Summer	174	76	−30	76	9	195
Autumn	254	0	56	103	41	55
Winter	112	0	−4	6	49	61
Total	743	123	23	184	104	555
2011–12 Deficit
Spring	203	47	16	0	1	233
Summer	174	147	−25	80	34	232
Autumn	254	17	37	105	36	94
Winter	112	0	−2	6	56	52
Total	743	211	26	192	127	610
2011–12 Irrigated
Spring	203	47	17	0	2	231
Summer	174	217	20	83	31	257
Autumn	254	35	14	108	53	114
Winter	112	0	−6	8	39	72
Total	743	299	44	199	125	673

ΔS, change in soil water storage; DP, deep drainage; ET_c, evapotranspiration.

The total amount of water applied (rainfall + irrigation) during the first year was 769, 899 and 1036 mm for rainfed, deficit and irrigated treatments, respectively. In the second year, similar amounts of water (866, 954 and 1042 mm) were applied to the respective treatments. Evapotranspiration values of rainfed, deficit and irrigated treatments in the second (‘off’) year were 11%, 11% and 7% lower than the corresponding values in the first (‘on’) year. In the first year, evapotranspiration from rainfed treatments was 9% and 14% lower compared with deficit and irrigated treatments, respectively. These values were 9% and 17% in the second year. During the irrigation months, evapotranspiration in irrigated trees was higher than that in the rainfed treatment trees. These results show that in rainfed trees, ET_c was limited by the low soil water content during the summer period. Lower evapotranspiration during the second (‘off’) season was not due to low water availability as more water was applied in the second year. The amount of water that was not evapotranspired was stored in the root zone at the end of the second year (Table 3). This is also evident from Fig. 5, where the soil water profile at the end of the first year (September 2011) was low compared with the ones at the end of the second year (September 2012). In the first year, about 28%, 32% and 37% of the water supplied (rainfall + irrigation) was lost as runoff and deep percolation in rainfed, deficit and irrigated treatments, respectively. However, in the second year there was almost no difference between the treatments in terms of the amount of water lost as runoff and deep drainage (average 33%).

Figure 5 Average soil water content variation: A–C, for the 0–60 cm and 60–120 cm layers for the three irrigation treatments rainfed, deficit and irrigated and, D, in the 0–120 cm soil layer. FC1, WP1, FC2 and WP2 refer to field capacity and permanent wilting point water contents of the upper (0–60 cm) and lower (60–120 cm) layers, respectively. High rainfall events are indicated by arrows. Standard error bars are also indicated.

Root water extraction variation with depth

The evolution of soil water content over the two seasons in the 0–60 cm and 60–120 cm soil layers presented in Fig. 5. The soil water content at field capacity and permanent wilting points is also indicated on the figure. The rainfed treatment has the lowest soil water content throughout the 2 years and it was the only treatment in which the water content fell below the permanent wilting point. After high rainfall events, the difference between the soil water contents of the three treatments decreased. The soil water content trend reflected the rainfall amount during the respective seasons. For example, the below-average rainfall of autumn and winter seasons resulted in low soil water content in the autumn and winter seasons of 2011 and autumn season of 2012. The highest soil water content values for all three treatments were in the spring of 2010 and winter/spring of 2012. Although the winter rainfall of 2012 was below average, the high rainfall in the autumn resulted in high soil water content at the end of the second year. Comparison of mean water content was carried out for the period during which the crop water use was maximum. For this analysis, soil water content data of 18 January 2011 and 27 January 2012 were used. There is a significant (P < 0.05) difference between the soil water contents of the upper layer and lower layer for all three water treatments. There was no interaction effect of the soil water content and the treatments. There was also significant (P < 0.05) difference between soil water contents of the three treatments on these dates (Fig. 5D). From Figs 5A, 5B and 5C it can be seen that the soil water content for the upper 60 cm layer was above field capacity (FC1) during the spring of 2010–11. However, the soil water content of layer 2 (60–120 cm) was lower than the field capacity (FC2). As the season progresses, the soil water content dropped below field capacity.

Soil water depletion relative to the water content at the beginning of the experimental period (20 September 2010) for each 30 cm soil depth is shown in Fig. 6. On the representative dates for the periods during which the olive tree water use was highest (18 January 2011 and 27 January 2012), there was significant difference (P < 0.05) between the water depletion when considering both 30 cm thick and 60 cm thick layers Irrespective of the water treatment, soil water depletion decreases with depth indicating no interaction effect. The highest soil water depletion was in the 30–60 cm layer (Figs 6A, 6B, 6C) and 0–60 cm layer (Figs 6D, 6E, 6F). The soil water depletion in the first year (2010–11) was significantly higher (P < 0.05) than in the second year (2011–12). This was in agreement with previous discussion (Figs 4–5; Table 3). There seemed to be no root activity below the 90 cm depth, with most root activity being in the top 60 cm depth. Below the 60 cm depth, the soil water content remained high and almost constant. Even in 2010–11, when there was high soil water depletion, there was almost no variation in soil water content in the 90–120 cm layer.

Figure 6 Average soil water content depletion relative to the soil water content at the beginning of the experimental period (20 September 2010) per 30 cm layers (A–C) and per 60 cm layers (D–F). Major rainfall events are indicated with broken line arrows. Standard error bars are also indicated. For the 30 cm layers, error bars of the major root zone (30–60 cm) and the deeper layer (90–120 cm) are presented.

Root water extraction variation in horizontal space

The soil zone affected and non-affected by drippers and the presence of roots was established according to differences in the soil water content profiles at different distances from the tree trunk. The soil water content at different depths and different positions from the tree trunk is shown in Fig. 7 for the beginning, middle and end of the experimental period. The change in soil water content was relatively higher in the top 60 cm compared with the deeper layers indicating higher root activity in this layer. The soil water content at the farthest monitoring access tube in the direction of the tree row (X200) was higher than that of the other access tubes. This shows that this zone received more water compared with the positions in the y-direction (Y75 and Y200) and had fewer roots compared with the monitoring position in the same direction but closest to the tree (X75).

Figure 7 Average soil water content at different depths and horizontal positions at the beginning, mid and end of the experimental period. X75 and X200 indicate soil water monitoring positions located 75 and 200 cm from the tree trunk in the direction of the olive orchard rows, respectively. Y75 and Y200 indicate similar positions but across the tree rows.

Figure 8 shows the average soil water content in the four access tubes under rainfed, deficit and irrigated treatments. The rainfed treatment shows that soil water content at Y200 was higher than that of the other monitoring positions. This was especially so in the first season during which the rainfed treatment was not irrigated. Since the drip lines are only 50 cm on either side of the tree rows, the effect of irrigation at Y200 is low. In the deficit and irrigated treatments, higher soil water content was observed at X200. This might be due to the fewer roots at X200 that received irrigation water equal to that at X75.

Figure 8 Soil water content at four positions around the olive trees subjected to three irrigation treatments (rainfed, deficit and irrigated) during the two hydrological years (2010–11 and 2011–12). Standard error bars are also indicated. For clarity, one-sided error bars of only two access tubes are shown.

In the 120 cm soil depth monitored in this study, the highest water content reached during the experimental period was 370, 380 and 395 mm for the rainfed, deficit and irrigated treatments, respectively. In all the cases, this maximum soil water content occurred for the access tube located 200 cm from the tree trunk in the direction perpendicular to the tree rows. The lowest water content was 237 mm in all the treatments. In the rainfed treatment this occurred at X75, whereas in the deficit and irrigated treatments it occurred at Y75. The former may be due to more roots in the X direction following drip irrigation line, whereas the latter may be due to less irrigation in the Y direction. A very high amount of rainfall in early autumn 2012 refilled the soil water content in a short period. Since the evapotranspiration requirement is low in the subsequent months, the soil water content remained high. The soil water content at Y200 of the rainfed treatment was almost always the highest. This can be due to low root density at this location.

Relative extractable water

The evolution of REW during the two seasons is presented in Fig. 9. The REW determined on 18 January 2011 (first year experiment) indicated significant differences (P < 0.05) between the water treatments and no significant difference for soil depth. However, for the second year experiment (27 January 2012), there was significant difference (P < 0.05) for both soil depth and water treatments. REW was higher than 0.6 at the beginning of the experimental period in all the treatments although REW in the rainfed treatment was close to zero during some of the periods. For the rainfed treatment, REW was <0.3 for most of the summer of 2011 and autumn, winter and spring of 2011–12 for the top 90 cm depth. REW was below 0.3 during the early autumn of 2012. However, it was <0.5 during the winter and spring of 2012. For the deficit treatment it dropped below 0.3 only during the winter of 2010–11 and early period of 2011–12. The irrigated treatment had REW > 0.3 throughout most of the 2 years except in the autumn of 2010–11.

REW of the deepest layer (90–120 cm) is more or less constant throughout the 2 years (0.5 in rainfed, 0.6 in deficit and 0.7 in irrigated treatments). The high REW during the autumn, winter and early spring of 2012 reflects the low evapotranspiration of the olives during the ‘off’ season. The average REW over the whole root zone profile (Fig. 9) shows that the rainfed treatment had REW < 0.3 for an extended period, whereas REW of the irrigated treatment did not drop below 0.3 at any time.

Figure 9 Relative extractable water (REW) in each of 30 cm soil layers (A–C) and the 120 cm soil layer (D) under the three irrigation treatments (rainfed, deficit and irrigated). Major rainfall events are indicated by broken line arrows. Standard error bars are also indicated.

Discussion and conclusion

The water balance analysis indicates that deep drainage in the rainfed treatment was lower than that of the deficit and irrigated treatments. In addition to the fact that irrigated trees received more water, Palomo et al. (2002) attributed such a phenomenon to the possibility that the rainfed trees were able to explore greater soil depths than the deficit and irrigated trees. The ET_c of fully irrigated olives was 723 and 673 mm in the first (‘on’) and second (‘off’) years of the study, respectively. For the treatments not irrigated during the pit-hardening period, these values were 620 and 529 mm, respectively. The evapotranspiration values estimated in this study indicate that olive water use varies with evaporative demand and water availability. For an environment in California with reference evapotranspiration, ET_o, of 1330 mm and rainfall of 533 mm, Grattan et al. (2006) determined olive evapotranspiration during the irrigation season to be about 600 mm for cv. Arbequina orchards. Michelakis (1990) found no significant difference in table and oil olive yield when 260–570 mm of irrigation water was applied. From a 3-year experiment in Spain, Moriana & Orgaz (2003) found average ET_c values of 830 and 566 mm in fully irrigated and regulated deficit irrigated olives, respectively. They attributed the big difference between the ET_c values to the difference between the amounts of water applied to the treatments: 565 vs. 131 mm. Different studies conducted under different environmental conditions have reported the following evapotranspiration values: Fereres (1995) (ET_c = 750 mm for ET_o = 1400 mm); Palomo et al. (2002) (ET_c = 653 mm for ET_o = 1400 mm); and Tognetti et al. (2006) (ET_c = 552 mm for ET_o = 1180 mm). A lower olive water use during the ‘off’ year can be a useful irrigation management strategy. Moriana & Orgaz (2003) withheld irrigation from ‘off’ olive trees and observed that it affected the following year's (‘on’ year) performance especially if there were drought and trees experienced water stress during autumn and spring. They observed that the evapotranspiration of ‘off’ year trees was lower than that of ‘on’ year trees. However, they noted the need for more study before concluding that not-irrigating in ‘off’ year can be used as a deficit irrigation strategy.

The soil water content at deeper depths (>60 cm) was more or less constant while there was more fluctuation at shallow depths indicating that deeper layers respond to the rain/irrigation input only slowly. The observations in this study show that olive roots are not uniformly distributed throughout the soil profile. The higher depletion in the upper layers can be due to evaporation, root extraction and drainage through light textured soil. On 20-year-old cv. Manzanilla, Fernandez et al. (1991) found higher values of root density in the direction of the irrigation emitters than in the direction perpendicular to the tree rows. They found maximum root activity at 0.5–0.6 m from the trunk and between 0.5 m and 1 m depth. This indicates that roots do not grow into the zone not influenced by drip irrigation. Palomo et al. (2002) also recorded soil water at 0.5, 1.5, 2.5 and 3.5 m from the tree trunk for an irrigated treatment. Values of soil water close to field capacity were recorded up to 0.5 m from the trunk. At 2 m from the trunk, soil water content in the deeper layers was quite high. This difference in soil water content might be due to differences in root densities as has also been observed by Fernandez et al. (1991). They measured root density at 0.5, 1.0 and 1.5 m distances from the tree trunk along the tree rows (X-direction) as 5.69 × 10⁻², 3.80 × 10⁻² and 0.78 × 10⁻² mm mm⁻³. Corresponding values across the tree rows (Y-direction) were 0.34 × 10⁻², 0.01 × 10⁻² and 0.01 × 10⁻² mm mm⁻³, respectively. The influence of localized irrigation on root distribution was also reported by Michelakis & Vougioucalou (1988) for olive trees, Black & Mitchell (1974) for pear trees, Willoughby & Cockroft (1974) for peach trees and Huguet (1976) for apple trees.

Stevens & Douglas (1994) have found that grapevine roots irrigated with drip irrigation were concentrated in the direction of vine row with 50% of the vine's roots less than 45 cm out in the row compared with 35% under microjet irrigation. Levin et al. (1979) found that most roots in an apple orchard in heavy clay soil were within 0.6 m of the drip line. Franco & Abrisqueta (1997) found that 80% of almond roots in a silt-loam soil were within 1 m of the drip line and in the top 60 cm. Michelakis et al. (1993) found that the percentage of roots of drip-irrigated avocado trees was higher in the upper 0.5 m soil layers and within 2 m from the drip line, where about 70% of the roots were located. They attributed higher root percentage in the areas close to the drip line to higher soil water content. Studies on the root distribution of olives have found that the main development of the olive root system occurs in the most superficial soil layers (Fernandez & Moreno 1999). Levin et al. (1979) indicated that the main feature that characterizes drip irrigation of fruit trees is the limited horizontal distribution of water in the area between tree rows.

REW of 0.3–0.4 is considered a threshold of water stress for many tree species (Breda et al. 1995; Fernandez et al. 1997). If the lower threshold (0.3) is considered for olive, it can be seen that the irrigated treatment was not stressed at any time (Figs 8–9). Although the REW value of rainfed trees indicated significant water stress, its evapotranspiration was not equally affected (Table 3). This shows that although the soil water content in the rainfed treatment was lower than that in the irrigated treatment, the root system of rainfed trees explored a greater volume of soil than the irrigated trees, as also reported by Fernandez et al. (1991).

Although it is a laborious and time-consuming exercise, the water balance method was successfully applied to determine olive evapotranspiration under three irrigation regimes. Evapotranspiration of fully irrigated mature olives in the semiarid region of Australia was 723 and 673 mm in ‘on’ and ‘off’ years, respectively. For olives not irrigated during the pit-hardening period, evapotranspiration was 620 and 555 mm in ‘on’ and ‘off’ years, respectively. The soil water extraction pattern indicated that olive roots were concentrated along the drip line and at shallow (<0.6 m) depths. These observations show the importance of having different irrigation managements in ‘on’ and ‘off’ years for perennial crops exhibiting alternative bearing.

Root distribution and root water extraction is affected by the layout of drip laterals. Placement of soil water monitoring instruments and fertilizer application in drip-irrigated horticultural trees need to take this into account. The difference in soil moisture distribution along the tree line and across the tree line indicates the importance of using three-dimensional models to understand the hydrology of drip-irrigated trees in arid and semiarid areas.