Local root abscisic acid (ABA) accumulation depends on the spatial distribution of soil moisture in potato: implications for ABA signalling under heterogeneous soil drying

Highlight Local root abscisic acid (ABA) accumulation depends on the spatial distribution of soil moisture in potato. Implications for ABA signalling under heterogeneous soil drying are discussed.


Introduction
Root-sourced abscisic acid (ABA) is considered a long-distance chemical signal that triggers physiological responses enhancing plant water economy (stomatal closure, decrease and contain the highest root density, and thus soil moisture increases progressively with depth (e.g. Zhang and Davies, 1989;Beff et al., 2013;Puértolas et al., 2014). Moreover, some irrigation techniques, such as partial root-zone drying (PRD), intend to induce lateral soil moisture heterogeneity by applying irrigation only to part of the root system (Stoll et al., 2000). Modelling of ABA signalling under heterogeneous soil moisture is essential to understand plant responses to soil drying under field conditions. This knowledge could inspire more efficient irrigation scheduling to maximize water-use efficiency (Dodd et al., 2008a, b;Liu et al., 2008).
Studies on 'one shoot-two root' grafted plants with only one of the root systems subjected to soil drying have shown that root xylem sap ABA concentration coming from the 'dry' root increased compared with the 'wet' root, initially increasing xylem sap ABA concentration above the graft union. Later, as water uptake from the dry side decreased, xylem sap ABA concentration ([X-ABA]) subsequently decreased (Dodd et al., 2008a), similar to observations in some PRD experiments (Khalil and Grace, 1993;Dry et al., 2000). However, the response of [X-ABA] to PRD is not uniform, as some studies report no increase at all in ABA concentration in the xylem sap or in the leaf (Blackman and Davies, 1985;Wakrim et al., 2005;Jensen et al., 2009;Einhorn et al., 2012). Further understanding of the factors causing these different responses to different soil moisture gradients is essential for modelling ABA signalling in response to soil drying.
Although 25-30% of the ABA in xylem sap might come from shoots due to recirculation of basipetally transported ABA in the phloem (Liang et al., 1997), the initial increase in xylem ABA concentration in response to water stress is believed to be triggered mainly by increased root ABA accumulation (Zhang and Davies, 1990;Neales and McLeod, 1991). Therefore, understanding ABA accumulation patterns within the root system is essential to predict long-distance ABA signalling responses to heterogeneous soil drying. Root ABA concentration increases in response to soil drying due to an increase in ABA biosynthesis by roots and ABA recirculation from shoots via phloem transport (Sauter et al., 2001). However, there are few studies on the effect of soil moisture heterogeneity on the spatial distribution of root ABA accumulation. Local soil moisture and root ABA concentration were tightly correlated under PRD in a split-root experiment in sycamore (Khalil and Grace, 1993). When maize was grown in drying soil columns, root ABA concentrations increased in drying soil layers, but continued soil drying later decreased root ABA concentrations (Zhang and Davies, 1989). In contrast, there was no clear relationship between root ABA concentration and local soil moisture in bean growing in drying soil columns (Trejo and Davies, 1991;Puértolas et al., 2013). Understanding the origin of these discrepancies is essential to predict xylem ABA concentration based on differences in local soil moisture.
ABA synthesis in root cells increases as root water potential (Ψ root ) decreases (Zhang and Davies, 1987;Simonneau et al., 1998). Therefore, heterogeneity in Ψ root throughout the root system might be the main factor explaining local differences in root ABA accumulation when soil moisture is distributed heterogeneously. Thus, for the same level of soil drying, ABA export from root to shoots could differ according to how soil moisture distribution affects Ψ root within the root zone.
If this hypothesis is confirmed, understanding how Ψ root is distributed across the root zone in response to soil moisture heterogeneity is essential to understand ABA signalling responses to soil drying. Some studies have shown differences in Ψ root across the root zone when soil moisture is heterogeneous (Dodd et al., 2008a;Simonneau and Habib, 1994), but other authors have also reported that, because of this water potential gradient, water can be redistributed within the plant in general and the root system in particular (de Kroon et al., 1996;Stoll et al., 2000). Therefore, the existence and extent of Ψ root heterogeneity across the root zone could be modulated by water redistribution and differences in hydraulic resistance within the root zone. Although water movement within the root system depends on water and osmotic potential gradients (Mao et al., 2009), it is plausible that vertical water recirculation might be more effective than lateral recirculation, as water transport through the vessels via the xylem might be more effective than cell-to-cell pathways (Steudle and Peterson, 1998). This could explain the observed homogeneity of Ψ root under vertical soil moisture gradients (Puértolas et al., 2013), unlike in split-root experiments subjected to lateral gradients (Simonneau and Habib, 1994;Dodd et al., 2008a). Equilibration of water potential within the root zone required night-time cessation of transpiration (Bauerle et al., 2008). This also suggests the possibility of diurnal fluctuations in water potential heterogeneity, which might influence root ABA accumulation and, in turn, root-to-shoot ABA signalling responses to soil moisture heterogeneity.
This study aimed to determine the relationship of local soil moisture with Ψ root , ABA concentration, and water uptake when soil moisture is heterogeneously distributed, and the influence of these variables on ABA export to the shoot. We assessed the following hypotheses: (i) ABA accumulation is higher in roots growing in drying soil than in wet soil; (ii) differences in Ψ root are associated with differential ABA accumulation within the root system, and are lower under vertical soil moisture gradients than under lateral gradients; and (ii) root water potential and ABA accumulation gradients within the root zone change during the day, with lower heterogeneity during night-time cessation of transpiration. To assess these hypotheses, root ABA concentrations ([ABA] root ) and Ψ root were measured in hydraulically isolated soil compartments at progressive degrees of soil drying when soil moisture heterogeneity was imposed laterally or vertically. The experiments used potato, since its ABA signalling has previously been investigated under lateral PRD (Liu et al., 2008), yet there are pronounced vertical gradients in soil moisture under field-grown potato crops even with the application of PRD (Puértolas et al., 2014). Understanding ABA signalling in the field therefore requires knowledge on how vertical and lateral soil moisture gradients affect root ABA accumulation.

Experiment 1: vertical PRD (VPRD) in soil columns
Potato seed tubers (Solanum tuberosum L. cv. 'Maris Piper') were sown in pots filled with a fertilized organic loam (John Innes No. 2; J. Arthur Bowers, UK). A moisture release curve for this substrate can be found in Puértolas et al. (2013). Pots were cylinders, 6.5 cm in diameter and 23 cm in length (0.75 cm 3 in volume), with stainless steel mesh (0.7 mm aperture) at the base to assist drainage, and were designed to fit tightly in a pressure chamber of the same volume (Soil Moisture Equipment Corp., Santa Barbara, CA, USA). A 1.5 cm thick quartzite gravel (15-20 mm) layer divided the upper from the lower half of the pot, preventing water ascent by capillarity from the lower to the upper part. Before filling, pots were cut in half lengthwise and the two halves were stuck together with duct tape to allow easy extraction of the intact soil column at harvest. Two pairs of 6 mm diameter holes (2 cm apart) were punched in the wall of the pot, at the midpoint of each pot compartment. Each pair of holes allowed insertion of the two rods of a soil moisture capacitance probe (Model SM-200; Delta-T, Burwell, UK) to estimate root water uptake (RWU) from each soil layer. Plants were placed in a controlled environment room, providing 400-600 μmol m -2 s -1 of irradiance, 14 h of photoperiod, and 24/19 °C day/night temperature. Five batches of 11 plants were sown every other day (55 plants in total were measured). All plants in each batch were measured within 2 d.
Plants were harvested 4 weeks after emergence. All plants were watered to field capacity in the morning by completely immersing the pots in water. Plants were randomly assigned to four groups, which were subjected to 0, 1, 2, or 3 d of VPRD. During PRD, only the bottom half of the pot was immersed once a day. Plants with 0 d of VPRD application were fully irrigated until the morning of their measurement date. Each treatment comprised 13-14 plants.
Measurements were taken 4-8 h after the lights were switched on in the controlled environment room. Before the measurements, the pot surface and base were covered with duct tape to minimize soil evaporation, weighed on a precision balance to 0.01 g (Adventurer Pro AV4102; Ohaus, Thetford, UK) and soil moisture sensors (Model SM200; Delta-T) inserted in each pair of holes (Fig. 1). After 1.5-2 h, the sensors were removed, the pot weighed again, and stomatal conductance (g s ) was measured in three mature leaflets with a porometer (AP4; Delta-T). Whole-plant water uptake was calculated from the initial and final pot weights. Evaporation was assessed by determining the water loss from a well-watered pot (without a plant) and ignored as negligible (<3% of the water loss of pots containing a plant). Water uptake from each compartment was calculated from soil moisture sensor readings as described by Puértolas et al. (2013) Soil volumetric water content (θ v ) in each compartment at the time of measurement was also recorded for each plant. One of the leaves in which stomatal conductance was measured was excised and leaf water potential (Ψ leaf ) was determined with a pressure chamber. The plant was then de-topped and the pot inserted in a pressure chamber with the stem protruding to determine Ψ root . Since sap flow rate can influence [X-ABA] (Dodd et al., 2008a), sap was collected at different overpressures above the balance pressure (0.4, 0.6, and 0.8 MPa) as described previously (Dodd et al., 2008a). Sap was collected in pre-weighed Eppendorf vials, frozen in liquid nitrogen immediately after being collected, and stored at -18 °C for ABA determination. Sap flow rate generated at each overpressure was calculated by weighing the vials before freezing them and recording the elapsed collection time. ABA concentration was determined in the sample with the closest sap flow rate to the actual transpirational flow rate (calculated by weighing the pot).
Each pot was opened and a root segment (3-6 mg in root dry weight) from the interior of each column compartment was excised, briefly (<10 s) washed in tap water to remove adhering soil particles, frozen in liquid nitrogen, and stored at -18 °C for ABA determination. The elapsed time between excision and freezing did not exceed 20 s.
Root samples for ABA determination were freeze dried and finely ground. Deionized water was added at 1:50 weight ratio. Both sap samples and root extracts were analysed by a radioimmunoassay (Quarrie et al., 1988) to obtain root xylem sap ABA concentration ([X-ABA] root ) and root ABA concentration ([ABA] root ). [X-ABA] root was measured in sap samples, while [ABA] root was measured in the aqueous extract obtained after incubating freeze-dried root tissue samples in a shaker at 4 °C overnight. Xylem sap of S. tuberosum does not present non-specific interference in the assay (Liu et al., 2006). No cross-reaction of the antibody with other compounds in root aqueous extracts was detected, based on a cross-reactivity test (Quarrie et al., 1988). [X-ABA] root decreased as collecting pressure increased (F=30.8, P<0.001).
Roots were extracted, oven dried, and weighed. No differences between the upper and bottom sides in root dry weight were observed (data not shown).

Experiment 2: VPRD vs horizontal PRD(HPRD)
The experiment was repeated twice. On each occasion, 26 potato seed tubers (S. tuberosum L. cv. 'Horizon') were sown in 0.25 l pots filled with the same soil used in Experiment 1. Plants were transplanted 5-7 d after emergence into 5 l pots (17.5 cm in height, 22.5 cm in upper diameter) divided into two compartments with equal capacity by a 3-4 cm layer of the same gravel type as in Experiment 1 to disrupt water capillarity between compartments. Half of the pots were divided by a vertical gravel layer (producing two lateral compartments of 1800 cm 3 capacity; HPRD). The rest of the pots were divided by a horizontal layer (producing two vertical compartments of ~1800 cm 3 capacity; VPRD) ( Fig. 1). In HPRD pots, a 5 × 5 cm acetate window was opened in the pot wall at mid-height on each side to allow root sampling, while in VPRD there was a single 14 cm high×5 cm wide window, which spanned the two vertical sides of the pot. Additionally, a pair of holes was punched in each compartment to insert SM-200 probes. In HPRD, each pair of holes was placed in the middle of the wall beside each opposite window, while in VPRD each pair of holes was located in the middle of each vertical compartment and close to the window. Windows were covered with the same cut piece of pot wall and fixed with duct tape until sampling.
Plants were grown for 5 weeks after emergence, before measurement under the same environmental conditions described in Experiment 1. They were watered to field capacity after transplanting until PRD was applied. For each plant, PRD was applied for a varying period of time before measurement (from 0 to 9 d) to obtain a wide range of soil moisture content on the dry side. During PRD application, plants were watered once a day. In HPRD, pots were inclined 20° and irrigation (to field capacity) was applied to the lower side of the pot to prevent water running through the layer towards the dry side. After 15 min, pots were placed back in the normal position. In VPRD, the bottom half of the pot was immersed in water for 5 s as in Experiment 1.
Half of the plants were measured at the end of the night period (pre-dawn measurements), and the rest between 6 and 8 h after the beginning of the photoperiod (midday measurements). Two or three plants were measured per day and time of day during a 7 d period. At midday, water uptake from the two compartments was measured as described in Experiment 1.
At both times of the day, a 2 cm root segment was excised from the edge of the pot after partially cutting the acetate window as described by Bauerle et al. (2008). Roots were sampled around the centre of each window in HPRD (at ~8.5 cm from the soil surface) and at the midpoint in each vertical compartment in VPRD (at ~4 cm from the soil surface and the bottom of the pot for upper and lower compartments, respectively). The root segment was tapped to remove adhering soil particles, briefly blotted with absorbent paper, and immediately placed and sealed inside a psychrometric chamber (C52; Wescor, Logan, UT, USA). From excision to sealing inside the chamber, the elapsed time was less than 10 s. Two samples were taken from each plant, one per compartment. The chamber was placed in a room at stable temperature and humidity conditions. Root water potential (Ψ root ) was measured after thermal equilibration by dew point psychrometry with a microvoltmeter (HR33-T; Wescor). Preliminary tests determined that thermal equilibration was attained after at least 6 h. Each chamber was calibrated previously with NaCl solutions of known water potential. Another root sample was excised through the opening in the wall and processed as in Experiment 1 to determine [ABA] root .
For midday measurements, stomatal conductance was measured in three fully expanded leaves located in the upper third of the canopy. For pre-dawn and midday measurements, Ψ leaf and leaf xylem sap ABA concentration ([X-ABA] leaf ) were determined as described for Experiment 1.
θ v and RWU were measured at midday with the soil moisture sensors as described in Experiment 1. For both pre-dawn and midday plants, a sample of soil from each compartment was weighed, dried at 80 °C for 4 d and the soil gravimetric water content (θ g ) was determined. A tight linear relationship between θ v and θ g was observed in midday plants [θ g (g g -1 )=0.94×θ v (cm 3 cm -3 )+0.02; r 2 =0.91; data not shown], which was used to convert θ g into θ v for all plants to allow comparison with previous studies (e.g. Dodd et al., 2008a, b;Liu et al., 2008;Puértolas et al., 2013).
No differences between wet and dry sides in root dry weight were observed in any of the two PRD distributions (data not shown).

Statistical analyses
No statistical differences in any of the studied variables were found between batches of plants (five batches in Experiment 1, two batches in Experiment 2), so all the plants were pooled in both experiments.
Visual inspection of the relationships between soil moisture and different variables, especially [ABA] root , revealed that altered physiological responses were only observed in plants with θ v <0.22 cm 3 cm -3 in the dry compartment. Plants then were separated into two groups based on θ v in the dry compartment: dry, θ v <0.22 cm 3 cm -3 ; wet, θ v >0.22 cm 3 cm -3 . Those groups were used to assess the effect of soil moisture in the dry compartment on [ABA] root , RWU, and Ψ root by repeated-measures analysis of variance (ANOVA) with PRD side (wet or dry) as the within factor (S) and dry compartment soil moisture (D) as the between factor in Experiment 1. For Experiment 2, S was considered the within factor, and D, measurement time of the day (T), and PRD spatial distribution (P -VPRD or HPRD) were the between factors. [X-ABA], Ψ leaf , and g s were assessed by ANOVA with D as the fixed factor in Experiment 1. For Experiment 2, T and P were the fixed factors. Relationships between different variables were assessed by linear regression. All statistical analyses were carried out with SPSS 20 (IBM, Armonk, NY, USA).

Local root responses
θ v was lower in the dry compartment, and decreased with time of water withholding in both experiments, although it was faster in the small columns of Experiment 1 (Fig. 2). Average soil water content in the dry side decreased with time of PRD application to ~0.10 cm 3 cm -3 in Experiment 2, but only to 0.19 cm 3 cm -3 in Experiment 1, although some individuals were close to 0.10 cm 3 cm -3 (Fig. 2). Daily irrigation of the wet side maintained θ v generally above 0.30 cm 3 cm -3 (data not shown) for both experiments. In Experiment 2, average (whole-pot) soil moisture (mean of wet and dry sides) was never below 0.16 cm 3 cm -3 . PRD distribution did not influence θ v , as it was similarly lower in the dry side for both HPRD and VPRD.
[ABA] root in the dry compartment increased but only when θ v was below 0.22 cm 3 cm -3 in both experiments (Figs  3 and 4). It did not increase in the wet side regardless of the soil moisture in the dry side (Fig. 4). In Experiment 2, this increase in the dry compartment was higher in HPRD than VPRD, despite statistically similar θ v (Fig. 4, P=0.008 for the compartment×soil moisture level×PRD distribution interaction; Table 1). However, no interaction with time of the day (pre-dawn versus midday) was detected (Fig. 4, P=0.78 for the side×soil moisture level×time of the day interaction). RWU was much higher in the wet compartment ( Table 1). The fraction of RWU (RWUF) from the dry compartment to total uptake was positively related to soil moisture in the dry side in both experiments (Fig. 5), but in Experiment 2 it was much lower in plants when the dry compartment had lower soil moisture (θ v <0.22 cm 3 cm -3 ; Fig. 6).
Ψ root in Experiment 2 was significantly lower in the dry side but only when soil moisture decreased below 0.22 cm 3 cm -3 (Table 1, Fig. 7). No effects of PRD distribution or time of the day were found.
[ABA] root in the dry side significantly increased as Ψ root decreased in HPRD but not in VPRD (Fig. 8).

Whole-plant responses
[X-ABA] root did not change with soil drying in Experiment 1 (Fig. 9). In Experiment 2, [X-ABA] leaf was higher at predawn, but VPRD and HPRD distributions had similar effects on [X-ABA] leaf.
Ψ leaf did not change with soil drying in any of the experiments (Table 2). It decreased by ~0.2 MPa from pre-dawn to midday in Experiment 2 (Table 2; Ψ leaf =-0.21 ± 0.01 and -0.45 ± 0.02 MPa for pre-dawn and midday, respectively), but there were no differences between PRD treatments. In Experiment 1, root xylem water potential did not differ significantly with different soil moisture in the dry compartment (F=1.2; P=0.25) The value of g s did not change across treatments or soil moisture levels in either experiment (Table 2).

Discussion
When compared against conventional deficit irrigation where water is applied homogeneously to the entire root zone, agronomical responses to PRD applied under field conditions are far from uniform (Dodd, 2009). Differences in root hormonal responses to soil moisture heterogeneity may partially cause this inconsistency. Our results indicated that the spatial distribution of soil moisture heterogeneity (whether vertical or horizontal) influences the impact of local soil moisture conditions on root ABA accumulation, which can influence not only root-to-shoot ABA signalling but other plant responses like root growth (Sharp and LeNoble, 2002) or root hydraulic conductivity (Hose et al., 2000). Although the roots sampled under VPRD or HPRD treatments were morphologically indistinguishable (despite being sampled at different depths in the soil profile), there were clear differences in their sensitivity of root ABA accumulation to Ψ root (Fig. 8). Nevertheless, these differences in local root ABA accumulation had minimal impacts on root-to-shoot ABA signalling (Fig. 9), since root ABA accumulation was coincident with limited water uptake (and thus sap flow) from roots in drying soil (Fig. 6).
Root ABA accumulation was higher in the dry part of the root system coincident with lower Ψ root in both experiments and PRD distributions (Fig. 4), consistent with the previously reported dependence of ABA synthesis on Ψ root (Simonneau et al., 1998). However, the increase in ABA accumulation with Fig. 3. Relationship between θ v in the dry compartment and [ABA] root for Experiment 1 (A) and each PRD distribution (HPRD and VPRD) in Experiment 2 (B). Since no differences between pre-dawn and midday were found, data was pooled for Experiment 2. The dashed line is drawn at θ v =0.22 cm 3 cm -3 . Each point is an individual root system. decreasing soil moisture and Ψ root was much higher in HPRD (Figs 3 and 8), which suggests that the capacity of roots to accumulate ABA could depend on their position, as roots in the dry side of VPRD were sampled from a higher position in the soil profile than in HPRD. Greater root maturity in upper soil layers may explain the lower response of [ABA] root to soil water potential observed in VPRD than in HPRD, as it has been reported that maize primary young roots and primary root tips start to accumulate ABA in response to soil drying at higher relative root water content than mature primary roots and secondary roots (Zhang and Tardieu, 1996). This explanation agrees with previous observations of root ABA accumulation in response to soil drying. While root ABA concentration was much greater in the dry side of sycamore seedlings growing in split pots (Khalil and Grace, 1993), it was not clearly higher in the upper layers of drying soil columns in bean (Trejo and Davies, 1991;Puértolas et al., 2013). In an extreme case, root ABA accumulation in the upper soil layers of maize growing in 1 m columns was lower than at intermediate depth, despite higher soil water content in the latter (Zhang and Davies, 1989). All these observations seem to support the hypothesis that the capacity of roots to accumulate ABA in response to soil drying decreases in the upper older roots compared with deeper younger roots. To our knowledge, these results are the first evidence of different sensitivity of root ABA accumulation to changes in Ψ root depending on root position in the root zone. This might explain the discrepancies in local root ABA accumulation responses to soil drying described above. Although the evidence found in this experiment strongly suggest the hypothesis of different sensitivity to water potential according to position within the root, this should be proven in an specific study, preferably in longer pots and with different species.
Equivalent Ψ root in the dry compartment in both HPRD and VPRD treatments (at equivalent θ v ) in Experiment 2  ( Fig. 7, Table 1) suggests that internal hydraulic redistribution is equally effective, regardless of how soil moisture heterogeneity is imposed. Although hydraulic redistribution should occur via night-time water potential equilibration (Bauerle et al., 2008), differences in Ψ root between dry and wet compartments occurred both pre-dawn and at midday (Fig. 7), indicating that the Ψ root disequilibrium was stable throughout the photoperiod. Therefore, root ABA accumulation seems to be affected only by the degree of soil drying, regardless of the spatial layout of soil moisture heterogeneity, and not by differential internal water redistribution.
The differential root ABA accumulation in drying soil depending on their position (or age) observed in Experiment 2 might affect [X-ABA], with potentially higher export in HPRD than in VPRD. However, no ABA differences between different soil moisture distributions were observed, as soil drying did not significantly affect [X-ABA] leaf (Fig. 9). Increased [X-ABA] leaf , observed only at pre-dawn in plants with high soil moisture (Fig. 9), was not associated with increased [ABA] root or decreased soil moisture in the dry compartment, and probably reflects a concentration effect due to low sap flow observed during the night period, as seen previously (Sobeih et al., 2004).
This absence of an [X-ABA] response to PRD (Fig. 9) contrasts with results reported elsewhere (Sobeih et al., 2004;Liu et al., 2006;Wang et al., 2010). However, the response of ABA status to PRD is not consistent, as some studies report no increase in xylem sap (Jensen et al., 2009) or leaf (Blackman and Davies, 1985;Wakrim et al., 2005;Einhorn et al., 2012) ABA concentration. The lack of response of [X-ABA] to PRD observed ( Fig. 9) seems to agree with a model predicting that [X-ABA] under PRD is determined by the average of root ABA in each side weighted by RWUF (Dodd et al., 2008a) (Fig. 10A, C). In the two experiments described (Fig. 1), even though ABA accumulates in roots in the dry compartment, [X-ABA] cannot increase in response to PRD since the increase in [ABA] root in the dry compartment is coincident with very low RWUF (<15%, Figs 4 and 6). Nevertheless, [X-ABA] of potato can increase in response to PRD (Liu et al., 2006(Liu et al., , 2008. Although these studies did not simultaneously measure RWU and [ABA] root RWU fraction dry side  during PRD, [X-ABA] increased (compared with fully irrigated plants) 2-4 d after applying PRD, when RWUF from the dry side exceeded 30% (Liu et al., 2008). When RWU fell below 20% on d 5, [X-ABA] fell accordingly. These observations suggest that when soil moisture in the wet side is close to saturation, ABA signalling can only increase when experimental conditions simultaneously allow appreciable RWUF (at least 20%) and ABA accumulation in the dry compartment. From a (PRD) management perspective, it becomes critical to determine experimental conditions that allow simultaneous root ABA accumulation and moderate water uptake from the dry side. The main difference between our study and previous studies in potato (Liu et al., 2006(Liu et al., , 2008 is soil texture. Relationships between soil and root water potential, and θ v and either root ABA levels or RWUF from drying roots under PRD conditions differ largely according to soil texture (Dodd et al., 2010). While RWU declines more sensitively with θ v in the organic loam used here than in a sandy soil (see Fig. 3a in Dodd et al., 2010), both substrates showed similar increases in [X-ABA] root as θ v declined. Thus, similar root ABA accumulation would allow greater export to the shoots in a sandy soil than in the soil used in our study, thereby increasing [X-ABA] (Fig. 10). The reasons for higher RWUF from dry sand compartments are not clear but may depend on the soil water potential of the wet side: lower soil matric potential in the wet compartment attenuates the decrease in RWUF from the dry compartment (Dodd et al., 2008b). Low matric potential in the wet side of sand-grown plants may be related to the difficulty of maintaining high matric potential in the wet compartment, as transpirational flow increases to balance the decrease in transpiration sourced from the dry compartment.
Our findings suggest that VPRD is less likely to increase [X-ABA] than HPRD, as [ABA] root did not respond to decreasing Ψ root in the former (Fig. 8). With a more moderate decrease in RWUF from the dry side with soil moisture than that observed (RWUF=0.25 when θ v <0.22), while maintaining the observed [ABA] root values in each side, the predicted weighted [ABA] root average would only clearly increase in the HPRD treatment of Experiment 2 (Fig. 10B, D). This implies that transient increases in [X-ABA] in response to heterogeneous soil drying would only be significant when soil drying affects parts of the root system that are better able to synthesize ABA in response to decreasing Ψ root .
The relative insensitivity of ABA synthesis to soil drying in some parts of the root system (Fig. 3A) should be considered in designing more efficient water-saving irrigation strategies. For instance, VPRD application could be intentionally achieved using simultaneously subsurface and surface drip irrigation, keeping the lower layers permanently wet while submitting the upper layers to drying and rewetting cycles (Kang et al., 2002). This may be a more efficient technique than HPRD, since the dry part of the root system would be the upper part, which produces less ABA (Fig. 4),  maintaining higher stomatal conductance (and photosynthesis) than HPRD for a similar degree of soil drying (and irrigation volume application). Identifying the spatial patterns of local root ABA accumulation in response to soil drying for different crop species may allow specific areas of the root system to be dried with minimal yield penalty.