Rhizosphere manipulations to maximize 'crop per drop' during deficit irrigation

Although much of global agriculture is rain-fed, production is frequently (and sometimes catastrophically) constrained by rainfall. Supplementary irrigation can stabilize yield from year-to-year and conventionally has aimed to meet full crop evapotranspiration (ET), since the relationship between ET and crop yield is near-linear at suboptimal water supply (Fereres and Soriano, 2007). Changes in climate (rainfall patterns) and/or resource management (irrigation quotas) will mean that future crops, either unintentionally or deliberately, will receive deficit irrigation (DI, less water than crop ET), necessarily drying the soil, limiting leaf expansion and gas exchange and, consequently, yield. Although decreased cellular turgor can limit leaf growth and gas exchange, under many circumstances plant roots can sense drying soil, and transmit chemical signals to the shoots to regulate their physiology (Davies and Zhang, 1991; Dodd et al., 1996). Much work has aimed to substantiate this ‘chemical signalling hypothesis’ by determining the production and distribution of various signals (e.g. the plant hormones ABA, cytokinins, ethylene), their role in regulating plant responses to soil drying (e.g. using mutants and/or transgenics altered in signal synthesis or sensitivity), and the importance of the root system as a signal source (e.g. using reciprocal grafts of wild-type plants and such mutants and/or transgenics; Dodd, 2005; Hartung and Wilkinson, 2009). This research area is largely ‘ABA-centric’, in part, due to its undoubted importance in regulating plant water use (Hartung and Wilkinson, 2009) and relative ease of measurement (Dodd et al., 1996). Other signals have been relatively ignored, even though mild soil drying can significantly change both plant cytokinin (Kudoyarova et al., 2007) and ethylene (Sobeih et al., 2004) status. High-throughput, multi-analyte physico-chemical techniques to quantify plant hormones in multiple plant organs (especially reproductive structures that directly influence crop yield) following rhizospheric stress (Albacete et al., 2008) need to be applied to real plants growing in the field under realistic soil drying scenarios. Such information will provide a sound physiological basis to underpin efforts aimed at manipulating long-distance hormonal signalling in planta. Several genetic manipulations have altered plant hormone signalling in crop plants, although the role of the root system (and its contribution to long-distance signalling) has not specifically been elucidated. A maize ACC synthase mutant with decreased leaf ethylene synthesis showed delayed leaf senescence and greater photosynthesis under drought compared to wild-type plants (Young et al., 2004). Brassica napus plants with genetically enhanced stomatal sensitivity to ABA showed increased seed yield under drought (Y Wang et al., 2005). Tomato plants with constitutive ABA overproduction showed increased leaf and whole plant water use efficiency (Thompson et al., 2007), but their delayed leaf area development (hence soil coverage to minimize evaporation) may diminish the expected gains in crop-level water use efficiency. Notwithstanding any physiological limitations of such technologies, socioeconomic factors may mitigate against their widespread adoption, such as consumer acceptance of GM crops and the significant biotechnological effort (and associated cost) required to make such genetic manipulations available in all crops/varieties. For these reasons, there remains a need for management options to minimize the yield penalties of deficit-irrigated crops. This article emphasises progress in understanding how rhizosphere manipulations, specifically partial rootzone drying and the introduction of plant growth-promoting rhizobacteria, alter plant root-to-shoot signalling to regulate leaf expansion and gas exchange and hence yield. The role of ABA in restricting transpiration was influential in designing a new deficit-irrigation technique, partial rootzone drying (PRD: Dry et al., 1996). The aim was to ensure that some roots were always exposed to drying soil, thus altering the production of chemical signals and their transmission to the shoots to restrict water use. This was achieved by alternately irrigating only one side of the crop row at a time (or one soil compartment in containerized plants) and allowing the other to dry the soil. Adoption of PRD (in favour of conventional DI where the entire rootzone is irrigated) requires evidence that it induces beneficial agronomic and physiological responses that differ from DI, when the same volumes of water are applied. Accordingly, the yield responses of PRD and DI plants have often been compared. A brief literature survey (15 separate studies comprising 10 different crop species) indicates that in no case did PRD significantly decrease yield and that, in six studies, PRD significantly increased yield by more than 15% (Fig. 1). However, the economic value of a crop may also depend on additional factors, and other agronomic advantages of PRD (compared to DI) have been reported. PRD promoted earlier crop maturity in tomato (Zegbe-Dominguez et al., 2003), increased fruit size in mango resulting in a more favourable fruit size distribution (Spreer et al., 2007), and increased berry skin anthocyanin concentrations independently of whether vines had greater (Antolin et al., 2006) or lesser (dos Santos et al., 2003) vegetative vigour. Crop yield ratio of PRD to DI at similar irrigation volumes (a ratio of 1 indicates that yield with both techniques is equivalent). The fraction of full irrigation is given as a thick line in each column. Columns are means ±SE of the number of experiments/seasons given in parentheses at the base of each column. Shaded columns denote pot experiments where the root system was confined. Significant (P <0.05) differences between PRD and DI are indicated with an asterisk (*). Studies are numbered thus: (1) Antolin et al., 2006, Vitis vinifera cv. Tempranillo; (2) dos Santos et al., 2003, Vitis vinifera cv. Moscatel; (3) Gu et al., 2004, Vitis vinifera cv. Sauvignon Blanc; (4) Leib et al., 2006, Malus domestica cv. Fuji; (5) Spreer et al., 2007, Mangifera indica cv. Chok Anan; (6) Dorji et al., 2005, Capsicum annum cv. Ancho St Luis; (7) Zegbe-Dominguez et al., 2003, Solanum lycopersicum cv. Petopride; (8) Kirda et al., 2004, Solanum lycopersicum cv. F1 Fantastic; (9) Gencoglan et al., 2006, Phaseolus vulgaris cv. Strike; (10) Liu et al., 2007, Fragaria×ananassa cv. Honeoye; (11) Saeed et al., 2008, Solanum tuberosum cv. Estima; (12) Shahnazari et al., 2008, Solanum tuberosum cv. Folva; (13) Du et al., 2006, Gossypium hirsutum; (14) Kang et al., 2000, Zea mays cv. Duanyu 13; (15) Kirda et al., 2005, Zea mays cv. Sele. At least three mechanisms may account for the differential responses of PRD crops compared to conventional deficit-irrigated crops, when both are supplied with the same irrigation volumes: (i) different soil water availability integrated over the entire season due to differences in soil evaporative losses; (ii) differences in root-to-shoot signalling; and (iii) different resource allocation or acquisition caused by repeated wetting/drying cycles in the rootzone. Each possibility is considered in turn. At the end of a growing season during which both treatments received the same irrigation volumes, more water remained at depth in the soil profile under PRD than DI (Gu et al., 2004; Leib et al., 2006), despite PRD inducing greater root proliferation (Mingo et al., 2004; L Wang et al., 2005). However, smaller evaporative losses from the soil under PRD plants (since less surface area of soil is wetted during each irrigation event) could only partially account for this (Leib et al., 2006), suggesting a greater restriction of water loss from leaves of PRD plants. Leaf stomatal conductance can be lower in PRD plants (de Souza et al., 2003; L Wang et al., 2005; Du et al., 2006), implying that irrigation placement causes differences in root-to-shoot signalling. Although drying part of the soil profile during PRD generates chemical signals, it also decreases sap flow from those roots in drying soil, thus signal transmission (and the consequential physiological effects) may be limited. When part of the root system of PRD-grown pepper was frequently irrigated (every 2 h), sap flow through the dry part of the root system declined linearly with soil water potential (Ψsoil) such that sap flow through those roots ceased when Ψsoil was <–100 kPa (Yao et al., 2001), thus stomatal closure was limited (less than 20%). It is thus necessary regularly to alternate the wet and dry parts of the rootzone during PRD in order to maintain a supply of soil drying-induced root-to-shoot signals to suppress crop water use (Stoll et al., 2000; Dodd et al., 2006), but decisions on when to alternate have thus far generally been arbitrary and in isolation from physiological understanding. Even without the alternation of wet and dry parts of the rootzone, PRD plants had a significantly different leaf xylem ABA concentration ([X-ABA]leaf) from DI plants, the direction of change depending on total soil water availability (Dodd, 2007). In attempting to explain these divergent responses, the contribution of different parts of the root system to [X-ABA]leaf (as an example chemical signal) was assessed. A novel ‘two root one shoot’ grafting procedure allowed xylem sap collection from both detached leaves and each individual root system. Specifically, it was hypothesized that decreasing sap flow from roots in drying soil would limit ABA export to the shoot during PRD, such that soil moisture heterogeneity would influence the relationship between [X-ABA]leaf and total soil water availability. More generally, multiple differences in xylem-borne signalling between DI and PRD plants would provide a possible physiological explanation for yield differences between plants supplied with the same irrigation volumes (Fig. 1). When the irrigated root system of PRD plants was adequately supplied with water (Ψsoil >–1 kPa), once Ψsoil of the dry root system decreased below a threshold, the fraction of sap flow from drying roots decreased linearly with Ψsoil. Root xylem ABA concentration ([X-ABA]root) increased in both DI and PRD plants as local Ψsoil declined. Although [X-ABA]leaf increased in DI plants as whole pot Ψsoil declined, in PRD plants [X-ABA]leaf actually decreased within a certain whole pot Ψsoil range (Dodd et al., 2008a). A simple model that weighted the ABA contributions of wet and dry root systems to [X-ABA]leaf according to the sap flow from each, better predicted [X-ABA]leaf of PRD plants than either the [X-ABA]root derived from wet or dry root systems, or their mean. This model revealed that, for the same whole pot soil water availability, simulated [X-ABA]leaf was higher in PRD plants than DI plants with moderate soil drying, but continued soil drying (such that sap flow from roots in drying soil ceased) resulted in the opposite effect (Dodd et al., 2008a), necessitating the alternation of wet and dry parts of the rootzone. However, in many field experiments with PRD, partial drying of the irrigated roots occurs if irrigation is infrequent (Kirda et al., 2004) and it is important to assess the implications for ABA signalling. In grafted plants, soil water status of the irrigated pot affected the relationship between the fraction of sap flow through the dry part of the root system and local Ψsoil: although Ψsoil of the irrigated pot determined the threshold Ψsoil at which sap flow from roots in drying soil decreased, the slope of this decrease was independent of the wet pot Ψsoil (Dodd et al., 2008b). Further modelling predicted that the specific Ψdry to maximize ABA signalling from the entire root system varied according to Ψwet (Dodd et al. 2008b), which has implications for decisions on when to alternate the wet and dry parts of the rootzone. Substantial field work is required to determine whether such ABA modelling may inform and complement standard plant- or soil-based methods of scheduling irrigation, to more reproducibly deliver desirable agronomic outcomes of PRD. Root growth and rhizosphere processes of plants grown with PRD have received relatively little attention. Several early studies showed that the root systems of PRD plants had a greater capacity for deeper and more widespread soil exploration than DI plants, and that rewatering of dry soil greatly enhanced biomass allocation to roots in that soil (Mingo et al., 2004). These root growth responses may partially explain why the soil under PRD irrigation had less nitrogen (N) than under conventional irrigation (Kirda et al., 2005; Shahnazari et al., 2008). In the absence of information on soil N mineralization and denitrification rates (see below), this probably indicates more efficient N foraging by PRD plants. Soil drying/rewetting cycles stimulate mineralization of organic N due to microbial death upon drying and subsequent mineralization of this microbial material (Vale et al., 2007). By contrast, gaseous losses of N via denitrification occur if soil water-filled pore space exceeds 60% (Grundmann and Rolston, 1987) and can be exacerbated if irrigation is frequent (Vale et al., 2007). Thus spatial and temporal differences in soil moisture status under PRD probably influence soil nutrient status by affecting microbial activity, although this has yet to be verified experimentally. Certainly, soil drying/rewetting cycles during PRD had a pronounced effect on microbial number. Under conventional irrigation, microbial number peaked at intermediate soil water content (θ) with drier and wetter soils showing 60% fewer culturable organisms. However, alternate PRD changed the relationship between θ and microbial number such that microbial number was maintained in wet soil (Wang et al., 2008). Since spatial and temporal differences in soil moisture status under different deficit-irrigation strategies affect microbial number, and their activity can potentially affect plant growth and development via both nutritional (see above) and other (see below) mechanisms, there is an opportunity to manage these microbial effects to achieve desirable physiological or agronomic outcomes. Thus soil inoculation with one or more plant growth promoting rhizobacteria (‘rhizosphere engineering’) has been a much prosecuted scientific goal, potentially augmenting or attenuating root-to-shoot signalling. Plant growth-promoting rhizobacteria are commonly found in the rhizosphere (on the root surface, in contrast to endophytic bacteria that invade plant tissues) and promote growth via several diverse mechanisms. Of particular significance to the discussion here are rhizobacteria that can impact on plant hormone signalling pathways by producing ABA (Cohen et al., 2008), auxins (Costacurta and Vanderleyen, 1995), and cytokinins (Timmusk et al., 1999) or by mediating plant ethylene levels (Glick et al., 1998). Much attention has focused on rhizobacteria containing the enzyme ACC deaminase (ACCd) that degrade the ethylene precursor ACC (Glick et al., 1998). Since a dynamic equilibrium of ACC concentration exists between root, rhizosphere, and bacterium, bacterial uptake of rhizospheric ACC (for use as a carbon and nitrogen source) decreases root ACC concentration and root ethylene evolution and can increase root growth (Glick et al., 1998; Penrose et al., 2001). To support the hypothesis that plant responses to rhizobacteria are mediated by plant hormones, rather than via alternate mechanisms (nutritional, antagonism of pathogens), mutant strains lacking the ability to influence plant hormone signalling have been created. ACCd-containing bacteria stimulated plant growth while mutants in the same genetic background but lacking ACCd did not (Glick et al., 1994; Belimov et al., 2009). A short-term (3 weeks) trial indicated that inoculation of pepper and tomato gnotobiotic seedlings with the ACCd-containing bacteria Achromobacter piechaudii ARV8 decreased stress-induced ethylene evolution and improved recovery of plants when watering was resumed (Mayak et al., 2004). Pot trials showed that the ACCd-containing bacteria Variovorax paradoxus 5C-2 attenuated a drought-induced increase in xylem ACC concentration in pea, and that bacterial rhizosphere colonization and effects persisted through the crop life cycle, thus increasing plant yield and water use efficiency (Belimov et al., 2009). Potential problems with rhizobacterial inoculation of field soils, due to competition between introduced and resident microbes (Strigul and Kravchenko, 2006), include poor inoculum persistence in the rhizosphere and, conversely, the displacement of indigenous microbes from the rhizosphere community. Both issues have been considered in research spanning >15 years with the rhizobacterium Azospirillum brasilense, where the impacts of inoculation on the microbial community structure were much less than environmental effects such as drought (Castro-Sowinski et al., 2007), but inoculum persistence remains an issue. Field studies demonstrated that soil inoculation (small volumes of a concentrated bacterial suspension were placed at the seedling base after transplanting) with the ACCd-containing bacteria V. paradoxus 5C-2 had minimal effects on microbial community structure and, despite rhizosphere inoculum populations decreasing by 60% over the season, pea pod biomass was significantly (20%) higher at harvest (RG Teijeiro, IC Dodd, WJ Davies, unpublished results). It will be important to repeat these results in many geographical locations under different environmental conditions (as achieved with Azospirillum: Diaz-Zorita and Fernandez-Canigia, 2009) before this strategy can be considered a viable field technique to ameliorate yield limitation induced by deficit irrigation. Advances in our fundamental understanding of long-distance signalling offer opportunities to intervene in the signalling process to minimize yield reductions under deficit irrigation. Since deficit irrigation both stimulates the production of hormonal growth inhibitors (thus suppressing their production might be desirable) and decreases the production of hormonal growth promoters (thus augmenting their production might be desirable), modelling the impacts of soil moisture heterogeneity on multiple xylem-borne signals (as elucidated above) may allow rational decisions as to when partial rootzone drying might beneficially alter plant hormone levels during the crop life cycle. Assuming that such model predictions are validated in the field, it may be possible to mitigate deleterious effects of deficit irrigation on plant hormone homeostasis by the targeted use of rhizobacteria. While there is much interest in determining the significance of specific hormonal mechanisms by which rhizobacteria affect growth (as discussed above), often rhizobacterial inoculation may affect several hormone signalling pathways. For example, soil inoculation with the cytokinin-producing bacterium Bacillus subtilis promoted the growth of lettuce plants grown in drying soil and approximately doubled both shoot cytokinin and ABA concentrations (Arkhipova et al., 2007). By contrast, early transgenic attempts at manipulating cytokinin biosynthesis via constitutive ipt expression (isopentenyl transferase, a gene for de novo cytokinin synthesis) limited root growth and caused water stress (Smigocki and Owens 1989). Thus multiple hormonal effects may be key to a successful rhizobacterial inoculation. With this aim in mind, co-inoculation of different rhizobacterial strains (that alter different hormone signalling pathways) may provide a ‘ready-made’ solution more easily achieved than genetically engineering a plant with alterations in several plant hormone signalling pathways. Depending on inoculum rhizobacteria may also provide an opportunity to alterations in plant hormone status to specific growth or under particular stress to the use of gene promoters to expression only at specific of development et al., 2007) or under specific environmental such can be and provide future for crop genetic remains an to and for to those work was not due to the limitations of