Control of leaf growth by abscisic acid: hydraulic or non-hydraulic processes?



    Corresponding author
    1. INRA, UMR759 Laboratoire d'Ecophysiologie des Plantes sous Stress Environnementaux, Place Viala, F-34060 Montpellier, France and
      François Tardieu. Fax: +33 4 67 52 21 16; e-mail:
    Search for more papers by this author

    1. Australian Centre for Plant Functional Genomics, Waite Campus, The University of Adelaide, PMB1, Glen Osmond, SA 5064, Australia
    Search for more papers by this author

    1. INRA, UMR759 Laboratoire d'Ecophysiologie des Plantes sous Stress Environnementaux, Place Viala, F-34060 Montpellier, France and
    Search for more papers by this author

François Tardieu. Fax: +33 4 67 52 21 16; e-mail:


Abscisic acid (ABA) affects plant metabolism and water transfers via multiple mechanisms at cell, organ and whole plant levels. These mechanisms translate into contradictory effects on leaf growth, so the literature reports positive, null or negative effects of ABA on leaf growth upon water deficit. We review evidences based on genetic manipulations of ABA biosynthesis, feeding the plant with artificial ABA or partial root drying and provide elements to avoid confusions of effects. We propose that ABA has mainly three effects on growth. (i) Via its controlling effect on stomatal aperture and transpiration rate, an increased concentration of ABA tends to buffer the day-night alternations of leaf growth rate and the negative effect of evaporative demand. (ii) ABA tends to improve leaf growth via an increase in the conductance to water transfer in the plant as a result of increased tissue hydraulic conductivity. (iii) ABA has also a modest non-hydraulic effect which is negative in plants subjected to water deficit, either manipulated for ABA synthesis or fed with artificial ABA, but can be positive in well watered plants deficient of ABA. The overall effect of increasing ABA biosynthesis depends on the relative weight of each of these effects under different environmental scenarios.


Abscisic acid (ABA) plays a central role in plant responses to water deficit via a large number of processes, with signalling pathways that are not yet fully understood in spite of the recent discovery of putative ABA receptors (Liu et al. 2007; Ma et al. 2009; Pandey, Nelson & Assmann 2009). It controls gene expression, either alone via ABA responsive elements (Shinozaki & Yamaguchi-Shinozaki 2000; Huang et al. 2008; Lenka et al. 2009), or in interaction with other compounds such as sugars (Finkelstein & Gibson 2002; Dekkers, Schuurmans & Smeekens 2008), auxin (Teale et al. 2008), gibberellins, (Weiss & Ori 2007) or brassinosteroids (Zhang, Cai & Wang 2009). At cell level, it controls the synthesis of enzymes that act in cell protection under severe stresses such as dehydration (Finkelstein & Gibson 2002; Li et al. 2002) or heat stresses (Rizhsky, Liang & Mittler 2002), but also in many other processes such as water transfer (Assmann 2003; Parent et al. 2009) or iron metabolism (Lobreaux, Hardy & Briat 1993). At an organ level, ABA is recognized to have a crucial role on stomatal movements (Assmann 2003; Christmann et al. 2007), on tissue hydraulic conductivity (Hose, Steudle & Hartung 2000; Parent et al. 2009), and on root and shoot growth (Munns & Cramer 1996; Sharp 2002). At the whole-plant level, ABA has been considered as a candidate for root–shoot communication during water or salt stresses (Davies & Zhang 1991), but it also interacts with other plant signals involved in inter-organ communication, such as sap pH (Wilkinson & Davies 1997; Bacon, Wilkinson & Davies 1998), hydraulic signals (Tardieu & Davies 1992), ethylene (Wilkinson and Davies, this issue) or other hormones (Acharya & Assmann 2009). Having such a large spectrum of actions at different scales of plant organization, ABA can be expected to have contradictory effects on an integrative process such as leaf growth depending on environmental or developmental context. Indeed, the literature on ABA effects on leaf growth at the plant level is contradictory, with positive or negative effects depending on the experimental setup and conditions (Sharp 2002) (also see further discussion).

Several causes may explain this complex situation. (1) The first of them is the confusion of the effects of ABA and of plant water relations. Plants fed with artificial ABA or genetically manipulated to overproduce ABA are typically at a higher degree of leaf hydration compared with controls (Ben Haj Salah & Tardieu 1997; Thompson et al. 2007; Parent et al. 2009), while the opposite occurs with mutants or transgenics that underproduce ABA (Dodd et al. 2009; Parent et al. 2009). Different experimental designs have been used to avoid this confusion of effects by independently manipulating ABA and plant water status (ABA feeding, use of transgenics, pressurization of the root system or partial root drying). These designs have often resulted in different results. (2) The balance between positive and negative effects of ABA of the different processes described earlier is likely to depend on experimental conditions, thereby causing overall positive or negative effects of ABA on growth in different experiments. (3) Leaf growth is defined in different ways in the literature, with different effects of ABA for each definition. The gain of leaf biomass depends on photosynthesis, respiration and allocation of assimilates to leaves, while the volumetric expansion of leaf tissues is largely independent of biomass accumulation on time scales of days to weeks (Tardieu, Granier & Muller 1999). These two definitions involve physiological processes that have different interactions with ABA.

In this review, we have attempted to disentangle the effects of ABA on traits and functions involved in the response of leaf growth (essentially expansive growth) to water deficit or salt stress. In particular, we consider separately hydraulic and non-hydraulic processes that may interact with ABA in different ways. Non-hydraulic processes have been considered to have a leading role in the control of leaf growth under water deficit and to involve ABA (Davies & Zhang 1991; Sharp 2002). We discuss here the possibility of an important role of hydraulic processes in the response of leaf growth to water deficit, and of an appreciable implication of ABA in them.


A hydraulic control is suggested by the effects of evaporative demand and of aquaporin activity on leaf growth

The reduction in leaf growth under water deficit is classically considered as caused by a decrease in cell turgor, which reduces the driving force for cell expansion (Lockhart 1965). This view has been challenged by several experiments, which demonstrate that leaf growth is inhibited by water deficit or salt stress in spite of a maintained turgor in growing tissues as a result of osmotic adjustment (Michelena & Boyer 1982; Termaat, Passioura & Munns 1985; Tang & Boyer 2002). However, the opposite behaviour has also been observed (Hsiao & Xu 2000). In particular, leaf growth is decreased by evaporative demand (Shackel et al. 1987; Schnyder & Nelson 1988; Sadok et al. 2007), with a simultaneous reduction in turgor (Shackel, Matthews & Morrison 1987; Bouchabke, Tardieu & Simonneau 2006). We have recently shown that chemical manipulation of root hydraulic conductivity also causes simultaneous effects on leaf growth rate and on cell turgor in the growing zone (Ehlert et al. 2009), thereby suggesting that turgor and growth are more coupled than often assumed.

In the last 10 years, Boyer and co-authors have developed the theory that water entry into cells is a major limiting factor of growth, primarily controlled by the gradient of water potential between the xylem and the growing cells (Westgate & Boyer 1984; Tang & Boyer 2002; Tang & Boyer 2003). Any change in water potential of the xylem affects this gradient and can cause changes in growth rate even in the absence of changes in turgor of growing cells. Hence, the hydraulic conductance for the water transfer from soil to growing leaves can have a crucial role in the maintenance of leaf growth under water deficit. Indeed, a change in the activity of aquaporins affects leaf elongation rate. Aquaporins are proteins that facilitate the water transport through membranes, thereby increasing the hydraulic conductivity of tissues when their channel is open, and rapidly respond to environmental conditions (Maurel et al. 2008). When aquaporin activity is affected in roots by acid load or anoxia, leaf elongation rate decreases and becomes more sensitive to changes in evaporative demand (Ehlert et al. 2009).

Rapid responses of growth to water deficit suggest hydraulic processes

Another potent argument in favour of hydraulic mediations of the effect of water deficit relies on the very short time constant of the changes in elongation rate in responses to evaporative demand or soil water status. A steep decrease in leaf elongation rate occurs when evaporative demand increases either naturally in the morning (Sadok et al. 2007) or artificially in controlled conditions (Munns et al. 2000). This can be visualized in Fig. 1, in which leaf elongation rate (compensated for temperature) decreased each morning in about 1 h in both well-watered and water deficit treatments. In the study of Sadok et al., the time constant of the decline in leaf elongation rate was linked to quantitative trait loci (QTLs) of growth sensitivity to evaporative demand. Salt or PEG addition in the root medium also causes almost immediate cessation of growth (Chazen & Neumann 1994; Fricke et al. 2004). In the same way, full recovery of leaf elongation rate of droughted plants occurs in about 1 h after soil rehydration (Figs 1 & 3).

Figure 1.

Time course of the changes in (a) transpiration rate, (b) leaf water potential, (c) concentration of ABA in the xylem sap, (d) ABA flux into the leaf and (e) leaf elongation rate, during a sequence when maize plants were either well watered (green), subjected to progressive soil water deficit (red) or fed with artificial ABA (blue). Leaf elongation rate is expressed in a temperature-compensated way (per unit °Cd). Error bars are presented every 12 h for better legibility. Re-elaborated from Ben Haj Salah & Tardieu (1997).

Figure 3.

Root hydraulic conductivity and time course of recovery of leaf water potential and of leaf elongation rate (LER) in transformed maize lines affected in the ABA synthesis pathway via modification of NCED/VP14 gene expression. (a) Time course of leaf water potential before and after rewatering. Symbols, experimental points, lines are the outputs of the model of water transfer presented in Fig. 2, taking into account the differences in root hydraulic conductivity presented in panel c. At time 0, plants were rapidly rewatered, lights were turned off and the air vapour pressure deficit, initially at 2.8 kPa was reduced to 0.9 kPa, thereby drastically reducing the plant transpiration rate; (b) Time course of leaf elongation rate before and after rewatering. Rates are expressed as the proportion of the leaf elongation rate from 10 to 18 h after rewatering; and (c) Root hydraulic conductivity measured in hydropony. Redrawn from Parent et al. (2009).

Some non-hydraulic processes may have these rapid reaction times, but a hydraulic signalling still remains likely. (1) Cell wall stiffening is observed in response to PEG, but the signalling itself is probably hydraulic because it still occurs when roots are killed by freezing and thawing (Chazen & Neumann 1994; Chazen, Hartung & Neumann 1995). (2) Rapid responses of growth to evaporative demand or to changes in hydraulic conductivity are abolished if leaves are maintained at full turgor by pressurization (Munns et al. 2000; Ehlert et al. 2009). (3) In rehydration experiments, the rate of recovery depends on the root hydraulic conductance (Martre et al. 2002; Parent et al. 2009). Nevertheless, longer-term non-hydraulic effects of water deficit have also been observed (Gowing, Davies & Jones 1990; Munns et al. 2000).


Relations between stomatal conductance and expansive growth

Increasing ABA concentration in leaves by feeding plants with artificial ABA or by genetic manipulation largely affects stomatal conductance (Zhang & Davies 1991; Iuchi et al. 2001), thereby decreasing photosynthesis rate and biomass accumulation. Conversely, it can increase expansive growth under water deficit via improvement of plant water relations (Sansberro, Mroginski & Bottini 2004). Our results in maize are in agreement with this indirect, hydraulic effect of ABA via stomata on plant water status and leaf growth (Fig. 1). Plants fed with artificial ABA underwent smaller day–night alternations of leaf growth than well-watered plants and, above all, than plants subjected to water deficit. This can be attributed to the alternations of leaf water potential, considerably buffered in plants fed with ABA compared with well-watered plants. Plants subjected to water deficit transpired with approximately the same rate as plants fed with ABA, but had a lower leaf water potential during the day because of a lower soil water potential. This translated into very low leaf elongation rates during the day. The values of growth observed during the night in ABA-fed plants were lower than those of well-watered plants, suggesting a negative non-hydraulic effect of ABA, analysed in the last section of this paper.

Stomatal closure also causes a more subtle effect on growth via changes in leaf water status that are not accessible to straightforward measurements. Classical methods, either pressure chamber or psychrometer, measure water potential of leaf cells and not that of the xylem vessels, while growth is sensitive to xylem water potential (see previous discussion). An appreciable resistance exists between the xylem and the evaporation sites, themselves in equilibrium with the leaf cells (Fig. 2). The gradient of water potential between the xylem and the evaporation sites increases with the water flux through the plant. Hence, even if a well-watered and a droughted plant have similar leaf water potentials, the water potential in the xylem is still more favourable in a well-watered plant with high stomatal conductance than in a droughted plant with low stomatal conductance. Because leaf growth is closely linked to the xylem leaf water potential, this may explain the larger differences in leaf growth than in leaf water potential between well-watered and droughted plants in Fig. 1.

Figure 2.

Model of water transfer in the plant, explaining the gradient of water potential between the xylem (determinant of leaf growth) and leaf cells (measured by pressure chamber or psychrometer). Plants in well watered conditions (green) or in water deficit (orange) can have similar leaf water potential via stomatal control (isohydric species). Even in this case, the xylem water potential, which controls leaf growth, is still higher in well-watered plants than in droughted plants because of the steeper gradient of water potential in well watered plants, caused by higher water flux This difference between xylem and leaf water potential may be an explanation of the effect of partial root drying on leaf growth. The plant is represented by 4 compartments, each at a water potential (Ψr, roots, Ψxyl, xylem, Ψevap, sites of evaporation, Ψcel: leaf cells), separated by resistances Rsp, resistance between the soil and roots, Rr, resistance from the soil root interface and the xylem, Rxl, resistance from the xylem to the evaporation sites. Leaf cells are in equilibrium with evaporation sites, and act as a capacitance which can either take up water upon rehydration or release water to the xylem upon dehydration. This water transfer accounts for the delay in recovery of water potential presented in Fig. 3.

The role of ABA on stomatal control is non-controversial, but its origin and interactions are discussed

It is well established that the control of stomatal aperture involves a wide range of phytohormones (Acharya & Assmann 2009). Among them, ABA has been considered as the dominating long-distance signal synthesized in roots and transported via the xylem stream where it mediates the effects of drought on stomatal conductance (Davies & Zhang 1991; Tardieu & Simonneau, 1998). This is supported by split-root experiments in which only part of the root system experiences a low soil water potential, thereby causing stomatal closure while leaf water potential is maintained by the water supply from the well-watered roots. An important demonstration has also been provided by Zhang & Davies (1991), who showed that the sap of droughted plants closes the stomata of well-hydrated leaves, but that this effect disappears if ABA is removed from it. However, evidences for root-sourced signals other than ABA have been put forward (Munns & King 1988; Borel et al. 2001a). Jiang & Hartung (2008) reviewed the chemical factors regulating the intensity of the ABA signal, whereas Christmann et al. (2007) recently re-examined the contribution of hydraulic signals of drought, suggesting that ABA acting on stomatal closure is synthesized locally in leaves, and not distally in roots. These recent insights still support a role of ABA on stomatal closure, but question whether it may act as the primary long-distance signal of drought in several ways.

It can be questioned whether roots have the capacity to synthesize and communicate the ABA amounts observed in the xylem sap. Christmann et al. (2007) did not detect any substantial increase in ABA levels in roots of Arabidopsis seedlings in response to a −1.0 MPa water stress. Our group was also puzzled by the relatively low (about 3.5-fold) increase in ABA accumulation rate in isolated maize roots in response to root dehydration from 0 to −1.0 MPa (Simonneau, Barrieu & Tardieu 1998), while a 50-fold increase in ABA concentration was observed in the xylem sap of intact plants subjected to equivalent root dehydration. We have calculated the capacity of the whole root system to synthesize ABA from our results on detached root segments, and compared it to the ABA flux in the xylem stream of intact plants at different root water potentials. Although roots show a relatively small increase in ABA synthesis in water deficit, they have the capacity to supply ABA at the rates observed in the xylem stream plants subjected to a wide range of soil water deficits. This is mainly because of the concomitant limitation of water flow in drought conditions, which concentrates the slight increase in ABA delivery rate by roots in these conditions. The range of ABA concentrations in the xylem stream of drought-stressed plants is consistent with typical values required for closing stomata in bioassays with detached leaves (Tardieu, Lafarge & Simonneau 1996; Borel et al. 2001b), although with some exceptions (Munns & King 1988). However, drought-induced stomatal closure has been observed without root-sourced ABA in shoots of sunflower, tomato and Arabidopsis grafted onto ABA-deficient root stocks (Fambrini et al. 1995; Holbrook et al. 2002; Christmann et al. 2007). ABA acting on stomata in these conditions likely originated from outside the roots. ABA can be synthesized in a wide range of tissues (Nambara & Marion-Poll 2005), including leaves where key enzymes involved in drought-induced stimulation of ABA biosynthesis are highly expressed (Endo et al. 2008). Xylem ABA can also be recycled from the phloem to the xylem (Jeschke et al. 1997) without any contribution of roots. Glucose-conjugated ABA represents one of these inactive pools, which may be hydrolysed to active ABA by a stress-inducible β-glucosidase in the endoplasmic reticulum (Lee et al. 2006). Finally, ABA trapped in alkaline cell compartments may also be released in drought-stressed plants because of the alkalinization of their apoplast (Wilkinson & Davies 2008).


ABA increases root hydraulic conductivity and aquaporin activities

ABA induces transcription factors that regulate the expression of plasma membrane intrinsic protein (PIP) aquaporins (Kaldenhoff, Kölling & Richter 1996; Shinozaki et al. 1998) and affects a larger number of PIP isoforms than water deficit (Jang et al. 2004), suggesting some degree of independence between ABA and drought signal transduction pathways (Mariaux et al. 1998). However, when feeding the plants with artificial ABA, the increase in PIP mRNA expression is often transient (Zhu et al. 2005; Beaudette et al. 2007), and does not necessarily result in an increase in PIP protein content (Morillon & Chrispeels 2001; Aroca et al. 2006).

Water deficit and salt stress usually decrease the root hydraulic conductivity (Lo Gullo et al. 1998; Zhang & Tyerman 1999; North, Martre & Nobel 2004; Boursiac et al. 2005; Vandeleur et al. 2009), but the specific effect of ABA is more discussed. At the cell level, artificial ABA causes a transient increase in root hydraulic conductivity, which often disappears after a few hours (Hose et al. 2000; Wan, Steudle & Hartung 2004; Lee, Chung & Steudle 2005). The same transient and somewhat contradictory effects have been observed at the whole root level (Quintero, Fournier & Benlloch 1999; Sauter, Abrams & Hartung 2002; Schraut, Heilmeier & Hartung 2005). In particular, experiments based on feeding plants with artificial ABA often resulted in no or a negative effect on hydraulic conductivity (Wan & Zwiazek 1999; Aroca et al. 2003), or in a transient effect (Hose et al. 2000), which depends on the ABA concentration (Beaudette et al. 2007).

Two recent studies involving the genetic manipulation of endogenous ABA have clarified the effect of ABA under water deficit at several levels of plant organization (Thompson et al. 2007; Parent et al. 2009). In our own study, lines with contrasting rates of ABA biosynthesis generated by overexpressing (sense lines) or repressing (antisense lines) a key enzyme in ABA biosynthesis were studied at a constant leaf water potential in order to avoid confusion of the effects of water deficit and of ABA. Overproduction of ABA caused an increase in the mRNA expression of most aquaporin PIP genes, while the opposite was observed in lines that underproduced ABA. The same pattern was observed for the protein contents of 4 PIPs. This resulted in more than sixfold differences between sense and antisense lines in root hydraulic conductivity (Fig. 3 inset), which translated into differences in whole-plant hydraulic conductance measured either in steady state or during a recovery of leaf water potential. After irrigation, sense plants with high hydraulic conductivity recovered more rapidly their water potential than wild-type (WT) plants and anti-sense plants, which had lower hydraulic conductivities (Fig. 3). The model of water transfer presented in Fig. 2 accounted for these differences when it took into account the measured differences in hydraulic conductivity. These differences in conductance in turn caused a faster recovery of leaf growth (Fig. 3). The study of Thompson et al. (2007) also revealed an effect of ABA overproduction on the hydraulic conductivity of roots, with a positive effect on leaf area.

Other contributions of ABA to the whole-plant hydraulic conductance

The overall conductance to the water transport from the soil to the growing cells of leaves depends on other traits in addition to root hydraulic conductivity.

Overall, and in spite of numerous unknowns, the most consensual effects of ABA on whole plant hydraulic conductance all lead to an increased water uptake, via an increase in hydraulic conductivity, and via maintenance of root growth and changes in the architecture of the root system. These responses tend to maintain the leaf water potential and the water flow from the xylem to the growing cells at a high level and, therefore, promote leaf growth under water deficit during the time of the day when transpiration rate is substantial (Ehlert et al. 2009). Conversely, none of these effects are expected to affect leaf growth during the night, while water fluxes are low in the plant.


Mechanisms for non-hydraulic effects on leaf growth

Cell walls of leaf tissues become less extensible when plants are exposed to water deficit, thereby reducing the tissue expansion rate (Matthews, van Volkenburgh & Boyer 1984). Cell wall stiffening is believed to be a major cause of the reduction in leaf growth when water deficits have a small effect on cell turgor. Three gene families are the main molecular candidates for changes in cell wall properties: expansins, xyloglycan endo transglycosylases (XET) and peroxidases (Cosgrove 2005). In particular, a proportion of the expansin family has an expression that is appreciably affected in case of water deficit in roots (Wu & Cosgrove 2000) and leaves (Muller et al. 2007). The latter study shows that the local expressions of four expansins in the growing zone of the maize leaf correlates with local leaf elongation rate under both well-watered and dry conditions. Can enzymatic changes in cell wall extensibility account for the fast responses of leaf expansion presented in Figs 1 and 2? Several hours are necessary for the whole process of protein synthesis and trafficking towards the cell wall, consistent with time responses of XET activity observed by Bacon, Thompson & Davies (1997). However, Chazen & Neumann (1994) found that cell stiffening can occur some minutes after the initiation of a water stress, suggesting rapid post-translational processes in the cell wall triggered by an hydraulic signal. If this was correct, ABA may trigger these post-translational processes either directly or via the hydraulic signal that drives their changes. Consistently, feeding plants with ABA causes an increase in yield threshold (the minimum force causing cell wall extension) in maize plants, but this was observed with high concentrations of ABA (Cramer, Krishnan & Abrams 1998). The resulting effect on leaf elongation rate was observed in about 2 h for both decrease in leaf growth with ABA feeding and recovery after ABA removal, i.e. with reaction times slower but which may still be compatible with the rapid time courses described earlier. It is noteworthy that this effect of ABA was observed even when roots were cut, i.e. without root hydraulic resistance.

Cell division rate is affected by water deficit in leaves (Schuppler et al. 1998; Aguirrezabal et al. 2006; Granier & Tardieu 2009), probably linked to the activity of cyclin dependent protein kinases, in particular the p34cdc2kinase (Granier, Inze & Tardieu 2000). ABA induces the expression of an inhibitor of this process, thereby potentially affecting the cell cycle (Wang et al. 1998). However, the short time responses of leaf growth are probably not compatible with an effect of changes in cell cycle duration, which is 15–30 h in most species (Tardieu & Granier 2000). This suggests that changes in cell cycle may follow, and not drive, those of leaf expansion in case of a water deficit.

Sugar availability and photosynthesis participate to the effect of soil water deficit on the limitation of leaf growth if the latter is defined in terms of biomass. ABA is involved in this decrease via its effect on stomatal control (Liang, Zhang & Wong 1997). The involvement of sugar availability on expansive growth under water deficit is more doubtful. Photosynthesis rate can affect leaf expansion during the first phases of leaf development (Cookson & Granier 2006; Wiese et al. 2007), but a direct role under water deficit is unlikely because the concentrations of soluble sugars are usually increased by water deficit (Tardieu et al. 1999; Kim et al. 2000; Trouverie et al. 2004). This is because of several potential mechanisms, including a lower utilization due to a lower supply of polysaccharides and pectins to the cell wall, which accompanies the decrease in expansive growth. It may also be linked to a lower dilution of sugars by the growth water, due to the fact that expansive growth is usually more affected than photosynthesis. ABA is involved in the synthesis of key enzymes of the sugar metabolism (Kim et al. 2000; Trouverie et al. 2004).

Evidences for a non-hydraulic negative role of ABA on growth

A first series of arguments relies on the synchrony of the changes in leaf growth rate with the concentrations of endogenous ABA in the xylem sap ([ABA]xyl]) or in growing organs of plants subjected to progressive water deficit. For instance, (Fricke et al. 2004) described simultaneous but contrasting changes in ABA concentration and in growth in different regions of barley leaves. This is also consistent with observations in maize that changes in leaf growth occur simultaneously with those of [ABA]xyl during scenarios of soil dehydration or of rehydration (Fig. 4).

Figure 4.

Response of maize leaf elongation rate to the concentration of ABA in the xylem sap. (a,b) plants were either subjected to progressive drought (empty symbols) or fed with increasing amounts of ABA (filled symbols). Growth rates are expressed as the proportion of the average rate of well watered plants; and (c) synthesis of results obtained with plants affected on the ABA biosynthesis via the NCED VP14 gene. Growth rates are expressed as a proportion of the elongation rate of wild type plants (WT). (a) Night growth rate in two series of experiments in the greenhouse (represented by different symbols, circles or squares). One of these experiments is presented in Fig. 1. Re-elaborated from Ben Haj Salah & Tardieu (1997); (b) Growth chamber, re-elaborated from Zhang & Davies (1990); and (c) Effect of the manipulation of the gene NCED VP14. Re-elaborated from Voisin et al. (2006) for six independent lines represented each by a different symbol (empty symbols), and unpublished data (filled symbols) obtained with the lines presented in Fig. 3.

A second series of arguments is based on the similarity of the effects of artificial ABA fed to plants and of endogenous ABA synthesized during drought episodes. Rice coleoptiles have a reduced growth upon ABA feeding, and this effect is alleviated by fluridone, an inhibitor or ABA synthesis (Hoffmannbenning & Kende 1992). These results are strikingly similar to those observed with endogenous ABA in maize mesocotyl subjected to water deficit (Saab et al. 1990). In sunflower, barley, soybean and maize leaves, similar effects were observed for fed and endogenous ABA measured either in the xylem sap (Zhang & Davies 1990; Ben Haj Salah & Tardieu 1997) or in the tissues of the leaf elongating zone (Creelman et al. 1990; Dodd & Davies 1996; Cramer et al. 1998; Cramer & Quarrie 2002). Two series of experiments are presented in Fig. 4 for maize, in which leaf elongation rate was similar at a given [ABA]xyl, whether this ABA was endogenous or fed to the plant. Because the compartmentation of ABA is largely different in well-watered and droughted plants (Hartung, Radin & Hendrix 1988; Bacon et al. 1998), the comparison of the effects of artificial ABA fed to well-watered plants to that of endogenous ABA of droughted plants may not be valid, and the similarity of response curves in Fig. 4a,b may be circumstantial. Furthermore, relationships between growth rate and [ABA]xyl applied during the night (Fig. 4a) or under very low evaporative demand (Fig. 4b), but not under high evaporative demand. It can be seen in Fig. 1 that plants subjected to water deficit and to ABA feeding had a markedly different behaviour during the day, as discussed in the section about stomatal control. This suggests that non-hydraulic mechanisms may have an appreciable role during the night but that they are overridden by hydraulic mechanisms in the presence of an appreciable evaporative demand.

The third series of arguments is based on the existence of positive root-sourced signals affecting leaf growth in split-root experiments, in which part of the root system grows in dry soil while another part grows in wet soil (Gowing et al. 1990; Liu et al. 2001; Sobeih et al. 2004). In all these experiments, leaf water potential and/or calculated turgor was indistinguishable in leaves of well watered plants and in those subjected to partial root drying (PRD), while leaf growth was largely affected by PRD, e.g. by 60% in apple trees or 30% in tomato. An interesting fact is that leaf growth recovered and almost reached the rate observed in well-watered plants when the roots that experienced a water deficit were excised, so the plant only relied on roots growing in wet soil (Gowing et al. 1990). This has been considered as a potent argument in favour of a root-sourced message that controls leaf growth under water deficit (Davies & Zhang 1991). ABA may contribute to this non-hydraulic root message evidenced in split root systems, although an increase in the concentration of [ABA]xyl of plants subjected to PRD may or may not be observed, depending on experiments and species (Zhang & Davies 1987; Sobeih et al. 2004; Dodd, Egea & Davies 2008). An alternative interpretation of split root experiments is proposed in further discussion.

Pressurization experiments and genetic studies do not support a key role of ABA-mediated non-hydraulic control of growth

Reductions in leaf growth upon water deficit are not observed any more when plants are kept fully turgid by placing their root system in a pressure chamber so that the xylem sap is maintained at atmospheric pressure. In this case, no effect on growth is observed upon water deficit, changes in evaporative demand, salt stress or decrease in root hydraulic conductivity (Munns et al. 2000; Tang & Boyer 2003; Ehlert et al. 2009). The absence of effects of water deficit on leaf growth of pressurized plants is a potent argument against non-hydraulic messages affecting leaf growth.

Genetic engineering of the plant ability for ABA synthesis suggests a weak effect of ABA in the control of growth under water deficit. Cramer (2002) found no difference in response to water deficit of leaf growth of A. thaliana plants, either WT or affected in genes of ABA biosynthesis. The respective effects of the genetic manipulation of ABA biosynthesis, of ABA synthesized by droughted plants and of ABA fed to well-watered plants are compared in Fig. 4. The NCED-VP14 gene was manipulated in six transformants in the experiment of Voisin et al. (2006) and in five transformants in that of Parent (unpublished). All transformants were compared to their respective null transformants (called WT hereafter) at a common soil water potential of −0.5 MPa. Hence, there was no confusion of effects between [ABA]xyl and leaf water status as in the cases of the studies presented in Fig. 4a,b. This resulted in a limited negative effect of [ABA]xyl (e.g. relative growths of 0.8, 1.0 and 1.1 in sense, WT and antisense plants in the study of Parent, Fig. 4c). This result only apparently contradicts studies that conclude to a promoting effect of ABA on leaf growth via a limitation of ethylene biosynthesis in mutants of tomato and of A. thaliana (Sharp et al. 2000; LeNoble, Spollen & Sharp 2004; Dodd et al. 2009). The opposite effects of ABA and ethylene result in the fact that a minimum biosynthesis of ABA is required for leaf growth in well-watered conditions. This does not necessarily imply that ABA also determines leaf growth maintenance under water deficit. Indeed, the compensation ABA–ethylene did not account for the effect of ABA on the leaf growth of maize plants subjected to a mild soil water deficit (Voisin et al. 2006).

Quantitative genetics does not provide compelling evidence in favour of a large non-hydraulic role of ABA on leaf growth either. In a maize mapping population, few common QTLs were observed for [ABA]xyl of plants maintained at a water potential of −0.4 MPa and for the sensitivity of leaf elongation rate to soil water status (Reymond et al. 2003). Furthermore, these co-localizations of QTLs were not confirmed in further studies (W. Sadok, unpublished data). This suggests that the genetic link between ABA and growth maintenance under water deficit is weak, consistent with the small effect observed in Fig. 4c. It is noteworthy that when soil water potential is not maintained at constant values for all lines, plants with largest leaf area or root system deplete soil more rapidly. To what extent this will translate into co-localization of QTLs of growth and of ABA will depend on water availability. In pot experiments, a co-localization of [ABA]xyl and leaf area was observed, with an apparent positive effect of the allele conferring high [ABA]xyl on leaf growth, because of rapid soil depletion by plants with largest leaf area (unpublished data). Consistently, a QTL for leaf ABA concentration (Landi et al. 2007) has been interpreted as a consequence of differences in root depth and distribution in the field (Giuliani et al. 2005).

Overall, the non-hydraulic controlling effect of endogenous ABA on leaf growth of droughted plants is probably weak, but still appreciable in maize, with a negative effect of high [ABA]xyl on leaf growth. In our experiments, ABA overproduction still had a large negative effect on the cumulated growth over several weeks. Conversely, null or positive effects of ABA on leaf growth of droughted plants may exist in other species (Sharp 2002; Sobeih et al. 2004) as they do in maize roots and mesocotyl (Sharp 2002).

An alternative interpretation of the reduction of leaf growth in split root systems

Alternative interpretations of the negative effect of partial root drying on leaf growth have to be elaborated if ABA is not the main responsible of the positive root message which limits growth. It is surprising that leaf water status is only maintained, not improved, in split root systems compared with well-watered plants. Plants in split root systems have access to water, so they can take up water, and they have a reduced transpiration rate because of lower stomatal conductance. The leaf water potential of plants with PRD should, therefore, be higher (closer to 0) than that of well-watered plants. This raises the possibility that an appreciable water transfer occurs from the wet to the dry compartment of soil via the root system, thereby lowering the leaf water potential because of a second water sink in addition to transpiration. If PRD plants have a low transpiration rate, their xylem water potential is close to the water potential of mature cells because of a slow flux between the xylem and the sites of transpiration (Fig. 2). The difference in water potential between the xylem and mature cells is greater in well watered plants which have a higher water flux. Xylem water potential is therefore lower in plants subjected to PRD than in well-watered plants, although leaf water potential measured with pressure chamber or psychrometers are similar. This can affect leaf growth rate via an hydraulic mechanism linked to changes in xylem water potential (Tang & Boyer 1996).


Manipulation of ABA biosynthesis has been considered as an avenue for improving drought tolerance (Iuchi et al. 2001; Xiong et al. 2006). The multiplicity of mechanisms and feedbacks described earlier, on leaf growth only, suggests that this strategy may cause unexpected effects. Furthermore, these mechanisms are only a part of the putative effects of ABA, in view of the large number of ABA-inducible genes (Matsui et al. 2008) and molecular interactions of ABA with growth (e.g. Baena-Gonzalez & Sheen 2008, Moes et al. 2008). Comprehensive networks of experiments would be necessary for testing the effect of ABA manipulation on plant behaviour under a large range of scenarios of water deficit for testing in which cases this strategy is risky. For the time being, we can propose that ABA affects leaf growth via at least three types of mechanisms.

  • 1A non-hydraulic, intrinsic effect. We propose that endogenous ABA tends to reduce leaf growth at least in maize, with an effect which is observed during the night or during days with very low evaporative demand. This non-hydraulic effect would therefore be opposite in leaves and roots in maize. The possibility also exists of a positive effect of ABA on leaf growth in other species via the opposition ethylene-ABA.
  • 2A growth promoting effect of ABA via stomatal closure and maintenance of leaf water status. It is noteworthy that this positive effect only applies to expansive growth, while the negative effect of ABA on biomass accumulation via stomatal closure is not controversial. Thus, this growth promoting effect is probably not sustainable on time courses longer than a few weeks.
  • 3A growth-promoting hydraulic effect of ABA on tissue hydraulic conductance in roots and leaves, via its effect on aquaporin activities and on the architecture of the root system. This is expected to have a high influence on leaf growth under high evaporative demand, but a lower influence in a climate with low evaporative demand.

These combinations of contradictory mechanisms can have counter intuitive overall effects (e.g. Tardieu 2003), thereby requiring a modelling approach aimed at testing the effects of ABA in a large range of climatic scenarios (Hammer et al. 2006).


We are grateful to Bill Davies, Robert Sharp, Ian Dodd and D Gowing for very helpful discussions during the redaction of this paper. This work has been partly funded by the Agence Nationale de la Recherche # ANR-08-GENM-003 ‘Dromadair’ in the National Programme Génoplante – ONIGC.