Drought-induced changes in development and function of grapevine (Vitis spp.) organs and in their hydraulic and non-hydraulic interactions at the whole-plant level: a physiological and molecular update

Additional keywords: abscisic acid, anisohydric, anthocyanins, aquaporin, genome, isohydric, polyphenols, proteomics, stomatal conductance, transcriptomics, water use efficiency.

Advances in the understanding of plant responses to water deficits require information from the molecular level to that of whole plant (Chaves et al. 2003). Grapevine has been used as model plant to study ecophysiological responses to water stress since the 1970s. Molecular studies began in the 1990s, and, more recently, applications of ‘-omics’ technologies, such as transcriptomics and proteomics, together with the availability of the grapevine genome sequence (Jaillón et al. 2007; Zharkikh et al. 2008) have provided a powerful molecular counterpart to physiological analyses (Troggio et al. 2008). This review uses a physiological platform to incorporate recent molecular developments to provide an understanding of how grapevine responds to and recovers from drought stress.

Grapevine is grown widely throughout the world, with its production making the top agriculture lists in many countries (Bisson et al. 2002). According to FAOSTAT time-series and cross-sectional data (http://faostat.fao.org, accessed September 2009) there were 66 271 676 tonnes of grapes produced on 7 501 872 ha in 2007.

Vitis vinifera L. cultivation is traditionally non-irrigated (especially in Europe) and spread widely across dry and semi-dry ecosystems. Yield and berry quality depend strongly on the vine adaptability to drought. Water stress does not imply exclusively negative effects, but a regulated water stress, which is the base of various agronomic practices (e.g. rootstock use, controlled cover crops, tillage, rescue irrigation techniques such as regulated deficit irrigation or partial root-zone drying) has been largely used to balance vine vegetative and reproductive growth with the aim of controlling berry quality. Understanding and manipulating plant–water relations and water-stress tolerance by means of physiology and molecular biology can significantly improve plant productivity and environmental quality, hence, is clearly of wide economic importance in viticulture.

In this review we examine root, shoot, leaf and berry, linking physiological and molecular knowledge to applied research. We focus on the role of aquaporins in water movement and how water constraints modify organ functioning and generate signals directed to other plant organs. Drought signalling among grapevine organs has a dual component: a hydraulic signal controlled by xylem physiology coexists with chemical signals (involving hormones), transported via xylem, phloem and parenchyma pathways (Loveys 1984b; Lovisolo et al. 2002a). Grapevine is a model species to study the different drought signalling mechanisms, as many cultivars and genotypes are available, including the different species of the genus Vitis.

When roots sense a soil water deficit, root cells respond in terms of their growth and differentiation, and also transmit a signal that is then perceived in the shoot. The response to drought involves first the acidification of the apoplast by protons pumped from the cytoplasm. This acidification stimulates the action of expansins, important in cell wall synthesis. Of the 10 expansin genes in grapevine, Vlexp2 isolated from Vitis labruscana (Vitis labrusca L. × V. vinifera L.) cv. Kyoho berries, showed high root expression level (Ishimaru et al. 2007). From V. vinifera cv. Arka Neelamani, Thomas et al. (2003) isolated VvPRP1 and VvPRP2, two genes encoding a distinct type of proline-rich proteins (PRPs), a major category of cell wall proteins in plants, involved in different developmental and environmental responses to abiotic stresses. Grape PRP genes are expressed in root tissues and not detected in shoot tip or leaf tissues; they play a role in the initiation of new roots on grape stem cuttings, probably by altering the cell wall mechanical properties to enable root emergence.

Although a reduction in leaf and shoot growth is one of the signs of grapevine water deficit (Stevens et al. 1995), the decrease in root growth is less than shoot growth (Dry et al. 2000a, 2000b). This results in a higher root : shoot ratio, which ensures adequate water and nutrient transport to the shoots. The tolerance of grapevines to drought has often been attributed to their ability to produce new roots selectively where soil water is available (Morlat and Jacquet 1993; Dry et al. 2000b). Droughted roots continue to grow into deeper, wetter soil layers, whereas the roots of irrigated plants proliferate mostly in the topsoil (Bauerle et al. 2008). As deeper roots procure soil-mobile nutrients, like nitrate (NO₃^–) (Keller 2005), this mechanism also enhances plant nitrogen uptake.

Shoot growth inhibition comprises inhibition of internode extension, leaf expansion and elongation of tendrils (Schultz and Matthews 1988; Hardie and Martin 2000) and has been used as a sensitive indicator of grapevine water status (Pellegrino et al. 2005; Lebon et al. 2006). In addition, comparably low water stress levels induce a decrease in the average diameter of grapevine vessels and a decrease of xylem hydraulic conductivity (Lovisolo and Schubert 1998; Lovisolo et al. 2002b). Although shoot growth inhibition limits transpiration, reduced vessel size prevents excessive loss of water by reducing xylem conductivity and may help to prevent embolisation, as smaller vessels are less susceptible to cavitation (Salleo et al. 1985). Extreme drought consequences at the shoot level involve shedding of leaves and cessation of secondary growth, as reviewed by Keller (2005).

V. vinifera Grenache and Chardonnay plants show a higher amount of suberin deposited in cell walls of droughted roots than in well watered roots in response to water stress, both at endodermal and exodermal levels. In Grenache, all cells of the endodermis appear to become suberised, whereas in Chardonnay some passage cells are still evident. These cellular adaptations assist in modulating stress responses in these two grape cultivars (Vandeleur et al. 2009).

In addition to root growth and apoplastic barrier formation, plants can modulate responses to soil water stress by comparmentalising the water transport pathways which are under metabolic control. This implies that roots can either modulate water-driving cell osmotic forces and/or regulate expression and activity of aquaporins (Galmés et al. 2007; Lovisolo et al. 2008b).

Aquaporins are thought to control the radial movement of water through roots (Maurel et al. 2008) and through living tissues adiacent to root and shoot xylem cells (Kaldenhoff et al. 2008). Some aquaporins are constitutively expressed (Johansson et al. 2000) but the expression of others is regulated by different stimuli, such as adverse environmental conditions (Vera-Estrella et al. 2004). Indeed, expression of different aquaporin genes may be stimulated, decreased or unchanged under abiotic stress (Kaldenhoff et al. 2008). Grapevine aquaporins have been cloned by Baiges et al. (2001), Picaud et al. (2003), Perrone et al. (2006), Reid et al. (2006), Galmés et al. (2007), Fouquet et al. (2008), Glissant et al. (2008), Schlosser et al. (2008), Shelden et al. (2009) and Vandeleur et al. (2009), and reported to be expressed either in roots or shoots.

The role of aquaporins in root water transport was investigated by comparing grapevine cultivars that differed in water use strategies, as described by Vandeleur et al. (2009). These workers studied gene expression of VvPIP1;1 and VvPIP2;2, the two PIP aquaporins with higher expression in roots of both Chardonnay and Grenache plants. VvPIP2;2 showed a constitutive expression in root regardless of cultivar or soil water condition, whereas VvPIP1;1 transcript expression increased during water stress in Chardonnay, but not in Grenache. Functional assays showed a gain of water permeability (P_os) when VvPIP2;2 and VvPIP1;1 cRNA were injected together in the oocytes but not when VvPIP2;2 was injected alone. Hence, it was proposed that Chardonnay plants show a minor reduction of root hydraulic conductance caused by water stress through increased transcellular component of radial water transport. This provides water while transpiration continues; thus, an anisohydric behaviour (see below) is reported for this cultivar. In this way, Chardonnay plants are able to maintain lower water potential differences between soil and xylem, possibly due to lower vulnerability to embolism formation compared with Grenache, as described by Alsina et al. (2007).

Grenache plants demonstrate a different strategy: the finding of no variation in transcript level of most important root PIP aquaporins and suberisation implies a lower hydraulic conductance in water deficit conditions. This supports the hypothesis of tight control on stomatal regulation that is typical of isohydric (see below for explaination) cultivars like Grenache, which aims to avoid excessively negative xylematic water potential and, therefore, cavitation (Schultz 2003a; Soar et al. 2006; Vandeleur et al. 2009). It is hypothesised that in the isohydric Grenache, the drought-induced root abscisic acid (ABA) biosynthesis increases apoplastic concentration because of a concomitance of events: an increase of suberisation of apoplastic barriers causes a reduction of water conductivity which is not compensated by aquaporin-mediated water transport.

Literature reviewed in this section shows a lack of experiments dealing with root hydraulics; grapevine root aquaporins are, in some cases, performed on Vitis spp. rootstocks grafting V. vinifera scions (e.g. Schultz 2003a; Alsina et al. 2007), whereas others are performed on own-rooted varieties of V. vinifera (e.g. Soar et al. 2006; Vandeleur et al. 2009). However, the iso- or anisohydric response is always detected at the scion level, showing that a strict causal linkage of the scion response with root aquaporins occurs only on own-rooted plants.

Embolism of xylem vessels as a result of water stress has been extensively studied in grapevines. Under high levels of tension, gas-filled xylem vessels may become disrupted by breakage of water columns, which causes embolism formation and drastically reduces hydraulic conductance (Schultz and Matthews 1988; Lovisolo and Schubert 1998). In different studies comparing root and shoot embolisms, roots have been shown to be more vulnerable to xylem cavitation than shoots (Lovisolo and Schubert 2006; Lovisolo et al. 2008a). This protects the stem from extreme xylem tensions during severe drought (Sperry et al. 2006). Froux et al. (2005) demonstrated that lateral roots seem to be more prone to embolism than the main root, suggesting a behaviour similar to ‘hydraulic segmentation’ in petioles. In plants with an anisohydric type response to drought, like Populus euphratica (Hukin et al. 2005), a lower cavitation vulnerability of the roots compared with shoots was found. In P. euphratica stomatal closure occured relatively late, well after shoot hydraulic conductance was significantly affected by embolisms, showing that in the anisohydric response to drought a lower vulnerability of roots to cavitation may be part of the survival strategy.

Although grapevine petioles and roots are more vulnerable to embolism than shoots (Lovisolo et al. 2008a), when plants are rewatered, either root or shoot and petioles recover ~35–40% of hydraulic conductivity within 24 h, suggesting that a common and coordinated mechanism of recovery among all plant organs occurs. The proposed mechanism is based on three points: (i) embolism repair occurs progressively in shoot and then in root and in petioles, following an almost full recovery of leaf water potential; (ii) hydraulic conductance recovery in all plant organs also occurs during diurnal transpiring hours when formation and repair of embolisms occurs in all plant organs; and (iii) a non-hydraulic stress-derived ABA residual signal in rehydrated leaves hinders stomatal opening even when leaf water potential and the overall plant hydraulics are recovered, suggesting that an ABA-induced transpiration control promotes gradual embolism repair in rehydrated grapevines (Lovisolo et al. 2008a).

To reintegrate vessel functionality, plants have developed different repair mechanisms, which, in some cases, are associated with positive root pressure, and in other cases involve active and energy-consuming processes in shoot conductive tissues (Salleo et al. 2004). In grapevine, experiments conducted with the metabolic and water transport inhibitor mercuric chloride point to the presence of an active mechanism involving the contribution of living cells (Lovisolo and Schubert 2006), and also possibly involving the contribution of aquaporins. A role of aquaporins in embolism repair was previously hypothesised based on asymmetrical aquaporin distribution, which appeared enriched at the interface between xylem vessels and associated living cells (Kaldenhoff et al. 2008). Aquaporin distribution in grapevine cells has not yet been described, although in water-stressed grapevine shoots the expression of an aquaporin (PIP1) was found to be downregulated (Cramer et al. 2007). However, aquaporins can be activated by post-translational modifications, e.g. by phosphorylation, thus, their role in embolism refilling cannot be ruled out on the basis of expression studies.

Aquaporins may be involved in embolism repair. In walnut (Juglans regia L.) PIP2 aquaporins, localised in xylem vessel parenchyma cells, were activated during spring embolism recovery (Sakr et al. 2003). PIP1 and PIP2 antisense Arabidopsis plants were slower than controls in the conductance recovery after rewatering (Martre et al. 2002). Tobacco (Nicotiana tabacum L.) RNAi plants showing an impaired expression of PIP1 or PIP2 genes led to a direct evidence of aquaporin involvement in embolism recovery. PIP1-RNAi and control plants repaired embolism in a few hours after rehydration, but PIP2-RNAi plants showed a delayed kinetics of recovery (Kaldenhoff et al. 2008).

Results of recent work suggest ABA/aquaporin interaction in embolism repair and in aquaporin activation during water stress (Kaldenhoff et al. 2008). ABA may be involved in gating mechanisms of water channels by facilitating their structural restoration, possibly acting from the cytoplasmatic side of aquaporins (Wan et al. 2004). Although downregulation of aquaporins after ABA treatment has been reported (Mariaux et al. 1998; Suga et al. 2002), more often aquaporins, if responsive, were upregulated by this hormone (Jang et al. 2004). Parent et al. (2009) obtained sense and antisense transgenic maize (Zea mays L.) plant for 9-cis-epoxycarotenoid dioxygenase NCED/VP14 gene that catalyses the first specific step in ABA biosynthesis. Expression levels of most root PIP isoforms were significantly higher in sense plants than in antisense plants, and analogous results were obtained for the influence on protein content. Moreover, a long-lasting effect was observed. This transgenic approach allows us to distinguish between ABA and non-ABA effects on aquaporins.

Under conditions of high irradiance and vapor pressure deficit (e.g. midday of clear summer days), water flow into grapevine leaves, as in many other species, is insufficient to compensate water losses through evapotranspiration, resulting in a midday to afternoon depression of leaf water potential (Schultz 2003a; Chaves et al. 2007). As a consequence, midday to afternoon depression of stomatal conductance (g_s) and net photosynthesis (A_N) has been reported in many cultivars, even under sufficient soil water availability (Gómez-del-Campo et al. 2004; Moutinho-Pereira et al. 2004). Differences among cultivars in stomatal responsiveness to midday conditions have been described (Winkel and Rambal 1990).

Midday stomatal closure is often associated with accumulation of ABA in the petiole xylem and leaves (Loveys 1984a; Rodrigues et al. 2008), although increased xylem pH (Rodrigues et al. 2008) and decreased plant hydraulic conductance (Salleo and Lo Gullo 1989; Vandeleur et al. 2009) could also be involved. However, stomatal closure is not the only cause of decreased photosynthesis during the midday depression. The depression involves both stomatal and non-stomatal factors, as reflected by lower photosynthesis during the afternoon than in the morning at any given stomatal conductance (Cuevas et al. 2006) or substomatal CO₂ concentration (Downton et al. 1987; Quereix et al. 2001). In addition, g_s in the afternoon is less sensitive to ABA and more sensitive to CO₂ than during the morning, also supporting the idea that reduced photosynthesis is limiting g_s in the afternoon and not vice versa (Düring 1991; Correia et al. 1995).

Non-stomatal limitations may also be partly responsible for the midday depression. Photoinhibition, feedback inhibition through source–sink interactions, and decreased mesophyll conductance to CO₂ have been suggested as important limiting factors, but none of these processes has been demonstrated to predominate (Flexas et al. 2008). For instance, Escalona et al. (2003) showed that, in irrigated plants, a midday depression occurs in the most exposed leaves of the canopy, but not in shaded leaves. Leaves exposed to constant high light, temperature and vapor pressure deficit had a maximum stomatal conductance 1 h after illumination, which declined thereafter (Correia et al. 1990; Lu et al. 2003). Also, using remote sensing of chlorophyll fluorescence, Flexas et al. (2000) showed that the quantum efficiency of PSII was lower during the afternoon than during the morning at any given light intensity. All these observations suggest a possible involvement of photoinhibition. Indeed, the maximum quantum efficiency of PSII (F_v/F_m) declines slightly at midday, although typically less than 20% (Quick et al. 1992; Bertamini and Nedunchezian 2004). Bertamini and Nedunchezian (2004) reported that this decline was initially (2 h after high light exposure) associated with a decrease in the concentration of the D1 protein (the core protein of PSII components), and later (4 h after high light exposure) with a decline of the 33KDa protein (the water-splitting complex) while D1 was recovered, suggesting that both acceptor side and donor side photoinhibition are involved in midday depression of photosynthesis. However, in other cases were F_v/F_m declines somewhat, it correlates better with the accumulation of de-epoxidated xanthophylls than with degradation of D1 protein (Chaumont et al. 1995). The mechanism of repair of D1 is very effective in grapevines; in its presence, the fraction of functional PSIIs is kept higher than 50% (and F_v/F_m between 0.6 and 0.8) even exposing leaves to photon exposures higher than those of a normal sunny day. Conversely, when the D1 repair is blocked with lincomycin the fraction of functional PSIIs and F_v/F_m rapidly decline to zero with photon exposures similar to those received by a grapevine at midday (Flexas et al. 2001). Moreover, in other experiments, F_v/F_m remained constant during the day (Iacono and Sommer 1996; de Souza et al. 2003), and there was no degradation of chlorophyll (Medrano et al. 2002a), suggesting that effective photoprotection is associated with midday depression of photosynthesis. Chlorophyll fluorescence data corroborate this view, showing midday to afternoon decline in the steady-state fluorescence emission and a constant or slightly declined in the effective quantum efficiency of PSII (φ_PSII) and rate of electron transport (ETR) (Correia et al. 1990; Flexas et al. 2000). Meanwhile, non-photochemical quenching parameters (qP, NPQ) largely increase and the photochemical reflectance index (PRI) decreases (Evain et al. 2004). These changes are associated with increased trans-thylakoid ΔpH and de-epoxidation of xanthophylls, reflecting safe thermal energy dissipation (Düring 1999; Medrano et al. 2002a; Evain et al. 2004). The rate of photorespiration increases or is kept constant in the afternoon in irrigated plants (Iacono and Sommer 1996; Flexas et al. 1999a, 2000), while the foliar pools of ascorbate and glutathione slightly increase, and their oxidised forms increase over the reduced forms (Chaumont et al. 1995). Together these mechanisms confer an effective photoprotection to grapevines leaves during the midday depression of photosynthesis.

Alternatively, a phloem-based feedback signal could be involved in the regulation of the balance between source and sinks activities, leading to an afternoon decline of photosynthesis (Quereix et al. 2001). This would be supported by the observed accumulation of sucrose and starch in leaves during the course of the day (Chaumont et al. 1994), although in other experiments they rather remain constant or even decrease (Quick et al. 1992). Alternatively, a decrease in mesophyll conductance to CO₂ (g_m) concomitant to decreased g_s, as observed by Moutinho-Pereira et al. (2004) in three different grapevine cultivars, may result in a further decrease in chloroplast CO₂ availability, explaining the non-stomatal component of photosynthesis reductions.

In the next section we provide a generalised view of grapevine leaf responses to water stress.

When soil water stress combines with high atmospheric water demand, reductions in leaf water potential and gas exchange become larger and longer-lasting (Liu et al. 1978; Flexas et al. 1998; Schultz 2003a; Pou et al. 2008). Based on their water potential behaviour in response to water stress, grapevine cultivars have been classified as isohydric or anisohydric (Schultz 2003a; Vandeleur et al. 2009). Isohydric cultivars are those that keep their leaf water potential above a certain threshold regardless of soil water availability or atmospheric water demand. Anisohydric cultivars are those in which leaf water potential drops with decreasing soil water availability or increasing atmospheric water demand (Fig. 1a). In isohydric grapevines, leaf water potential rarely drops below –1.5 MPa (Fig. 1a). This is close to the threshold for severe cavitation in this species (Salleo and Lo Gullo 1989; Lovisolo et al. 2008a), although some cavitation occurs at lower water potentials in petioles, shoot nodes and internodes and roots (Schultz and Matthews 1988; Salleo and Lo Gullo 1989; Schultz 2003a; Lovisolo et al. 2008a). According to observations in other isohydric species as laurels (Laurus nobilis L.) (Salleo et al. 2000), it may be argued that these cultivars present a fine co-regulation of g_s and leaf hydraulic conductivity, allowing them to avoid cavitation. Grapevines typically reported as isohydric include the hybrid species V. labruscana (Vitis labrusca × V. vinifera) and the rootstock Richter-110 (Vitis berlandieri Planch. × Vitis rupestris Scheele), widely used V. vinifera cultivars such as Grenache, Trincadeira Preta and Tempranillo, as well as some cultivars native to dry viticultural areas. Anisohydric cultivars, by contrast, drop leaf water potential through the day as a function of soil water deficit (Fig. 1a), which is often achieved by means of osmotic adjustment (Downton 1983; Düring 1984; Schultz and Matthews 1993; Patakas and Noitsakis 1997), although may also be through changes in cell wall elasticity (Patakas and Noitsakis 1997). Grapevines typically reported as anisohydric include the species V. labrusca and Vitis californica Benth., as well as many V. vinifera cultivars, including Chardonnay, Cabernet Franc, Cabernet Sauvignon, Syrah, Riesling, Carignan, Muscat, Thomson seedless, Touriga Nacional, as it appears from a literature survey (references in the legend of Fig. 1).

Although it has been suggested that differences between iso- and anisohydric cultivars may include different stomatal responses and water use (Schultz 2003a), on average, they show similar decreases in g_s (Fig. 1b) and A_N (data not shown) in response to water stress. As a consequence, both iso- and anisohydric cultivars present similar values of leaf intrinsic water-use-efficiency (A_N/g_s) both under irrigation and under water stress (Fig. 1c). Indeed, differences among grapevine cultivars in water use efficiency (WUE) have been reported, based on instantaneous gas-exchange data (Gómez-del-Campo et al. 2004), isotopic composition (¹³C/¹²C) of leaf and/or fruit dry matter (Gibberd et al. 2001; Gaudillère et al. 2002) or biomass accumulation per unit of water used (Gibberd et al. 2001). However, no clear pattern of correlation is observed in these studies between WUE and the iso- or anisohydric character of the studied cultivars.

Moreover, the same cultivar can behave as iso- or anisohydric, depending on the conditions. For instance, V. labruscana was reported to be anisohydric by Liu et al. (1978), but as isohydric by Naor and Wample (1994). Pinot Noir behaves as anisohydric when water stress is applied pre-véraison and as isohydric when it is applied post-véraison (Poni et al. 1993). In contrast, during most of the growing season (June–July), Tempranillo and Manto Negro are often reported as isohydric (Flexas et al. 1998; Medrano et al. 2003), but later in the season (August), however, they behave as iso- or anisohydric, depending on the year (J. Flexas and H. Medrano, unpubl. data). However, the cavitation threshold of –1.5 MPa described for grapevines (Salleo and Lo Gullo 1989) is rarely and barely reached even during the ‘anisohydric years’, suggesting that grapevines are effective in avoiding catastrophic cavitation.

Stomatal closure is one of the first processes that occurs in the leaves in response to soil water stress. Many signals are involved in sensing environmental variations during soil drought-induced stomatal closure. They are related more to ABA metabolism, hydraulic signals (cavitation), regulation of expression and activity of aquaporins, and/or electric signals, than changes in leaf water status, measured as leaf water potential. Indeed, with some exceptions, stem water potential has been described as a more preferable indicator of grapevine water status than leaf water potential (Choné et al. 2001; Intrigliolo and Castel 2006; Williams and Baeza 2007), and g_s often correlated better with leaf water potential determined at pre-dawn (Ψ_pre-dawn) than Ψ_midday (Winkel and Rambal 1990; Schultz 2003b; Intrigliolo and Castel 2006). This is also observed in the survey of iso- and anisohydric cultivars used in Fig. 1 (Fig. 2). Although a significant correlation is found when plotting midday g_s against Ψ_pre-dawn, pooling all the data together (Fig. 2a), no clear relationship is observed between g_s and Ψ_midday (Fig. 2b), even if only anisohydric cultivars (triangles) are considered. These observations suggest that g_s in grapevines decreases under water stress in response to some root- or stem-based signal(s). Indeed, this is the basis of the partial root drying technique used to reduce vegetative growth and water use and improve WUE in grapevines (Dry and Loveys 1998; Chaves et al. 2007).

Grapevines were among the first plant species in which a direct role of ABA in stomatal closure was demonstrated (Loveys and Kriedemann 1974; Liu et al. 1978; Loveys 1984a, 1984b). In different grapevine genotypes during the gradual imposition of soil water stress (non-irrigation) or partial root drying, tight negative correlations are often observed between g_s and either xylem (Pou et al. 2008; Rodrigues et al. 2008) or leaf tissue (Loveys and Kriedemann 1974; Liu et al. 1978; Lovisolo et al. 2002a) ABA contents. On the basis of these observations it is often assumed that root ABA synthesis in response to water stress and transport through the xylem into leaves mediates most of the stomatal response in grapevines, although a reverse transport, i.e. from leaves to roots has also been shown (Loveys 1984b). However, gradients in both xylem and leaf ABA along shoots of grapevines, from higher concentrations close to the apex to lower concentrations downwards have been observed (Soar et al. 2004). These gradients, which resulted in a gradient of g_s among leaves with different positions along the shoot, are inconsistent with the concept of root-derived ABA. Analysing the patterns of expression along the shoots of two genes involved in ABA synthesis, Soar et al. (2004) concluded that the observed differences in ABA were due to differences in the in situ synthesis of ABA in shoots and leaves. Soar et al. (2006) further confirmed that regulation of gene expression in response to increased vapour pressure deficit in leaf tissue and not in roots was associated with the higher xylem ABA concentrations found in Grenache than in Syrah. In Arabidopsis, it has been clearly demonstrated that g_s response to soil water stress is due to increased ABA synthesis in the shoots, not in the roots (Christmann et al. 2007). The observations of Soar et al. (2004, 2006), although not conducted on water stressed plants, suggest that this may also be the case in grapevines.

That ABA synthesis in the shoots and leaves increases in response to soil water stress implies that some other root-based signal may trigger this response. Other hormonal signals have been suggested, such as phaseic acid (Loveys and Kriedemann 1974) or cytokinins (Stoll et al. 2000), as well as the influence of xylem sap pH (Rodrigues et al. 2008), but no such roles have been confirmed. In contrast, in Arabidopsis the signal triggering an increase of shoot ABA synthesis has been shown to be of hydraulic nature (Christmann et al. 2007). Furthermore, there is a gradient of hydraulic conductivity in nodes and internodes along grapevine shoots, similar to the gradient described for ABA synthesis (Salleo et al. 1982, 1985; Lovisolo and Schubert 1998). In internodes, both xylem hydraulic conductivity (K_h) and leaf specific conductivity (LSC) are lower in the most apical parts, but in lower internodes they are similar (Salleo et al. 1982, 1985) or increase slightly from the shoot base to intermediate shoot positions, thereafter declining towards the apex (Lovisolo and Schubert 1998). In the nodes, LSC is often lower than in internodes and continuously declines from the base to the apex (Salleo et al. 1982). This distribution of hydraulic conductivity could explain why apical leaves synthesize more ABA and have lower g_s (Soar et al. 2004). There is evidence that hydraulic conductivity of roots, shoot nodes and internodes and petioles, as well as whole-plant conductivity decline in grapevines subjected to water stress (Schultz and Matthews 1988; Salleo and Lo Gullo 1989; Winkel and Rambal 1993; Lovisolo and Schubert 1998, 2006; Lovisolo et al. 2002a, 2008a, 2008b; Schultz 2003a; Pou et al. 2008). This decline can be due to water stress-induced changes in xylem development (Mapfumo et al. 1993; Mapfumo and Aspinall 1994), but most often occurs by means of drought-induced cavitation (Schultz and Matthews 1988; Salleo and Lo Gullo 1989; Schultz 2003a). Roots and leaf petioles appear to be more sensitive than shoots to drought-induced cavitation (Schultz 2003a; Lovisolo et al. 2008a), although in the shoots, internodes are much more sensitive than nodes (Salleo and Lo Gullo 1989). In addition to cavitation, aquaporins have been suggested to play a role in the regulation of hydraulic conductivity in grapevines. All known aquaporins in grapevines are much more abundant in roots than in any other tissue, although they are still present in shoots and leaves. Experiments of gene expression during drought and recovery in Richter-110 have shown that several PIP and TIP aquaporins are upregulated at early stages of water stress in roots, while they are mostly downregulated in leaves (Galmés et al. 2007). In contrast, after re-watering stressed plants, most aquaporins are upregulated in leaves but not in roots (Galmés et al. 2007). Using mercury as an inhibitor of the activity of some aquaporins, it has been suggested that aquaporins are involved in the recovery after water stress of shoot (Lovisolo and Schubert 2006) and root (Lovisolo et al. 2008b) hydraulic conductivity. Using this technique it has been shown that in drought-resistant rootstocks, aquaporin-regulated decrease of K_h is more important than cavitation, the opposite being true for drought-sensitive rootstocks. Moreover, it has been suggested that a reduced transpiration induced by ABA after re-watering promotes aquaporin-mediated embolism repair after water stress (Lovisolo and Schubert 2006; Lovisolo et al. 2008a).

Regardless of the mechanism for its regulation, whole-plant hydraulic conductivity often correlates well with g_s during drought imposition (Winkel and Rambal 1993; Schultz 2003a; Pou et al. 2008), and much better during recovery after water stress (Lovisolo et al. 2008a; Pou et al. 2008), suggesting that g_s may be regulated by hydraulic signals during water stress. However, in a factorial experiment involving water stress, partial root drying and downwards shoot position, Lovisolo et al. (2002a) presented clear evidence that it was leaf ABA and not whole-plant hydraulic conductivity that determines g_s in grapevines. Further work is required to understand the role of hydraulics on stomatal regulation in grapevines.

Alternatively, electrical signalling could be involved in stomatal regulation in grapevines, as suggested in other species (Fromm and Fei 1998; Grams et al. 2007). In grapevines, the two types of wound-induced electrical signals, namely variation potentials (VPs) and action potentials (APs), propagate very fast (~3 mm s^–1 and 100 mm s^–1 for VPs and APs, respectively) in leaf tissues (Mancuso 1999). Preliminary data (J. Fromm and J. Flexas, unpubl. data) showed that re-watering water stressed grapevine plants generates APs that are transmitted very quickly (less than 10 s) from roots to leaves, similar to the observations in Zea mays, where the electrical rather than the hydraulic signal was shown to induce stomatal re-aperture (Grams et al. 2007).

As in other species, a tight curvilinear correlation between g_s and A_N has been described in grapevines (Flexas et al. 2002a). This relationship is also observed when pooling together all data of the survey described in previous sections (Fig. 2c). The observed dispersion may depend on temperature and atmospheric vapour pressure deficit (Zufferey et al. 2000), leaf age (Zufferey et al. 2000) and time of the day or season (Gómez-del-Campo et al. 2004; Cuevas et al. 2006), but primarily due to intrinsic differences among cultivars (Düring 1987; Bota et al. 2001).

Due to the tightness of this correlation and to the fact that intrinsic WUE (i.e. photosynthesis to g_s ratio) increases as g_s decreases (Fig. 2d), it is often assumed that the drought-induced decrease of photosynthesis is mediated by stomatal closure. Indeed, under mild water stress this may be the case. Using the daily maximum value of g_s as an indicator of water stress that allows comparison of plants with iso- and anisohydric behaviours, Flexas et al. (2002b) and Medrano et al. (2002b) defined several stages of photosynthesis regulation in grapevines subjected to progressive soil water stress on the basis of the main causes of A_N decline. These stages are general regardless of the cultivar under study and are as follows:

1. Stage 1. Mild water stress: g_s decreases from a maximum (typically 200–500 mmol H₂O m^–2 s^–1 in grapevines) to 150 mmol H₂O m^–2 s^–1.

2. Stage 2. Moderate water stress (or transition stage): g_s ranges between 50 and 150 mmol H₂O m^–2 s^–1.

3. Stage 3. Severe water stress: g_s drops below 50 mmol H₂O m^–2 s^–1.

At Stage 1, the effects of water stress consist in a relatively small decline of A_N, which results in a progressive increase of intrinsic WUE (Fig. 2d) and a slight decline of substomatal CO₂ concentration (C_i). As a consequence of decreased CO₂ availability in the mesophyll, the rate of photorespiration increases somewhat (Flexas et al. 1999a, 1999b; de Souza et al. 2005b). Under these conditions there is no increase in photoprotection mechanisms or heat energy dissipation (Flexas et al. 2002a; Medrano et al. 2002a) and there is no effect on parameters reflecting photosynthetic capacity, such as F_v/F_m (Flexas et al. 1998; de Souza et al. 2003, 2005b), ETR (Düring 1998; de Souza et al. 2003, 2005b) or the apparent maximum capacity of carboxylation (V_{c,max_Ci}) or for electron transport (J_{max_Ci}) both derived from A_N-C_i curves (de Souza et al. 2003). There is also no inhibition of photosynthetic enzymes such as Ribulose-1,5-biphosphate carboxylase/oxygenase (Rubisco) (Bota et al. 2004b), glyceraldehydes-3-phosphate dehydrogenase (G3PDH), ribulose-5-phosphatase kinase (Ru5PK) or fructose-1,6-biphosphate phosphatase (FruBPase) (de Souza et al. 2005b). Therefore, at this stage, diffusional limitations are the only cause for decreased photosynthesis. These are mostly due to stomatal closure, although if the stage is prolonged, mesophyll conductance to CO₂ also starts declining, contributing to restricted CO₂ diffusion (Flexas et al. 2002a; Moutinho-Pereira et al. 2004), particularly under conditions of high irradiance and temperature (Flexas et al. 2009).

Stage 2 represents a transition phase between predominant stomatal to non-stomatal limitations, and occurs invariably when g_s ranges between 50 and 150 mmol H₂O m^–2 s^–1. During this phase a further reduction in A_N occurs and WUE usually increases to reach maximum levels, and C_i decreases to minimum levels (Flexas et al. 2002a; Maroco et al. 2002). This indicates that stomatal limitations still dominate, although in some cultivars WUE starts decreasing at this stage, indicating predominant non-stomatal limitations (e.g. Naor et al. 1994). For instance, V_{c,max_Ci} decreases sometimes, suggesting impairment of Rubisco, which has been shown to decrease its activity (along with activities of G3PDH, Ru5PK and FruBPase) at this stage in some experiments (Maroco et al. 2002). In other experiments, however, Rubisco activity is unaffected (Bota et al. 2004b) and decreases of V_{c,max_Ci} are due to decreased g_m, since when it is calculated on a chloroplast CO₂ (C_c) rather than a C_i basis (V_{c,max_Cc}) it remains unaffected by stress (Flexas et al. 2002a). F_v/F_m remains unaffected (Flexas et al. 1998). There is also no net feedback inhibition of photosynthesis by sugar accumulation (Maroco et al. 2002). Instead, ETR characteristically declines during this phase (Flexas et al. 1999a, 1999b, 2002b), concomitant with increased NPQ, decreased PRI and decreased steady-state chlorophyll fluorescence (F_s) under high light (Flexas et al. 2000; Medrano et al. 2002a; Evain et al. 2004). These changes do not represent permanent damage to photosystems, but rather, they are dynamic and reverse immediately upon shading, e.g. during the passage of a cloud (Evain et al. 2004). As a consequence of both diffusional (i.e. reduced CO₂ availability) and biochemical (i.e. impaired Rubisco, decreased ETR) limitations, the rate of photorespiration often declines at this stage back to values similar to those of non-stressed plants (Guan et al. 2004).

Stage 3, when g_s drops below 50 mmol H₂O m^–2 s^–1, results in more generalised and dominant non-stomatal limitations to photosynthesis, particularly under conditions where water stress is accompanied by very high temperature and irradiance. This is demonstrated by the fact that photosynthesis cannot be restored even using very high CO₂ concentrations (Quick et al. 1992; Flexas et al. 1999b), and there is a drought-induced depression of gross oxygen evolution as seen using O₂ isotope discrimination in a mass spectrometer (Flexas et al. 1999b). At this stage, steeper reductions of A_N, ETR and V_{c,max_Ci} occur and NPQ further increases. During this phase A_N/g_s often decreases and C_i increases steeply (Düring 1987; Flexas et al. 2002a; Sivilotti et al. 2005), indicating that non-stomatal limitations to photosynthesis become dominant. Nevertheless, this is not always observed (see e.g. Fig. 2d). The rate of photorespiration is decreased, but the ratio of photorespiration to photosynthesis still increases, maintaining ETR relatively high with respect to A_N (Flexas et al. 1999a, 1999b; de Souza et al. 2005b). Under these conditions, changes in the expression of genes and proteins associated with stomatal functioning and photosynthesis occur (Cramer et al. 2007; Vincent et al. 2007).

Even under these conditions, F_v/F_m is extremely resistant to water stress (Gamon and Pearcy 1990; Flexas et al. 1998). Bertamini et al. (2007) showed that severe water stress does not affect the concentration of the D1 protein and reduces only slightly the electron transport capacities of different components of the electron transport chain. Also, decreases of V_{c,max_Ci} are sometimes due only to decreased g_m (Flexas et al. 2002a) and to errors in the calculation of C_i, as demonstrated by constant V_{c,max_Cc}. The errors in the calculation of C_i are due to the interference of heterogeneous stomatal closure, which has been shown to occur in grapevine leaves fed with ABA (Downton et al. 1988a) or subject to water stress (Downton et al. 1988b) or salinity (Downton et al. 1990). These errors become important (>10% error in the calculation of C_i) when average g_s drops below 30 mmol H₂O m^–2 s^–1 (Flexas et al. 2002a, 2009).

In many other conditions, however, Rubisco activity is truly impaired (Bota et al. 2004b), mostly due to decreases in its concentration (Bota et al. 2004b; Bertamini et al. 2006) and, to a lesser extent, to decreased activation state (Bota et al. 2004b). Moreover, in some cultivars there is a large decrease in the capacity for photoassimilate export out of the leaves (Bota et al. 2004a) resulting in accumulation of soluble sugars in leaves, which can induce feedback inhibition of photosynthesis. In some other cultivars and/or conditions, however, the decrease in the capacity for photoassimilate export is minor and sugars do not accumulate as a consequence of severe water stress (Quick et al. 1992).

The carbon balance of a plant during a period of water stress and recovery may depend as much on the velocity and degree of photosynthetic recovery as on the degree and velocity of photosynthesis decline during water depletion (Flexas et al. 2006). Relatively few studies have addressed the rate and limiting factors for the recovery of grapevine leaves after water stress. Still, there are some indications suggesting that previous water stress intensity is a crucial factor affecting both the velocity and the extent of recovery after re-watering. For instance, Tempranillo grapevines subjected to ‘Stage 2’ water stress (i.e. maximum stomatal conductance among 0.1–0.15 mol H₂O m^–2 s^–1) recovered completely overnight after re-watering, but ‘Stage 3’ plants of the same cultivar recovered only slowly during the next week, and did not reach the maximum photosynthesis rates reached before water stress (Flexas et al. 2009).

In addition to stress severity, differences associated with both cultivar and environmental conditions may affect recovery. For instance, in cultivars Airén and Chardonnay, g_s and A_N recovered completely but slowly (3–5 days to complete recovery) after a ‘Stage 2’ water stress. Plants of the rootstock Richter-110 recovered slowly (2 weeks for complete recovery) even after a ‘Stage 1’ water stress (Pou et al. 2008; Flexas et al. 2009). In contrast, several cultivars and rootstocks including Cabernet Sauvignon showed almost complete recovery 2 days after rewatering of ‘Stage 3’ water stressed plants (Guan et al. 2004), similar to ‘Stage 3’ V. labruscana (Liu et al. 1978). Furthermore, g_m, Rubisco and ETR recover quickly (1–3 days) after re-watering, although g_s remains lower for longer, becoming the most limiting factor for photosynthesis recovery (Flexas et al. 2009).

The reasons for sustained low g_s are unclear. Both in cases of rapid recovery of photosynthesis and those in which photosynthesis lasts few days after re-watering, leaf water relations recover fast after re-watering, being totally reversed the day after (Lovisolo et al. 2008a; Pou et al. 2008). In contrast, free cis-trans ABA also recovers control values quickly (1–3 days after re-watering), although hydrolisable cis-trans ABA and phaseic acid remain high for longer (Liu et al. 1978; Pou et al. 2008). Hydraulic conductivity recovers quickly (1 day) in some experiments (Lovisolo and Schubert 2006) but slowly in others (Pou et al. 2008). Lovisolo et al. (2008a) proposed that ABA-induced g_s decrease allows embolism repair during the day after re-watering, but this mechanism may not work in cases where ABA is fully reversed in few days while conductivity and g_s are not (Pou et al. 2008). Clearly more studies are required to understand the dynamics and mechanisms involved in recovery of grapevine cultivars after water stress.

Following historical linkages between research on grapevine physiology and viticulture techniques, there is a wide and heterogeneous literature related to drought effects on grape berry development, as reviewed by Ollat et al. (2002) and Keller (2005). Knowledge of the phenomenon focuses on (i) effects of an impaired plant metabolism (especially photosynthesis and transpiration) on the accumulation of sugars and secondary metabolites in berry, (ii) consequences at the berry level of both the chemically-mediated long distance signalling between root and shoot (essentially ABA and cytokinin) and the whole-plant hydraulic control via both the xylem and the phloem from root to berry and (iii) berry metabolism adaptations to severe osmotic stress in berry cells.

Water influx into fruits occurs through xylem and phloem and during ripening water flow via the xylem markedly decreases from the onset of véraison when the main source of water to berries becomes the phloem sap. Evidence for ceased xylem functionality in berries has been reported from studies of apoplastic dye perfusion through the pedicel (Greenspan et al. 1996; Rogiers et al. 2001; Bondada et al. 2005; Keller et al. 2006; Tilbrook and Tyerman 2009) even though some studies have been doubtful about the validity of the interpretation of dye uptake experiments (Tyerman et al. 2004). Nevertheless, the hydraulic isolation of the berry due to development of a xylem discontinuity in the pedicel or inside the berry has been considered a way to prevent the loss of solutes in berries (Sarry et al. 2004). Keller et al. (2006) have demonstrated that berry xylem functionality is retained during ripening; the detected decline in xylem water influx into ripening grape berries is due to the apoplastic phloem unloading coupled with solute accumulation in the berry apoplast. More recently, Tilbrook and Tyerman (2009) showed that, according to varietal differences, the activity of semi permeable membranes in mesocarp cells is the key controller of flow in to and out of the berry, dictated by xylem flow. As drought lowers leaf water potential by concentrating osmolites in leaf, drought effects drive water flows between leaf and berry.

Drought effects on accumulation of secondary metabolites in berry were thoroughly investigated since the 1980s. Several studies have measured the effects of water stress on polyphenols (Kennedy et al. 2002; Ojeda et al. 2002; Koundouras et al. 2006; Pedreira dos Santos et al. 2007; Poni et al. 2007) and flavors (Oliveira et al. 2003; Koundouras et al. 2006; Bindon et al. 2007; Pedreira dos Santos et al. 2007). Matthews and Anderson (1988) stated that in the anisohydric Cabernet Franc berry, skin polyphenol and anthocyanin concentrations increased as a consequence of berry volume reduction due to water stress, even though, also expressing data on the basis of surface area, an increase in polyphenol concentration was detected. Roby et al. (2004) found that the concentration of polyphenols such as anthocyanins and proanthocyanidins of berry skins increased after water stress conditions independently from the differences in berry size due to water availability. There was still uncertainty in whether the increases in skin anthocyanins and proanthocyanidins resulted from a differential growth of exocarp cells compared with pulp cells rather than direct effects on phenolic biosynthesis. Recent researches on anisohydric Cabernet Sauvignon (Castellarin et al. 2007a) showed that the expression of some enzymes of the phenylpropoanoid biosynthetic pathway increased as a consequence of early and late water stress application, and also, that this increase was more effective on the biosynthesis of tri-hydroxylated anthocyanins. Early applied water stress was also proved to accelerate the accumulation of sugar and the onset of ripening.

ABA production in drying roots and its translocation and accumulation in the shoot drive several physiological mechanisms in grapevine leaves (Lovisolo et al. 2002a, 2008a), as reviewed above. In grape berries, ABA is considered a promoter of ripening as its concentration increases at the beginning of véraison (Gagné et al. 2006; Wheeler et al. 2009). Exogenous ABA, used to promote ripening in berries (Peppi et al. 2006; Cantín et al. 2007; Peppi and Fidelibus 2008) induces the accumulation of the mRNA of VvmybA1, a regulatory gene of anthocyanin biosynthesis of grape, leading to an increase of anthocyanin accumulation in Cabernet Sauvignon berry skins (Jeong et al. 2004). ABA also activates invertases (Pan et al. 2005) and VvHT1 (V. vinifera hexose transporter 1) (Çakir et al. 2003). The effect of ABA on UFGT (UDPglucose:flavonoid 3-O-glucosiltransferase), a key-enzyme of the flavonoid biosynthesis, has recently been proved, as well as its influence on the quality of color, the skins of ABA treated Crimson Seedless berries having lower lightness and hue than ABA-untreated control plants (Peppi et al. 2008).

ABA application has little or no effect on berry sugar content at harvest (Jeong et al. 2004; Peppi et al. 2006). The accumulation of sugars in berries requires the coordinated expression of sucrose transporters, invertases, and monosaccharide transporters. The expression of the glucose transporter homologue (VvHT1, V. vinifera hexose transporter 1), isolated from grape berries at véraison, is regulated by sugars and ABA (Vignault et al. 2005). Phloem influx into the berry is accompanied by a decrease in cell turgor (Thomas et al. 2006), which influences the expression of many genes at the onset of ripening (Deluc et al. 2006). Water stress, lowering cell turgor, particularly if applied pre-véraison (Thomas et al. 2006), may induce an increase in sugar influx and ABA, influencing several key steps of the phenylpropanoid biosynthetic pathway. In particular, the expression of genes F3H, F3′5’H, LDOX and DFR involved in the biosynthesis of anthocyanins, proanthocyanidins and flavonols, increases in water-deficit conditions (Castellarin et al. 2007a, 2007b). The increase in total anthocyanins results predominantly from an increase of the F3′5’H expression, responsible for the biosynthesis of tri-hydroxylated anthocyanins. During ripening, the cumulative expression of genes strictly associated to the anthocyanin accumulation UFGT and GST (the latter probably playing a role in the transport of anthocyanins into the vacuole (Ageorges et al. 2006)), is strongly upregulated under water stress conditions (Castellarin et al. 2007a).

The ripening process per se is not accelerated by water deficit in Cabernet Sauvignon or Merlot, since no changes in timing of berry growth and sugar accumulation occur after water deficit (Castellarin et al. 2007a). Furthermore, no earlier downregulation of the genes responsible for proanthocyanidin biosynthesis, which is expected when an early ripening occurs, is detected when water deficit is applied (Castellarin et al. 2007b). The effects of early or late water stress on proanthocyanidin and flavonol concentration and on the expression of genes responsible for their synthesis are still unclear.

Myb transcription factors MybA and Myb5a are higher in water-stressed vines respect to control vines in Merlot, as well as MybC which is upregulated from post-véraison to harvest; MybB and MybD are not influenced by water availability (Castellarin et al. 2007b). In conditions where the accumulation of ABA is higher, the m-RNAs of VvmybA1 are also higher (Yamane et al. 2006); these results suggest that the endogenous ABA level affects the expression of VvmybA1 that controls the expression of the anthocyanin biosynthetic enzyme genes. In Merlot berry skins, the genes NCED1 and NCED2 are only transiently upregulated (Castellarin et al. 2007b) at the onset of véraison after water stress. Moreover, ACPK1 and rd22, both correlated with the ABA metabolism, the former in grape berries (Yu et al. 2006), the latter in other species (Iwasaki et al. 1995), are not at all or only partially upregulated in the experiments reported by Castellarin et al. (2007a, 2007b).

The root and shoot ABA-mediated responses to water stress conditions, or, more generally, to abiotic stresses, are relevant to vine productivity and yield. As described earlier, water stress influences ABA accumulation at the root, shoot and leaf level, and also affects berry quality (Keller 2005). However, a connection between ABA and berry quality has not yet been clarified.

Cytokinin signalling downregulation in ripening berries may represent another point of cross-talk between water stress responses and developmentally-driven ripening processes. Cytokinin concentration sharply decreases at the onset of ripening (Alleweldt et al. 1975), and a zeatin O-glucosyltransferase (Zhang et al. 2008), as well as a gene encoding a cytokinin synthase (Carra et al. 2009), are downregulated in mature berries. Since the repression of cytokinin metabolism is a known response to water deficit in grapevine shoots (Stoll et al. 2000), the relative abundance of ABA and cytokinin may be regulated in the berries by water stress-like signals, and may be important for the coordination of the ripening process. However, direct evidence of drought-induced repression of cytokinin metabolism in berries is still lacking.

High sugar accumulation in ripening grapes generates a severe osmotic stress in berry cells and is accompanied by the expression of thaumatin-like proteins similar to osmotins (Tattersall et al. 1997; Salzman et al. 1998; Davies and Robinson 2000). Osmotins are induced by ABA, thus, it could be that ABA and high sugar concentration may interact to trigger in the berry a defence response, which may also be triggered by water stress. Supporting this idea, transcripts of genes associated with responses to pathogens were shown to be significantly upregulated in the skin of water-stressed plants compared with well watered controls (Grimplet et al. 2007). ABA and sugar signals converge in the induction of VvMSA, an ASR (ABA-, stress- and ripening-induced) protein that acts as a transcriptional activator of the hexose transporter VvHT1 (Çakir et al. 2003). ASR proteins are also induced by drought and salt stress, further supporting the idea that processes characterising berry development and ripening overlap with water stress responses.

Mature grape berries accumulate high amounts of free proline, a well known compatible osmolyte that accumulates in tissues of water stressed plants. Although accumulation of free proline is likely to exert a protective action on grape cellular processes, the molecular determinants of proline homeostasis in berries are largely unknown. In most plant species the reciprocal rates of proline biosynthesis and catabolism are controlled by transcriptional regulation of the rate limiting biosynthetic enzyme Δ¹-pyrroline-5-carboxylate synthetase (P5CS) and the catabolic enzyme proline dehydrogenase (PDH) (Verbruggen and Hermans 2008). In grape berries, however, the expression of P5CS and PDH remains relatively unchanged along ripening (Stines et al. 1999; Deluc et al. 2006), despite the rise in concentration of ABA, which is required for the expression of P5CS genes in Arabidopsis (Strizhov et al. 1997). By contrast, in water stressed grapevine, shoot and berry increase in free proline is accompanied by upregulation of both P5CS and PDH (Cramer et al. 2007; Deluc et al. 2009), as well by upregulation of the key ABA biosynthetic gene NCED and of several ABA-responsive genes. Stines et al. (1999) cloned a single P5CS gene from grapevine, and suggested that its protein product may be regulated post-translationally. However, the publication of the grapevine genome sequence draft evidenced that P5CS may be represented by a family of two or three genes in grapevine (Jaillón et al. 2007). Of two family members included in the Arabidopsis genome, P5CS1 is induced by osmotic stress while P5CS2 is developmentally regulated and is required for embryo vitality (Székely et al. 2008). Thus, a yet uncharacterised member of the P5CS family may be responsible for developmentally regulated proline accumulation in ripening grape berries. Alternatively, proline might be translocated to the berry through the phloem.

Knowledge about chemical and hydraulic root signals induced by soil water stress has stimulated new irrigation strategies. Partial root-zone drying (PRD) was developed to improve yield-to-irrigation ratios. PRD is designed to expose part of the root system to drying soil in order to produce the root drought signal, while the remaining roots in wet soil can maintain water supply and, therefore, leaf water potential (Dry and Loveys 1999). PRD enhances root hydraulic conductance in fruit trees; during PRD treatment, roots show higher uptake capacity than in whole root-zone irrigation treatment (Kang et al. 2002). Putative aquaporin stimulation by ABA produced by PRD may be involved. Prolonged exposure of roots to drying soil may cause anatomical changes as epidermis suberisation, collapse of cortex and loss of secondary roots. Alternate watering, after a period of soil drying, may improve this situation by inducing new secondary roots (Kang and Zhang 2004). In the field during a typical drying cycle (10–15 days), only roots near to soil surface feel dry soil whereas deeper roots extract water from wetter soil layers. Maybe this reduces the synthesis of drought chemical signals, like ABA, and, hence, probably reduces PRD effect. Also, the water redistribution process from wet to dry roots (the so-called ‘hydraulic lift’) in response to water potential gradients can contribute to decrease of ABA biosynthesis (De la Hera et al. 2007). This phenomenon has been observed in several grapevines subjected to dry soil conditions or PRD treatment (Smart et al. 2005; Bauerle et al. 2008).

In order to improve PRD technology, further studies may be focused on intra-specific variation in the mechanisms of control of transpiration and their relative sensitivity to soil water deficit, i.e. isohydric and anisohydric behaviour of different grapevine cultivars (Schultz 2003a; Sadras 2009). Genotypically different responses to water stress, such as stomatal sensitivity to non-hydraulic signalling or the ability of the xylem to supply ABA (as described for Grenache and Chardonnay, see above) or rootstock vigour, could be very important factors in determining intensity of PRD response (Antolín et al. 2006; De la Hera et al. 2007). For example, de Souza et al. (2005a) suggest that PRD appears more successful with the more drought-responsive wine grape varieties.

Besides PRD, other techniques aiming at controlling the water balance in vineyards are being adopted in viticultural practice, as discussed by Keller (2005) and Chaves et al. (2007).

We have addressed grapevine responses to water stress by examining perturbations to physiological and molecular processes at the root, shoot, leaf and berry levels (Fig. 3).

Vitis genotypes have been described in relation to their isohydric or anisohydric response to water stress, linked to stomatal behaviour and non-stomatal effects. Stomatal regulation of grapevine is under ABA and hydraulic control; the latter linked to embolism formation and recovery in xylem tissues upstream the stomata. We have focused on ABA effects on stomata and their interrelationship with plant hydraulics from the root towards leaves and with photosynthetic assimilation. Using the daily maximum value of stomatal conductance as an indicator of water stress that allows comparison of plants with iso- and anisohydric behaviours, we have defined three stages of photosynthesis regulation in grapevines subjected to progressive water stress on the basis of the main causes of assimilation decline.

We have shown that in grapevine, xylem embolism occurs and repairs during diurnal cycles under ABA control, and that an almost full recovery of water potential is needed to promote repair mechanisms.

Aquaporins play a fundamental role in the control of plant water status. Different drought-defence strategies between iso- and anisohydric cultivars have been highlighted on the basis of drought-induced root ABA biosynthesis: an increase of suberisation of apoplastic barriers causes a reduction of water conductivity, but this is not compensated by an enhanced aquaporin-mediated cell-to-cell water transport.

Reverse genetics study on key genes of molecular pathways could provide a better understanding of drought tolerance mechanisms in grapevine. The Vitis genus is not usual for genetic transformation (Vivier and Pretorius 2002). However, progress has been recently made in grapevine transformation (Bouquet et al. 2006; Carmona et al. 2008; Burger et al. 2009). Today, a transgenic approach to grapevine improvement is more attractive than a classical breeding approach because a transfer of individual traits as single genes with a minimum disruption of the original genome would leave the traditional characteristics of the cultivar intact (Bouquet et al. 2006).

All aspects reviewed in this paper, when taken together, show that grapevine fits well as a complex, modern model plant for molecular and physiological studies on both plant drought avoidance and tolerance. In particular, knowledge of drought effects on (i) differential development of root and shoot, (ii) long-distance water transport and hydraulic (involving xylem embolism) and hormone signalling, (iii) adaptations of the phosynthetic machinery and of stomatal behaviour and (iv) fruit ripening and technological quality for wine making, highlights the fundamental role that grapevine can play as a model crop. All these aspects are spread on different genotypes either used as rootstocks or as scions, which enlarge the genetic background facing peculiarities known in viticultural practice.