Photosynthetic carbon assimilation and associated metabolism in relation to water deficits in higher plants


D. W. Lawlor. Fax: + 44 (0)1582 76 3010, E-mail:


Experimental studies on CO2 assimilation of mesophytic C3 plants in relation to relative water content (RWC) are discussed. Decreasing RWC slows the actual rate of photosynthetic CO2 assimilation (A) and decreases the potential rate (Apot). Generally, as RWC falls from c. 100 to c. 75%, the stomatal conductance (gs) decreases, and with it A. However, there are two general types of relation of Apot to RWC, which are called Type 1 and Type 2. Type 1 has two main phases. As RWC decreases from 100 to c. 75%, Apot is unaffected, but decreasing stomatal conductance (gs) results in smaller A, and lower CO2 concentration inside the leaf (Ci) and in the chloroplast (Cc), the latter falling possibly to the compensation point. Down-regulation of electron transport occurs by energy quenching mechanisms, and changes in carbohydrate and nitrogen metabolism are considered acclimatory, caused by low Ci and reversible by elevated CO2. Below 75% RWC, there is metabolic inhibition of Apot, inhibition of A then being partly (but progressively less) reversible by elevated CO2; gs regulates A progressively less, and Ci and CO2 compensation point, Γ rise. It is suggested that this is the true stress phase, where the decrease in Apot is caused by decreased ATP synthesis and a consequent decreased synthesis of RuBP. In the Type 2 response, Apot decreases progressively at RWC 100 to 75%, with A being progressively less restored to the unstressed value by elevated CO2. Decreased gs leads to a lower Ci and Cc but they probably do not reach compensation point: gs becomes progressively less important and metabolic limitations more important as RWC falls. The primary effect of low RWC on Apot is most probably caused by limited RuBP synthesis, as a result of decreased ATP synthesis, either through inhibition of Coupling Factor activity or amount due to increased ion concentration. Carbohydrate synthesis and accumulation decrease. Type 2 response is considered equivalent to Type 1 at RWC below c. 75%, with Apot inhibited by limited ATP and RuBP synthesis, respiratory metabolism dominates and Ci and Γ rise. The importance of inhibited ATP synthesis as a primary cause of decreasing Apot is discussed. Factors determining the Type 1 and Type 2 responses are unknown. Electron transport is maintained (but down-regulated) in Types 1 and 2 over a wide range of RWC, and a large reduced/oxidized adenylate ratio results. Metabolic imbalance results in amino acid accumulation and decreased and altered protein synthesis. These conditions profoundly affect cell functions and ultimately cause cell death. Type 1 and 2 responses may reflect differences in gs and in sensitivity of metabolism to decreasing RWC.


This review considers how decreasing cellular water content (measured as relative water content, RWC%= fresh mass – dry mass/water saturated mass – dry mass × 100)) of leaves of mesophytic higher plants affects their photosynthetic physiology, particularly carbon assimilation and energy use. As the topic has been reviewed many times (Chaves 1991; Lawlor & Uprety 1991; Cornic 1994; Lawlor 1995; Cornic & Massacci 1996), the approach is selective in its consideration of the literature. We assess the current state of knowledge and views about stomatal versus metabolic determination of A, where there are conflicting concepts (Lawlor 1995; Tezara et al. 1999; Cornic & Massacci 1996; Cornic 2000). Response to water deficiency is a syndrome, a group of concurrent symptoms, with as yet unclear causes and mechanisms, involving complex interaction of physical and metabolic processes in cells, tissues and organs (Kramer & Boyer 1995). The aim is to clarify and explain the phenomena and by so doing to aid future developments, both scientific and applied. Clarification of changes in photosynthesis as a consequence of water deficiency is essential, and of practical importance; for example, production of plants for human food depends greatly on water supply. Adaptation of plants to use less water, increase water use efficiency and to be more tolerant and resistant to water deficits by selection breeding has had limited success (Evans 1998); genetic modification is regarded as a way of achieving these aims, but currently has very unspecific targets.

Plant water balance and RWC

Plant and cell water balance is determined by water lost in evaporation (transpiration) to the atmosphere and water absorption from the soil. When transpiration exceeds absorption, cell turgor (P) falls as RWC and cell volume decrease, whilst the concentration of cellular contents increases, so osmotic potential (π) and water potential (ψ) fall. Osmotic adjustment may modify the relationships but we do not consider this. Low P and RWC slow growth and decrease the stomatal conductance for H2O (gs). A satisfactory basis for relating cellular water status to metabolism is RWC, an easily measured, robust indicator of water status for comparison of tissues and species, which ‘normalizes’ water content by expressing it relative to the fully turgid (hydrated) state. RWC is a measure of relative change in cell volume; ψ is the resultant of P and π, and thus depends both on solutes concentration and cell wall rigidity and does not relate directly to cell volume (Kaiser 1987; Kramer & Boyer 1995; Lawlor 1995). It is desirable to know the relation between π, ψ, P and RWC for assessing changes induced by water deficits, because the relative magnitude of components of cellular water balance is species-specific, depending on acclimatory processes (e.g. osmotic adjustment), duration of the deficit and on growth conditions. There is no unique relationship, and experiments must be interpreted with this in mind. Water deficiency covers the range from fully hydrated cells (100% RWC and maximum π, P and ψ), which we assume provides the control state with metabolism functioning at the potential rate (with other conditions optimal), to very dehydrated cells (50% RWC or less, zero P and very small π and ψ) at which the cell will not recover when rehydrated. Mention of a particular RWC should be treated as a generalization, owing to the factors mentioned above, and is done to aid the analysis; it is perhaps more relevant to take the lower limit of RWC as that at which irreversible changes in metabolism occur, but this is time dependent and is not an easily defined value. In mesophytes experiencing water deficits, many changes in physiology and metabolism occur, related to regulation (adjustment) of metabolism to conditions within the cell and tissue. When regulation is inadequate and metabolism does not adjust to maintain functions, damage and eventually death ensue. Regulation is very dependent upon the species and on environment, and functions within particular limits, involving changes in biochemistry (e.g. synthesis and accumulation of osmotically active solutes (Delauney & Verma 1993), protein composition (e.g. amounts and proportions of enzymes; Riccardi et al. 1998; Deleu et al. 1999) and alterations to cell, organ and whole plant structure (Kramer & Boyer 1995). Comprehensive understanding of the interactions between these processes must await more detailed information from studies addressing the problem in an integrated way. The need to understand the interactions between water and plant functions is inescapable, both for breeding and, particularly, genetic engineering, if improvements in ‘drought tolerance and resistance’ and production under water-limiting conditions are to be achieved. Genetic engineering requires clearly identified targets but many genetically determined processes are involved and there is uncertainty as to the most important: there is no ‘gene for drought resistance’!

Causes of decreased photosynthesis under water deficiency

Progressive decrease in RWC decreases A of leaves, eventually inhibiting it, and there is net CO2 release from respiration: A depends on the activity of Rubisco per unit leaf, the rate of RuBP synthesis (hence on capture of photosynthetically active radiation (PAR)) and on the CO2 supply, determined by gs and the ambient CO2 concentration (Ca) (Lawlor, 2001). We first consider the response of leaves of decreasing RWC to CO2 supply.

The A/Ci response curve

CO2 supply to the PCR cycle in the chloroplast is determined by Ca and conductances of the pathway for diffusion between air and enzyme active sites, principally gs in the gas phase and gm in the liquid phase, which includes all physico-chemical and biochemical factors (see von Caemmerer 2000). The CO2 concentration within the leaf, Ci, depends on A, gs and Ca:

inline image

The CO2 concentration at the active sites of Rubisco in the chloroplasts, Cc, is given by (von Caemmerer & Evans 1991):

inline image

By measuring A as a function of Ci (or, better, Cc) under standard PAR flux (here considered as providing RuBP sufficient to allow the maximum steady-state rate of the PCR cycle) the limitations to A may be assessed. We define the maximum rate of A under saturating CO2 (Ca, Ci and Cc) and light in fully hydrated leaves as Apot. To achieve the same Apot at small RWC as at large RWC, Cc must saturate Rubisco and so Ca must be sufficient to overcome the limitation of gs, etc. If Apot at small RWC does not attain the value of Apot at large RWC, despite CO2 saturation, then metabolism is inhibited. For C3 leaves in bright light in the current atmosphere with Ca 360 µmol mol−1 CO2 and 0·21 mol mol−1 O2, and A of 30 µmol m−2 s−1, Ci is c. 250 µmol mol−1 with gs = 0·3 mol m−2 s−1, i.e. Ci/Ca is c. 0·7. A is not saturated at this Ci and increasing Ca to 800 µmol mol−1 increases A to Apot if gm is not changed when Ca increases. At smaller gs (as with low RWC), Ci decreases: for gs of 0·1 mol m−2 s−1 and Apot of 30 µmol m−2 s−1, Ci would decrease to 60 µmol mol−1, which is the compensation point for C3 leaves. At gs of 0·1 or 0·05 mol m−2 s−1, a Ca of 1000 or 1400 µmol mol−1 CO2 would maintain Apot of 40 µmol m−2 s−1. Thus, in theory, a five-fold increase in Ca should restore A in leaves with low gs due to small RWC. Of course, if gm decreases as a consequence of stress, then a greater Ca would be needed.

Measured A/Ci response curves of leaves over a range of RWC show the effect on carboxylation efficiency and Apot, independent of gs. In C3 leaves, such as those of wheat (Lawlor & Khanna-Chopra 1984) and sunflower (Giménez, Mitchell & Lawlor 1992; Pankovíc et al. 1999; Tezara et al. 1999), at 100 to 90% RWC under saturating light, Apot is achieved at Ci of some 400 µmol mol−1, but with decreasing RWC Apot decreases and in extremis, CO2 is evolved. This progressive decrease in Apot is not reversible, in the studies cited above, by application of large Ca (up to 2500 µmol mol−1 CO2 in open-gas exchange systems). There is also no indication of slowly increasing A as Ci increased in leaves at RWC of c. 75%) as expected if Ca was still not sufficiently concentrated to achieve full Apot. Analyses using 5–10% CO2 also failed to return Apot to the control value (Tezara et al. 1999). The decrease in Apot is primary evidence for metabolic inhibition of A (Lawlor 1995).

Calculation of Ci is based on gs calculated from measured fluxes of H2O and temperature at the evaporating surfaces, and the ratio of the diffusivities of the CO2 and H2O in air. With the difficulties in measuring very small fluxes and differences in concentration of H2O at small gs, plus inaccuracy in measuring leaf temperature, calculation of gs may be inaccurate and thus Ci erroneous (Renou et al. 1990; Cornic 2000). However, careful calibration of measurement systems and comparisons between different methods decreases such errors. Heterogeneous distribution of stomatal apertures and gs over the leaf surface (stomatal ‘patchiness’) may also invalidate calculated Ci (Terashima 1992). Evidence for patchiness and how it affects calculation of gs has been assessed and reviewed (Weyers & Lawson 1997; Pospí silová & Santrúcek 1997; Mott & Buckley 2000). For leaves stressed relatively slowly on the plant, and for longer periods of gas exchange, patchiness has been observed in some studies (Stuhlfauth et al. 1990) but not others (Giménez et al. 1992; Gunasekera & Berkowitz 1992). Poor correlation of heterogeneity of fluorescence and stomatal aperture in water-stressed Potentilla leaves (Osmond et al. 1999) suggests that patchiness is not a universal effect of significance to CO2 assimilation and that there is a metabolic limitation (Martin & Ruiz-Torres 1992). Patchiness should also be overcome by very large Ca, yet elevated CO2 did not restore A to Apot in some studies (Escalona et al. 1999; Pankovièet al. 1999; Tezara et al. 1999) and Osmond et al. (1999) were unable to eliminate fluorescence heterogeneity by elevated Ca. Such evidence shows that decreased A was a metabolic effect of low RWC and not due to patchiness, which is probably much less important than once thought.

In some studies (see Pankovi ´c et al. 1999; Tezara et al. 1999) A was not restored to Apot by elevated CO2 (within the range 100% to c. 75% RWC), but it was in others, even in Helianthus annuus (Cornic & Massacci 1996), which generally is unresponsive. Also, O2 evolution in the oxygen electrode was restored to the maximum rate (Quick et al. 1992). This is a fundamental difference between studies. However, a common feature in the analyses is the inability of large Ca to restore A to Apot of the well-watered plants at large RWC, either as RWC decreased progressively or as it fell below a threshold RWC, showing that small or heterogeneous gs is not the cause of decreased Apot.

When stomata are closed, transport of CO2 and H2O is predominantly through the cuticle, not in the gas phase. Transport characteristics of the two gases in the cuticle differ from those of air, so invalidating standard methods of calculating Ci under such conditions (Boyer et al. 1997). Conductance of grape leaf cuticle to H2O was much smaller than that of CO2 (Boyer et al. 1997) so the effect on calculation of Ci is potentially large. The importance of this potential error was assessed at small RWC at high Ca with A about 15% of the maximum. The Ci was underestimated by 10% and did not change the interpretation of the A/Ci relationship greatly. However, the effect on the slope of the relationship was much larger, about 40%, so the validity of the carboxylation efficiency is in question. Also, a decrease (c. 50%) in Apot was observed at high Ca in sunflower, when stomatal conductance was not minimal (Tezara et al. 1999), when the flux of CO2 is predominantly stomatal and not cuticular. Thus, the differential conductance of H2O and CO2 does not explain decreased Apot, a conclusion also reached by Flexas & Medrano (2002). As with stomatal patchiness, very large Ca should overcome low (cuticular) conductance, but even 10% CO2, which inhibited O2 evolution in leaves with RWC < 75%, did not increase A. Further evaluation is required. Supply of CO2 to Rubisco could be limiting because of physical alterations in the structure of the intercellular spaces due to leaf shrinkage, and by changes in any biochemical reactions (e.g. impaired transport of CO2 or conversion of bicarbonate to CO2) at low RWC, but no mechanisms have been identified. Physical limitations should be overcome by large Ca, and the latter would be a metabolic limitation. More detailed analysis of these limitations is required. In conclusion, the problems of measuring A, Ci and Cc in water-stressed leaves are potentially important and require further analysis, but in practice several lines of evidence, including A/Ci responses, suggest that there is either a progressive decrease in Apot with decreasing RWC or below a threshold value of RWC: the apparent differences between studies require explanation.

Relative magnitude of gs and gm

From A/Ci responses the relative magnitudes of the stomatal and metabolic limitations may be calculated (Cornic et al. 1983; Jones 1985; Assmann, 1988). For experiments where Apot (i.e. at saturating CO2) is unaffected by decreasing RWC from 100%, where gm is defined as maximum, to c. 75% there is no effect on gm, and all decrease in A is due to gs. With further decrease in RWC Apot decreases, so the gm limitation increases and that of gs decreases. Where Apot decreases progressively with loss of RWC, gm limitation also increases progressively from zero to almost 100% and gs decreases from about 25% to approaching zero. The response of grape leaves to drought in the field shows that their photosynthetic metabolism was impaired below c. 75% RWC (Escalona et al. 1999). Thus, generally decreasing RWC below c. 75% progressively impairs photosynthetic metabolism, but this does depend on the validity of the A/Ci responses.

Response of Apot to large Ca

The use of large Ca to overcome small gs and provide Rubisco with sufficient substrate to restore A to Apot has been a key point in water relations studies. Indeed, maintenance of O2 evolution at control rates in leaves in oxygen electrodes at 5% CO2 is primary evidence that Apot is unimpaired by low RWC or that the inhibition of Apot is sensitive to CO2 and can be relieved at high CO2. It should be noted that we define Apot as CO2 assimilation not O2 evolution; although they are very closely related under most conditions, they may not be identical under severe stress. Fundamental differences in the relative roles of gs limitation of CO2 supply and metabolic limitation of Apot have appeared in experiments when large Ca (5–15%) is applied to restore A to Apot. Basically the data fall into two groups, which we call Type 1 and Type 2 responses. They are shown in Fig. 1 and are characterized by the response to CO2.

Figure 1.

Schematic of the responses of A to RWC. In the Type 1 response (1) Apot is unaffected by decreased RWC from 100 to 75% then drops progressively to zero with further loss of RWC. A in air (360 µmol mol−1 CO2) falls with stress but is restored to Apot by elevated CO2. At smaller RWC A is not restored as metabolic effects limit Apot. In the Type 2 response (2) Apot decreases linearly with loss of RWC; some restoration of A occurs with increased Ca but the effect of loss of RWC is a shift from stomatal control to metabolic control. In Types 1 and 2, initial regulation of A is by stomatal closure, but the low RWC effect is due to metabolic changes, probably related to decreased ATP. The reason for the Type 1 and 2 responses and what determines the distinct transition between the stomatal and metabolic phases in 1 or the progressive response in 2 is unknown.

Type 1 response

With RWC = 90% to c. 75%, increasing Ca to 5% restores A fully to the Apot of control leaves (Kaiser 1987; Tourneux & Peltier 1995; Cornic & Massacci 1996). Some studies required up to 15% to achieve this, although in others, where C4 plants where used, 10% CO2 inhibits metabolism (Saccardy et al. 1996). This concentration also inhibited A in sunflower (Graan & Boyer 1990), at high as well as low RWC (Tezara et al. 1999). At RWC < 75% restoration of Apot to the value at 100% RWC is not achieved and the response to CO2 becomes progressively smaller. However, A in air (360 µmol mol−1) could be increased, showing that under these conditions gs still controlled A.

Type 2 response

Elevated CO2 increases A to Apot in unstressed leaves (RWC 100 to 75%), but A is progressively less stimulated as RWC decreases, i.e. A is not restored to the unstressed Apot, so the potential rate of CO2 assimilation is decreased (e.g. Pankovíc et al. 1999; Tezara et al. 1999). Failure to restore A at low RWC to control values by increasing Ca has been regarded as indicative of insufficiently large Ca, rather than of inhibition of Apot by low RWC. However, inhibition of A with 10% or greater Ca, rather than restoration, shows that metabolism is affected by elevated CO2, but as A is not restored to Apot, photosynthetic metabolism must be impaired.

The ability to restore Apot appears equivocal in some studies. Graan & Boyer (1990), for example, partially restored A of mildly to severely stressed sunflower leaves (RWC not given, ψ− 1·3 MPa) with 3000 Pa (30000 µL L−1 or 3%) CO2. (Note: µmol mol−1 and µL L−1 are used here interchangeably, or Pa is quoted, reflecting the use in the original studies). Possibly this partial restoration was caused by insufficient CO2 (indicated by extrapolation of Fig. 2a in Graan & Boyer 1990). However, O2 evolution was partially inhibited by increased water stress even with Ca of 10 000 Pa (10%) CO2; larger Ca substantially decreased O2 evolution at all stresses. The evidence is therefore of partial metabolic inhibition of A by moderate stress and substantial inhibition at more severe stress. In contrast, low gs caused by abscisic acid application was reversed by elevated CO2. Leaves with ψ of − 1·3 MPa may be in the transition between the stomatal phase of regulation (large RWC) and a phase (small RWC) where Apot is limited by metabolism. Other studies (e.g. Tezara et al. 1999) show only partial restoration of Apot in sunflower (Tezara et al. 1999). This may be a characteristic of sunflower (however, Cornic & Massacci (1996) found a Type 1 response), but it also occurred in wheat (D.W. Lawlor, unpublished data). Apot is certainly inhibited by metabolic factors at RWC < 75% in most studies; the discrepancy comes with some studies showing progressive loss of Apot and others showing no effect at RWC 100 to 75%. There is no information from tissue water relations or photosynthetic metabolism to explain this difference.

Effects of removing leaf epidermis

By stripping the epidermis from leaves, gs diffusion limitations are removed, permitting CO2 to diffuse into the intercellular spaces (although these may shrink with loss of RWC, restricting entry) so that much smaller Ca is required to saturate A. Although physical damage to the mesophyll may occur, in some species the epidermis is removed relatively easily. Schwab et al. (1989) and Dietz & Heber (1983) applied this technique to Ramonda mykoni and Primula palinuri, respectively, and showed most of the decline in A, induced by decreasing RWC from about 100 to 50%, was explained by stomatal closure. This is clear evidence of a Type 1 response. In other studies (Tang et al. 2002) with stripped sunflower leaves, A decreased progressively as RWC fell, a Type 2 response. Given the potential for damage, for possible drying of exposed cells surfaces, which may decrease the conductance to gases, and the need to maintain atmospheric humidity (with potential for artificial rehydration of tissues), such studies require assessment in the context of other methods of analysis.

Other estimates of Ci

Evidence for decreasing Ci in stressed leaves based on changes in fluorescence has been reviewed (Cornic & Massacci 1996). Essentially, the changes in fluorescence observed are identical to those produced by decreasing Ca: ergo the effect of low RWC and small gs on A is due to low Ci. Renou et al. (1990) calculated Ci from gas exchange and Cc from O2 isotope exchange, and showed a discrepancy: the former indicated high and constant Ci, whereas the latter showed very small Ci and Cc. Given the uncertainties with the gas exchange, low Cc was regarded as more realistic. Tourneux & Peltier (1995), using O2 isotope exchange measurements, showed that Ci reached compensation point, and Cc was 50 or 75 µL L−1 depending on whether day respiration was assumed to be zero or equal to the O2 uptake rate at large RWC, respectively. With the latter assumption, Cc was smallest at about 70% RWC and increased at smaller RWC. This is similar to gas exchange data measured under steady-state conditions (e.g. Bodribb 1996; Tezara et al. 1999), with Ci dropping, although not to the compensation point, as RWC decreased from c. 90 to 75%, and then increasing with further decrease in RWC. Tezara et al. (1999) observed a Ci/Ca of 0·8 at RWC = 90%, 0·7 at 75% and 1·0 at 45%. Direct measurements of Ci in leaves were made (Lauer & Boyer 1992) by sealing a chamber to the surface and circulating the gas through an IRGA to measure the equilibrium CO2 concentration. In unstressed sunflower leaves in air with c. 350 µL L−1, Ci was about 250 µL L−1, a ratio of 0·8, as expected. Drying the plants for 10 h from high water content to rather low, decreased gs, and Ci reached about 180 µL L−1 (Ci/Ca = 0·51). During the following dark period, Ci increased to 400 µL L−1 as stress increased. However, on illumination, Ci only decreased to c. 350 µL L−1 (Ci/Ca = 1). Soybean and bush bean responded similarly. The technique may be questioned because of the difficulties of sealing the chamber to the leaf, etc. Even very small leaks may greatly affect Ci when A is decreased; however, such errors were evaluated and appear minor. The study shows that Ci does not decrease to the compensation point (Γ) and that Ci increases at small RWC. Note, however, that Cc is the critical value. This cannot be estimated in such systems, but is probably close to it under equilibrium conditions. In conifers, Ci also decreased with the initial decrease in RWC but rose at more severe stress (Bodribb 1996), consistent with measurements on wheat and sunflower (Lawlor 1995). The increase in Ci when RWC is relatively large is indirectly supported by Duranceau et al. (1999) from measurements of carbon isotope ratios of CO2 produced in Phaseolus vulgaris as RWC decreased. As RWC fell, so sucrose and starch became enriched in 13C, as expected if A is consuming CO2 recycled within the intercellular spaces with decreasing Ci. However, at very small RWC and A, the carbohydrates became lighter, possibly due to increasing Ci, although other explanations are advanced. Again, what matters is the decline in Cc, which must be assessed either from fluorescence or oxygen isotope studies but in many studies it is unlikely that Cc attains the compensation point at small RWC.

Measurement of CO2 compensation point Γ

From A/Ci curves the CO2 compensation point (Γ) rises with decreasing RWC. Also, using closed gas exchange systems without pressure differentials (to minimize leakage) to measure Γ, Lawlor (1976) and Tezara et al. (1999) showed that Γ increased exponentially as RWC decreased below c. 80%. Lauer & Boyer (1992) showed that Γ increased, suggesting that CO2 evolution becomes relatively much greater than A at low RWC. The source may be respiratory CO2 production in the light by tricarboxylic acid (TCA) cycle metabolism, which is insensitive to O2 concentrations above 2%, and is called ‘day respiration’ (DR). As the respiration continues in 2% O2, the increase in Γ is not likely to be caused by photorespiration (Tezara et al. 1999), although it was once thought to be (Lawlor 1976). The evidence is that A is indeed decreased at low RWC, with respiratory CO2 increasing Γ and Ci. If A is decreased because metabolism is inhibited, then photorespiration (PR) would also decrease, even if the ratio of ribulose bisphuphate oxygenase to carboxylase activity increases with low Ci at relatively constant O2, so O2 insensitivity results. The increase in Γ would then depend on the relative magnitudes of CO2 evolution (PR and DR), and A.

Day and dark respiration under stress

At high RWC with large A, day respiration is a small proportion of A but at severe stress (RWC < 70%) the proportion increases, and CO2 is evolved in the light in some experiments (Lawlor & Fock 1975; Lawlor 1976). We assume that it is the same process as respiration measured in darkness, hence called dark respiration (DR) although the rates may differ, as it is insensitive to O2. Dark respiration is more resistant to stress than A and is associated with increasing Γ of stressed leaves (mentioned earlier). Progressively more of the carbon evolved is derived from stored carbohydrates (sucrose and starch) as RWC decreases, as shown by the decrease in specific radioactivity of CO2 evolved during 14CO2 feeding of sunflower leaves (Lawlor & Fock 1975, 1977a). This, plus the decreased sucrose content at low RWC (Lawlor & Fock 1977b), suggests that stored carbon, not immediate photosynthate contributed to CO2 release. If the evolved CO2 were only from rapidly turning-over pools of metabolites directly supplied with C from photosynthesis, then the rate of CO2 evolution might have decreased but the specific activity would have remained unaffected and as at high RWC. Similar effects of stress are shown by Duranceau et al. (1999) from the carbon isotope ratios of CO2 respired by stressed P. vulgaris, suggesting that increased consumption of reserve carbohydrates is related to an increase in Ci. Also, inclusion of DR in the calculation of Cc by Tourneux & Peltier (1995) increased Cc at RWC < 70%. Thus, there is strong evidence at low RWC of maintenance of DR which consumes stored carbohydrates and evolves CO2, which could contribute to the increase in Ci and Γ as RWC falls. This suggests that Apot is small or zero and so cannot assimilate the respiratory CO2. Although DR is a small proportion of A (10%) at large RWC, it becomes infinite when A ceases at extremely low RWC. Interestingly, the increased importance of mitochondrial metabolism at small RWC is indicated by the accumulation of glutamate and proline, associated with mitochondrial metabolism (see later). In summary: CO2 evolution by DR is relatively constant or decreases slightly as RWC decreases, but increases substantially as a proportion of A and utilizes stored carbohydrates. If A decreases proportionally more than respiration or gs, then Ci will rise. Thus, an absolute increase in respiration is not required to increase Ci, and given that Γ is an equilibrium determination and so independent of gs, decreased A is probably responsible for increasing Ci and Γ of stressed leaves.

Conclusions and general assessment of CO2 exchange

  • 1A/Ci responses to decreasing RWC indicate that, despite some uncertainty over methods, A, gs, Ci and Cc decrease in all experiments. Fluorescence and O2 isotope exchange studies show that Ci and Cc decrease to the compensation point at RWC of c. 75%. But under steady-state gas exchange Cc may not reach the compensation point and Ci rises at low RWC. The equilibrium CO2 compensation point, Γ, also increases with decreasing RWC.
  • 2Application of large CO2 concentrations to leaves with low A, at RWC between 100 and c. 75%, restores A to Apot in some studies but not others. In all studies, Amax decreases progressively at RWC < 75% and cannot be returned to control Apot by elevated CO2.
  • 3Removal of the epidermis restores A to Apot in some studies but not in others.
  • 4Day respiration increases as a proportion of CO2 exchange as RWC falls and is responsible for the increased Ci and Γ.

We suggested that there are two types of response of Apot to decreasing RWC: Type 1 with no decrease as RWC falls from 100 to c. 75% and Type 2 with progressive decrease over this range (Fig. 1). Further decrease in RWC results in smaller Apot in both types. In some studies, A/Ci responses and measurements of Ci and Γ show that there is progressive metabolic inhibition of Apot as RWC decreases. However, as RWC drops gs may decrease more than Apot, so Ci decreases but not to the compensation point. In other studies, Ci and Cc reach the compensation point, showing that Apot is not impaired. In general, further decrease in RWC increases Ci. The Γ rises a little with a small decrease in RWC but substantially with further loss of RWC. The increase is O2-insensitive and reflects the maintenance of respiration and decreased A. In those experiments where the Apot is unaffected and Ci falls to the compensation point, the RWC at which Apot starts to become inhibited is generally very small, 60% and below, and Apot stops at c. 20% RWC. Thus, the evidence suggests either that decreasing RWC has no effect on photosynthetic metabolism until a threshold is reached, below which it is impaired, or a progressive inhibition of metabolism. Kaiser (1987) considered that photosynthetic metabolism was inhibited at low RWC because it was impossible to restore Apot. However, as A was much lower than Apot in this situation, he concluded that, with current atmospheric Ca, small gs still determined A, i.e. stomata still strongly limit photosynthesis in normal air. The causes of the differences between the responses is not understood. The Type 1 and 2 responses may be related to: (a) different experimental approaches, although no clear pattern has emerged that provides a technical explanation; (b) differences in metabolism depend on particular characteristics of the species: Type 2 responses have been observed frequently in sunflower (but Type 1 has also been observed) and both occur in wheat. The Type 1 and 2 responses appear quantitatively very different but are qualitatively similar; they may reflect differences in the cell water balance, for example, related to differences in cell elastic modulus. Alternatively, they could reflect differences in sensitivity of a basic process to changing cellular conditions in different species or under different conditions, e.g. if ionic concentration in chloroplasts differed and thus resulted in different rates of if ATP synthesis (see later).

Figure 2.

Schematic of the interacting processes in C3 photosynthesis (A) that are implicated in the water stress syndrome. With decreasing RWC from c. 100 to 75% stomatal conductance, gs, decreases, so A falls as a consequence of lower Ci according to the ‘stomatal control model’, and is fully reversed by increasing the ambient CO2 in many studies. Further decrease in RWC causes a decrease in Apot, which is not reversible by elevating CO2 concentration; this is metabolic limitation. Light harvesting and PSI and II functions are impaired only by very low RWC, but electron transport in the absence of inhibitors even at high RWC is down-regulated by increased qN quenching. Electron transport to oxygen via photorespiration (the major flux), Mehler reaction and dark respiration (small fluxes) remove excitation energy when CO2 assimilation decreases. Stomatal control determines the rate of CO2 assimilation and slows and minimizes the development of stress to the system over a range of RWC, before metabolism is disrupted. Metabolic limitation includes decreased ATP synthesis because of loss of Coupling Factor amount and activity. This slows RuBP synthesis and hence CO2 assimilation (Apot, see Fig. 1), Many of the metabolic effects of low RWC are explicable by ATP limitation in metabolism. Alternative interpretations, principally the effect of low Ci are considered.

Photosynthetic mechanisms at different RWC

An overview of the parts of photosynthetic metabolism that may, potentially, be affected by RWC is given in Fig. 2; the interactions between them are crucial to understanding regulation under water deficiency (Lawlor 2002). The main factors determining A and Apot are:

  • 1Light capture: PSII activity: electron transport and reduction of acceptors leading to synthesis of NADPH and ATP.
  • 2Synthesis of RuBP: determined by PCR cycle capacity and activity and ATP and NADPH supply.
  • 3Carboxylation of RuBP catalysed by Rubisco: determined by RuBP and CO2 supply and metabolic regulation.
  • 4CO2 supply to the PCR cycle: determined by atmospheric CO2 and stomatal and other conductances.
  • 5Consumption of assimilated carbon: determined by export of carbon from the chloroplast, rates of sucrose and starch synthesis and consumption of assimilates in photo- and dark respiration.
  • 6Consumption of electrons: transfer to alternative sinks, particularly oxygen: energy dissipated by non-photochemical quenching.

These processes are listed sequentially for clarity; however, they are linked in a feed-forward, feed-back network, making it difficult to untangle cause and effect, e.g. small Ca and decreased supply of CO2 to the chloroplast when Rubisco is limiting (at saturating light when RuBP synthesis is maximal) removes the main sink for electrons, slowing electron transfer through PSII. At the same time, RuBP and ATP accumulate. The resultant decrease in pH of the thylakoid lumen, and increased reductant state of the electron acceptor of PSII and of the transport pathway, stimulates non-photochemical (qNP) dissipation of energy by the xanthophyll cycle, generally substantially (Saccardy et al. 1998). A decrease in photochemical quenching (qP) may result, depending on the relative magnitude of the energy flux and the CO2 supply. If the PAR flux is low, the electron transport is small, so CO2 supply can decrease more than at high light before effects on electron transport, qP and qNP are observed (Cornic & Briantais 1991). Moreover, for a given amount and activity of Rubisco per unit leaf area, Apot (note the earlier definition as light and CO2 saturated rate of A) will require a particular rate of RuBP synthesis and CO2 supply. If these are inadequate, Apot is not achieved. Alternatively, if Pi is restricted, then ATP supply falls and with it synthesis of RuBP, but as electron transport potential is maintained, the consequences for energy dissipation are the same as if Ca were reduced. Such interactions are crucial to understanding what determines A and Apot with decreasing RWC.

Effects of water deficits on energy capture, transduction and electron transport

Captured photon energy (as excitons in the light harvesting chlorophyll-protein complexes) is either used to excite the reaction centre and initiate electron transport and chemical reactions, or is dissipated as heat. The main route is electron transport to CO2 and its reduction to carbohydrates at large RWC and photon flux which saturate A. Excess electrons are also transferred to O2, via PR, the Mehler-ascorbate pathway, dark respiration or nitrate and sulphate reduction. Imbalance between photon flux captured by the light-harvesting system, electron flux and availability of terminal acceptors (particularly CO2) as a consequence of decreased RWC, is potentially damaging to the photosynthetic system under many conditions. Primarily this is because electrons are transferred via PSI to oxygen in the Mehler reaction, forming superoxide and other ‘active oxygen’ species, which react with and destroy components of the thylakoid membranes.

Energy capture, use and dissipation are regulated in several ways (Cornic 1994; Osmond et al. 1997). An effective mechanism for decreasing the photon flux captured is wilting of leaves. With loss of turgor, surface area of leaves decreases and they becomes parallel to the solar rays, dicot leaves generally hanging down and monocot leaves rolling or folding vertically, substantially decreasing the photon flux reaching the surface and thereby the energy load. Photons captured by the antennae excite electrons in chlorophyll: the excitation energy is used to drive electron transport, is dissipated, or long-lived energized states of chlorophyll are formed.

There are three ways for estimating the rate of electron transfer through the whole chain in intact leaves: (1) measurement of chlorophyll a fluorescence emission; (2) measurement of O2 exchange using 18O2 and mass spectrophotometry; or (3) measurement of CO2 assimilation. Considering (1), electron transport is estimated from the quantum yield of PSII photochemistry (φPSII), which is linearly related to the quantum yield of gross CO2 assimilation (φCO2). A measure of the number of photons absorbed by the photosystems is required. In (2), O2 uptake (Uo) is measured and added to net O2 evolution to obtain gross O2 evolution (Eo). As first shown by Genty et al. (1989), and later by Ruuska et al. (2000) and Badger et al. (2000), the agreement between the two methods is very good, although Biehler & Fock 1996) did not obtain good agreement between the two and regarded Eo as a better indicator of electron transport rates than those derived from fluorescence. For (3), electron transport may be determined from gas exchange by the Farquhar model of photosynthesis, and requires the CO2 compensation point (Γ*) in the absence of mitochondrial respiration’. The three methods agree well, allowing whole chain electron transport to be assessed and, together with net CO2 assimilation, providing information on the magnitude of sinks for electrons.

Net CO2 assimilation declines more rapidly than whole chain electron transport and net O2 evolution. Even with A close to zero, electron transfer was 80% of the control value at RWC of 70% (Tourneux & Peltier 1995). Biehler & Fock (1996) observed a decrease in A of 75% in wheat leaves at − 3 MPa ψ but electron transport was only decreased by 10%, at PAR of 850 µmol m−2 s−1. At very small photon flux (90 µmol m−2 s−1), A decreased by 50% but Eo was unaffected. Similarly, Cornic (1994) reported, for sunflower under a limiting PFD (300 µmol m−2 s−1), 80% decrease of A with no changes in whole chain electron transport rate. Moreover, in this case, O2 could entirely replace CO2 as an electron acceptor. Generalizing, in saturating photon flux, electron flow exceeds A, particularly during drought. Photon capture, excitation transfer, charge separation, reduction of acceptors and electron transport, all of which largely occur in the lipid–protein matrix of thylakoid membranes, are insensitive to decreasing RWC within the physiologically relevant range (100 to 50% RWC).

Photon capture, reaction centre function and water splitting

Photon capture in the chlorophyll–protein matrix of the light harvesting complexes, and transfer of excitation energy to the reaction centres, are unaffected by decreasing RWC. The basal rate of electron transport, Fo (measured on dark-adapted leaves) decreases in some experiments (Haupt-Herting & Fock 2000) but increased in others (Saccardy et al. 1996), suggesting that exciton transfer from antenna to reaction centre may be altered but that it is not a general phenomenon. There is no evidence of reduced reaction centre capacity (measured by the maximum quantum yield of PSII, derived from fluorescence measurements Fv/Fm) as a consequence of decreasing RWC in the physiologically relevant range, although severe loss of RWC did decrease photosystem activity and alter the structure of PSII (Giardi et al. 1996). The ability of reaction centres to use excitons depends on the state of the PSII acceptor, QA; when the acceptor is oxidized (reaction centre ‘open’) it can accept electrons, using the excitation energy and little fluorescence is emitted. This is not so if QA is reduced (reaction centre ‘closed’), excitation energy is not used and fluorescence is maximal (Fm′) with minimal efficiency of PSII.

The function of the water splitting complexes and thereby electron transport to the open reaction centres and evolution of O2 are not impaired by low but physiologically relevant RWC. Oxygen evolution is determined by the rate of electron transport (Badger 1985) and is not considered a limitation per se to A under water deficit. As mentioned, maintenance of high rates of O2 evolution in stressed leaf samples in the oxygen electrode at 5% CO2 is prime evidence that Apot is unimpaired. Continuation of O2 evolution at very small RWC when A (measured as net CO2 assimilation) is inhibited and cannot be stimulated by elevated CO2, must depend on transfer of electrons from water oxidation to acceptors, principally O2, via PR or the Mehler reaction. Several lines of evidence show that electron transport capacity is unimpaired. From 100 to 75% RWC, changes in fluorescence parameters at equivalent values of A were the same as in unstressed leaves but with altered CO2 (Cornic & Briantais 1991). This shows that decreased A is due to small gs and low Ci, not to decreased photosynthetic electron transport and O2 evolution (Cornic et al. 1992). Over a wide range of RWC, and in different experiments where there may or may not be response to a large concentration of ambient CO2, the relative concentration of open PSII centres, given by qP (qP = (Fm′ − Fs)/(Fm′ − Fo′)), is not decreased (Cornic & Massacci 1996; Tezara et al. 1999). This shows that the reduction state of QA is unaffected even when increasing Ca does not overcome small gs and increase A. Thus, electrons must be transferred to sinks other than CO2. Generally, the quantum efficiency of electron transport in PSII, and qP, decrease only at RWC below c. 75%. The maximum photochemical efficiency of PSII, given by Fv/Fm, is little affected by decreasing RWC (between 100 and 50%) in any studies indicating absence of photoinhibition (Sharp & Boyer, 1986); Tezara et al. 1999)) but does decrease at lower RWC (e.g. Cornic et al. 1989; Tourneux & Peltier 1995; Tezara et al. 1999). Similarly, maximum quantum yield of PSI is not affected (Cornic et al. 1989). Thus, with decreasing RWC, the functions of PSII, PSI and of electron transport are largely maintained, even where A and Apot are strongly inhibited.

Consumption of electrons with decreasing RWC

Reduction of CO2 to carbohydrates by the PCR cycle consumes by far the largest proportion of electrons derived from water oxidation. However, when A is decreased, as at low RWC, the electron transport chain becomes strongly reduced and electron transfer to O2 increases, producing active oxygen, which is very damaging unless alternative pathways for removal are available. The potential routes (Cornic & Briantais 1991) are PR, DR and Mehler-ascorbate pathway reactions, all of which result in electron transfer to oxygen, forming water, and have been considered in the discussion on oxygen emission and uptake. In addition, nitrate and sulphate reduction, and chemical syntheses, consume electrons.


This is the second most important mechanism for removal of electrons in C3 plants. Reaction of RuBP with O2, catalysed by Rubisco, generates phosphoglycollate, which is metabolized to glycerate by the glycolate pathway. In the process, CO2 is released as PR and electrons enter the mitochondrial electron transport chain and are transferred to O2. At 100% RWC, in the current atmosphere, PR uses about 25% of the electrons and protects metabolism when energy supply exceeds consumption (Kozaki & Takeba 1996; Osmond et al. 1997). Decreasing RWC has long been known to increase the ratio of PR to A (Lawlor & Fock 1975: Lawlor 1976), but no increase in the absolute rate of PR was reported from net CO2 exchange measurements (Lawlor & Fock 1975, 1977a; Lawlor 1976; Gerbaud & André 1980). However, as there is internal recycling of CO2, the magnitude of PR and proportion related to other processes were not known. Analysis of O2 exchange showed that PR is a substantial sink for electrons in leaves at high and low RWC (Tourneux & Peltier 1995). Indeed, O2 can entirely replace CO2 as an electron acceptor under limiting light so PR will increase in this situation, allowing a constant flux of electrons despite the decrease of A (Cornic 1994; Biehler & Fock 1996). Clearly, the PFD during measurement is crucial. Fluorescence and O2 isotope analysis show that in stressed leaves PR/A increases (perhaps by up to 50%; Stuhlfauth et al. 1990) but the absolute rate of PR may not be as large as in unstressed leaves, so fewer electrons will be used (Biehler & Fock 1996). Cornic & Massacci (1996) estimate that CO2 and O2 fixation (mainly by PR, with the Mehler reaction being negligible) accounted for 45% of energy at large RWC and 20% at small, and the proportion decreased under large photon flux (Brestic et al. 1995). However, Brestic et al. (1995) demonstrated, on dehydrated leaves showing little CO2 exchange, that inhibition of PR in bean leaves during a high light treatment did not change the Fv/Fm ratio (i.e. there was no photoinhibition): obviously they could cope with high light using other protective mechanisms. Generally, PR consumes a substantial proportion of the total electron flux and plays an important part in redox regulation even if the rate of electron flux is decreased by down-regulation at low A (Biehler & Fock 1996; Cornic & Massacci 1996; Haupt-Herting & Fock 2000, 2002).

Mitochondrial respiration

The mitochondrial electron transport chain accepts electrons from the TCA cycle and exogenous NAD(P)H, with O2 as the terminal acceptor. There is rapid exchange of reducing equivalents between the different forms and pools of the pyridine nucleotides, so when the NADPH/NADP+ ratio is large in the chloroplast, malate dehydrogenase is activated and reducing equivalents are exported as malate, ultimately forming NADH, which is oxidized by the mitochondria. In the light, the photorespiratory glycollate pathway (including the glycine decarboxylase conversion of glycine to serine) releases electrons, which are transported by the mitochondrial electron transport chain to O2; this is coupled to ATP synthesis (Siedow & Umbach 1995). The relatively large PR flux will maintain a large NADH pool and NADH/NAD+ ratio in mitochondria, and probably cytosol and peroxisomes. This may explain the large tissue NADH concentration and the high NADH/NAD+ ratio but relatively constant NADPH content at low RWC (Lawlor & Khanna-Chopra 1984). Although the flux of electrons derived from PR does produce ATP, PR also consumes ATP in synthesis of RuBP as substrate for the oxygenase reaction. Possibly the amount and site of ATP synthesis allows metabolism to continue in the mitochondria and cytosol but not in the chloroplast. The relevance of mitochondrial TCA cycle respiration to redox regulation at low RWC is unclear. The magnitude is not well established, although generally small (Lawlor 1976; Tourneux & Peltier 1995), particularly when compared to A in unstressed leaves. However, when A approaches zero and PR also, then mitochondrial activity as a sink for electrons will be relatively very large. Accumulation of products associated with mitochondrial metabolism, e.g. proline (Delauney & Verma 1993), at low RWC suggests that the shift from photosynthetic to respiratory mode is important in metabolism, possibly associated with the increased role of mitochondria and changes in nucleotides.

Mehler-peroxidase reaction

Rates of production of reduced O2 species (‘active O2’) with decreasing RWC have not been quantified, but because of their reactivity and potential for damaging thylakoids, etc, even small fluxes may be critical and rapid removal is essential. The elaborate, multiple mechanisms (e.g. the ascorbate pathway) with large capacity for rapid detoxification are testimony to this need. The magnitude of electron consumption by the Mehler-peroxidase reaction and by DR is generally considered negligible (Rey & Peltier 1989; Renou et al. 1990; Tourneux & Peltier 1995) over a wide range of RWC at photon fluxes saturating A (Ruuska et al. 2000). Calculation of PR flux derived from fluorescence studies (Cornic & Briantais 1991) fully accounted for O2 uptake. Tourneux & Peltier (1995) were able to calculate the specificity factor for Rubisco from O2 exchange without accounting for any additional flux to O2. However, there are differences between studies. Biehler & Fock (1996) concluded that the Mehler reaction was more substantial at larger photon flux, although Haupt-Herting & Fock (2002) considered the reaction to be less important.

Nitrate and sulphate reduction

Estimates of the proportion of the electron flux used for reduction of nitrate (sulphate reduction is negligible) in unstressed tissues range from 5 to 20%. It may be smaller at low than high RWC, as nitrate reductase (NR) activity is inhibited by stress. Alternative sinks for electrons in amino acid metabolism is considered later. Other chemical reactions, such as synthesis of secondary metabolites, may consume electrons and be important in long-term adaptation to stress but are probably negligible in utilization of electrons from the photosynthetic electron chain.

Summary of energy and electron use

Consumption of electrons at low RWC when A is small is primarily by PR, with negligible contribution from other processes. There is consensus that the potential capacity for light harvesting, energy transduction, electron transfer in reaction centres of the photosystems (PSIII and PSI) and electron transport in thylakoids are unaffected by a wide range of RWC. The rate of electron transport at saturating photon flux is determined by sink capacity for electrons, principally A at large RWC. PR is substantial in current atmospheric CO2 and O2 concentrations, and with A decreasing and PR rising (possibly absolutely but certainly as a proportion of A) at low RWC. Decreased sink capacity for electrons results in increased non-photochemical energy dissipation in Type 1 and 2 responses. However, maintenance of Apot in the Type 1 response will enable greater CO2 recycling and energy use within the tissue than if Apot decreases as in Type 2. At low RWC, where Apot decreases in both responses, electron transport is decreased because of biochemical, as opposed to biophysical, limitations. At low RWC, electron consumption and the redox state may be important in relation to mitochondrial activity, even if not as a large sink for electrons.

Oxygen exchange at low RWC

Studies on the effects of decreasing RWC on O2 metabolism are considered specifically as they have shown insensitivity of Apot with elevated CO2 and also the sinks for electrons. Tourneux & Peltier (1995) showed three stages in O2 evolution (Eo) using potato leaf discs in saturating CO2: (1) no change in Eo as RWC decreased from 100 to 80%; (2) slight decrease as RWC decreased further to 70%; (3) progressive decrease below 70% ceasing at about 30% RWC. In air (360 µmol CO2 mol−1) during stage (1) Eo decreased as gs declined, reaching a rather constant value, which was maintained in stage (2) before decreasing in stage (3) in the same way as in elevated CO2. Thus, A in stage (1) is determined by stomatal conductance, (2) includes metabolic limitation and (3) is limited by metabolism. If Eo at RWC 100% is equated with Apot, then CO2 assimilation is initially insensitive to loss of RWC but becomes progressively more inhibited as RWC falls. This corresponds to the Type 1 response discussed earlier.

Uptake of oxygen (Uo), as already mentioned, results from PR in C3 leaves, with the Mehler reaction and dark respiration small or negligible. Biehler & Fock (1996) showed that Uo increased with increasing water stress: in saturating light it was 30% of Eo in unstressed wheat leaves increasing to c. 55% at − 2·8 MPa ψ (and similarly in tomato, where the ratio of Uo/Eo increased from 50% to 67%; Biehler et al. 1997). Renou et al. (1990) also observed increased O2 uptake with water deficits. However, Haupt-Herting & Fock (2002) measured a decrease in Uo with stress. Tourneux & Peltier (1995) showed that the Uo/Eo ratio of potato leaf discs was about 0·5 at 90% RWC, increased to 1 at 80% RWC due to increased PR as Ci fell with stomatal closure, then remained relatively constant until RWC < 70%, when it decreased in parallel with Eo. Thus, there was greatly increased dissipation of electrons by O2 reduction as RWC decreased, related to PR. Flexas et al. (1999a) observed an increased ratio of electron transport to net O2 evolution, even at large Ca, due to increased O2 uptake in stressed grape vine. The flux of electrons to Uo may be absolutely greater in stress conditions than in unstressed and also relatively larger when compared with Eo (Thomas & André 1982; Renou et al. 1990). However, the decrease in Uo in parallel to Eo as RWC decreases suggests a limited sink capacity, which is not reversed by elevated CO2. Photosynthetic CO2 assimilation is the most likely candidate for the site of inhibition due to decreased synthesis of RuBP or slowed Rubisco rate (see later). If Apot decreases then the rate of the carboxylase reaction will fall, thus decreasing the ability to re-assimilate CO2 released within the tissue (hence leading to increased Ci and Γ). However, as PR/A depends on the CO2/O2 ratio, the flux of electrons to O2 via PR will change with conditions. For example, Tourneux & Peltier (1995) showed that O2 uptake increased as RWC dropped from 100 to 80%, remained relatively constant (and equal to O2 evolution) with a further fall to c. 60% and then fell in parallel to O2 evolution. This suggests that at low RWC, metabolism limits both Apot and PR (see Fig. 4c, Tourneux & Peltier (1995)).

Excitation energy dissipation

With decreased A, and thus decreased consumption of electrons, there is down-regulation of electron transport and a concomitant increase in excitation energy dissipation (Scheuermann et al. 1991; Saccardy et al. 1998; Tezara et al. 1999). This occurs by thermal dissipation in non-photochemical quenching, qN (qN = 1 − ((F′m − F′o)/(Fm − Fo))), related to activity of the xanthophyll cycle, with conversion of violaxanthin to zeaxanthin, resulting from epoxidation caused by an epoxidase enzyme. When ΔpH in the thylakoids is large, the enzyme is stimulated, thus increasing qN. With decreasing RWC and decreased A, qN rises strongly (Saccardy et al. 1998; Tezara et al. 1999). Substantial qN requires a large ΔpH and hence qN rises with photon flux. However, in very bright light violaxanthin is fully de-epoxidated even at 100% RWC (Saccardy et al. 1998). In very low light the capacity to use electrons exceeds the electron flux, and so qN is small. The inability to dissipate excess energy at large photon fluxes explains why water deficits have more detrimental effects on leaves in bright than dim light. Large qN is evidence for continued electron transport and a large ΔpH at low RWC. Any condition that decreases A without decreasing ΔpH will increase qN, e.g. low Ci and Cc and also inadequate ATP caused by lack of Pi (Jacob & Lawlor 1992). Conversely, dissipation of ΔpH by proton ionophores (e.g. phloridzin, see Lawlor 2001) prevents ATP synthesis, stops A and decreases qN.

The increase in qN with decreasing RWC resembles the effect of low A due to inadequate Ci caused by stomatal closure or low Ca at large RWC and can be reversed by large Ca. However, the association is not always found. Flexas et al. (1998, 1999b) observed, with rather limited decrease in RWC, decreased activity (‘down-regulation’) of PSII with increased qNP, with electron transport linked to CO2 assimilation. However, at small RWC, PSII was drastically down-regulated and electron transport and A were dissociated. Tezara et al. (1999) observed large qN at low A but were unable to increase A with elevated Ca. In all studies, at RWC below c. 75%, large qN cannot be fully reversed by elevated Ca and Ci, suggesting that factors other than CO2 supply limit A or that the quenching induced relaxes slowly. In conclusion, as A decreases at low RWC due either to low Cc or metabolic inhibition, energy capture continues and potential for electron transport is maintained, but the rate decreases and PSII is down-regulated because of limitation of acceptors, despite increased PR. Excess energy is dissipated in the thylakoid membranes by qN quenching, as a consequence of increased ΔpH.

Reductant content of leaves

The redox system under water stress is probably in a reduced state due to continued electron transport and absence of sinks. This applies to ferredoxin, which is important in nitrate reduction, and for regulation of PCR cycle enzymes by the thioredoxin system (see Lawlor 2001), but no measurements have been reported. Pyridine nucleotides are in a reduced state under stress, NADPH remains constant and NADH increases, so the reduced/oxidized ratios increase over a wide range of RWC (Lawlor & Khanna-Chopra 1984) as expected if sinks for electrons are limited. Dynamics of the pools of reductant in different leaf compartments, and with different conditions, are complex (Siedow & Umbach 1995) and there is no information about specific pools in relation to RWC. It is unclear if the normally very rapid and effective transfer of reducing equivalents between organelles continues at low RWC. Increased NADH may be explained by increased mitochondrial activity and respiration (both PR and DR) as A is decreased, and a relative increase in PR would increase the peroxisomal NADH pool. NADPH is essential for reduction of glycerate 1,3-bisphosphate to glyceraldehyde 3-phosphate by NADP glyceraldehyde phosphate dehydrogenase, the only reduction step in the PCR cycle and essential for RuBP synthesis (Lawlor 2001). Thus, inadequate reductant supply is not likely to decrease A at low RWC. There is no information that can be related to the Type 1 and 2 responses of Apot.

ATP synthesis in chloroplasts and content in tissues

The role of changes in ATP content and ATP synthesis (by ATP synthase, also called Coupling Factor, CF1-CF0) in the decrease in Apot caused by water stress is controversial (Tezara et al. 1999; Cornic 2000; Lawlor, 2002). Inhibition of CF and photophosphorylation was considered by Keck & Boyer (1974) to be the main cause of metabolic inhibition of A in droughted leaves, based on evidence from isolated chloroplasts. Activity of isolated CF1 (the part of CF extrinsic to the thylakoid membrane) was inhibited by large Mg2+ concentrations (Younis et al. 1979), which are expected in chloroplasts from cells with low RWC. Also the relaxation of flash-induced absorbance changes at 518 nm indicated inhibition of CF with mild to severe water stress (Boyer et al. 1987). This absorbance change results from an electrochromic shift in carotenoids, which is related to ΔpH and thus to phosphorylation rate and thereby the activity of CF (Ortiz-Lopez et al. 1991). Ortiz-Lopez et al. (1991) examined the effects of activation of CF by pre-illumination in very weak light (when the gamma subunit of CF is activated by thioredoxin f) by the flash-kinetic method, showing no changes associated with decreased RWC. A longer time constant for the relaxation of the spectroscopic signal occurred in severely stressed leaves after 90 s pre-illumination in weak light sufficient to fully activate CF1, suggesting a slower rate of catalysis by CF1, but this was not significant at light intensities relevant to the normal physiological activities of photosynthesis. Wise et al. (1990) used this technique to measure changes in field-grown sunflower leaves of contrasted water status. A slower decrease in the rate of change of signal within the first 100 ms after applying the actinic flash for droughted leaves in the afternoon (when A was smallest) was negligible, so photophosphorylation capacity was considered to be unimpaired. The observation that A decreased, whereas Ci increased in field-grown sunflower was explained by ‘down-regulation’ of photosynthesis to balance stomatal and non-stomatal factors (Wise et al. 1990). The responses may (Ort, Oxborough & Wise 1994) differ substantially between field-grown and slowly stressed (and acclimated?) plants and those grown with ample water in a greenhouse and then rapidly dehydrated. Therefore, proton transport and ATP synthesis associated with activity of CF in the light, were assumed to be unaffected. If ΔpH is maintained in thylakoids at low RWC, as shown by the large qN, and CF is activated (shown by the flash kinetics study), then ATP synthesis should occur and ATP contents should not decrease. The conclusion of Ortiz-Lopez et al. (1991) was ‘Thus, after more than 10 years during which photophosphorylation has been considered a major candidate in limiting photosynthesis at low ψ, it is now clear that this is not the cause and that attention should be focussed on other factors.’ None of the studies of the impact of RWC on ATP synthesis, either directed to whole plant responses or with isolated chloroplasts, included measurements of ATP (or ADP or AMP, important in understanding adenylate regulation), or of CF amount or activity (despite the obvious relevance). Rejection of the role of ATP in determining A and Apot is based solely on the spectrophotometric method.

However, substantial evidence from biochemical analyses suggests that decreased ATP synthesis under water deficits plays an important role in decreased Apot. Low RWC was shown, by de Kouchkovsky & Meyer (1992), to decrease ATP synthesis and Meyer et al. (1992) considered changes in thylakoid lipids to be important in the inhibition of photophosphorylation. ATP content decreased in stressed wheat (Lawlor & Khanna-Chopra 1984), and in sunflower, Tezara et al. (1999) showed that the ATP content decreased substantially (50%), and approximately linearly, over the range of Ψ (− 0·3 to −2·5 MPa), equivalent to 100–60% RWC, but ATP was still present at the smallest RWC. The decrease in A was not strongly correlated with that in ATP measured on whole leaves (Tezara et al. 1999), possibly due to variability in sampling and analysis, thus raising doubts about the role of ATP in determination of A. As ATP is synthesized in chloroplasts and mitochondria, variability may be associated with different sources. The amount of CF1 has only been measured in one study (Tezara et al. 1999), its decrease providing an explanation for smaller ATP concentration in the same tissue, and also for decreased photophosphorylation in stressed and isolated chloroplasts. In contrast, mild water deficit had no effect on ATP content of leaves in studies by Sharkey & Seeman (1989), although no measurements were made at more severe stress. Also, Sharkey & Badger (1982) conclude that ATP content and PCR cycle intermediates could not limit photosynthesis in osmotically stressed Xanthium cells. Thus, there is discrepancy between the conclusions about the effect of decreasing RWC drawn from the flash-kinetic measurements (no effect on CF activity and therefore ATP synthesis over a range of RWC), measurement of ATP content (loss of ATP over the whole range of RWC with more substantial reduction at RWC < 70%) and amount of CF (loss of CF at small and very small RWC).

In seeking to explain the discrepancy, which is very fundamental, the techniques of measuring ATP content and growing conditions must be considered:

  • 1Is the method of plant culture and stressing responsible? Plants were grown and stressed in soil, those of Ortiz-Lopez et al. (1991) more rapidly (4–6 d) than those of Tezara et al. (1999) (10–12 d), but such a fundamental difference in effect on CF is not likely to occur, as the same plant (sunflower) was used, although growing conditions and rates and degrees of stress differed.
  • 2Are the 515 nm kinetic measurements a reliable estimate of ATP synthesis (Ortiz-Lopez et al. 1991; Haraux & de Kouchkovsky 1998)? The altered signal should include any change in the amount of CF as well as activity as a slower rate of decay of the signal. However, correlation of the signal with amount and activity of CF and ATP content in defined conditions relevant to the studies was not made (Ortiz-Lopez et al. 1991) but would have strengthened the argument.
  • 3Is measurement of ATP in stressed leaves reliable? They may be erroneous because of the small and very dynamic pools and because of increased hydrolysis of ATP by phosphatases in stressed cells. Fast freeze-clamping of leaves under steady-state conditions and cold-extraction in denaturing medium (Lawlor & Khanna-Chopra 1984; Tezara et al. 1999), which are standard and accepted in other systems (Jacob & Lawlor 1992; Paul et al. 1995), should have minimize the effects of rapid ATP turnover and differential loss of ATP at different RWCs.
  • 4Could decreased amount of CF1 per unit leaf area at low RWC (Tezara et al. 1999) be caused by differential proteolysis? Use of inhibitors should have prevented this; if general loss of proteins was responsible for the change in CF, then much larger loss of Rubisco protein should have occurred.

If the validity of techniques for measuring ATP is accepted, then the data lead to the conclusion that low RWC does decrease ATP concentration, which is a general feature of stressed plants (Flexas & Medrano, 2002), probably by reducing synthesis as a consequence of loss of CF. If chloroplastic CF activity is not inhibited by decreasing RWC, then ATP synthesis in chloroplasts should be maintained as ΔpH is large. With very low A, ATP concentration should be maintained or rise, as observed by Paul et al. (1995) when PCR cycle activity, and thus RuBP synthesis, was decreased by removal of 95% of the activity of phosphoribulokinase. Thus, the decrease in ATP at low RWC (Tezara et al. 1999) is probably not caused by reduction of the capacity of or rate of flux through, the PCR cycle (Price et al. 1995; von Caemmerer 2000). Experimental evidence about the changes in ATP, etc. in relation to the Type 1 and 2 responses is crucial to understanding events in metabolism. In experiments where Ci decreases to the compensation point, ATP content increased, as expected from studies in which RuBP increased when Ci was lowered (von Caemmerer & Edmundson 1986). If ATP were available, but not CO2, then ATP export (via shuttles) from the chloroplast should increase ATP in the cytosol, providing more energy for sucrose and protein synthesis, ion regulation, etc. However, there is general metabolic inhibition, e.g. lower SPS activity and protein synthesis (see later), at RWC below c. 80%, so limiting demand for ATP.

Inorganic phosphate limitation

Inadequate Pi supply to the chloroplast would limit ATP synthesis and also explain the insensitivity to O2 observed at low RWC (Sharkey 1985). With small A and generally smaller content of phosphorylated intermediates, export of triosephosphate from the chloroplast would be small. However, if it is not used in sucrose synthesis, as SPS activity is decreased (Vassey & Sharkey 1989), and sucrose contents also in some experiments (e.g. Lawlor & Fock 1977b), then exported triose might be consumed in other reactions and the Pi not recycled back to the chloroplast. Limitation of Pi (e.g. due to malfunction of the triosephosphate-Pi transporter) cannot be eliminated. However, this is unlikely, as the triosephosphate translocator operates in counter-exchange of Pi and triosphosphate. Accumulation of phosphorylated PCR cycle intermediates in the chloroplast would limit ATP synthesis, but the content generally falls (Sharkey & Seeman 1989). At large RWC, accumulation of intermediates leads to starch accumulation, with recycling of Pi, but at low RWC and A, starch does not accumulate. Increased PR would recycle Pi in the chloroplast, so helping to maintain ATP synthesis and RuBP regeneration; it would be most important when Apot is least affected by low RWC, and diminish as Apot falls. Analysis of Pi in chloroplasts from leaves over a range of RWC in relation to A, phosphorylate intermediates, etc., is required.

ATP synthesis in mitochondria

No analysis has been made of ATP production in mitochondria, or of the roles of different cellular adenylate pools and how they affect A at low RWC. As PR and DR continue at low RWC and result in mitochondrial electron transport, which is coupled to ATP synthesis, it is likely that mitochondria produce ATP, and probably a larger proportion of cellular ATP, at low RWC. Regulation of ATP synthesis in the mitochondria is complex (Vedel et al. 1999) and there is little understanding of what might occur at low RWC. ATP from mitochondria might increase the ATP/ADP ratio in the cytosol (at high RWC it is required for metabolism, e.g. sucrose synthesis), but transfer of ATP from mitochondria to chloroplasts by shuttle mechanisms is presumably too small to prevent cessation of RuBP synthesis, if ATP shortage is the cause. Continued DR when A is very small might explain why ATP concentration did not become zero at low RWC, even when A (and Apot) did, in the study by Tezara et al. (1999). This suggests that inhibition of ATP synthesis in chloroplasts might be more sensitive to low RWC than that in mitochondria: CF from the two organelles do differ in structure (Vedel et al. 1999) and therefore potentially in other ways. Information about these processes in relation to the Type 1 and 2 responses would be valuable.

Synthesis of RuBP under water stress

The capacity for RuBP synthesis is a key factor in the assimilation of CO2, and depends on the supply of ATP and NADPH and the function of the PCR cycle. A strong sigmoidal relation between A and RuBP was demonstrated by Giménez et al. (1992) in stressed sunflower leaves, where Apot was progressively inhibited by falling RWC, suggesting that A depended on RuBP supply, not on CO2. There is a general decline in contents of PCR cycle intermediates during dehydration, so decreased RuBP might be caused by the general run-down of the cycle and decrease in A. The large ratio of 3PGA/RuBP suggests limitation in the RuBP regeneration part of the PCR cycle, either caused by enzyme limitations or inadequate ATP, although Tezara et al. (1999) interpreted it as evidence of 3PGA production by mitochondria. Gunasekera & Berkowitz (1993) also identified RuBP synthesis as a limitation but assumed that ATP was not limiting, so the cause was inhibition of enzyme activity. Decrease in Apot with stress was explained by loss of PCR cycle enzyme activity in early studies (see Chaves 1991). However, currently there is little evidence of limitation caused by loss of enzyme activity, either due to loss of protein or down-regulation. The large 3PGA/RuBP ratio suggests that Rubisco is not limiting, otherwise the ratio should decrease. As discussed earlier, the measured decrease in ATP suggests that the PCR cycle (or any particular enzyme in it) is not responsible for the decrease in RuBP and Apot. If the decrease in A were due only to CO2 depletion, then RuBP content should increase, as observed by von Caemmerer & Edmondson (1986) at low CO2 and saturating irradiance. The duration of stress might affect results and interpretation, e.g. the decrease in ATP and RuBP observed by Tezara et al. (1999) on leaves held at low RWC for some days, may be caused by low Ci inhibiting sucrose synthesis (see later). However, the decrease in ATP shown by Lawlor & Khanna-Chopra (1984) was a rapid response to decreased RWC. Small RuBP content is best explained by inadequate supply of ATP resulting from dehydration of the chloroplast, but analysis of the dynamics of relationships between ATP, Pi, RuBP, 3PGA and A under different RWCs and in relation to A, Apot (Type 1 and 2 responses) and Ci is desirable.

Activity of PCR cycle enzymes under water stress

Although PCR cycle enzymes are not implicated directly in the decrease of Apot, they play an important role in metabolism under stress. Rubisco has received considerable attention: amounts of protein and initial, total and maximal activities do not change substantially between 90 and 70% RWC or less (Sharkey & Seemann 1989, Gunasekera & Berkowitz 1993), although Tezara et al. (1999) found greater reduction, which did not correlate with loss of Apot. Medrano et al. (1997) showed that initial and total activity of Rubisco decreased (but not amount of Rubisco protein) with stress compared to well-watered subterranean clover grown in the field, i.e. the specific activity of the enzyme rather than activation state decreased, suggesting that the catalytic sites were blocked by inhibitors. Such differences associated with stress may reflect altered metabolism of inhibitors of importance in regulation of Rubisco. Activation of Rubisco in the light requires Rubisco activase and ATP: Robinson & Portis (1988) showed that activation decreased in proportion to the concentration of ATP in the chloroplast stroma. Sage et al. (1988) showed that high Ci may decrease Rubisco activity via ATP supply. Rubisco active sites are likely to be blocked by inhibitors, analogues of RuBP generated in metabolism, when RuBP concentration is low (Edmonson et al. 1990). Thus, limited ATP availability for the Rubisco activase reactions is consistent with the c. 65% decrease in Rubisco activity as RWC decreased from 100 to 50% (Tezara et al. 1999). Parry et al. (2002) demonstrated decreased activase and lower activation state at low RWC, suggesting that ATP may play an important role in regulation of A and in determining Apot under stress.

Utilization of assimilates and metabolism associated with photosynthesis

Metabolism of assimilates is required to avoid accumulation of products and inhibition of A. Considerable changes occur in amounts of carbohydrates and amino acids at low RWC, so there is likely to be substantial interaction with photosynthesis. Some aspects are considered.

Carbohydrate metabolism

As A falls with decreasing RWC, the amount of assimilate available for export as triosephosphate from chloroplast to cytosol diminishes and sucrose synthesis also. Sucrose content in leaves fell in rapidly stressed leaves at RWC < 80%, caused by low A and continued respiration, plus synthesis of amino acids (Lawlor & Fock 1977b). Thus, it is very unlikely that accumulation of assimilates would result in feedback inhibition of A or that the capacity of the triosephosphate-Pi transporter in the chloroplast envelope is affected by low RWC (see previous discussion). The rate of sucrose synthesis also depends on the activity of sucrose phosphate synthase (SPS), which is greatly decreased by even small loss of RWC. Low SPS activity could reduce the flux of triosephosphate from chloroplasts; accumulation of phosphorylated intermediates indicates this but there is no evidence of impaired capacity for transport. Loss of SPS activity is thought to depend on Ci rather than on RWC per se (Vassey et al. 1991). In Phaseolus leaves in air, a decrease in RWC from 100 to 80% decreased gs, Ci and SPS activity, the latter by more than 60% (Vassey & Sharkey 1989). Castrillo (1992) also measured decreased SPS activity but increased activation in the same species under mild stress. However, Quick et al. (1989) found that SPS activity increased with stress in wilted plants but under saturating CO2. The different response of SPS in these experiments was related to Ci, with several hours of exposure to elevated Ca reversing the effects of low Ci (Vassey et al. 1991). Rapidly stressing spinach leaf discs and slowly drying plants in soil (Zrenner & Stitt 1991) increased SPS activation (selective assay) but not total activity (saturating assay), and there was an interaction with light. In leaves of two Zea genotypes with RWC decreased from 100 to 75%, SPS activity decreased rapidly by 60–80% (rather independent of assay conditions) as gs, A and Ci decreased (Pelleschi et al. 1997). Also SPS activity decreased in rapidly stressed Zea, related to decreased A not to RWC, during the initial phase of drying, but at about −1 MPa, A decreased by 40% and maximal extractable SPS by 60% (Foyer et al. 1998). Thus, a decreased flux of triosephosphate from the chloroplast and loss of SPS activity seem to be the cause of slow sucrose synthesis and low content.

SPS is subject to complex control, including allosteric modulation by G6P and phosphorylation by a protein kinase using ATP (Huber & Huber 1996); the latter activates SPS under osmotic stress, suggesting that inactivation may be related to decreased ATP content. Given the complexity of regulation and differences between species, variable response are expected, because of different conditions in experiment. Clearly, SPS activity is rapidly lost with small change in water status and regulation probably depends directly on Ci (Vassey et al. 1991) although the mechanism is unclear, and Tang et al. (2002) argue that SPS is not dependent on Ci but on RWC. Many questions remain about regulation of sucrose synthesis and SPS at low RWC, particularly the relative roles of decreasing substrate with small A and changes in, e.g. decreased 3PGA and triosephosphate concentrations but increased 3PGA/triosephosphate ratio, and of Pi concentration effectors (see Sharkey & Seemann 1989).

Water deficits change the proportion of different carbohydrates: starch, glucose and fructose concentrations increased in bean with mild drought after 7 d but sucrose changed little (Pelleschi et al. 1997). Such changes may be adaptive (osmoregulation). This was associated with increased soluble (vacuolar) acid invertase activity in bean but not neutral invertase in Zea (a C4 plant) when gs and A decreased slightly but RWC was unaffected (Castrillo 1992; Pelleschi et al. 1997). In Zea, which is particularly sensitive to low RWC, the increase in sucrose, fructose and glucose suggests a general accumulation in carbohydrates due to inhibition of growth. As changes in RWC are small when stomatal control operates efficiently, it is likely that changes in sink capacity, turgor-based sensing mechanisms or Ci are more important than RWC. Zrenner & Stitt (1991) suggest that the change in proportions of carbohydrates is due to increased Pi concentration caused by smaller cell volume and perhaps to changes in starch remobilizing enzymes, which produce the mono- and disaccharides accumulated in some studies at very mild stress. The significance of these changes for photosynthesis are unclear, but may be important for adaptation to stress.

An important but little addressed problem related to carbohydrate content is the role of sink/source capacity over the range of water stress. With reduction of RWC from c. 100 to 80%, there is substantial, if not total, reduction in organ expansion, which is often accompanied by increased carbohydrates, although respiration, which uses some 50% of A in unstressed plants, may use more in stressed plants. Carbohydrate accumulation suggests that total A is less affected than total demand, so there is a need to consider the whole plant, not just the leaf. The capacity for translocation from leaves is probably unaffected by low RWC, but when sucrose synthesis is very limited, translocation could decrease sucrose accumulation in leaves, as observed by Lawlor & Fock 1977b). However, under some stress conditions, carbohydrate consumption may be limited by sink capacity, rather than translocation, and carbohydrates accumulate in leaves (Quick et al. 1989). Under other conditions, respiration could consume sucrose. Differences in these processes may cause of much of the variation in responses to RWC.

From the evidence, there is not a simple relation between A and carbohydrate content. Also, there is too little information to relate clearly to the Type 1 and 2 responses of A and Apot. There may be several phases dependent on RWC. (1) From 100 to 80% RWC, gs and A are relatively large and respiration low. In the Type 1 response photosynthetic metabolism is not affected, although consumption of assimilate could be inhibited by loss of SPS activity caused by low Ci. However, maintenance of Apot suggests that this effect is either not operative or rapidly overcome by elevated CO2 (Vassey et al. 1991). In the Type 2 response, the effect would be more progressive. In both, expansion growth decreases, limiting demand and possibly increasing starch and sucrose contents and starch/sucrose ratio with small hexose concentrations; there may be some loss of sucrose phosphate synthase (SPS), which does not prevent sucrose and carbohydrate accumulation. (2) At RWC c. 75% and below, in both Type 1 and 2 responses, growth is strongly decreased or stopped, with small gs and A and respiration/A large. SPS is inhibited due to a direct effect of RWC or to low Ci on metabolism or to changes in substrate/effector concentrations related to limited assimilate fluxes. Synthesis of starch and sucrose decrease strongly or stop but invertase activity and accumulation of hexoses increases so storage carbohydrates are consumed, ultimately being used in respiration and amino acid synthesis.

Nitrate assimilation

Water deficits substantially alter all aspects of nitrogen assimilation. Indeed, accumulation of amino acids is characteristic of low RWC (see Yamaya et al. 1986; Delauney & Verma 1993), with glutamate and particularly proline increasing greatly below a well-defined threshold, due to increased rates of synthesis relative to consumption (see Lawlor & Fock 1977b). This threshold appears to be at or below 75% RWC so plants with Type 1 and 2 responses may respond similarly, and there is no evidence of correlation between the changing response of Apot and amino acid accumulation. Changes in amino acid content are loosely associated with very small A (and Apot) and with rising Ci and Γ, suggesting that there is a relationship between altered CO2 assimilation, cell energetics and mitochondrial metabolism. Despite much analysis of the impacts of stress on N metabolism, many points are unclear. There is no evidence that supply of N (as NO3) to plants is limiting even at small RWC, although flux of NO3 to roots is by mass flow, so decreasing transpiration may decrease uptake. Probably the major limitation is in reduction of NO3, because small RWC substantially and rapidly decreases nitrate reductase (NR) activity in leaves, whereas rehydration quickly re-establishes it (Kaiser & Foster 1989; Foyer et al. 1998; Ferrario -Méry et al. 1998). As with SPS, NR activity was decreased by low Ci and increased by large Ci (Ca 10–20%), rather than low RWC, but activation state was not altered by CO2 (Kaiser & Brendel-Behnisch 1991). Regulation of NR amount is complex, with interactions between nitrate, sucrose, organic acids and amino acids determining the transcription and translation of the gene and protein activation: NR is rapidly turned over. Carbon metabolism is intimately linked with NO3 assimilation, and increasing sucrose and glucose concentrations stimulate NR-gene transcription and accumulation of transcripts (Foyer et al. 1998) and activation. Protein synthesis is also dependent on ATP supply. Supply of reductant is unlikely to limit NO3 conversion to nitrite. Synthesis of nitrite by NR is probably substantially decreased by low RWC, and so unlikely to accumulate and consequently ammonia synthesis is also probably decreased. At large RWC, NH3 is assimilated by the GS/GOGAT cycle, which requires ATP, but at low RWC lack of NH3 and ATP would prevent cycle function; however, there is no evidence. Effects of RWC on NH3 concentrations will depend on the relative changes in NO3 supply, NR activity, reductant supply and ATP concentration, and on GS/GOGAT activity (Morot-Gaudry et al. 2001). The increase in NH3 concentration sometimes measured at low RWC may be due to increased release by metabolism, e.g. in conversion of glycine to serine in the mitochondria as a consequence of PR. Concentrations of glycine and serine increase as RWC falls (this showed increased glycollate pathway flux under water deficits; Lawlor 1976; Lawlor & Fock 1977a,b) but if the glycollate pathway flux slows, then they must be derived from other sources and with loss of NR the NH3 required cannot come from NO3 reduction. Catabolism of proteins and amino acids in severely stressed tissues may provide NH3 but refixation might be limited if GS/GOGAT cycle activity is limited by low ATP.

Nitrogen metabolism under stress may be much less dependent on chloroplast and much more on mitochondrial metabolism, particularly as it relates to synthesis of amino acids (Morot-Gaudry et al. 2001). Amino acids accumulate at low RWC, and NH3 may be recycled by mechanisms other than the GS/GOGAT cycle. Glutamate dehydrogenase (GDH) catalyses the reversible reaction between α-ketoglutarate and NH3 to yield glutamate in isolated mitochondria (Yamaya et al. 1986; Hirel & Lea 2001) and could recycle NH3, using α-ketoglutarate from the TCA cycle. GDH requires a larger concentration of NH3 to synthesize glutamate than does GS/GOGAT (see Hirel & Lea 2001), but not ATP. As accumulation of amino acids is characteristic of low RWC, where DR is much more important than A, it is probably associated with mitochondrial activity. Glycolysis and TCA cycle organic acids, not assimilates directly from the chloroplast, provide the C-skeletons for synthesize of glutamate, explaining why 14CO2 does not enter glutamate under stress conditions (Lawlor & Fock 1977a). Glutamate is a precursor of proline, which accumulates because the rate of synthesis increases at low RWC, due to stimulation of transcription and translation of genes of enzymes involved in its synthesis (e.g. 1Δ-pyrroline-5-carboxylase reductase). Also, breakdown slows, as degradation by proline dehydrogenase is slowed or prevented by down-regulation of gene expression. As proline synthesis requires reductant (NADPH), it has been suggested to consume reductant but it is a relatively small sink and thus also a limited source of reductant, N and C upon rehydration. Accumulation is probably more important for long-term adaptation and as a protective, neutral osmolyte and compatible solute (Delauney & Verma 1993), although as proline accumulates mainly at very severe stress its role as a protectant is equivocal. In relation to photosynthetic metabolism, accumulation of proline and other amino acids is a probably a consequence of imbalance between continued photosynthetic electron flow, low A, maintenance of mitochondrial activity and altered regulation under extreme cellular conditions (low ATP). Thus, accumulates may be regarded as a consequence of altered cellular metabolism rather than a primary protective response. Despite considerable attention to proline synthesis, detailed knowledge of the conditions at low RWC responsible for synthesis/accumulation of such ‘stress metabolites’ and their role in regulation of energy, carbon metabolism, etc. is limited.

Protein synthesis in relation to water content

Because the effect of low RWC on A probably involves changes in CF, and also in Rubisco, changes to protein synthesis may important. Protein synthesis is generally substantially inhibited (see Deleu et al. 1999, and references therein), related to decreased polyribosome content (see Kramer & Boyer 1995), but the RWC at which this occurs is not clear. Even at quite large RWC polyribosomes are lost, suggesting that the process is sensitive to water deficits. The conditions responsible are not established. ATP is essential, so decreased ATP content would have profound effects, but the relation between ATP content and protein synthesis in photosynthetic tissue has not been examined. Changes in specific proteins at different stresses have been established, many are inhibited but there is no evidence that CF is affected by such a mechanism and Rubisco seems unaffected. Other proteins increase; these may have particular functions, such as the dehydrins and aquaporins, which are generally regarded as having a role in water transport (Deleu et al. 1999). Heat-shock proteins are synthesized and accumulate in very dehydrated tissues; they are molecular chaperones and function in protein folding (Schöffl et al. 1998). Synthesis is increased when proteins are not correctly formed and ATP supply is limited (Kabakov & Gabai 1997). Possibly they are associated with particular organelles under more extreme conditions and protect against dehydration. It would be interesting to establish if they have a role in maintenance of CF activity, as Rubisco activase does with Rubisco.

In conclusion, nitrogen and protein metabolism are greatly affected by water deficits, yet there is no evidence for a link with Type 1 and Type 2 responses of Apot. Probably most of the effects occur at lower RWCs and are not directly related to Apot. There are decreased sinks for amino acids (protein synthesis and growth), and inhibition of NR (and protein synthesis) under mild stress and large accumulations of amino acids at low RWC. Great changes in carbon and energy metabolism occur over the range of RWC with altered PR and, at lower RWC, mitochondrial DR probably a key process, related to use of reductant and to amino acid synthesis. This results from the particular conditions in very dehydrated cells, probably due to limited ATP.


Experimental studies on CO2 assimilation of C3 plants under decreasing relative water content show that decreasing RWC slows A and decreases the potential rate (Apot). The response of Apot to RWC is generalized into Type 1 and 2 responses. In Type 1, there are two main, relatively distinct, phases in the relation, with a transition between them. The stomatal phase occurs at RWC between 100 and 75%, without effect on Apot, so that A may be restored to Apot by large concentration of CO2. With decreasing gs, A, Ci and Cc fall, the latter possibly to the compensation point. Down-regulation of electron transport occurs by energy quenching mechanisms, and there are changes in carbohydrate and nitrogen metabolism, which may be acclimatory and probably a consequence of low Ci (although other conditions in the cell may have a role), with the effects reversible by elevated CO2. However, as Apot is not affected, gs regulates A via Ci and Cc. This is essentially normal regulatory metabolism, serving to maintain large RWC, and not a ‘stress’ phase with respect to metabolism. Below c. 75% RWC, metabolic inhibition occurs, which decreases Apot (progressively less reversible by elevated CO2 as RWC falls). However, A may still be lower than Apot, indicating that diffusion limits A because elevated CO2 concentrations raise the activity to Apot.

In Type 2, the phases are not distinct but progressive, lacking the two clearly distinguished phases of Type 1; gs and A decrease as in Type 1, but Apot decreases progressively and is only partially increased by elevated CO2, with the effect diminishing as RWC falls. Stomatal regulation, i.e. decreased gs, dominates at relatively large RWC, leading to a lower Ci and Cc but not (on evidence of gas exchange studies) reaching compensation point: gs becomes progressively less important and metabolic limitations more important as RWC falls. Type 1 and 2 responses reflect the balance between the sensitivity of gs and of metabolism regulating Apot to decreasing RWC. Type 1 has low gs but maintains Apot until RWC drops substantially. In Type 2, Apot is more sensitive to decrease in RWC than in Type 1. The Type 2 response is regarded as equivalent to the Type 1 response below the threshold of RWC, with the primary cause of the decrease in Apot in the Type 1 and Type 2 responses probably limitation of RuBP synthesis, caused by ATP deficiency related to loss of synthetic capacity. Inadequate ATP, not enzyme of the PCR cycle, limits RuBP synthesis. The alternative view is that decreased Apot is caused by low Ci. As Apot is inhibited by limited ATP and RuBP synthesis, respiratory metabolism become more important, Ci rises and Γ also. Electron transport is maintained (but down-regulated) in Type 1 and 2 over a wide range of RWC and there is a large reduced/oxidized adenylate ratio. As a consequence of such metabolic imbalance amino acids accumulate, and synthesis of some proteins decreases but others accumulate. These conditions profoundly affect cell functions and ultimately cause cell death.

Received 18 June 2001; received in revised form 9 October 2001; accepted for publication 9 October 2001