We review the current understanding of the temperature responses of C3 and C4 photosynthesis across thermal ranges that do not harm the photosynthetic apparatus. In C3 species, photosynthesis is classically considered to be limited by the capacities of ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco), ribulose bisphosphate (RuBP) regeneration or Pi regeneration. Using both theoretical and empirical evidence, we describe the temperature response of instantaneous net CO2 assimilation rate (A) in terms of these limitations, and evaluate possible limitations on A at elevated temperatures arising from heat-induced lability of Rubisco activase. In C3 plants, Rubisco capacity is the predominant limitation on A across a wide range of temperatures at low CO2 (<300 µbar), while at elevated CO2, the limitation shifts to Pi regeneration capacity at suboptimal temperatures, and either electron transport capacity or Rubisco activase capacity at supraoptimal temperatures. In C4 plants, Rubisco capacity limits A below 20 °C in chilling-tolerant species, but the control over A at elevated temperature remains uncertain. Acclimation of C3 photosynthesis to suboptimal growth temperature is commonly associated with a disproportional enhancement of the Pi regeneration capacity. Above the thermal optimum, acclimation of A to increasing growth temperature is associated with increased electron transport capacity and/or greater heat stability of Rubisco activase. In many C4 species from warm habitats, acclimation to cooler growth conditions increases levels of Rubisco and C4 cycle enzymes which then enhance A below the thermal optimum. By contrast, few C4 species adapted to cooler habitats increase Rubisco content during acclimation to reduced growth temperature; as a result, A changes little at suboptimal temperatures. Global change is likely to cause a widespread shift in patterns of photosynthetic limitation in higher plants. Limitations in electron transport and Rubisco activase capacity should be more common in the warmer, high CO2 conditions expected by the end of the century.
Temperature is one of the principle controls over plant distribution and productivity, with large effects on physiological activity at all spatial and temporal scales. Its central role in species success was apparent to the earliest biologists, and its great influence over crop yield and plant fitness has led to extensive research on temperature effects throughout the modern history of plant biology. While the majority of the work has focused on stress responses of plants to extremes of temperature, there has been steady progress in developing an understanding of the photosynthetic response to the range of temperatures where injury is not apparent. This range varies with species and growth conditions, and is loosely defined as that where the photosynthetic rate is fully reversible after short-term excursions to non-optimal temperature.
In general, photosynthesis can function without harm between 0 and 30 °C in cold-adapted plants that are active in winter and early spring, or grow at high altitude and latitude (Regehr & Bazzaz 1976; Mawson, Svoboda & Cummins 1986; Larcher 2003). In plants from equitable habitats (e.g. warm season crops), photosynthesis operates safely between 7 and 40 °C, and in plants from hot environments, such as tropical species and summer species in the Mojave Desert, photosynthesis operates between 15 and 45 °C with no apparent problem (Berry & Raison 1981; Downton, Berry & Seemann 1984; Bunce 2000). In all cases, photosynthesis shows an optimum temperature that roughly corresponds to the middle of the non-harmful range, and drops off with increasing slope as temperatures rise above the thermal optimum. With changes in growth conditions, the thermal optimum can shift, typically by one-third to one-half the number of degrees as the shift in growth temperature (Berry & Björkman 1980). Some species, particularly those in more equitable habits, can acclimate to temperature change as indicated by shifts in the thermal optimum and enhanced assimilation rates at the new growth temperature (Mawson et al. 1986; Atkin, Scheurwater & Pons 2006). Other species, such as specialists for extreme environments, show less potential to acclimate, and exhibit similar thermal responses in warm or cool growing conditions (Atkin et al. 2006). Eventually, at temperatures near the extreme end of the functional range, injury occurs, and the rate of photosynthesis is irreversibly impaired.
Over the past 30 years, numerous ideas have been proposed to explain the biochemical controls over C3 and C4 photosynthesis, and sophisticated theoretical models havebeen developed to test the possibilities (Farquhar, von Caemmerer & Berry 1980; von Caemmerer & Furbank 1999; von Caemmerer 2000). As a result, we have a good understanding of the limitations over photosynthesis at the thermal optimum in C3 plants (von Caemmerer 2000); however, the biochemical controls over the rate of photosynthesis at non-optimal temperatures are less clear. Advances in model parameterization and increased experimental attention have in recent years improved our understanding of the mechanisms controlling photosynthetic responses to both high and low temperatures (Bernacchi et al. 2001, 2002; Medlyn et al. 2002; Bernacchi, Pimentel & Long 2003; Salvucci & Crafts-Brandner 2004a). These advances occur at a time when society is increasingly concerned about global climatic change, and we are now observing the first wave of responses to climatic change in the form of range shifts, phenological adjustments and population collapses (Jump, Hunt & Penuelas 2006; Menzel et al. 2006; Mouthon & Daufresne 2006). To be able to predict the effects of climatic change, and to adapt agricultural systems and land management to a warmer world, it is imperative to understand how temperature affects photosynthetic carbon gain. This review will synthesize our current understanding of temperature effects on the photosynthetic biochemistry, and how these effects determine the thermal response of whole-leaf photosynthesis in C3 and C4 plants. This understanding will then be used to evaluate the mechanisms by which plants acclimate to different thermal environments.
THEORETICAL CONTROLS ON PHOTOSYNTHETIC CAPACITY IN C3 PLANTS
The rate of net CO2 assimilation (A) in the leaves of C3 plants is generally characterized as being under the control of three distinct processes: the capacity of ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco) to consume ribulose bisphosphate (RuBP) (RuBP-saturated photosynthesis or Rubisco-limited photosynthesis), the capacity of the Calvin cycle and the thylakoid reactions to regenerate RuBP (RuBP regeneration-limited photosynthesis) and the capacity of starch and sucrose synthesis to consume triose phosphates and regenerate inorganic phosphate for photophosphorylation (Pi regeneration-limited photosynthesis or triose phosphate use-limited photosynthesis) (Harley & Sharkey 1991; von Caemmerer 2000). The Rubisco capacity to consume RuBP is generally the predominant limitation on A at light saturation and CO2 levels below the current ambient partial pressure of 380 µbar. Because of this, the initial slope of the response of net photosynthesis to intercellular CO2 partial pressure (the A/Ci response) is determined by Rubisco capacity (Vcmax) at light saturation (Farquhar & von Caemmerer 1982; Hikosaka et al. 2006). RuBP regeneration capacity (as affected by the rate of light harvesting) becomes limiting at subsaturating light intensities at all levels of CO2 (Sage, Sharkey & Seemann 1990b). At saturating light intensities, moderate temperature (25–30 °C) and elevated CO2, either the capacity of RuBP regeneration or the Pi regeneration capacity limits A (Sharkey 1985a). RuBP regeneration capacity at light saturation generally reflects limitations in electron transport capacity (Jmax), a feature that allows for the estimation of Jmaxin vivo using gas exchange measurements at high CO2 (von Caemmerer & Farquhar 1981; von Caemmerer 2000; Cen & Sage 2005).
Characteristic responses of A to CO2 and O2 variation in C3 leaves indicate whether A is limited by the RuBP regeneration capacity versus the Pi regeneration capacity. When A is controlled by the RuBP regeneration capacity, it is stimulated by an increase in CO2 or a reduction in O2 supply under conditions where photorespiration occurs (von Caemmerer & Farquhar 1981; Sharkey 1988). This is normally the case above 5 °C, and below a CO2 level of 500 µbar (cooler temperatures, <15 °C) to over 1000 µbar CO2 (warmer temperatures, >35 °C) (Ehleringer et al. 1991). The sensitivity of RuBP regeneration-limited A to CO2 and O2 reflects the competition between these gases for RuBP; as CO2 levels increase, oxygenase activity is increasingly inhibited and A rises with decreasing slope to a plateau that begins at the CO2 level where photorespiration becomes nil. By contrast, when A is limited by the Pi regeneration capacity, there is little if any stimulation following CO2 enrichment or O2 reduction at CO2 levels where photorespiration is present (Sharkey 1985a,b, 1988). This lack of CO2 or O2 sensitivity occurs because Pi is used as fast as it becomes available, a situation that is not altered by reducing photorespiration (Sharkey 1985b). In practice, the easiest approach to detect a Pi regeneration limitation is to model the predicted sensitivity of A to variation in O2 and CO2, and compare this with observed sensitivities (Sharkey 1988). O2 sensitivity of A can be defined as (A21 − A2)/A2 where A21 is net CO2 uptake at current O2 levels (21% O2) and A2 is A at 2% O2 where photorespiration is nil (Sage & Sharkey 1987). CO2 sensitivity, defined as (Aair − Asat)/Asat, compares A in a lower CO2 atmosphere (Aair) with CO2-saturated A (Sage, Sharkey & Pearcy 1990a). If RuBP regeneration capacity is limiting, the observed O2 or CO2 sensitivity will be equivalent to the respective theoretical sensitivity; if the Pi regeneration capacity is limiting, the ratio will be well below the theoretical value, and will often be near zero or negative (Fig. 1) (see also Mawson et al. 1986; Sage, Sharkey & Seemann 1989; Cowling & Sage 1998; Sun, Edwards & Okita 1999; Hendrickson, Chow & Furbank 2004b; Kubien & Sage 2004b).
Temperature and the CO2 response of photosynthesis
A common approach for analyzing environmental effects on photosynthetic limitations in intact leaves has been to measure the response of A to intercellular partial pressures of CO2 (the A/Ci curve). This is because the three major classes of limitation described earlier show characteristic responses to CO2 variation that allow for their identification and quantification (von Caemmerer & Farquhar 1981; Hikosaka et al. 2006). The initial slope of the A/Ci response at light saturation usually is a direct reflection of Rubisco capacity, while the slope of the A/Ci response at elevated CO2 distinguishes between an RuBP regeneration limitation (if a positive slope is present) and the Pi regeneration limitation (if the slope is absent or negative) (Sharkey 1985a; Sage et al. 1990a). Often, at intermediate CO2 levels, there is a region of curvature in the response where it is not obvious which process is limiting. This may occur because of colimitation by multiple processes, or because the CO2 level corresponds to regions where both the Rubisco limitation and the RuBP regeneration limitation show similar responses to CO2 variation (e.g. Wise et al. 2004). To evaluate these possibilities, it is necessary to model the A/Ci response using estimated Rubisco Vcmax (from either gas exchange or in vitro assessments), estimated Jmax (from either gas exchange, chlorophyll fluorescence or in vitro assessments) and the Pi regeneration capacity, which is estimated from the CO2 saturated rate of A under conditions where photorespiration occurs (Sharkey 1985a,b). Some caution is required, because there can be substantial variation in the values of kinetic constants for Rubisco, mesophyll conductance and light usage (von Caemmerer & Quick 2000; Bernacchi et al. 2002; Yamori et al. 2006b). In addition, it is possible that processes not parameterized in the classical version of the Farquhar et al. model and its derivatives (e.g. Harley & Sharkey 1991; Medlyn, Loustau & Delzon 2002) may become limiting. For example, the activation state of Rubisco may become limiting in certain circumstances such as low O2 and low CO2, or high temperature (Sage, Cen & Li 2002; Crafts-Brandner & Salvucci 2004). In such cases, the models must be adjusted to account for new types of limitation if they are to retain predictive power. One must also be sure the Vcmax and Jmax estimates are reasonably correct. Failure to account for the possibility that Pi regeneration is limiting at the CO2 saturation point can lead to serious errors in Jmax estimates, and assumptions that the initial slope of the A/Ci response are determined by Vcmax can be in error at low light or extreme temperatures (Sage et al. 1990b; Harley & Sharkey 1991).
To demonstrate how temperature theoretically affects the biochemical limitations on A, we have modelled A/Ci responses at four temperatures using input parameters from tobacco (Fig. 2) (Bernacchi et al. 2001, 2002, 2003). In Fig. 2, the predicted A/Ci curve shows a modest response of the initial slope to temperature variation below 20 °C, but little response above 20 °C. Photorespiration and mitochondrial respiration increase with rising temperature (Brooks & Farquhar 1985; Sharkey 1988; Sage et al. 1990a), increasing the CO2 compensation point and shifting the A/Ci curve to higher Ci. The most pronounced effect of increasing temperature is on the CO2-saturated plateau, which typically rises with a Q10 near 2 below the thermal optimum (Sage 2002). This rise in the plateau reflects either a rise in the electron transport capacity or a rise in the Pi regeneration capacity, both of which have a high thermal dependence below the photosynthetic thermal optimum (Sage, Santrucek & Grise 1995; Leegood & Edwards 1996; Yamasaki et al. 2002; Cen & Sage 2005). In the simulation in Fig. 2, the Pi regeneration capacity sets a ceiling on A at 10 and 20 °C, but rises above the RuBP regeneration capacity at 30 °C and above. Pi regeneration has a greater thermal dependence than RuBP regeneration because of a high Q10 of sucrose and starch synthesis (Pollock & Lloyd 1987; Stitt & Grosse 1988; Leegood & Edwards 1996). As a result, the A/Ci response shifts from being completely CO2 insensitive above a Ci of 200–400 µbar at 10 and 20 °C, respectively, to exhibiting CO2 sensitivity of gradually decreasing slope above a Ci of 400 µbar at 30 and 40 °C. At 10 °C, the ceiling imposed on A by the Pi regeneration capacity lowers the CO2 saturation point below the operational Ci, causing A to become CO2 insensitive at current atmospheric CO2 levels and below.
Models of the temperature response of C3 photosynthesis
The theoretical temperature responses of A that correspond to the A/Ci responses in Fig. 2 are shown for intercellular CO2 partial pressures corresponding to the Pleistocene minimum level (180 µbar), current levels of CO2 (380 µbar) and future CO2 levels (700 µbar) (Fig. 3). At a Ci of 150 µbar, our model demonstrates how the A/T curve is controlled by Rubisco capacity down to 10 °C, at which point Pi regeneration capacity becomes limiting. Electron transport capacity is non-limiting at 150 µbar except above 45 °C, where a strong decline in the electron transport capacity can cause the RuBP regeneration capacity to fall below the Rubisco capacity.
At all three CO2 levels, the modelled temperature responses of Pi regeneration capacity are identical because CO2 supply does not affect the triose phosphate use rate established by starch and sucrose synthesis (Fig. 3) (Sharkey 1985a,b). By contrast, the rise in CO2 from 150 to 300 and 600 µbar stimulates the Rubisco and RuBP regeneration-limited values of A, such that the limitation by the Pi regeneration capacity controls A to higher temperature. At a Ci of 300 µbar, A is limited by Pi regeneration up to 18 °C, while at a Ci of 600 µbar, Pi regeneration limits A up to 23 °C. As the limitation caused by the Pi regeneration capacity becomes pronounced, the A/T response develops a steeper thermal response, reflecting a high Q10 of starch and sucrose synthesis.
Rubisco capacity limits A at intermediate temperatures at 300 µbar, forming the broad thermal optimum often seen in the A/T response of C3 plants at current CO2 levels (Ferrar, Slatyer & Vranjic 1989; Hikosaka, Murakami & Hirose 1999; Sage 2002; Atkin et al. 2006). Our simulation shows that Rubisco-limited A gradually declines above the thermal optimum at 300 µbar, until about 40 °C, where the electron transport capacity asserts control. At this point, the drop-off in A with further increases in temperature, is pronounced, following the rapid decline modelled for the electron transport capacity.
Increasing CO2 stimulates Rubisco-limited A through the combined effect of the increased substrate availability and the suppression of photorespiration; rising CO2 enhances RuBP regeneration-limited A only by suppressing photorespiration (von Caemmerer 2000). Consequently, Rubisco-limited A increases more with rising CO2 than RuBP regeneration-limited A. At 600 µbar, the elevated CO2 level allows Rubisco-limited A to be in excess at all temperatures, and as a result, the response of A to temperature is dominated by electron transport capacity above the temperature where Pi regeneration capacity is limiting (Fig. 3). The thermal optimum narrows and forms a sharp peak that often mirrors the thermal optimum observed for electron transport in vitro (Mawson & Cummins 1989; Sage et al. 1995; Yamasaki et al. 2002; Cen & Sage 2005). The thermal optimum has also shifted to higher temperature, reflecting the reduced impact of photorespiration at elevated CO2.
The short-term response to temperature reduction below the thermal optimum has been widely linked to limitations in Pi regeneration, as indicated by (1) O2 and CO2 insensitivity of steady-state photosynthesis; (2) oscillations in A following a change in CO2 or O2 in leaves at low temperature; (3) a high ratio of phosphoglycerate (PGA) to triose phosphates, and elevated levels of phosphorylated immediates in the carbon metabolic pathway; (4) high sensitivity of A to the application of Pi-sequestering agents to the leaf; (5) the Pi optimum for A in isolated chloroplasts increases at low temperature; and (6) feeding Pi to leaves transferred to low temperatures enhances A (Leegood & Edwards 1996, and references therein; Strand et al. 1999; Hendrickson et al. 2004b). Pi regeneration limitations are associated with declines in RuBP pool size; however, deactivation of Rubisco in response to limitations in Pi regeneration can allow metabolite levels to recover to levels observed when Rubisco capacity is limiting (Sage, Sharkey & Seemann 1988; Sharkey & Vanderveer 1989). Limitations by Pi regeneration capacity at low temperature are not universal. Species adapted or acclimated to cool conditions are less likely to be limited by the Pi regeneration capacity (Sage & Sharkey 1987; Makino et al. 1994; Strand et al. 1997, 1999; Savitch, Harney & Huner 2000). In winter- and spring-active S. desertorum plants, for example, O2 sensitivity of A showed that Pi regeneration limitations were not present above 10 °C, while in P. fremontii, a deciduous species that is active in late spring and summer, Pi regeneration was limiting below 20 ° (Fig. 1).
The limitation that predominates at suboptimal temperatures when Pi regeneration capacity is excessive is not always clear. It is probably Rubisco capacity at low CO2 levels (<380 µbar) and electron transport capacity at higher CO2 levels, as indicated by analysis of published A/Ci curves (Sage & Sharkey 1987; Cen & Sage 2005). At current CO2 levels, Rubisco capacity limits A in cool-grown oak and Plantago leaves below the thermal optimum (Hikosaka et al. 2006). By contrast, at high CO2, the capacity of electron transport shows a sharp decline below the thermal optimum of photosynthesis that is often well correlated with the temperature response of A (Mawson & Cummins 1989; Makino et al. 1994; Yamasaki et al. 2002).
In sweet potato, Cen & Sage (2005) estimated the temperature response of Pi regeneration capacity, electron transport capacity, Rubisco Vcmax and the Rubisco activation state. With these estimates, they were able to effectively model the A/T response from 9 to 41 °C (Fig. 4). Rubisco-limited A matched observed A across most of the temperature range at a Ci of 140 µbar, and at the thermal optimum at current levels of CO2. Rising CO2 increased the temperature sensitivity of A in sweet potato, narrowed the breadth of the thermal optimum and raised the thermal optimum, as predicted by a shift in limitation from Rubisco to electron transport capacity. A limitation in the Pi regeneration capacity was more pronounced at elevated CO2 in sweet potato than that predicted for tobacco in Fig. 3, such that Pi regeneration controlled A up to the thermal optimum at elevated CO2, at which point the model predicted electron transport capacity became limiting.
Contrary to the mentioned observations are proposals that A declines above the thermal optimum because the capacity of Rubisco activase to maintain Rubisco in an activated state declines to limiting levels (Salvucci & Crafts-Brandner 2004a, and the references therein). As a result, Rubisco deactivates to a point where its ability to consume RuBP limits CO2 assimilation. Reductions in the activation state of Rubisco are well correlated with reductions in A above the thermal optimum in spinach, cotton, tobacco, oak, pea, Antarctic hairgrass and Larrea (Salvucci et al. 2001; Haldimann & Feller 2004, 2005; Salvucci & Crafts-Brandner 2004a,b). The issue of Rubisco activase versus electron transport limitation above the thermal optimum is discussed further below.
MECHANISMS CONTROLLING THE TEMPERATURE RESPONSE OF THE PHOTOSYNTHETIC BIOCHEMISTRY IN C3 PLANTS
The response of Rubisco-limited photosynthesis to increasing temperature has two general explanations: (1) a change in carboxylation capacity caused by thermal effects on the Km and kcat of Rubisco, and an increase in oxygenase activity that reflects reductions in both the CO2/O2 ratio in solution and the relative specificity of Rubisco for CO2 versus O2 (Jordan & Ogren 1984; von Caemmerer & Quick 2000). The decline in the CO2/O2 solubility with rising temperature accounts for about a third of the rise in photorespiration, while the remainder is caused by the reduction in relative specificity. The decline in relative specificity with increasing temperature is largely driven by a greater increase in the Rubisco Km for CO2 than the Km for O2 (von Caemmerer & Quick 2000).
The response of the electron transport capacity to temperature is complex, because of the large number of individual steps, a requirement for coordinated turnover of these steps, the need for rapid diffusion in the lipid matrix and the requirement for the membranes to be fluid yet viscous enough to separate the stroma from the thylakoid lumen. As a result, it has been difficult to pinpoint specific limiting steps that control the temperature response of electron transport. At suboptimal temperatures, electron transport capacity rises with a Q10 near 2, and shows a sharp transition to a narrow thermal optimum (Berry & Björkman 1980; Mawson & Cummins 1989; Sage et al. 1995; Yamasaki et al. 2002; June, Evans & Farquhar 2004; Cen & Sage 2005). This rise in the whole chain electron transport rate with increasing temperature has been correlated with temperature stimulation of energy flow through photosystem II (PSII), and electron flow from the quinones to photosystem I (PSI) (Mawson & Cummins 1989 for S. cernua, Makino et al. 1994 for rice and Yamasaki et al. 2002 for winter wheat). Electron transport capacity is closely correlated with cytochrome f (cyt f) content in rice and spinach, implying that this may be a key rate-limiting step (Makino et al. 1994; Yamori et al. 2005). If so, the thermal response of electron flow through cyt f may be a leading controller of the thermal response of A at high CO2, as indicated by results from pea and winter wheat (Mitchell & Barber 1986; Yamasaki et al. 2002).
The mechanism causing the decline in the electron transport rate above the thermal optimum remains uncertain (June et al. 2004). A leading proposal is that cyclic electron transport is activated at elevated temperature at the expense of linear electron transport, thereby causing a shortage of NADPH (Bukhov et al. 1999; Sharkey & Schrader 2006). In barley, elevated temperature is proposed to activate cyclic electron flow by diverting electrons from the NADPH pool to the plastoquinone pool (Egorova & Bukhov 2002; Egorova et al. 2003; Bukhov, Dzhibladze & Egorova 2005). In pima cotton, enhanced cyclic photophosphorylation may explain reductions in the stromal oxidation state observed above the thermal optimum (Schrader et al. 2004). The rise in cyclic electron flow above the thermal optimum increases the thylakoid pH gradient, thereby activating photoprotective quenching and causing a dissociation of the outer light harvesting antennae from PSII (Sharkey & Schrader 2006). Electron flow through PSII decreases above the thermal optimum in a pattern that mimics a decline in whole chain electron transport (Yamasaki et al. 2002 for winter wheat); by contrast, electron flow rate through PSI is stable between the thermal optimum for photosynthesis and 40 °C, indicating it has high capacity to support enhanced cyclic electron flow at elevated temperature (Berry & Björkman 1980; Havaux 1993a; Yamasaki et al. 2002). Non-photochemical quenching of PSII is widely observed at temperatures where electron transport capacity slows with rising temperature, indicating that the reduction in electron flow through PSII is a regulatory response to limitations further down the electron transport chain (Yamasaki et al. 2002; Salvucci & Crafts-Brandner 2004b; Schrader et al. 2004). Consistently, Yamasaki et al. (2002) demonstrated that the capacity for electron transfer from plastoquinone to P700 declines above the thermal optimum, implicating electron flow between the photosystems as a possible cause for the decline in the electron transport rate.
Inactivation of the water-spitting complex is also implicated as a cause of heat-induced reductions in electron transport capacity, particularly at high temperatures (above 38 °C in potato and above 40 °C in spinach) (Havaux 1993a; Enami et al. 1994). At moderately warm temperatures, this lesion is probably not significant, as leaves can readily alter PSII properties to reduce heat sensitivity of the water-splitting complex (Havaux 1993b). Furthermore, the thylakoid membrane becomes leaky at higher temperature, potentially uncoupling electron flow from photophosphorylation (Schrader et al. 2004). This is not thought to be a direct cause of the decline in A at elevated temperature, because ATP levels and the pH gradient remain high due to the rerouting of electrons from NADPH back to PSI (Sharkey & Schrader 2006). The activation of cyclic electron flow at the expense of linear flow is proposed to protect PSII from damage at high temperature by inducing photoprotective quenching, stabilizing the thylakoid membrane through enhanced zeaxanthin formation, and reducing the size of the PSII light-harvesting antennae (Tardy & Havaux 1997; Bukhov et al. 1999; Schrader et al. 2006; Sharkey & Schrader 2006).
Calvin cycle capacity
Calvin cycle limitations on A have been explored at the thermal optimum, most directly through the use of antisense transgenics deficient in key enzymes of the pathway (Raines 2003). Of the examined Calvin cycle enzymes, sedoheptulose 1,7-bisphosphate (SBPase) exerts the greatest control, with modest (<35%) reductions in SBPase levels causing a significant decline in A at air levels of CO2, and larger reductions at elevated levels of CO2 (Harrison et al. 2001; Olcer, Lloyd & Raines 2001). Transketolase and aldolase also show significant control over A at air levels of CO2 (Raines 2003). To our knowledge, none of the antisense lines have been used to study the response of the Calvin cycle control coefficients to temperature, although it is conceivable they could have significant control over the response given their ability to affect A at the thermal optimum. Stitt & Grosse (1988) observed low metabolite levels in the Calvin cycle above the thermal optimum in spinach, leading them to suggest that excessive sucrose synthesis capacity at elevated temperature may drain the Calvin cycle of intermediates, slowing its turnover. Levels of SBPase are high in chilled tomato leaves, suggesting that fructose 1,6-bisphosphatase (FBPase) and SBPase activities are limiting (Sassenrath, Ort & Portis 1990. Brüggemann et al. (1994), however, did not observe SBPase or FBPase limitations in numerous tomato lines grown at low temperature and low light, and suggested that the results of Sassenrath et al. (1990) reflect regulatory deactivation of the bisphosphatases in response to photoinhibition of PSII in high light.
Pi regeneration capacity
At low temperature, the capacity of sucrose synthesis is generally observed to be the predominant limitation over the Pi regeneration capacity (Leegood & Edwards 1996; Strand et al. 1997, 1999), although in potato a restriction in the chloroplast phosphoglucose isomerase causes starch synthesis to be the main limitation (Sharkey & Vassey 1989). The high Q10 for sucrose synthesis occurs because of the direct thermal sensitivity of the rate-limiting steps in the pathway, and temperature effects on the sensitivity of sucrose phosphate synthase (SPS) and FBPase to regulatory molecules (Leegood & Edwards 1996). In spinach, SPS has a Q10 of 2.4 in the light, and cytosolic FBPase showed high thermal sensitivity in its response to effectors; inhibition of FBPase caused by Pi fell threefold, fructose-2,6-bisphosphate inhibition fell fourfold and AMP inhibition declined 30-fold between 8 and 35 °C (Stitt & Grosse 1988). More triose-P was also needed to activate FBPase at low temperature in spinach (Stitt & Grosse 1988).
At the thermal optimum, the activation state of Rubisco declines in response to shading or elevated CO2 (Perchorowicz, Raynes & Jensen 1981; Kobza & Seemann 1988; Sage et al. 1988; Woodrow & Berry 1988). In both cases, Rubisco deactivation is considered a regulatory response to a shift in limitation away from Rubisco capacity to either RuBP or Pi regeneration capacity (Mott et al. 1984; Sage 1990; von Caemmerer & Quick 2000). The identification of Rubisco activase as the main regulatory protein for Rubisco provided a mechanistic explanation for the patterns of Rubisco deactivation in response to shade or CO2 enrichment (Spreitzer & Salvucci 2002; Portis 2003). Rubisco activase consumes ATP and reducing power in a reaction sequence that frees tightly bound phosphorylated sugars from Rubisco catalytic sites. Removal of the sugar phosphates allows for spontaneous carbamylation, which then activates the catalytic site, or frees the carbamylated catalytic site of bound inhibitors. ADP is an important inhibitor of activase, and activase requires ATP and reducing power to exist in its most active configuration. Active activase typically occurs in aggregates of 8 to 16 subunits, which are assembled from inactive dimers and monomers using ATP and reducing power in a thioredoxin-dependent reaction (Zhang & Portis 1999; Zhang et al. 2002).
The elegance of the activase lability hypothesis led to its rapid and widespread acceptance, and efforts are now underway to improve heat tolerance of photosynthesis by enhancing the thermal tolerance of activase (Spreitzer & Salvucci 2002; Zhu et al. 2005; Wu et al. 2006). We are not convinced, however, that heat lability of Rubisco activase always explains the decline in A above the thermal optimum, largely because no study has demonstrated activation state limitations using photosynthetic models that have parameterized electron transport capacity and Rubisco capacity in its fully active and deactivated condition. In all studies where an activation state limitation has been proposed, only the response of Rubisco-limited A was modelled, if any modelled analysis was conducted at all. Consequently, we cannot rule out an electron transport limitation on A, with the decline in the Rubisco activation state occurring as a regulatory response. By contrast, studies using models with electron transport algorithms demonstrate the thermal response of measured A at elevated temperature is consistent with the thermal response of electron transport-limited A (Bernacchi et al. 2001, 2003; Wise et al. 2004; Cen & Sage 2005; Hikosaka et al. 2006).
Proponents of the activase hypothesis claim to have ruled out the potential for electron transport limitations based on a rise in RuBP : PGA ratios, sustained ATP/ADP pools in leaves, a rise in non-photochemical quenching above the thermal optimum, and the presence of a CO2 stimulation of A at 10 mbar O2 and high temperature (Woodrow & Berry 1988; Law & Crafts-Brandner 1999; Portis 2003; Crafts-Brandner & Salvucci 2004; Salvucci & Crafts-Brandner 2004a; Kim & Portis 2005). The RuBP : PGA increase is perhaps the best evidence in support of an activase limitation; however, RuBP : PGA can increase when RuBP or Pi regeneration capacities are limiting because of deactivation of Rubisco and other Calvin cycle enzymes (Mott et al. 1984; Sage et al. 1988; Schrader et al. 2004). To evaluate the true limitation, the response of RuBP pools should be followed immediately after a rapid increase in temperature. The evidence here is uncertain, with reports showing both a decline (Schrader et al. 2004) and a rise in RuBP pools (Crafts-Brandner & Salvucci 2004) following a sudden increase in temperature. The evidence for a rise in non-photochemical quenching is not definitive, because the heat-induced diversion of PSI complexes from linear to cyclic electron transport would maintain high ATP/ADP and a high pH gradient, thereby promoting non-photochemical quenching at PSII (Sharkey & Schrader 2006).
The proposal that a CO2 stimulation of A at high temperature and low O2 is evidence of an activase limitation should be evaluated with theoritical models of photosynthesis to rule out other possible limitations. The importance of a modelled assessment is demonstrated in Fig. 5, which presents an A/Ci response measured for cotton at 42 °C and 10 mbar O2 (table 5 in Crafts-Brandner & Salvucci 2004). We modelled the A/Ci response corresponding to these data using the Rubisco activity reported for cotton by Crafts-Brandner & Salvucci (2004) and a Jmax value estimated from the gas exchange curve in Fig. 5. No deactivation of Rubisco was assumed in our model. Crafts-Brandner & Salvucci (2004) argued that because A in cotton at 42 °C and low O2 was stimulated by CO2 enrichment, RuBP regeneration capacity could not be limiting, given that photorespiration would have been nil. Instead, they reasoned, A had to reflect a limitation in the Rubisco activation state. As shown in Fig. 5, our modelled Rubisco-limited response matches the observed A/Ci response below 450 µbar, where A is CO2 sensitive. The modelled RuBP regeneration rate matches the observed A above 450 µbar, where A is CO2 insensitive. This analysis demonstrates that at 42 °C and 10 mbar O2, cotton is not limited by deactivation of Rubisco.
Recently, Cen & Sage (2005) examined limitations on A above the thermal optimum using sweet potato. They hypothesized that if the activation state of Rubisco were regulated in response to limitations in RuBP regeneration capacity at elevated temperature, then any perturbation in the ratio of RuBP regeneration to RuBP consumption capacity would alter the activation state of Rubisco in a predictable manner. By contrast, if the heat lability of Rubisco activase were limiting A, then the activation state would decline regardless of changes in RuBP regeneration/consumption. Lowering CO2 causes a rise in RuBP regeneration/consumption, and eventually allows Rubisco capacity to become limiting (Fig. 6a). If Rubisco were deactivated at high temperature and high CO2 in response to a limitation in electron transport capacity, then CO2 reduction should reactivate Rubisco. In sweet potato, the activation state of Rubisco markedly declined above 30 °C at a Ci of 500 µbar, and above 35 °C at current CO2 levels (Fig. 6b). Deactivation was also apparent below 15 °C at each of these CO2 levels. At low CO2 (a Ci of 140 µbar), the activation state recovered to near maximum values at high and low temperatures, consistent with modelled predictions that reactivation would occur when RuBP regeneration capacity was limiting. Lowering CO2 also increases the activation state of Rubisco in heated cotton leaves (Crafts-Brandner & Salvucci 2000). The slight deactivation that was observed in sweet potato at high temperature in the low CO2 treatment corresponded to a predicted limitation in electron transport capacity in the low CO2 conditions above 40 °C (Cen & Sage 2005). Notably, in the low CO2 region where Rubisco capacity was predicted to be limiting (see Fig. 4), Cen & Sage (2005) were able to effectively model the observed initial slope of A using Vcmax values determined in vitro. Furthermore, the observed RuBP pool declined slightly with rising temperature, which is consistent with an RuBP regeneration limitation on A above the thermal optimum (Cen & Sage 2005).
The discrepancies between the results of Cen & Sage (2005) and prior studies proposing an activase limitation on A at high temperature may simply reflect species variation in the sensitivity of activase to elevated temperature. Sweet potato is relatively heat tolerant as it is a warm-season crop, and thus may have a heat-stable form of activase. Spinach is not heat tolerant, and exhibits contrasting responses that are consistent with an activase limitation. For example, RuBP pools rise with temperature in spinach (Weis 1981a), and its initial slope of the A/Ci response rapidly declines above the thermal optimum of A in concert with a loss of Rubisco activation (Yamori et al. 2006a). Alternatively, the apparent heat lability of activase may reflect a regulatory response to declining electron transport capacity. This possibility is supported by results that show increased heat stability of the active form of activase at high ATP levels and redox status (Crafts-Brandner, van de Loo & Salvucci 1997; Portis 2003). In tobacco extracts, the addition of ATPγS increased the temperature where activase formed insoluble aggregates from 37 to 45 °C (Salvucci et al. 2001). One possibility that has not been fully evaluated is that Rubisco deactivation between the thermal optimum and 40–42 °C may be a regulatory response to a limitation in electron transport capacity, while above 42 °C, denaturation of activase may reduce the activation state of Rubisco to such a degree that it directly limits A. Cen & Sage (2005) did not evaluate this possibility because they examined responses below 42 °C.
ACCLIMATION OF C3 PHOTOSYNTHESIS TO CHANGES IN GROWTH TEMPERATURE
Most plant species are able to acclimate to changes in growth temperature by modifying the photosynthetic apparatus in a manner that improves performance in the new growth environment. In general, following a change in growth conditions by 5–10 °C, the thermal optimum shifts in the direction of the new growth temperature, and the rate of A increases at the growth temperature while the rate on the other side of the thermal optimum decreases (Fig. 7) (Regehr & Bazzaz 1976; Berry & Björkman 1980; Mawson et al. 1986; Yamori et al. 2005). Variations are apparent depending on growth conditions and species. For example, A often declines at all measurement temperatures in C3 plants grown at temperatures approaching their high or low temperature limit (Kemp & Williams 1980 for Agropyron smithii grown at 20 versus 35 °C days; Kubien & Sage 2004b for Calamagrostis canadensis grown at 26 versus 14 °C days). Furthermore, acclimation that shifts the thermal optimum appears to require shifting growth regimes to non-optimal temperatures. In C. album (Sage et al. 1995) and bean (Cowling & Sage 1998), the thermal optimum was unaffected between plants grown slightly below (23–25 °C) and above (34–36 °C) the thermal optimum of A. Variation in the acclimation response caused by species differences may reflect the degree of thermal specialization (Osmond, Björkman & Anderson 1980). In the genus Plantago, fast-growing species from lowland environments show greater acclimation than slow-growing species from high elevations, indicating that specialization for extreme environments may restrict the potential for thermal acclimation (Atkin et al. 2006).
An important component of thermal acclimation is an alteration in membrane composition to shift the thermal range where membranes are fluid yet stable towards the growth temperature. Between 20 and 40 °C, this is accomplished by altering the ratio of saturated to unsaturated fatty acids in the lipid matrix (Huner 1988; Mikami & Murata 2003). Saturation of fatty acids increases the number of hydrophobic interactions between adjacent fatty acids, thereby increasing the rigidity of the membrane (Hochachka & Somero 2002). For this reason, increasing saturation of fatty acids is considered an acclimation response to higher temperatures and is associated with increased thermotolerance in a wide range of species (Sharkey & Schrader 2006). The cost of increased saturation of fatty acids could be a loss of fluidity at lower temperatures, however, with a potential for reduced stability and turnover of membrane-bound enzymes (Mitchell & Barber 1986).
Acclimation associated with changes in fatty acid composition is relatively slow, requiring many hours to a few days to be effective. As such, this mechanism of acclimation is too slow to response to rapid changes in leaf temperature over the course of the day, as may occur in response to sudden enhancements in light level or a reduction in wind speed (Wise et al. 2004; Sharkey 2005). Over the short term (minutes to hours), membrane fluidity is stabilized by the rapid production of zeaxanthin and small molecules that increase membrane hydrophobicity, such as isoprenes (Sharkey 2005). Zeaxanthin levels are stimulated by the high pH gradient associated with induction of cyclic photophosphorylation by heat (Tardy & Havaux 1997). Zeaxanthin increases the rigidity of the thylakoid membrane in addition to serving a photoprotective role, and may thus help maintain membrane integrity and electron transport capacity above the thermal optimum (Havaux 1998).
The rate of respiration in the light also acclimates to rising temperature. Respiration rates generally rise with measurement temperature; however, day respiration declines in many species following growth at elevated temperature, and may approach the rate observed at the original growth temperature (Atkin et al. 2005, 2006). This change affects the thermal response of A, particularly in plants with thick leaves and low photosynthetic capacities, such as conifers (Way & Sage, unpublished data). In such cases, the respiration rate must be estimated and factored out (e.g. by evaluating responses of gross photosynthesis) if one is to evaluate acclimation of the photosynthetic biochemistry. For detailed discussion of the thermal acclimation of respiration, see Atkin et al. (2005).
Mechanisms of photosynthetic acclimation to temperature variation
The rise in A during acclimation to low temperature is rarely associated with just changes in Pi status and sucrose synthesis capacity, which complicate the understanding of the predominant limitations on A after acclimation (Leegood & Edwards 1996; Strand et al. 1999). A general rule is that at low temperature, the kcat of enzymes is slowed substantially because of a reduced fraction of enzymes having sufficient internal energy to meet activation energy thresholds for catalysis (Hochachka & Somero 2002). To compensate, plants can increase enzyme levels or shift protein expression to produce isoforms with improved performance at low temperature. The former commonly occurs, and there is some evidence that the latter can also be of significance. At lower growth temperature (4/2 °C day/night), winter rye expresses an isoform of Rubisco with higher affinity for CO2 than warm- (25/20 °C) grown rye (Huner & Macdowall 1979). In spinach, an isoform of Rubisco expressed in warm growth conditions (30/25 °C) has greater specificity at elevated temperature, but a lower specificity at cooler temperatures than an isoform expressed in plants grown at 15/10 °C (Yamori et al. 2006b). The warm-grown isoform also exhibits a break in the Arrhenius curve at 15 °C, indicating that its activation energy is increased at low temperature. Consistently, CO2 compensation points differed between plants expressing the two isoforms, being greater at high temperature and lower at low temperature in the cool-grown plants (Yamori et al. 2006b).
In addition to the enzymes of sucrose synthesis, Rubisco activity often increases during acclimation to cooler temperatures as seen in Arabidopsis (Strand et al. 1997), winter rape (Hurry et al. 1995b), rye (Hurry et al. 1995a), spinach (Holaday et al. 1992; Yamori et al. 2005) and wheat (Hurry et al. 1995b; Savitch et al. 1997). Calvin cycle enzymes (d-glyceraldehyde-3-phosphate dehydrogenase, aldolase, transketolase) also increase during cold acclimation in Arabidopsis (Strand et al. 1999). Whether these differences affect patterns of limitation depend upon whether the increase is disproportional, and here the record is less clear. Recently, Yamori et al. (2005) provided a detailed assessment of thermal acclimation in spinach. In their plants, the leaf N content was approximately 50% greater in leaves grown at 15/10 °C versus those produced at 30/25 °C. Associated with this was a near doubling in Rubisco and cyt f levels, and a near threefold increase in leaf mass per area. These changes did not correlate with increased Rubisco capacity in vivo, as the maximum value of the initial slope was similar in the two growth conditions. Instead, the CO2-saturated rate of A increased in concert with a rise in cyt f content, and the electron transport capacity rose relative to the Rubisco Vmax. In rice, a warm season grass, there was no change in Rubisco or leaf N content with a shift in growth temperature from 30/23 to 18/15 °C (Makino et al. 1994). At a given leaf N level, the content of cyt f and the coupling factor were unchanged between the temperature treatments, while the content of light-harvesting complex II (LHCII) and chlorophyll decreased in the cool-grown plants. This indicates a reduction in electron transport capacity in cool- versus warm-grown rice. With the rise in FBPase and SPS, it appears that cool-grown rice has reallocated its internal resources to boost Pi regeneration capacity at low temperature at the expense of the electron transport capacity.
A common (but not universal) acclimation response to changes in growth temperature is for the electron transport curve to shift towards the new growth temperature in a pattern that is often identical to the shift in A, particularly when measured at elevated CO2. In winter wheat, the thermal optimum of CO2-saturated A of 15 °C grown leaves was 15–20 °C, as was the rate of whole chain electron transport (Yamasaki et al. 2002). At a growth temperature of 35 °C, the thermal optimum for both A and whole chain electron transport in winter wheat had shifted to near 35 °C. The optimum for electron transport through PSII shifted in concert with A and whole chain electron transport; however, electron flow through PSI did not show a shift in the thermal optimum with changing growth temperature (Yamasaki et al. 2002). In the subpolar species S. cernua, a reduction in growth temperature from 20 to 10 °C caused a decline in the thermal optimum of A from near 20 to 10 °C (Mawson et al. 1986). This was accompanied by a reduction in the thermal optimum of whole chain electron flow, and electron flow through PSII, from near 25 to 10 °C (Mawson & Cummins 1989). As observed with wheat, electron flow through PSI in S. cernua did not correlate with the changed thermal response of A or whole chain electron transport. Winter rye also shows a large increase in whole chain electron transport at low temperature following a reduction in growth temperature from 20 to 5 °C; this increase corresponds to increases in the rate of electron transport through PSI in the cold-acclimated plants (Huner 1988; Huner et al. 1993). In two species from hot, dry climates, Larrea divaricata (Armond, Schreiber & Björkman 1978) and Nerium oleander (Badger et al. 1982), growth temperature (45 versus 20 °C days) did not affect the thermal optimum of whole chain electron transport, while the thermal optimum of A was reduced about 10 °C. However, electron transport capacity was stimulated below 35 °C in the cool relative to the warm-grown plants. In pea, electron transport capacity increased across the measurement range (4–20 °C) in plants grown at 7 °C relative to 17 °C, apparently because of an increase in turnover rate at a step between the photosystems (Mitchell & Barber 1986). Mitchell & Barber (1986) suggest that changes in membrane fluidity at low growth temperature accelerate interactions between cyt f and the plastoquinones, allowing for the rise in the electron transport capacity in the cold-grown plants.
In plants transferred to hot conditions, a different isoform of Rubisco activase can be produced that confers heat stability. In spinach, a 45 kDa isoform is produced by alternative splicing in hot conditions, while in mild conditions, a 41 kDa isoform is the only form synthesized (Crafts-Brandner et al. 1997). The longer isoform is heat stable to 45 °C, while the shorter isoform is heat stable to about 32 °C in vitro. In cotton, heat stress promoted the synthesis of a 46 kDa isoform to complement existing 47 and 43 kDa forms (Law, Crafts-Brandner & Salvucci 2001). In wheat, heat acclimation of activase stability was associated with a slight reduction of a 46 kDa isoform, and a large increase of a 42 and 41 kDa isoform (Law & Crafts-Brandner 2001). Mixtures of the isoforms in these species enhance thermal stability of activase activity in vitro (Portis 2003). In contrast to the species with multiple isoforms differing in heat sensitivity, tobacco only produces one isoform (Salvucci et al. 2001), while Arabidopsis produces two isoforms of similar thermal sensitivity (Kallis, Ewy & Portis 2000). Given that spinach has a large acclimation response of A (Yamori et al. 2005, 2006b), it would be interesting to know if the thermal acclimation response of tobacco is constrained by having only one isoform of activase.
The diffusion of CO2 into the leaf and chloroplast is directly dependent on temperature via diffusivity effects, stomatal control, solubilization and membrane permeability. Stomatal responses to temperature will not be covered in detail here, as they are highly variable and would require a lengthy review to provide a cogent synthesis. Depending upon species and growth conditions, stomata can open with rising temperature (a common response when vapor pressure deficit is low), close (often in response to increasing vapor pressure deficit with rising temperature) or remain unaffected (Kemp & Williams 1980; Monson et al. 1982; Tenhunen et al. 1984; Sage & Sharkey 1987; Santrucek & Sage 1996; Cowling & Sage 1998; Yamori et al. 2006a). It is worth noting that the sensitivity of A to variation in stomatal conductance generally increases at warmer temperatures, because the biochemical controls over A at high temperature are more sensitive to changes in Ci. If A is Pi-regeneration limited at cooler temperatures, large changes in stomatal conductance have little affect on A; by contrast, a Rubisco limitation on A at elevated temperature creates a steep A/Ci response such that substantial changes in A would result from small changes in stomatal conductance. For this reason, stomatal limitations are generally greater at elevated temperature, regardless of the stomatal response (Sage & Sharkey 1987; Hendrickson et al. 2004a).
Mesophyll conductance (the conductance of CO2 from the stomata to the chloroplast stroma) affects stromal CO2 levels and as such, potentially contributes to the thermal response of A. Mesophyll conductance is highly sensitive to temperature, showing a Q10 near 2 below the thermal optimum, and a thermal optimum that is similar to the thermal optimum of A (Bernacchi et al. 2002; Warren & Dreyer 2006; Yamori et al. 2006a). This high Q10 indicates that the temperature response of mesophyll conductance is largely controlled by proteins (Yamori et al. 2006a). Both aquaporins and carbonic anhydrase are known to be important facilitators of CO2 entry into the cell and chloroplast (Evans & Loreto 2000; Terashima et al. 2006; Yamori et al. 2006a). Unlike the response of mesophyll conductance below the thermal optimum of A, above the thermal optimum, there is no clear pattern. Substantial reductions in mesophyll conductance at elevated temperature are reported for tobacco and cool-grown spinach (Bernacchi et al. 2002; Warren and Dreyer 2006; Yamori et al. 2006a), but little if any reductions occur in oak and warm-grown spinach between 25 and 35 °C (Warren & Dreyer 2006). No study has examined the response of mesophyll conductance above 40 °C, so its contribution in the thermal range where electron transport or Rubisco activase may assert control is unknown. If mesophyll conductance substantially declines above 40 °C, then it could be a major limitation on A at elevated temperatures.
Mesophyll conductance also acclimates to variation in growth temperature. In rice, plants had a threefold higher mesophyll conductance when grown at 32 °C relative to 25 °C (Makino et al. 1994). In spinach, the mesophyll conductance increased in warm-grown plants (30/25 °C) relative to cool grown (15/10 °C), but only at measurement temperatures above 20 °C; plants from both growth conditions had the same thermal response of mesophyll conductance below 20 °C (Yamori et al. 2006a).
TEMPERATURE RESPONSE OF C4 PHOTOSYNTHESIS
In C4 plants, photosynthetic carbon assimilation reflects the activity of Rubisco in vivo, as it does in C3 plants. However, because Rubisco is enclosed in a sealed compartment where CO2 is concentrated to near-saturating levels, the dynamics of photosynthetic limitation differ. At low CO2, A is theoretically predicted to be limited by the capacity of phosphoenolpyruvate (PEP) carboxylase (PEPCase) to fix bicarbonate for movement into the bundle sheath as part of a C4 acid (von Caemmerer 2000). PEPCase activity isindependent of temperature at low CO2 (Laisk & Edwards 1997). Hence, the initial slope of the CO2 response of A in C4 plants is largely insensitive to temperature, except in chilling conditions where injury occurs (Fig. 8) (Long & Woolhouse 1978; Sage 2002). Here, the initial slope may decline, possibly reflecting increases in the activation energy of PEPCase at low temperature (Pittermann & Sage 2000; Kubien & Sage 2004b) or chilling lability of PEPCase in vivo (Krall & Edwards 1993; Matsuba et al. 1997). Oxygenase activity is largely suppressed in C4 plants, and as a result, the CO2 compensation point does not rise significantly with temperature as it does in C3 plants. This, in combination with the thermally insensitive response of the initial slope, causes the temperature response of A at low CO2 to be relatively flat in C4 plants (Sage 2002). This commonly occurs at low atmospheric CO2 levels corresponding to late-Pleistocene (180 µbar) and pre-Industrial (270 µbar) times. To exhibit high thermal sensitivity, the photosynthesis rate in C4 plants has to be above the CO2 saturation point (Sage 2002).
As with C3 plants, the CO2-saturated plateau of the C4A/Ci response rises with temperature up to the thermal optimum that is typically between 38 and 45 °C (Fig. 8) (Pittermann & Sage 2000; Sage 2002). This plateau can be determined by numerous limitations: Rubisco capacity, RuBP regeneration capacity, PEP regeneration as controlled by pyruvate–Pi dikinase capacity and Pi regeneration capacity (von Caemmerer & Furbank 1999; von Caemmerer 2000). At cooler temperatures (<20 °C), the CO2-saturated plateau is generally limited by Rubisco capacity in chilling-tolerant C4 species (Sage 2002; Kubien et al. 2003; Kubien & Sage 2004b). C4 plants are predisposed to encounter Rubisco limitations in cooler conditions because they have low Rubisco content compared to C3 plants of similar ecological habit (about 20–35%) (Sage, Pearcy & Seemann 1987; Sage 2002). At elevated temperature, warm temperature stimulates the Rubisco turnover rate enough to allow Rubisco capacity to be non-limiting; at cool temperatures by contrast; the turnover rate is comparatively sluggish, and given the relatively low enzyme content, Rubisco capacity is commonly limiting for A (Fig. 8) (Sage 2002; Kubien et al. 2003). Above 20–25 °C, Rubisco capacity becomes non-limiting; but it is unclear what process is the principle limitation on A. If the operational Ci in air has shifted to the initial slope region of the A/Ci curve at warmer temperatures, the limitation over the temperature response of A could be PEPCase capacity. Above the CO2 saturation point, PEP regeneration and electron transport capacity are leading possibilities for controlling A above 25 °C (Pittermann & Sage 2001; Crafts-Brandner & Salvucci 2002; Sage 2002; Kubien et al. 2003; Dwyer et al. 2007). At supraoptimal temperatures, the activation state of Rubisco may become limiting in C4 plants, as indicated by recent data from maize that shows a parallel decline in the activation state of Rubisco and A at high temperature (Crafts-Brandner & Salvucci 2002). As in C3 plants, however, this response may also reflect a regulated reduction in the Rubisco activation state in response to a limitation elsewhere in the photosynthetic apparatus (Dwyer et al. 2007).
C4 photosynthesis has been suggested to be inherently chilling sensitive because of cold lability of the C4 cycle enzymes, notably PEPCase and pyruvate orthosphosphate dikinase (PPDK) (Long 1983; Leegood & Edwards 1996; Du, Nose & Wasano 1999a). This chilling sensitivity was hypothesized to explain the general exclusion of C4 plants from cold climates (Long 1983). While most C4 species are chilling sensitive as a result of being adapted to warm climates, it is now recognized that there are many chilling-tolerant C4 species that show no obvious failure of the C4 cycle enzymes at low temperature (Osmond et al. 1980; Krall & Edwards 1993; Simon & Hatch 1994; Matsuba et al. 1997; Du et al. 1999b; Pittermann & Sage 2000, 2001; Cavaco, de Silva & Arrabaca 2003; Naidu et al. 2003; Kubien & Sage 2004a,b). C4 grasses and dicots occur in the high alpine (>3800 m), and in boreal fens where summer chilling is routine (Long 1999; Sage, Wedin & Li 1999; Kubien & Sage 2003). Alpine C4 species are frequently exposed to summer freezing, without apparent harm (Sage & Sage 2002; Sage unpublished data). The relatively low ceiling imposed on A by a limited Rubisco capacity in these species does appear to be an important consideration for their low temperature performance, however (Pittermann & Sage 2000, 2001; Sage & Sage 2002). In addition to directly limiting A, the low ceiling predisposes the C4 plants to photoinhibition, which necessitates a higher investment in photoprotection than seen in C3 plants (Long 1983, 1999; Kubien et al. 2003; Kubien & Sage 2004a; Sage unpublished results). C4 photosynthesis also becomes less efficient at low temperature because the Rubisco limitation causes the bundle sheath CO2 level to rise at low temperature, and a greater proportion of the CO2 in the bundle sheath leaks out of the leaf (Kubien et al. 2003; Kubien & Sage 2004b). As a result of these additional costs, C4 plants at high elevation and latitude have to escape the consequences of a cool climate by growing in microsites where solar heating allows for favourable daytime temperatures for photosynthesis (Sage & Sage 2002; Kubien & Sage 2003). The low frequency of C4 plants in cool climates is not likely the result of an inherent physiological intolerance at low temperature, but is instead a combination of lower quantum yield, a greater probability of high light stress that in turn increases photoprotection costs, a limited number of favourable microsites, and competition from C3 plants (Sage & Pearcy 2000).
Thermal acclimation of C4 photosynthesis
C4 photosynthesis shows three general patterns of thermal acclimation. The first is observed in generalist species such as found in warm desert Atriplex shrubs that are active year-round. Here, the breadth of the thermal response and the values of A at the thermal optimum change a little with variation in growth temperature; instead, the thermal optimum shifts in the direction of the growth temperature (Osmond et al. 1980). The second response has largely been observed in C4 species from cool environments [Atriplex confertifolia, Bouteloua gracilis, Miscanthus, Muhlenbergia species, Paspalum dilatatum (Caldwell et al. 1977; Kemp & Williams 1980; Sage 2002; Cavaco et al. 2003; Naidu et al. 2003)], but can occur in summer active weeds [Atriplex rosea (Björkman & Pearcy 1971)]. Here, the thermal response of A changes little below measurement temperatures of 20 °C following growth in a cooler environment, while it changes at elevated temperature (see, e.g. Fig. 8). In the third type of response, A changes at all measurement temperatures following thermal acclimation; species with this type of response often increase A following transfer to cooler conditions that are not stressful [e.g. warm-season grasses and Atriplex sabulosa (Osmond et al. 1980; Dwyer et al. 2007)]. In contrast to these acclimation responses, there is also the response of chilling-sensitive C4 plants such as maize to low growth temperature. These species show a loss in photosynthetic capacity, membrane integrity is impaired and enzyme activity is reduced at all measurement temperatures following prolonged chilling (Long 1983, 1999; Naidu et al. 2003). These are considered to be stress responses that lead to a loss of vigour and possibly death.
The mechanistic explanation for acclimation in C4 photosynthesis remains unclear, although recent progress in our understanding of the biochemical controls over C4 photosynthesis allows for some partial evaluations. Because Rubisco imposes a strong limitation on A at low temperature, the most obvious means of thermal acclimation to suboptimal temperatures would be to enhance Rubisco content. Studies to date indicate some species do acclimate in this manner, while numerous others do not. Acclimation involving a change in Rubisco content occurs in the warm-season grasses Panicum coloratum and Cenchrus ciliaris, and the warm-season dicot Flaveria bidentis grown at 25/20 and 35/30 °C; plants grown at the cooler temperatures had 30–40% more Rubisco than plants grown at the warmer temperatures, which was associated with a rise in A below 25 °C (Dwyer et al. 2007). In C4Atriplex lentiformis from Death Valley grown at 23 °C days, Rubisco content was more than double the levels measured in plants grown at 43 °C, and this was associated with a near doubling of measured A at 10 °C (Pearcy 1977). In maize grown at 19 versus 31 °C, Rubisco activity was by 21% greater while PPDK and NADP–malic enzyme activities were 64–100% greater in the 19 °C-grown plants (Ward 1987). These responses are distinct from responses of warm-season C4 grasses in chilling conditions (10–15 °C), where Rubisco and other protein levels gradually decline following chilling exposure (Du et al. 1999a; Naidu & Long 2004).
In chilling-tolerant, cool-season C4 grasses, there is no change in Rubisco content nor A below 20 °C, indicating minimal acclimation capacity to cool conditions (Fig. 8) (Pittermann & Sage 2001; Cavaco et al. 2003; Naidu et al. 2003; Kubien & Sage 2004a). Instead, acclimation responses are mostly observed at the thermal optimum. Cool- and warm-grown plants of the boreal C4 grass Muhlenbergia glomerata exhibited identical A/T responses below 23 °C, while the warm-grown grass had 50% higher A at the thermal optimum and a 5 °C higher thermal optimum (Kubien & Sage 2004a). Miscanthus × giganteus, a cool-tolerant C4 grass from the mountains of Taiwan, shows the same A/T response below 20 °C in warm- and cool-grown plants, and an enhancement in A at the thermal optimum in the warm-grown plants (Naidu et al. 2003). Cool-grown Paspalum dilitatum also showed no change in the A/T response and Rubisco content below 20 °C compared to warm-grown plants, but had a higher A at the thermal optimum and less heat tolerance (Cavaco et al. 2003). The higher rate at elevated temperature in the cool-grown P. dilitatum was associated with greater total protein in the leaves; however, PEPCase activity was unchanged by the treatment (Cavaco et al. 2003).
The most comprehensive C4 acclimation study is by Dwyer et al. (2007) with three warm-season C4 species (C. ciliaris, F. bidentis and P. coloratum) grown at 25/20 and 35/20 °C. In these species, A was reduced at the thermal optimum, and the thermal optimum for A was about 1–3 °C higher, in the warmer-grown plants (Dwyer et al. 2007). In P. coloratum and C. ciliaris, A showed a pronounced decline above 35 °C in the plants grown at 25/20 but not 35/30 °C. In the cooler-grown plants, the increase in A at the thermal optimum was associated with increased Rubisco capacity, a rise in carbonic anhydrase activity, a rise in leaf nitrogen content and the production of leaves with higher leaf mass per area. The two grass species also had greater cyt f content in the cool-relative to the warm-grown plants, indicating that there may have been an increase in electron transport capacity that supported an increase in A at the thermal optimum. In all three species, PEPCase activity, PSII content and chlorophyll content were unchanged. Dwyer et al. (2007) did not examine Rubisco activation state or activase lability.
GLOBAL CHANGE AND THE TEMPERATURE RESPONSE OF PHOTOSYNTHESIS
With the rise in atmospheric CO2 and the associated warming of the global climate, it is becoming clear that plants of the future will face fundamentally different patterns of control over photosynthetic carbon gain. In the past 400 millennia, when CO2 levels ranged between 180 and 300 µbar (Sage & Coleman 2001), light-saturated photosynthesis would have largely been limited by Rubisco capacity in C3 plants and PEPCase capacity in C4 plants over much of the thermal range encountered during the growing season. At higher CO2 in a warmer world, Rubisco will be less limiting unless there is a strong, disproportional reduction in Rubisco content. If electron transport capacity is the predominant limitation in warm, high CO2 environments of the future, natural selection should favour species and genotypes with increased heat stability of membrane and proteins. Alternatively, if activase is a strong controller of A, then heat-stable forms of activase should be favoured. Through breeding and genetic engineering, humans could get a jump on climatic change by directly selecting for traits that will preadapt species to warmer, CO2-enriched environments. To do this effectively, however, we will need to clearly identify the main limitations on A above the thermal optimum and how they vary in natural and agricultural populations.
While much of the discussion of climatic change effects on plants has addressed heat effects, the greater effect will actually occur at the low-temperature end of the growing range. This is because most of the warming will occur at higher latitudes, during winter and at night (IPCC 2001). The moderation of winter temperatures in temperate to boreal latitudes will allow for longer cool growing seasons in spring and even winter, which will favour species adapted for photosynthesis in cooler conditions. In these situations, the ability of plants to avoid Pi regeneration limitations will be important. Paradoxically, if humans are to best exploit the opportunities and minimize the danger of global climatic change, it is important to understand the biochemical responses controlling A in cool conditions as well as increasingly warm conditions.
In summary, this review has assessed our understanding of the mechanisms controlling the temperature response of photosynthesis in land plants. While there is a good general knowledge of the potential limitations, major areas of uncertainty remain, particularly with respect to limitations at supraoptimal temperatures. In the immediate future, efforts should be made to clarify the limitations on A above the thermal optimum, in particular, the importance of electron transport versus lability of Rubisco activase. Better understanding is also needed of the main limitations on A following thermal acclimation to low temperature in C3 plants. Over the longer term, we need to develop a better understanding of the acclimation and adaptation potential of photosynthesis in natural populations. For example, the greatest warming will tend to affect the more thermally stressed environments, where highly specialized species live. Do these species generally lack the acclimation potential to allow them to persist in a warmer environment? With an improved understanding, we will be in a much better position to predict and mitigate the effects of global climatic change.