• The stomata of angiosperms respond to changes in ambient atmospheric concentrations of CO2 (Ca) in ways that appear to optimize water-use efficiency. It is unknown where in the history of land plants this important stomatal control mechanism evolved. Here, we test the hypothesis that major clades of plants have distinct stomatal sensitivities to Ca reflecting a relatively recent evolution of water-use optimization in derived angiosperms.
• Responses of stomatal conductance (gs) to step changes between elevated, ambient and low Ca (600, 380 and 100 µmol mol−1, respectively) were compared in a phylogenetically and ecologically diverse range of higher angiosperms, conifers, ferns and lycopods.
• All species responded to low Ca by increasing gs but only angiosperm stomata demonstrated a significant closing response when Ca was elevated to 600 µmol mol−1. As a result, angiosperms showed significantly greater increases in water-use efficiency under elevated Ca than the other lineages.
• The data suggest that the angiosperms have mechanisms for detecting and responding to increases in Ca that are absent from earlier diverging lineages, and these mechanisms impart a greater capacity to optimize water-use efficiency.
The global preoccupation with the steadily rising concentration of atmospheric CO2 has produced prolific research, which demonstrates impacts ranging in scale from individual leaves (Sage, 1994; Medlyn et al., 2001; Ainsworth & Rogers, 2007) to canopy (Wullschleger et al., 2002) and ecosystem levels (Betts et al., 1997; Betts et al., 2007). The first and most direct response of plants to CO2 perturbation occurs at the leaf surface, where stomatal aperture is sensitive to ambient CO2 concentration (Ca) as it affects the internal CO2 environment in the leaf (Mott, 1988). The sensitivity of stomatal guard cells to CO2 concentration appears to have evolved as a means of enabling leaves to optimize water use during photosynthesis (Cowan & Farquhar, 1977). As such, the ‘CO2 response’ of stomata is a centrally important process both in terms of land plant evolution, and for the hydrological balance of terrestrial ecosystems. Despite its central importance, our understanding of the feedback mechanism remains far from complete. In particular, current information about these processes is entirely based on studies of a number of derived eudicot angiosperms.
The evolutionary pathway that has led to the water-use-efficient stomatal control present in all eudicot angiosperms sampled to date is unknown (Morison, 1987). Given the apparent complexity of stomatal control physiology it seems plausible that the acquisition of different components of the guard cell ‘program’ may have been an incremental process over the 450 or so million years since stomata first evolved (Raven, 1977; Edwards et al., 1998). Recent evidence alludes to such a possibility because fundamental differences have been observed between the stomatal responses of the fern Adiantum capillaris-veneris and angiosperms to blue light (Doi et al., 2006). In addition, there is evidence that selection to increase the rate of stomatal response to water status of the leaf has led to changes in the configuration of the guard cell complex in grasses (Franks & Farquhar, 2007). Meta-analyses of the reams of gas exchange data from manipulated Ca experiments in chambers and free-air enrichment plots have suggested that conifers and angiosperm trees also show different CO2 sensitivity (Curtis & Wang, 1998; Medlyn et al., 2001). To date, however, there has been no broadly comparative analysis of how stomatal physiology may have changed over the course of evolution across a phylogenetically diverse sample of land plants.
The ratio of water lost relative to CO2 uptake during photosynthesis (water-use efficiency, or WUE) is directly related to the concentration of CO2 in the leaf substomatal cavity (Ci). Under standard conditions of vapour pressure and Ca, a low Ci corresponds to a high instantaneous WUE because stomatal conductance to water vapour (gs) is low relative to the rate of electron transport and CO2 fixation (A) (Von Caemmerer & Farquhar, 1981). Interestingly, unstressed plants tend to express a rather conservative range of Ci during steady-state photosynthesis (Wong et al., 1979; Yoshie, 1986; Franks & Farquhar, 2007). Conservative Ci suggests that the regulation of gas exchange by the stomata of vascular plants involves a common ‘program’ for optimizing the balance between photosynthesis and water lost from the leaf. As such, plants should be expected to regulate stomatal conductance to CO2 according to the RuBP regeneration capacity of the photosynthetic light reactions (Raschke, 1979; Messinger et al., 2006). The control mechanism responsible for matching photosynthetic capacity with stomatal aperture remains elusive, but appears to involve Ci as an instantaneous proxy for the ratio of gs : A (Farquhar & Sharkey, 1982). Among the relatively wide selection of woody and herbaceous angiosperms, changes in light, nutrients, temperature and oxygen all appear to induce a similar Ci mediated regulation of stomatal aperture suggesting a general strategy of maintaining optimal water use with respect to carbon gain (Cowan & Farquhar, 1977).
Here we contrast the stomatal response of different species to perturbations in Ca (and Ci) with the aim of finding where in land plant evolution water-use-efficient regulation of photosynthetic gas exchange first appeared. Our hypothesis was that major clades of plants should have distinct stomatal sensitivities to CO2 that reflect a relatively recent evolution of water-use optimization.
Materials and Methods
Fifteen species were selected to represent an evolutionary cross-section of vascular land plant diversity, including members of all the major vascular plant lineages and incorporating representatives from major clades within these lineages (Table 1). Within lineages, species were also selected to include a range of ecological types, including sun and shade plants, and species from very wet to very dry habitats (Table 1). Twelve species were each represented by three potted individuals housed in glasshouses of the School of Plant Science, University of Tasmania, Australia. These plants were acclimatized to controlled glasshouse conditions for c. 2 wk before gas exchange measurements in order to maximize their stomatal sensitivity to CO2 (Talbott et al., 2003). Some studies have suggested that low light and high humidity usually present in growth chambers attenuate the responsiveness of stomata to CO2 (Raschke, 1975) so plants were transferred to a glasshouse with close to natural light conditions and 50% relative humidity. Glasshouse conditions were 25°C during the day and 15°C at night; with a 16-h photoperiod. Plants received a maximum quantum photosynthetic photon flux density (PPFD) of 1300 µmol quanta m−2 s−1; and relative humidity was maintained by a dehumidifier coupled to a humidity probe. Two tropical species (Table 1) were also included and these were measured in situ in the field during the wet season in Costa Rica. Field conditions during the sampling period (11:00 h to 1400 h) were between 28°C and 33°C and 55% and 72% RH. Also, one fern species was sampled from field conditions in Knoxville, TN, USA during the summertime (July). Field conditions during the sampling period (11:00 h to 14:00 h) were 28–30°C and 55–68% RH.
Table 1. Experimental species, showing family, native habitat, mean assimilation rate and stomatal conductance (both at 380 µmol mol−1 CO2), stomatal sensitivity to a step change in Ca from 380 to 600 µmol mol−1, and stomatal closure half times in response to darkening leaves and severing leaves
A (µmol m−2 s−1)
gs (mol m−2 s−1)
CO2 sens. (%)
Stomatal closure half-time(s)
Species measured in the field in Knoxville, Tennessee.
A portable infrared gas analyser (Li-6400; Li-Cor) was used to determine stomatal conductance (gs) and CO2 assimilation rates (A) at different Ca. All other variables within the leaf chamber of the Li-6400 were standardized during measurements; leaf temperature was maintained at 20°C for glasshouse plants and 30°C for species sampled in the field, and PPFD at 1000 µmol quanta m−2 s−1. The light intensity used was sufficient to saturate photosynthesis in most species, but without damaging shade tolerant species. Vapour pressure deficit (VPD) was set between 1.2 kPa and 1.5 kPa and maintained within a range of ±0.1 kPa by manually tuning the proportion of inlet air directed through a desiccant column. Manual control of VPD enabled a standard air flow of 500 ml min−1 to be set for all measurements. The Ca in the leaf chamber was controlled for the duration of the experiment by a gas injection system (Li-6400-01; Li-Cor) regulating the concentration of CO2 in the air supply line.
Two leaves were sampled from each of the three individuals of each species. These leaves were measured at three CO2 concentrations: low (100 µmol mol−1, selected as the lowest workable Ca without risking damage to the leaf (Bunce, 2006); current ambient (380 µmol mol−1); and high (600 µmol mol−1). So as to avoid the possibility of patterned responses of stomata to non-CO2 signals, particularly circadian interactions, transitions between the three CO2 concentrations were randomized such that leaves of each species were exposed to all possible step transitions of increasing and decreasing Ca. Step changes between CO2 concentrations were applied in random sequences such that both the size and direction of the CO2 perturbation were non-sequential. Using this procedure, we focused on the absolute relationship between Ca and stomatal conductance rather than possible artefacts associated with different types of stomatal perturbation. At the commencement of measurements each day a single healthy, fully-expanded leaf was enclosed in the leaf chamber and allowed to equilibrate at the initial Ca. The gs, A and leaf environmental parameters were logged every 2 min. Leaves remained at each CO2 concentration until gas exchange parameters stabilized. Stability was defined as less than a 3% change over 8 min. Once stability was achieved, a second step change in Ca was applied, and the process repeated until three to four transitions had been made. After a maximum of four transitions the leaf was removed from the chamber and a new leaf inserted. Leaves were measured between 10:00 h and 15:00 h during which time it was generally possible to measure 8–10 transitions.
For each plant, maximum gs and A were taken from the initial steady-state measurements at a Ca of 380 µmol mol−1 and subsequent stomatal conductances were converted to unitless fractions of this initial state for comparison between species. The relative conductances at each Ca were averaged for each species and compared. Although gs was expressed relative to the initial steady state when Ca = 380 µmol mol−1, relative values of gs at 380 µmol mol−1 CO2 were taken from subsequent transitions and hence were not necessarily equal to 1. Relative sensitivity to CO2 was quantified as the linear regression between Ca and gs. The WUE at each Ca was calculated as the ratio of A : E, and relative changes expressed within replicate leaves to ensure minimal variation in VPD.
Stomatal kinetics test
The kinetics of stomatal responses to light and rapid desiccation were measured in all species to determine whether they followed similar patterns to those observed in response to perturbation of Ca. Light-to-dark transitions were carried out on two leaves per species, whereby gs was tracked after the chamber illumination was turned off while maintaining Ca at 380 µmol mol−1 and all other parameters were stable. Similarly, stomatal response to leaf desiccation was tested by allowing leaves to achieve steady gas exchange at 1000 µmol quanta m−2 s−1 then cutting leaves off at the petiole and logging the subsequent decline in gs.
For various physiological traits we tested the differences between groups (ferns and fern allies; conifers; angiosperms), between levels of Ca (100, 380 and 600 ppm) and the variation among species within groups. These analyses were executed using restricted maximum likelihood mixed model analyses of variance with groups and Ca as fixed effects in a fully factorial design and species nested within groups, and levels of Ca as a random effect. Since the interaction between group and Ca was highly significant for each variable, we ran separate analyses among levels of Ca for each group and among groups for each level of Ca, with species as a random effect nested within the fixed effect. All analyses were implemented in Proc Mixed of SAS 9.1 (SAS Institute Inc., Cary, NC, USA). The data used were plant means of the physiological characteristics. All data was checked for normality of residuals and homogeneity of variances. Fixed effects were tested using F-tests adjusted for multiple comparison effects using the Tukey–Kramer method, random effects were tested with likelihood ratio tests.
Apart from the gross differences among groups, these analyses do not take phylogenetic relationships among species into account. However, the results of the analyses can still be used to make valid inferences. In particular, very small species-within-group variance components (i.e. minimal variation among species) imply little phylogenetic contribution to the variation among species, which would make tests comparing groups valid for that variable. Also, the sole lycopod sampled (Selaginella pallescens) was grouped with the ferns, even though the resulting group is not monophyletic. However, the gas exchange parameters for S. pallescens were nested within those of the ferns, and exclusion of this species did not alter the pattern or statistical significance of the results.
Stomatal responses to changes in atmospheric CO2 concentration (Ca) were variable between species (Fig. 1). All species responded to a decline in Ca below ambient by increasing stomatal conductance (gs) relative to the gs value observed at an ambient Ca of 380 µmol mol−1 (gs380). Also, all species showed similar kinetics of the gs response, having a half-time of between 600 s and 1200 s. However, in angiosperm and conifer species, the increase in gs was larger (averages of 64% and 41%, respectively) than the response in the ferns and Selaginella (average of 25%; Fig. 2). The mean per cent responses of angiosperms and conifers) were not significantly different from each other (P > 0.05), but these groups were both significantly greater than the ferns and Selaginella (P < 0.05). In particular, there was no detectable variation among species within the group (variance component estimates of 0) in the responses to either 100 µmol mol−1 or 600 µmol mol−1, suggesting that the analysis of variance was not confounded by phylogenetic relationships within the groups.
In contrast to the effects of lowering Ca, increasing CO2 concentration to 600 µmol mol−1 produced a different pattern of responses, though once again the responses were clearly demarcated according to major lineages. Thus, the angiosperm species sampled were all highly sensitive to increased Ca, closing stomata such that gs600 was reduced by an average of 40% relative to gs380. The decrease in gs was proportional to the increase in gs observed at 100 µmol mol−1 indicating an approximately linear relationship between Ca and gs over the concentration range 100–600 µmol mol−1. However, the stomata in both conifers and the ferns and lycopods were significantly less responsive (P < 0.05) to increased Ca above 380 µmol mol−1 than those of angiosperms. Neither group showed a detectable decline in gs on average (Fig. 2). This relative insensitivity to Ca above 380 µmol mol−1 led to an asymmetrical relationship between Ca and gs whereby stomatal sensitivity to Ca declined close to zero as Ca as increased above the current atmospheric concentration of 380 µmol mol−1 (Fig. 2). Transitions from 100 µmol mol−1 to 600 µmol mol−1 yielded similar final gs as 380 µmol mol−1 to 600 µmol mol−1 transitions, and step changes in Ca from 600 µmol mol−1 to 380 µmol mol−1 and the reverse caused very weak or no change in gs.
Thus, the three groups showed three distinct patterns of sensitivity: angiosperm stomata had high sensitivity to both increased and decreased Ca; conifer stomata had moderately high sensitivity to reduced Ca but little sensitivity to greater Ca; and the stomata of ferns and the lycopod were insensitive to elevated Ca and showed weak responses to below present-day Ca.
The different sensitivities of stomata to Ca were not related to stomatal kinetic responses to desiccation and only weakly related to the light response. Half-times for stomatal closure during rapid desiccation ranged from 216 s to 1777 s (Table 1) and there was no difference in the means of each group (one-way anova on log-transformed data; P > 0.05). Stomatal closure in response to dark was approximately four times slower in the fern + Selaginella group (geometric mean = 3155 s) compared with angiosperms (856 s) and conifers (653 s; P < 0.05, one-way anova on log-transformed data). The slow response to light in ferns and Selaginella corresponded with their lack of response to elevated Ca, although there was only a weak correlation between half-time for stomatal closure in the dark and stomatal sensitivity to Ca (r2 = 0.25, P < 0.05).
WUE and assimilation
There was no difference in the mean assimilation or gs between groups, and neither of these measures of rates of gas exchange were significantly correlated with sensitivity to CO2 perturbation (Table 1). Furthermore there was large variation among the species within groups for both of these parameters. These results suggest that there was no significant ecological bias among groups in species selection, and that the groups did include high levels of physiological variation, so that the observed pattern is likely to apply across a wide range of ecological conditions. By contrast, changes in WUE were very distinct between groups, with angiosperms achieving a significantly larger increase in WUE at 600 µmol mol−1 than either conifers or ferns and Selaginella (100% versus 43% and 47%, respectively; P < 0.01; Fig. 3). The larger increase in WUE achieved in angiosperms was a result of pronounced reductions in gs at 600 µmol mol−1 CO2. These reductions resulted in significantly lower photosynthetic enhancement in angiosperms at elevated Ca compared with the ferns and lycopods (P < 0.05; Fig. 3). Low Ca produced similar effects on WUE and assimilation in all species and although angiosperms produced the largest mean change in WUE at Ca = 100 µmol mol−1 these differences were too small to be significant.
The instantaneous responses of gas exchange to changes in Ca in angiosperms differed significantly from those of conifers, ferns and Selaginella. In the taxa studied, angiosperm stomata were highly sensitive to both decreased and elevated Ca. However, conifers, ferns and Selaginella were minimally responsive to elevated CO2. These data suggest that the regulatory pathway responsible for matching water loss with instantaneous photosynthetic rate in angiosperm leaves may be absent from other major vascular plant lineages.
The outstanding feature of angiosperm stomata when compared here with those of other vascular plants was not an overall sensitivity to CO2, but rather the ability to engage a closing response to Ca above 380 µmol mol−1. Nearly all leaves demonstrated some degree of stomatal opening when Ca was lowered to 100 µmol mol−1, but only angiosperm leaves were highly responsive to the transition to super-atmospheric CO2 concentration (600 µmol mol−1). The lack of responsiveness of stomatal conductance in fern and Selaginella leaves to high CO2 is in agreement with a previous study that found the stomata of Adiantum capillis-veneris were insensitive to CO2 concentration(Doi & Shimazaki, 2008). The species-specific differences observed here were not related to the maximum conductivity of the stomata (Table 1) and, as a result of careful sampling to maximize photosynthetic and ecological overlap between groups, there was no group-related bias in gs. Neither was CO2 sensitivity related to cuticular conductance, since all of the species sampled possessed similar minimum conductances after stomatal closure (i.e., between 5% and 10% of maximum gs; unpublished). The rate of stomatal closure was not significantly correlated (P > 0.05) with the sensitivity of stomata to high CO2 although the ferns tended to be slower in closing after darkening the leaf (Table 1). Indeed, the kinetics of the opening response of fern and conifer stomata were very similar in dynamics to those of the angiosperms sampled when exposed to low Ca (Fig. 1). Given the absence of any interaction with cuticular conductance or stomatal kinetics, we propose that the different responsiveness of angiosperm stomata relative to nonangiosperms is caused by differences in the stomatal control process.
The angiosperm stomatal response to perturbation of CO2 whereby stomata open after a drop in Ca and close in response to an increase has been described many times, yet its physiological control remains poorly understood. Controversy on the CO2 response of gs surrounds the location as well as the signalling and transduction cascades associated with stomatal regulation relative to photosynthetic rate. Stomatal response to Ca has been variously attributed to a control signal within the guard cells generated from the balance of RuBP carboxylation and RuBP regeneration (Buckley et al., 2003) or, alternatively, to the epoxidation state of zeaxanthin (Zhu et al., 1998). However, recent experimental evidence suggests that the sensor for light and CO2 regulation of stomatal aperture resides in the mesophyll and is transferred to the epidermis, possibly by diffusion (Mott et al., 2008), though antisense manipulations of Rubisco and RuBP regeneration do not support this interpretation (Von Caemmerer et al., 2004; Lawson et al., 2008). Irrespective of the identity of the specific mechanisms involved in the CO2 sensing by angiosperms, our results provide strong evidence that these mechanisms are nonfunctional or completely absent in stomata of nonangiosperm taxa sampled here. Furthermore a recent study has shown that stomatal opening in the fern Adiantum capillis-veneris is insensitive to blue light, another important activator of stomatal opening in angiosperm species (Doi et al., 2006).
Together, these data portray nonangiosperm stomata as being more autonomous and less integrated into the functioning of the leaf than angiosperm stomata. The source of these functional differences may be the presence of a mesophyll to guard cell signal in angiosperms that is absent in other major lineages of vascular plants. This is in accord with the fact that the CO2 response in angiosperm stomata is absent in epidermes that have been isolated from the mesophyll (Mott et al., 2008). Furthermore, similar stomatal functionality in response to nonmetabolic stimuli such as soil and atmospheric drying in all sampled taxa (Table 1) highlights the specific peculiarity of the CO2-related closing response in angiosperms.
One observation that requires further explanation is that, despite an insensitivity to high Ca, the stomata of nonangiosperm species (Fig. 2) opened in response to low Ca, albeit to a somewhat lower degree in ferns and Selaginella. This suggests that there may be fundamental differences in the physiology of opening and closing responses of stomata to CO2 concentration. The functional advantages derived from the stomatal opening response to low Ca are rather different from the advantages of closing in response to high Ca, and it is plausible that the physiological pathways and evolution of these two responses are distinct. Opening stomata when Ci decreases is necessary to avoid the potentially damaging situation of an excess of photoreductive power without sufficient CO2 substrate, while stomatal closure as ci increases is associated with water saving. The adaptive pressures behind the evolution of these opposite responses to CO2 are very distinct, making independent evolution of distinct processes a distinct possibility.
The closing response of stomata to high Ca (sensed via Ci) is an essential component of the WUE regulation function in angiosperms (Farquhar & Sharkey, 1982). However the absence of a stomatal closure response to Ca in our sample of ferns and Selaginella indicates that these plants are likely to be incapable of optimizing the ratio of water lost to CO2 assimilated. This means that ferns and lycopods should have substantially lower WUE than angiosperms under conditions that depress assimilation (and elevate Ci), such as photoinhibition, and fluctuating light and temperature (Fig. 3). This prediction is consistent with observations of significantly greater 13C fractionation (low WUE) in tropical fern leaves relative to angiosperms (Watkins et al., 2007).The reduced capacity for WUE regulation in ferns is correlated with a high sensitivity and rapid closure of fern stomata in response to desiccation (Table 1). Ferns have also been shown to be more conservative than angiosperms in the timing of stomatal closure during water stress, which means that compared with angiosperms fern stomata close with a very large ‘safety margin’ before water stress-induced xylem failure (Brodribb & Holbrook, 2004). This suggests that a lack of Ci regulation in ferns is compensated by a high responsiveness to desiccation and humidity (Brodribb & Jordan, 2008). High stomatal responsiveness to drought puts ferns at a considerable disadvantage relative to angiosperms because their stomata are forced to close and productivity is lost under mildly desiccating conditions that are still favourable for the photosynthesis and growth of angiosperms.
We therefore propose that the lack of water use optimization in fern leaves, as shown by an absence of stomatal response to elevated Ca, represents a major functional constraint on ferns and lycopods when competing with angiosperms. In the light of recent data indicating that the gametophyte generation of ferns is often highly desiccation tolerant (Watkins et al., 2007), it is possible that the inefficiency of water management in the fern sporophyte may explain the confinement of virtually all homoiohydric ferns and lycopods to humid habitats (Dyer & Page, 1985). It should be noted that conifers are not so confined, despite sharing a low stomatal sensitivity to elevated Ca (Enright & Hill, 1995; Richardson & Rundel, 1998). Clearly, conifers and the few homoiohydric arid-adapted ferns have other features (such as the torus and margo structure in wood; Hacke et al., 2004) that allow them to compete with angiosperms in very dry environments.
It is interesting to note that differences in the CO2 sensitivity of fern and conifer stomata relative to angiosperms led to profound differences in the degree to which instantaneous WUE and assimilation change after an increase in Ca from a present-day concentration of 380 µmol mol−1 of CO2 to 600 µmol mol−1. Although the angiosperms sampled here were able to greatly enhance WUE via stomatal closure at 600 µmol mol−1 CO2, they did so at the expense of photosynthetic gain (Fig. 3). The stomatal closure response has the interesting consequence that fern and conifer assimilation rates benefit differentially more under enhanced CO2 than that of angiosperms. The implications of these responses integrated over evolutionary time are unknown, but it is pertinent to note that conifers, ferns and lycopods likely evolved under atmospheric Ca substantially higher than 600 µmol mol−1. By contrast, the greater part of the diversification of angiosperms almost certainly occurred under conditions of atmospheric Ca much closer to the present day (Robinson, 1994; Berner, 2006).
A key component of the CO2 response of angiosperm stomata is the apparent ability of some species to modify the responsiveness of stomata to CO2 when grown under different humidities (Frechilla et al., 2002; Talbott et al., 2003). The loss of CO2 sensitivity is a fascinating observation because it suggests that angiosperms are capable of shedding this water-conserving trait when water is in abundance. Such behaviour suggests an extraordinary dynamism in the way angiosperms regulate water loss, and provides evolutionary flexibility that may have contributed to the success of angiosperms in an extraordinarily diverse range of habitats, particularly in the current low CO2 world (Berner, 2006).
We thank Ian Cummings for excellent glasshouse facilities. This work was supported by ARC Discovery Grant DP0878342 (T.J.B. and G.J.J.) and NSF grant IOB-0714156 (T.S.F.).