Physiological framework for adaptation of stomata to CO₂ from glacial to future concentrations

Abstract

In response to short-term fluctuations in atmospheric CO₂ concentration, c_a, plants adjust leaf diffusive conductance to CO₂, g_c, via feedback regulation of stomatal aperture as part of a mechanism for optimizing CO₂ uptake with respect to water loss. The operational range of this elaborate control mechanism is determined by the maximum diffusive conductance to CO₂, g_c(max), which is set by the size (S) and density (number per unit area, D) of stomata on the leaf surface. Here, we show that, in response to long-term exposure to elevated or subambient c_a, plants alter g_c(max) in the direction of the short-term feedback response of g_c to c_a via adjustment of S and D. This adaptive feedback response to c_a, consistent with long-term optimization of leaf gas exchange, was observed in four species spanning a diverse taxonomic range (the lycophyte Selaginella uncinata, the fern Osmunda regalis and the angiosperms Commelina communis and Vicia faba). Furthermore, using direct observation as well as flow cytometry, we observed correlated increases in S, guard cell nucleus size and average apparent 1C DNA amount in epidermal cell nuclei with increasing c_a, suggesting that stomatal and leaf adaptation to c_a is linked to genome scaling.

1. Introduction

As an important greenhouse gas and the substrate for photosynthesis, atmospheric CO₂ concentration, c_a, is one of several important components in the feedback interaction between vegetation and the atmosphere [1,2]. Throughout the history of plants (approx. 400 Myr) on land, c_a is estimated to have fluctuated from as high as several thousand ppm in the Devonian [3], to as low as around 180 ppm in the Last Glacial Maximum (LGM) [4]. Currently, it stands at approximately 390 ppm (National Oceanic and Atmospheric Administration data) but could exceed 1000 ppm by the end of this century according to Intergovernmental Panel on Climate Change estimates. Pre-historic CO₂ fluctuations have been linked to alternating greenhouse/icehouse global climates [3,5] as well as major episodes of plant evolution and diversification [6,7]. Along with current concerns about the impact of future elevated atmospheric CO₂ concentrations on global climate, the direct effects of the ongoing rapid rise in c_a on plant photosynthetic productivity and water use is likely to be significant [1,8,9].

Changes in c_a have a direct and immediate effect on the rate of CO₂ assimilation [10–13]. To counteract or mitigate these effects in the short term (seconds to minutes), plants employ a physiological feedback mechanism involving interaction between the diffusive conductance of the leaf epidermis, governed by stomata, and the rate of photosynthesis occurring in the mesophyll (figure 1) [14–16]. This short-term feedback mechanism controls the operational stomatal conductance to CO₂, g_c(op), at a value between zero and the maximum, g_c(max). However, it is the physical attributes of the stomata that determine g_c(max) (figure 2). At any given average guard cell turgor pressure P_g, g_c(op) will be a function of the average stomatal aperture, a, the depth of the stomatal pore, l, and the number of stomata per unit area (or density, D) [17,18]. Ultimately, as P_g approaches its maximum and the a versus P_g relationship saturates, average stomatal aperture will reach its maximum, a_max (figure 2a), causing g_c(op) to reach its maximum, g_c(max) (figure 2b). Because the overall size of stomata determine both a_max and l, g_c(max) can be described conveniently as a function of stomatal size, S, and D [19,20]. When g_c(max) adapts to new environmental conditions (figure 2c), it necessarily involves adaptation of S and/or D [19,20].

Figure 1.

A schematic of the physiological feedback loop for stomatal response to CO₂. The arrows trace the feedback signal flow following a change in atmospheric CO₂ concentration, Δc_a, showing the effect on CO₂ assimilation rate, A, via the stomatal response. The sloped lines in boxes represent the sensitivities of the key variables involved, indicating a generally positive or negative relationship between them: A, CO₂ assimilation rate; c_i, leaf intercellular CO₂ concentration; g_c, stomatal conductance to CO₂. Overall, the final change in assimilation rate, ΔA_(final), is the sum of the initial change occurring without any stomatal response, ΔA_(initial), plus the change due to stomatal response, ΔA_(stomatal). The negative relationship between g_c and c_i (box with dashed line border in the feedback loop) indicates that ΔA_(stomatal) is opposite in sign to ΔA_(initial).

Figure 2.

Guard cell pressure drives stomatal aperture and stomatal conductance. (a) Stomatal aperture, a and, consequently, (b) stomatal conductance to CO₂ diffusion, g_c, increase with guard cell pressure P_g in a saturation fashion. Stomatal aperture control is achieved through osmotic regulation of P_g. The operational g_c, g_c(op), is typically regulated via P_g within the high sensitivity region of the curve (double arrow). Maximum conductance, g_c(max), is determined by the number and size of stomata (see equation (2.1)). Data are for Tradescantia virginiana, adapted from Franks & Farquhar [17]. (c) Illustrating how altering g_c(max) keeps g_c(op) in the high sensitivity region of the P_g versus g_c curve. Starting at the initial operating point (point a), a sustained change in c_a from ambient (amb.; solid line) to elevated CO₂ concentration results in adjustment of g_c to a lower g_c(op) (point b). The plant then changes stomatal size and density in new leaves, altering the g_c versus P_g curve (dashed line, elevated CO₂; elev.), to return g_c(op) to its optimal position (point c). Similarly, following a sustained shift from ambient to subambient c_a, the plant adjusts g_c to a higher g_c(op) (point d). New leaves are produced with altered stomatal size and density, shifting to a new g_c versus P_g curve (dotted line, subambient CO₂; sub.) and increasing g_c(max) to return g_c(op) to its optimal position (point e). Curves in (c) are representative, based on (b).

(a). Stomatal conductance and CO₂

Sustained exposure to elevated or subambient c_a invokes feedback regulation of the biochemical capacity for CO₂ assimilation [21–26]. Over a gradient of growth c_a treatments from subambient to elevated CO₂ concentration, the maximum rate of carboxylation, V_cmax, typically declines [24,26–28], consistent with a mechanism of adaptive feedback regulation of photosynthesis via adjustment of the amount of key photosynthetic enzymes [21,23]. Regulation of CO₂ assimilation rate under changing c_a therefore involves stomatal and non-stomatal feedbacks operating over several timescales. However, despite significant advances in understanding the non-stomatal feedbacks, much less is known about the long-term response of stomata to CO₂ (weeks to millennia).

Our hypothesis is that over the long term, the leaf epidermis adapts to keep g_c(op) and g_c(max) aligned optimally with the leaf's biochemical capacity for carbon assimilation under the prevailing c_a. Taking the short-term feedback response to c_a as an analogue for the long term, we predict that following sustained exposure to a significant change in c_a, new leaves or plants will exhibit altered g_c(max) in the direction of the short-term g_c(op) response, through adjustment of S and/or D. This adjustment is a practical necessity, ensuring that normal g_c(op) is within the high sensitivity region of the g_c(op) versus P_g curve (figure 2b). Furthermore, there is a strong body of evidence suggesting that both across and within species, S versus D forms a negative linear log–log relationship, such that higher g_c(max) is achieved through smaller S and higher D [19,20,29,30]. We predict, therefore, that in plants grown under low c_a, g_c(max) will be higher, and this will be attributed to smaller S and higher D. If our hypothesis is correct, then over evolutionary timescales (millions of years), sustained shifts in c_a could select for g_c(max) on the basis of adaptation of S and D. Reconstructions of g_c(max) from S and D in fossil leaves show a significant negative correlation with reconstructed c_a over geological time [20], suggesting that g_c(max) and c_a have covaried in the long term. However, the pattern of adaptation of S, D and g_c(max) has not been examined closely under independent control of c_a.

(b). Stomatal size and genome size

Although S and D can be modified in many different combinations to alter g_c(max), there are constraints. For example, allocation of more epidermal space to stomata may be at the expense of other important epidermal structures [20]. The tradeoffs involved in altering S and D may therefore be complex. There is, however, a readily apparent constraint with regard to changing S. The size of the guard cell nucleus is proportional to the width of the guard cell (figure 3a) resulting in a correlation between the two across species [31]. Because guard cell nuclei are typically diploid [32], this scaling of the size of a guard cell and its nucleus leads to a correlation between stomatal size and plant genome size (or C-value, measured as haploid or 1C DNA amount) [31,33,34]. This suggests that the large changes in S over geological time were accompanied by changes in plant genome size, and that both S and plant genome size co-evolved with c_a. This of course would not mean that c_a alone drove the evolution of S and/or plant genome size. Other environmental factors, including drought and light intensity, are known to affect S [19,35,36]. However, if S, g_c(max) and 1C DNA respond to independent control of c_a, then a firmer physiological framework can be established for the adaptation, natural selection and evolution of plants in relation to past and future changes in global c_a. The objective of this study was to test for sensitivity, in the form predicted earlier, of g_c(max), S and 1C DNA amount to controlled, long-term changes in c_a, from a subambient concentration (180 ppm) approximating that in the LGM, to an elevated concentration (1000 ppm) predicted for the end of this century. The timescale of sensitivity in this case is that of plant growth and development, i.e. in the order of several weeks.

Figure 3.

Stomatal guard cells. (a) Guard cell nuclei. Fluorescing guard cell nuclei of V. faba after preparation with a DNA fluorphore (see Materials and Methods). Suspensions of similarly stained nuclei isolated from bulk leaf tissue were used to measure genome size (1C DNA amount). Scale bar, 20 µm. (b) Maximum stomatal aperture, a_max. Fully inflated guard cells of C. communis forming maximum stomatal apertures approximating circular pores. This is common in angiosperms but less so in non-angiosperms. Scale bar, 50 µm.

2. Materials and methods

(a). Plant material and treatments

Plants were grown in controlled environment chambers under standard conditions: 1000 µmol m⁻² s⁻¹ photosynthetically active radiation (500 µmol m⁻² s⁻¹ for Selaginella uncinata); 10 h photoperiod; 25°C; and well-watered, commercial compost soil. Growth chamber CO₂ concentration (c_a) was controlled at 180 or 240 ppm for ‘subambient CO₂’, representing glacial atmospheric conditions, 450 ppm for ‘ambient CO₂’, representing current day conditions, and 1000 ppm for ‘elevated CO₂’, projected to occur this century. Commelina communis and Vicia faba were grown from seed sown directly into pots in their respective growth chambers; Osmunda regalis was grown from spores and S. uncinata cloned by layering and dividing parent plants. All measurements are reported as the mean ± s.e. Significance of difference between means, at the 0.05 level, was determined with one-way analysis of variance (ANOVA) and post hoc means comparison tests (Tukey test) using standard statistical software (OriginPro v. 8.0, OriginLab Corporation, Northampton).

(b). Stomatal size and density

Epidermal peels were dissected from leaves and mounted on glass slides for viewing with a light microscope. Digital images of stomata were recorded at 200× and 630× magnification. Guard cell length and width were measured using image analysis software (ImageJ), and stomatal size (S) was calculated as guard cell length multiplied by the width of the guard cell pair (not including aperture) [20]. Guard cell dimensions were measured on 20 stomata from three epidermal peels from three plants per CO₂ treatment (n = 60). Stomatal density (D) was calculated as the number of stomata per square millimetre of epidermis, measured in ten 0.09 mm² fields on epidermis from three plants (n = 30). The volume of a single guard cell was approximated as that of a cylinder the same width as the guard cell, with hemispherical ends. A simplified volumetric measure of guard cell size, guard cell width cubed, is also reported.

(c). Maximum stomatal conductance, g_c(max)

Maximum stomatal conductance to CO₂ (g_c(max), mol m⁻² s⁻¹) was estimated using a modified version of the Brown and Escombe equation [17,18,20]:

2.1

where d_c is the diffusivity of CO₂ in air (m² s⁻¹), D is the stomatal density (m⁻²), a_max is the mean maximum stomatal pore area (m²), v is the molar volume of air (m³ mol⁻¹), l is the depth of the stomatal pore (m, approximated as the width of a single guard cell [37]) and π is the mathematical constant (typically rounded to 3.142).

Note that a_max relates to fully inflated guard cells that, for many species, can form stomatal pores approximating circular dimensions (figure 3b). It is therefore convenient in some cases to estimate a_max as the area of a circle, π(p/2)², where p is the stomatal pore length [20]. Alternatively, where images of stomata with fully inflated guard cells are available, either for the species of interest or for a species with similar guard cell geometry, a_max can be estimated as a fraction α of S, i.e. a_max = αS [19,20]. We adopted the latter approach here, estimating a_max as 0.08S for S. uncinata (based on the lycophyte Huperzia prolifera [37]), 0.1S for O. regalis (based on the fern Nephrolepis exaltata [37]) and 0.33S for both C. communis and V. faba (based on the herbaceous angiosperm Tradescantia virginiana [37]). Importantly, a_max and hence g_c(max) occur only under the somewhat artificial conditions that promote fully inflated guard cells, such as the combination of low atmospheric CO₂ concentration (below that at which plants grew), high (photosynthesis-saturating) light intensity and 100 per cent humidity, or for certain in vitro preparations using epidermal peels [37]. Under more typical conditions, the highest operating stomatal conductance, g_c(op), is likely to be around half of g_c(max) [19]. Owing to the infinite combination of natural conditions under which g_c could be operating, g_c(max) is a more informative standard measure of the capacity of the leaf epidermis to support the diffusion of CO₂ from the atmosphere to the leaf interior, as it is determined solely on the basis of static anatomical dimensions.

(d). DNA flow cytometry

Plant nuclear genome size was estimated for bulk epidermal tissue (in epidermal peels) using flow cytometry following the method described in Doležel et al. [38], using LB01 buffer and the DNA fluorophore propidium iodide. To minimize potential error due to variations in the standard, the same tissue was used as the standard across treatments for each species. For O. regalis and V. faba, the standard was Allium cepa cv. ‘Ailsa Craig’, and for C. communis it was Solanum lycopersicum cv. ‘Gardener's Delight’, propagated from seed at the Royal Botanic Gardens, Kew. The mean 1C-value includes both epidermal and guard cell nuclei. Guard cell and nucleus size were measured in fresh epidermal peels of O. regalis, C. communis and V. faba using standard light and fluorescence microscopy techniques (excited with 488 nm wavelength light, and detected with a 562–588 nm band pass filter), after treating peels with the same fluorescent nuclear stain and buffer as in the flow cytometry. Guard cell nuclei of S. uncinata could not be viewed clearly with this method and were not examined.

3. Results and discussion

(a). Stomatal size, density and g_c(max) response to CO₂

The results show a consistent pattern of correlated adaptation of S and D to c_a that may serve as a physiological framework for adaptation, selection and evolution of g_c(max) under changing global c_a regimes. In response to growth across a gradient of increasing c_a, all four species developed leaves with increasingly large guard cells and lower stomatal density (figure 4). This translated into lower g_c(max) with increasing c_a, as calculated from S and D measurements using equation (2.1) (figure 5). In all cases, g_c(max) and D were significantly lower, and S significantly larger, at elevated relative to subambient c_a. The consistent response across these four species, representing a broad taxonomic range (lycophyte, fern and angiosperm), suggests that this mode of adaptation to c_a is widespread in vascular plants and its genetic basis may therefore be highly conserved through plant lineages.

Figure 4.

Stomatal size and density covary with CO₂. (a–d) Negative relationship between stomatal size (S) and density (D) in (a) S. uncinata (lycophyte), (b) O. regalis (Fern), (c) V. faba (angiosperm) and (d) C. communis (angiosperm), respectively, when grown at three different atmospheric CO₂ concentrations. Treating c_a as a continuous variable, each datum point (mean ± s.e.m.) is connected by lines in the order of highest to lowest c_a treatment. (a–d) Stomata were significantly smaller and more numerous at subambient compared with elevated atmospheric CO₂ concentration. Means with different adjacent letters, along respective axes, are significantly different. Filled circles, subambient CO₂; unfilled squares, ambient CO₂; filled triangles, elevated CO₂.

Figure 5.

Maximum stomatal conductance, g_c(max), is altered by CO₂. The observed changes in stomatal size and density (figure 4) translate to significant changes in g_c(max) when calculated from stomatal dimensions using equation (2.1). (a–d) g_c(max) decreases with atmospheric CO₂ concentration in S. uncinata, O. regalis, V. faba and C. communis, respectively. (a–d) g_c(max) in plants grown at elevated atmospheric CO₂ concentration is significantly lower than in plants grown at subambient CO₂ concentration. Symbols are the same as for figure 4; mean ± s.e.m.

The CO₂-induced shift in g_c(max) is in the direction of the short-term feedback adjustment of operational stomatal conductance resulting from stomatal aperture changes in response to CO₂ [14]. This correction of g_c(max) via anatomical adjustment of S and D indicates a realignment of the g_c versus P_g relationship (figure 2b) that keeps the new operational g_c in the same region of the curve as before the CO₂ change. This adaptive feedback response is consistent with a general mechanism of feedback adaptation of operational stomatal conductance and g_c(max) to environmental changes. For example, in response to growth under treatment with the drought hormone abscisic acid, Tradescantia virginiana leaves altered S and D, which lowered g_c(max) and increased water-use efficiency, while maintaining operational g_c in the same region of the g_c versus P_g curve [17].

Although the response to c_a was significant, the change in g_c(max) resulting from adaptation of S and D was less than 10 per cent in three of the four species (S. uncinata, V. faba and C. communis) over the subambient to elevated CO₂ gradient. In O. regalis, it was closer to 40 per cent. The sensitivity of g_c(max) to growth at different c_a is therefore variable and may, in some cases, be small or negligible. This is reflected in a survey of stomatal density response [26], where the change from ambient (mean 363 ppm) to elevated (mean 571 ppm) c_a was a mean reduction of 5 per cent, with the most frequent response being a reduction of less than 10 per cent. The same survey reported a reduction in operating stomatal conductance of 22 per cent, suggesting partial stomatal closure in addition to a reduction in D. However, the magnitude of the c_a treatment or change should be considered when assessing the sensitivity of S, D and g_c(max) to c_a. The survey by Ainsworth & Rogers [26] spanned a 1.5-fold range of c_a, averaging −2.4 per cent change in D per 100 ppm; in this study the c_a treatment spanned a fivefold range, averaging –3.2 per cent change in D per 100 ppm; and for the survey of fossils in Franks & Beerling [20], where c_a spanned a greater than eightfold range over 400 Myr, there was an average −4.2 per cent change in D per 100 ppm increase in c_a (noting that the response over large ranges of c_a appears to be nonlinear [20]). The sensitivity of D to c_a is therefore typically in the order of minus a few per cent per 100 ppm increase in CO₂, but this may be difficult to detect over smaller ranges of CO₂, both in studies of the geological record and in experimental treatments. Furthermore, the response appears to be nonlinear over a large range in c_a [20,39], with greater sensitivity at low c_a. The study of herbarium samples by Woodward [40], spanning the approximately 100 ppm rise in c_a over the last 200 years (i.e. subambient conditions), reports a −41 per cent change in D and a −61 per cent change for growth chamber experiments simulating similar conditions. This is supported by a recent study of leaf material preserved in peat deposits, together with herbarium samples, for the same period [30], which shows an average −28 per cent change in D.

A negative relationship between S and D appears to characterize the general adaptation, plasticity or variability in g_c(max) in response to various environmental variables. It is expressed within the canopy of individual species as part of the natural variability in g_c(max) [19] and holds for the pooled data of hundreds of fossil species across the geological record and the present [20]. Adaptation or selection of g_c(max) by large global changes in c_a appears to operate within a general framework of down- or upregulation along a negative S versus D relationship in order to optimize plant carbon/water balance.

(b). Stomatal size and nucleus response to CO₂

Direct observations of guard cell nucleus size, made following fluorescent DNA staining, indicate significant increases in nucleus volume at elevated relative to subambient CO₂ concentration in all three species (figure 6a–c). Accompanying these changes were correlative increases in the volume of the guard cells (figure 6d), with guard cell volume and nucleus volume both significantly larger at elevated compared with subambient c_a. This pattern is reflected as an increase in the apparent 1C DNA amount in epidermal tissue with increasing growth c_a (figure 7a–c). In all three species, 1C DNA amount in epidermal tissue was significantly greater at elevated relative to subambient c_a. This suggests that the physiological adaptation of stomatal size to c_a involves a coordinated change in guard cell nucleus size and genome size.

Figure 6.

Atmospheric CO₂ influences the size of guard cell nuclei. Guard cell nuclei are significantly larger in plants grown under elevated compared with subambient atmospheric CO₂ concentration for (a) O. regalis, (b) C. comunis and (c) V. faba. (d) Guard cell nucleus volume increasing with guard cell volume for all three species grown under the different atmospheric CO₂ concentrations. Symbols are as given for figure 4; in (d) the solid line is O. regalis, dashed line is C. comunis and dotted line is V. faba; mean ± s.e.m.

Figure 7.

Effect of growth CO₂ on genome size. 1C DNA amount in epidermal tissue is significantly larger in plants grown at elevated compared with subambient atmospheric CO₂ concentration for (a) O. regalis, (b) C. communis and (c) V. faba. To illustrate the relationship to guard cell size, 1C DNA amount is plotted against guard cell width cubed, a simplified volumetric measure of guard cell size. Symbols are as given for figure 4; mean ± s.e.m.

In contrast to the clear physiological consequences of adaptation of S, D and g_c(max) to c_a, the physiological benefit of an accompanying change in nucleus size or genome size is unclear. The minimum size of guard cells is of course constrained physically by the size of the guard cell nucleus (small guard cells have small nuclei), but the close coupling between guard cell size, nucleus size and apparent genome size in figures 6 and 7 appears to relate to subcellular scaling processes rather than the diffusion physics of leaf gas exchange. It has long been observed that there is a fundamental positive correlation between cell size (specifically the cytoplasm volume) and the size of the cell nucleus, corresponding broadly to a correlation between cell size and genome size [41–43]. The underlying mechanism for this has been the subject of considerable debate [43,44], but one model with an advanced biophysical basis is the ‘skeletal DNA theory’ [43]. The principle of the skeletal DNA theory is that the correlation of cell cytoplasm and nucleus volume maintains a balance between the rate of protein synthesis (cytoplasm-based) and the rate of RNA synthesis and processing (nucleus-based). It is possible that this energetic balancing of cell size and genome size applies to small, plastic changes in cell size within species, as observed in this study with guard cells. However, considerably more work is required to prove this. Until recently, even the possibility of variation in genome size within species (variation between populations and within individuals, as observed in this study) was hotly debated, but there is now overwhelming evidence to show that variability and plasticity in genome size is common within species [45]. Assuming that within-species differences in genome size are linked to the fitness of the phenotype, as shown here in relation to the feedback adjustment of g_c(max) involving guard cell size, then it is possible for natural selection to act on these individuals, i.e. for c_a to select for genome size [31].

The size of the nucleus comprises the total amount of DNA in the nucleus (i.e. genome size, or ‘C-value’) and the degree of chromatin packing, and our results suggest that c_a may play a role in influencing either one or both of these. Over the short time frames of the CO₂ experiments (up to three months), the observed increase in guard cell nuclear volume at elevated c_a (figure 6a–c) is likely mediated through changes in the states of euchromatin and/or heterochromatin, which are known to be dynamic and responsive to developmental and environmental cues [46,47]. The results may indicate an epigenetic response [48], whereby adaptation of guard cell size to c_a to facilitate the optimization of leaf gas exchange invokes chromatin repackaging possibly mediated via RNA interference [49]. The qualitatively consistent results across species suggest at least that c_a is an external cue to which the plant nucleus responds, either directly or as a correlated character with guard cell size. An alternative possibility is that the differences in the amount of DNA detected result from altered binding of the fluorochrome to the DNA owing to changes in the condensation state of chromatin [50], rather than absolute changes in genome size. However, it is noteworthy that the same result, as predicted from physiological and biophysical theory, was observed across three species.

4. Conclusions

The results provide a physiological framework for analysing and predicting the adaptation of stomata and maximum leaf diffusive conductance to atmospheric CO₂ concentrations ranging from glacial to the elevated levels forecast for this century. The results suggest that for many species g_c(max) exhibits developmental plasticity through a negative relationship between S and D that shifts the operational range of leaf diffusive conductance in the direction of the short-term feedback response of stomatal aperture to c_a. Stomata therefore appear to exhibit both short-term dynamic and long-term adaptive feedback responses to c_a. The plasticity in S appears also to be linked to correlated plasticity in the size of the guard cell nucleus as well as the apparent 1C DNA amount in the cells of epidermal tissue. The results also suggest that the potential productivity of plants, as governed by c_a, S, D and consequently g_c(max), is tightly coupled to plant genome size. However, the mechanism underlying this coupling remains unclear. Over the much longer time frame of millions of years, in which natural selection and evolution can operate [44,51–53], our results suggest that large changes in c_a may, in selecting for the phenotypic benefits of optimal g_c(max), select also for correlated changes in S, D and genome size.