Introduction

Stomata are the pores on a leaf surface that allow plants to balance CO2 uptake for photosynthesis against water loss through transpiration. A reduction in stomatal conductance (G s) is commonly observed in response to an increase in the atmospheric concentration of carbon dioxide ([CO2]) to enhance plant water use efficiency (Woodward 1987). This stomatal control is achieved by the regulation of stomatal aperture through changes in guard cell turgor, and by alteration of stomatal density through modification of stomatal initiation and leaf expansion during leaf development. Active stomatal control is considered to represent physiological control of stomatal aperture via active guard cell ion transport, regulated by plant signalling mechanisms such as abscisic acid, that permit rapid stomatal movements. Passive stomatal behaviour does not involve physiological control, instead guard cell turgor passively reflects leaf water status (Doi and Shimazaki 2008; Brodribb and McAdam 2011; Ruszala et al. 2011). Species with passive stomatal control are unresponsive to the plant stress hormone abscisic acid, and do not exhibit rapid stomatal movements (Brodribb and McAdam 2011; McAdam et al. 2011). These components of stomatal control are likely to have played a critical role in plant evolution and the interaction of plants with their atmospheric environment over earth history (Robinson 1994; Hetherington and Woodward 2003; Franks and Beerling 2009; Berry et al. 2010; Haworth et al. 2011b), and in the response of vegetation to current climate change (Drake et al. 1997; de Boer et al. 2011; Lammertsma et al. 2011; Franks et al. 2012).

A diverse range of physiological and morphological stomatal responses to [CO2] are observed in controlled environment (e.g. Woodward and Kelly 1995; Beerling et al. 1998a; Hirano et al. 2012), free air carbon enrichment (e.g. Ainsworth and Rogers 2007; Bernacchi et al. 2007) and herbarium studies (e.g. Kouwenberg et al. 2003; Miller-Rushing et al. 2009; Haworth et al. 2010). It has been suggested that the stomata of more recently derived angiosperms exhibit different physiological responses to environmental stimuli such as [CO2], light quality or water vapour pressure deficit than more ancient groups such as conifers, ferns and lycophytes (Doi et al. 2006; Doi and Shimazaki 2008; Brodribb et al. 2009; Brodribb and McAdam 2011; McAdam et al. 2011; McAdam and Brodribb 2012). In an analysis of major plant clades, angiosperms, conifers, ferns and lycophytes all exhibited increases in G s in response to [CO2] reduced below current ambient (~380 ppm); however, only angiosperms reduced stomatal conductance when [CO2] was increased above ambient (Brodribb et al. 2009). Moreover, the presence of the plant drought stress hormone abscisic acid was found to increase stomatal sensitivity to [CO2] in the angiosperm Senecio minimus, but not in two conifer species, Callitris rhomboidea and Pinus radiata (McAdam et al. 2011). This divergence in physiological responses between more recently derived angiosperms and plant groups with more ancient lineages has led to the suggestion of an evolutionary transition from passive to active metabolic stomatal control (Brodribb and McAdam 2011; McAdam and Brodribb 2012). However, evidence of abscisic acid and [CO2] sensitivity in the stomatal aperture response of the ancient lycophyte Selaginella uncinata (Ruszala et al. 2011) and moss Physcomitrella patens (Chater et al. 2011) do not support this interpretation.

The relationship between stomatal density and/or stomatal index (a ratio of the number of stomata to epidermal cells) and the atmospheric [CO2] in which a leaf developed also differs between plant species, in direction, strength and the range of [CO2] over which stomatal initiation is modified (Woodward 1987; Kürschner et al. 1997; Royer et al. 2001; Kouwenberg et al. 2003; Haworth et al. 2011c; Hirano et al. 2012). Many angiosperm species exhibit a “ceiling of response” at 350–400 ppm [CO2], above which stomatal density and index no longer respond (Woodward 1987; Kürschner et al. 1997, 2008, Bettarini et al. 1998), whereas many conifers with ancient evolutionary origins often continue to reduce stomatal initiation at [CO2] levels above current ambient (Kouwenberg et al. 2003; Haworth et al. 2010, 2011a; Grein et al. 2011). A similar stomatal density and stomatal index response to [CO2] above 400 ppm is observed in Ginkgo biloba (Beerling et al. 1998a; Royer et al. 2001); however, atmospheric [CO2] does not influence stomatal initiation in Cycadaceae (Haworth et al. 2011c). The lower ceiling of response observed in angiosperms (Woodward 1987; Kürschner et al. 1997) relative to conifers (Kouwenberg et al. 2003; Grein et al. 2011) may be associated with greater stomatal aperture control of angiosperms at [CO2] above 400 ppm (Brodribb et al. 2009), possibly indicating a degree of co-ordination between physiological and morphological control of stomatal conductance in response to [CO2] (Haworth et al. 2011b).

In addition to fluctuations in [CO2], levels of atmospheric oxygen ([O2]) have also varied throughout earth history (Berner 2006, 2009; Belcher et al. 2010). Ribulose-1,5-bisphosphate carboxylase-oxygenase displays an affinity for both CO2 and O2 as part of the competing processes of photosynthesis and photorespiration (Miziorko and Llorimer 1983). The level of [O2] may therefore possibly affect stomatal initiation through changes in the photosynthetic availability of CO2 expressed by the atmospheric CO2:O2 ratio (Beerling and Woodward 1997; Beerling et al. 1998b). Oxygen may also influence stomatal function via the respiratory costs associated with stomatal opening and closing (Mawson 1993; Srivastava et al. 1995). The expansion of angiosperms during the Late Cretaceous and Tertiary has been associated with falling levels of atmospheric [CO2] (McElwain et al. 2004; Heimhofer et al. 2005). However, this period in earth history also coincides with rising levels of atmospheric [O2], possibly reducing the respiratory costs associated with more functional stomata, and thus favouring plants with more effective stomatal control (Haworth et al. 2011b), and accounting for the apparent shift towards active stomatal control in more recently derived plant groups (Brodribb and McAdam 2011).

This study intends to test the hypothesis that vascular plants show a co-ordination of physiological (via stomatal aperture) and morphological (via changes in stomatal initiation) control of leaf gas exchange in response to [CO2] and [O2] (Table 1). Specifically, we aim to investigate: (1) stomatal sensitivity to fluctuations in external atmospheric [CO2] concentration (C a) across a range of plants with divergent evolutionary lineages; (2) the effect of growth at elevated [CO2] and sub-ambient [O2] on stomatal sensitivity to C a; (3) the stomatal density, index and pore length responses to growth at elevated [CO2] and sub-ambient [O2]; and (4) possible co-ordination of stomatal functional and morphological responses to [CO2] in the control of leaf gas exchange.

Table 1 Levels of atmospheric [CO2] and [O2] used during growth treatments experienced by plant species in this study and CO2:O2 ratio of respective atmospheres
Full size table

Materials and methods

Controlled environment experiments

Lepidozamia peroffskyana (cycad), Hordeum vulgare (angiosperm) and Solanum lycopersicum (angiosperm) grown from seed, 6-month old specimens of Osmunda regalis (fern) and 2-year-old specimens of Ginkgo biloba (Ginkgoaceae), Nageia nagi (conifer), Podocarpus macrophyllus (conifer) and Agathis australis (conifer) were potted in 4-l square pots (15 × 15 × 23 cm) with 80 % compost (2 kg m−3 15:10:20 N:P:K; Bord na Móna, Newbridge, County Kildare, Ireland), 20 % vermiculite and 2.5 g l−1 slow release Osmocote fertilizer (15 % N, 10 % P2O5, 10 % K2O, 2 % MgO, plus trace elements; Scotts, Marysville, OH, USA). Plants were grown in four Conviron BDW-40 (Winnipeg, Manitoba, Canada) walk-in growth chambers in UCD’s PÉAC facility at Thornfield (see Table 1 for atmospheric growth conditions). Plants were grown under experimental atmospheric conditions for 18 months with the exception of H. vulgare and S. lycopersicum that were grown for 3 months. To avoid chamber effects, plants were rotated between chambers every 3 months (Hirano et al. 2012). Atmospheric concentration of [CO2] within the chambers was monitored by a PP-systems WMA-4 IRGA (PP-Systems, Amesbury, MA, USA) and supplemented by compressed CO2 to increase [CO2] above ambient (BOC, Guildford, Surrey, UK). Atmospheric oxygen level was monitored by a PP-systems OP-1 Oxygen Sensor. To reduce [O2], the nitrogen level in the chambers was supplemented via a compressed air line from a nitrogen generator (Dalco Engineering, Dunshaughlin, County Meath, Ireland). All other growth conditions remained constant, with plants experiencing 16 h of light per day in a simulated day/night program (0500–0600 hours, dawn; 0600–0900 hours, light intensity rises from 300 to 600 μmol m−2 s−1; 0900–1700 hours, midday light intensity of 600 μmol m−2 s−1; 1700–2000 hours, light intensity decreases 600 to 300 μmol m−2 s−1; 2000–2100 hours, dusk), temperature regime (nighttime temperature of 18 °C rising to a midday peak of 28 °C), relative humidity of 80 %, downward ventilation to ensure mixing of atmospheric gases and receiving 60 ml of water each day. In order to avoid mutual shading plants were randomised within areas of identical canopy height within the growth chambers (Hammer and Hopper 1997; Sager and McFarlane 1997). After full leaf development and expansion, the uppermost leaves receiving full irradiance and not affected by self-shading were used for stomatal [CO2] sensitivity analysis through analysis of G s response to instantaneous step changes in C a, and then destructively sampled for stomatal counts.

Measurement of stomatal conductance sensitivity to [CO2]

Stomatal conductance (G s) measurements were conducted on a minimum of three replicates per species from each atmospheric treatment. Plants were removed from the growth chamber and measurements were recorded in a well-ventilated room maintained at a constant temperature of 25 °C under ambient levels of [CO2] and [O2]. Timings of day/night programs on the plant growth chambers were staggered to allow the maximum number of plants to be analysed at the optimal time of the day/night program for photosynthetic activity, and thus avoid the influence of circadian stomatal behaviour; particularly where stomata close at midday or during the early afternoon when temperatures rise and leaf water potentials decrease. Stomatal conductance responses to fluctuations in external [CO2] concentration (C a) measurements were taken between 0900 and 1100 hours using a PP-Systems Ciras-2 attached to a PLC6(U) leaf cuvette and LED light unit (PP-Systems) under saturating light intensity calculated from PAR (photosynthesis response curves) (Parsons et al. 1998). Temperature within the cuvette was maintained at 25 °C. Leaves were allowed to stabilise within the cuvette for approximately 20 min at 380 ppm [CO2], before step changes in C a (200, 400, 750, 1,000 and 2,000 ppm [CO2]) occurred. At each C a value G s was allowed to stabilise and then recorded after G s had remained stable for ~10 min. Vapour pressure deficit in the leaf cuvette was maintained constant throughout each C a step change analysis at 1.3 ± 0.1 kPa. This protocol was used to examine the extent to which G s was actively controlled by changes in guard cell turgor, whereby a change in G s was used to infer “active stomatal control”, and no change in G s to infer “passive stomatal control” (see Fig. 1; Table 2).

Fig. 1
figure 1

Relative stomatal conductance response to an increase of external atmospheric [CO2] (200, 400, 750, 1,000, 2,000 ppm [CO2]) of an evolutionary cross-section of plants grown in atmospheres of elevated [CO2] and sub-ambient [O2] in comparison to control atmospheric conditions (see Table 1): control (open squares); low [O2] (open diamonds); high [CO2] (open circles); and combined low [O2]/high [CO2] (open triangles). Error bars one standard error either side of the mean

Full size image
Table 2 One-way ANOVA results for stomatal conductance of plant species grown in atmospheres of elevated [CO2] and sub-ambient [O2] when exposed to sub-ambient and super-ambient levels of C a (see Fig. 1)
Full size table

Stomatal density and index counts

Conifer, G. biloba and L. peroffskyana leaf cuticles were macerated using a 50:50 solution of glacial acetic acid and 30 % H2O2 at 70 °C, stained using safranin-O solution and mounted in glycerol on glass slides. Osmunda regalis, S. lycopersicum and H. vulgare leaf impressions were taken using dental impression gel (Coltène President Light Body Material), and nail varnish “positives” mounted onto glass slides (Weyers and Lawson 1985). Cuticle images were taken under transmitted light using a Leica DM2500 microscope attached to a Leica DFC300FX camera (Leica Microsystems, Wetzlar, Germany) and Syncroscopy Automontage (Syncroscopy, Cambridge, Cambridgeshire, UK). As an indicator of stomatal aperture size, the stomatal pore length (Wagner et al. 1996; Hetherington and Woodward 2003) of ~20 stomata was measured using Automontage, with the average taken to represent the treatment value for a given species. A 0.09-mm2 grid (Poole and Kürschner 1999) was superimposed on the images for stomatal and epidermal counts using Syncroscopy AcQuis. In the controlled environment study, five stomata/epidermal cell counts were performed on each of three leaves from a plant, with the average of 15 counts taken to represent the mean stomatal density and stomatal index of an individual plant (except for L. peroffskyana where 9 counts were averaged). Stomata and epidermal cells were counted on 1,548 images in total, with 9 or 15 images counted per plant and then the average of three plants taken to represent the mean stomatal density or stomatal index value for a species in a given atmospheric treatment. The abaxial surface was analysed for stomatal counts in all species, with the exception of H. vulgare where stomatal counts were taken from the abaxial and adaxial surfaces; these were broadly similar and the mean was taken to produce an average H. vulgare value for Figs. 3 and 4 (individual abaxial and adaxial values are given in supplementary data tables). The percentage area of stomatous regions of the cuticle available as stomatal pore during maximal stomatal opening (A %) was calculated assuming elliptical stomatal pore geometry and assuming stomatal width at full stomatal opening was equivalent to 0.5 stomatal pore length (Beerling and Chaloner 1993). Relative changes in stomatal density, stomatal index and A % (Δ stomatal density, Δ stomatal index and ΔA %) between the control and each treatment were then calculated and plotted against relative changes in stomatal conductance (Δ stomatal conductance) sensitivity to C a increases from 400 to 2,000 ppm [CO2]. One-way Bonferroni method ANOVAs were performed using SPSS 20 (IBM, New York, USA) to test whether G s, stomatal index, stomatal density and stomatal pore length values of the plants differed significantly between treatments to identify any affects of elevated [CO2] and sub-ambient [O2] on stomatal morphology and conductance (for full details and results of post hoc analysis, see supplementary data).

Results

As [CO2] was increased the evolutionary cross-section of plants studied showed a diverse range of physiological responses to C a, from no change (passive) to pronounced reductions (active) in G s (Fig. 1; Table 2). The fern O. regalis, ginkgoalean G. biloba and conifers P. macrophyllus and A. australis exhibited passive stomatal control, with no reduction in G s to C a above 400 ppm (Fig. 2; Table 1). In contrast, the cycad L. peroffskyana and conifer N. nagi grown in atmospheres of ambient [CO2] exhibited pronounced reductions in G s (−58.5 and −40.2 %, respectively) as C a was increased from 400 to 2,000 ppm [CO2]. However, when grown in atmospheres of elevated [CO2], both L. peroffskyana and N. nagi no longer altered G s in response to changes in C a, suggesting the loss of stomatal sensitivity to [CO2] (Figs. 1e, f, 3; Table 2). Stomatal sensitivity to C a in G. biloba appeared to be enhanced by growth in atmospheres of 13.0 % [O2], possibly suggesting a respiratory requirement for physiological control of stomatal aperture (Fig. 2b) as G s sensitivity to C a was measured under ambient [O2] (G s reduction from C a 400 to 2,000 ppm [CO2]: control −15.5 %; low [O2] −23.0 %; high [CO2] −11.8 %, and; combined low [O2]/high [CO2] −23.7 %). Active stomatal control of G s in response to C a is also evident in the angiosperms S. lycopersicum and H. vulgare (Figs. 1g, h, 3c, d). The atmospheric growth conditions of the two angiosperms did not affect the relative changes in G s in response to C a, suggesting that [CO2] sensitivity was not impaired (see supplementary data) by growth at elevated [CO2] or sub-ambient [O2] for these taxa.

Fig. 2
figure 2

Physiological and morphological responses of plant species with passive stomatal behaviour grown in atmospheres of elevated [CO2] (1,500 ppm) and sub-ambient [O2] (13.0 %), relative to control conditions of ambient [CO2] (380 ppm) and [O2] (20.9 %). Line graphs indicate stomatal conductance response to external atmospheric [CO2] (200, 400, 750, 1,000, 2,000 ppm CO2) of plants grown in control (open squares); low [O2] (open diamonds); high [CO2] (open circles); and combined low [O2]/high [CO2] (open triangles): error bars one standard error either side of the mean; italicised letters significant differences between treatments at each C a level (a ≥ 0.05; b ≤ 0.05; c ≤ 0.01; d ≤ 0.001). Statistical analyses of differences in G s of plants grown in the same atmospheric treatments in response to changes in C a are reported in Table 2. Histograms indicate stomatal density, index and pore length responses of plants to growth in atmospheres of elevated [CO2] and sub-ambient [O2] relative to control atmospheres (see Table 1). Histogram error bars one standard deviation either side of mean, letters significant difference between treatments using Bonferroni method ANOVA

Full size image
Fig. 3
figure 3

Physiological and morphological responses of plant species with active stomatal behaviour grown in atmospheres of elevated [CO2] (1,500 ppm) and sub-ambient [O2] (13.0 %), relative to control conditions of ambient [CO2] (380 ppm) and [O2] (20.9 %). Line graphs indicate stomatal conductance response to external atmospheric [CO2] (200, 400, 750, 1,000, 2,000 ppm CO2) of plants grown in control (open squares); low [O2] (open diamonds); high [CO2] (open circles); and combined low [O2]/high [CO2] (open triangles): error bars one standard error either side of the mean; italicised letters significant difference between treatments at each C a level (a ≥ 0.05; b ≤ 0.05; c ≤ 0.01; d ≤ 0.001). Statistical analyses of differences in G s of plants grown in the same atmospheric treatments in response to changes in C a are reported in Table 2. Histograms indicate stomatal density, index and pore length responses of plants to growth in atmospheres of elevated [CO2] and sub-ambient [O2] relative to control atmospheres (see Table 1). Histogram error bars one standard deviation either side of mean, letters significant difference between treatments using Bonferroni method ANOVA

Full size image

Those species with passive stomatal control in response to increased C a above ambient (O. regalis, G. biloba, P. macrophyllus and A. australis) did, however, show an effect of atmospheric conditions during growth on G s (Fig. 2). Osmunda regalis displayed significantly higher G s when grown in atmospheres of ambient [CO2] compared to atmospheres of elevated [CO2]. When grown in high [CO2], the stomatal density values of O. regalis were 35.9 % lower than those in control atmospheres (Fig. 2a). Podocarpus macrophyllus and A. australis also exhibited 12.1 and 23.9 % reductions in stomatal density when [CO2] was increased to 1,500 ppm (Fig. 2c, d). In contrast, the angiosperms S. lycopersicum and H. vulgare exhibited 7.1 and 9.9 % (abaxial +10.9 %; adaxial +8.8 %) increases in stomatal density when grown in atmospheres of elevated [CO2] (Fig. 3c, d). However, L. peroffskyana that displayed active stomatal control to C a when grown in 380 ppm [CO2], appeared to become passive in response to instantaneous increases in C a following growth in elevated [CO2] (Table 1). Lepidozamia peroffskyana showed a reduction in G s when grown at elevated [CO2] that was not accompanied by a reduction in stomatal density (+6.6 %) or stomatal pore length (+20.4 %). In combination, these results suggest reduced physiological stomatal function of L. peroffskyana grown in elevated [CO2], and perhaps that impairment of stomatal opening resulted in reduced G s (Fig. 3a). Solanum lycopersicum and H. vulgare exhibited greater conductance rates at all C a values when grown in atmospheres of sub-ambient [O2] and ambient [CO2], while growth at elevated [CO2] did not significantly affect G s relative to control atmospheres (Fig. 3c, d).

The proportional change in physiological stomatal response to [CO2] above ambient (G s change between C a 400 to 2,000 ppm [CO2]) of plants grown in low [O2], high [CO2] and combined low [O2]/high [CO2] was compared in relation to plants grown in control atmospheres (Δ stomatal conductance); this was then plotted against the relative changes in morphological stomatal responses (Δ stomatal density, ΔA % and Δ stomatal index) between plants grown in control atmospheres and those grown in atmospheres of low [O2], high [CO2] and combined low [O2]/high [CO2] (Fig. 4). A strong correlation between stomatal density and G s sensitivity to C a was observed in plants grown in atmospheres of elevated [CO2] (R 2 = 0.611; regression P = 0.022) and combined low [O2]/high [CO2] (R 2 = 0.636; regression P = 0.018), suggesting the co-ordination of physiological and morphological responses to increases in [CO2] (Fig. 4b, c). This suggests that as [CO2] is increased above ambient species with active stomatal sensitivity to C a are less likely to reduce stomatal density than species with passive stomata. Relative changes in A % and G s when grown in atmospheres of elevated [CO2] compared to control atmospheres showed a significant correlation (R 2 = 0.765; regression P = 0.045) (Fig. 4e). However, similar relationships between shifts in A % and G s are not observed when plants are grown in atmospheres of low [O2] (R 2 = 0.168; regression P = 0.314) and combined low [O2]/high [CO2] (R 2 = 0.028; regression P = 0.695) (Fig. 4d, f).

Fig. 4
figure 4

Changes in stomatal morphology following long term exposure to different atmospheric treatments (Δ stomatal density; ΔA %; Δ stomatal index) versus physiological response of stomata of an evolutionary cross-section of plants with passive (open symbols) and active (closed symbols) stomatal control (see Figs. 2, 3 for species’ morphological and G s response) to an instantaneous increase in C a from 400 to 2,000 ppm CO2 when grown in atmospheres of low [O2], high [CO2] and combined low [O2]/high [CO2] in comparison to plants grown in control atmospheres (Δ stomatal conductance): a Δ stomatal conductance and Δ stomatal density of plants grown in atmospheres of low [O2]: y = −0.2514x + 1.2146; R 2 = 0.0372, regression P = 0.154; b Δ stomatal conductance and Δ stomatal density of plants grown in atmospheres of high [CO2]: R 2 = 0.611, regression P = 0.0220; y = −0.6954x − 17.468); c Δ stomatal conductance and Δ stomatal density of plants grown in atmospheres of combined low [O2]/high [CO2]: R 2 = 0.636, regression P = 0.0178; y = −1.7633x − 3.1869); d Δ stomatal conductance and ΔA % of plants grown in atmospheres of low [O2]: R 2 = 0.168, regression P = 0.314; y = 0.1508x + 0.0971); e Δ stomatal conductance and ΔA % of plants grown in atmospheres of high [CO2]: R 2 = 0.765, regression P = 0.00449; y = −0.4813x − 16.228); f Δ stomatal conductance and ΔA % of plants grown in atmospheres of combined low [O2]/high [CO2]: R 2 = 0.0275, regression P = 0.695; y = −0.1615x − 13.298; g Δ stomatal conductance and Δ stomatal index of plants grown in atmospheres of low [O2]: R 2 = 0.0494, regression P = 0.597; y = 0.2605x + 1.8152); h Δ stomatal conductance and Δ stomatal index of plants grown in atmospheres of high [CO2]: R 2 = 0.445, regression P = 0.0708; y = −1.2494x − 27.777), and; i Δ stomatal conductance and Δ stomatal index of plants grown in atmospheres of combined low [O2]/high [CO2]: R 2 = 0.00180, regression P = 0.922; y = 0.0832x − 13.261). Error bars one standard error either side of mean

Full size image

The correlations of stomatal density and A % with G s response to C a suggest a co-ordination of stomatal functionality and morphology in the control of transpiration in response to increased [CO2]. However, a statistically significant relationship at the 95 % confidence level between stomatal index and reduction in G s to C a above ambient is not observed when plants are grown in atmospheres of elevated [CO2] (R 2 = 0.445; regression P = 0.071) (Fig. 4h), suggesting that stomatal initiation was influenced to a lesser extent than stomatal density and size, both of which directly determine limits of transpirative water loss and photosynthetic [CO2] uptake. Growth in sub-ambient [O2] did not appear to induce a co-ordinated function and morphological stomatal response in the plant groups studied (Fig. 4a, d, g). Tables of stomatal density, index and pore length values alongside statistical tests are presented in supplementary information.

Discussion

The plants analysed in this study exhibited a diverse range of morphological and physiological stomatal responses to [CO2] and [O2]. The gymnosperms L. peroffskyana and N. nagi all exhibited sensitivity to C a above ambient, consistent with the findings of Ruszala et al. (2011) and Chater et al. (2011) that active stomatal control is not restricted to more recently derived angiosperms (cf. Brodribb et al. 2009; Brodribb and McAdam 2011; McAdam et al. 2011; McAdam and Brodribb 2012). The atmospheric growth environment of the plants also had a significant effect on stomatal function in terms of G s response to C a. Lepidozamia peroffskyana and N. nagi when grown in atmospheres of elevated [CO2] no longer altered G s to increased C a (Figs. 1, 3a, b) possibly suggesting impairment of stomatal sensing of [CO2] (e.g. Wheeler et al. 1999; Levine et al. 2009) or guard cell turgor modification responsible for stomatal control (e.g. Meidner 1968; Franks and Farquhar 2007). The loss of stomatal sensitivity and guard cell movement in response to increased C a when L. peroffskyana and N. nagi were grown in atmospheres of elevated [CO2] may also suggest that, at high [CO2], the costs of maintaining effective stomatal function, via the operation of physiological systems for [CO2] sensing and stomatal control, are no longer beneficial in terms of enhancing water use efficiency when compared to growth at lower ambient [CO2] levels. However, stomatal sensitivity to C a was unaffected in the angiosperm species (S. lycopersicum and H. vulgare) by growth in 1,500 ppm [CO2] and 13.0 % [O2]. This divergence in responses of the species with active stomatal control to growth [CO2] and [O2], suggests that these species may possess different physiological mechanisms for both [CO2] sensing and stomatal movements (Hetherington and Woodward 2003; Franks and Farquhar 2007; Doi and Shimazaki 2008; McAdam et al. 2011).

Oxygen plays an important role in stomatal movements, as stomatal opening and closing are energetically expensive processes. Mitochondria within the guard cells provide the energy for transport of potassium ions across the cell membrane (Walker and Zelitch 1963; Willmer and Mansfield 1970; Raghavendra 1981). In Gossypium barbadense, increased stomatal conductance rates occur alongside elevated guard cell respiration (Srivastava et al. 1995). The two angiosperms analysed within this study both exhibited enhanced G s rates when grown under sub-ambient [O2] and ambient [CO2] (Fig. 3c, d). As G s response to C a was measured under ambient [O2], this enhanced capacity for G s may represent an elevated respiratory capacity (Walker and Zelitch 1963) incurred through growth in atmospheres of 13.0 % [O2] prompting greater stomatal opening at 20.9 % [O2]. A similar pattern of increased G s caused by growth in atmospheres of low [O2] was not apparent in the gymnosperm species (with the possible exception of L. peroffskyana at C a 750 and 1,000 ppm CO2) and fern O. regalis, possibly suggesting differential mechanisms for stomatal movements such as guard cell chloroplast photosynthesis (Doi et al. 2006; Doi and Shimazaki 2008) or a lower respiratory requirement for guard cell turgor control in these species (Srivastava et al. 1995). The apparent greater respiratory demand for stomatal control in the two angiosperm species may indicate that rising [O2] in the mid- to Late Cretaceous favoured species with greater guard cell respiratory demands and therefore played a role in the expansion and diversification of the angiosperms (Haworth et al. 2011b). In contrast, in the gymnosperm species analysed, the greatest G s rates were observed in plants grown in control atmospheres of ambient [CO2] and [O2] at ambient C a levels of 400 ppm [CO2], suggesting that the CO2:O2 ratio of the growth atmosphere may play a more significant role in determining gymnosperm capacity for G s.

Those species with passive stomatal control to increased C a were more likely to exhibit reductions in stomatal density and A % when grown in atmospheres of elevated [CO2] than species with active stomatal control in response to C a (Fig. 4). This suggests that the morphological and physiological response of stomata to [CO2] are linked, with plants with active stomatal control less likely to alter stomatal density to elevated [CO2] than plants with passive stomatal control (Fig. 4b). The co-ordination of stomatal sensitivity and morphology to increased [CO2] may represent a trade-off between different strategies of stomatal control, determined by the selective pressures exerted by their associated costs and benefits (Haworth et al. 2011b). Active stomatal control to C a requires investment in mechanisms for sensing of [CO2], co-ordination of signals and resulting stomatal movements (Heath 1950; Hetherington and Woodward 2003; Hu et al. 2010). Plant species with highly sensitive stomata to short-term stimuli, such as members of the Cycadaceae (Fig. 3a), may not alter stomatal initiation, and thus modify stomatal density and/or stomatal index, to changes in their atmospheric environment (Marler and Willis 1997; Haworth et al. 2011c). Therefore, uptake of CO2 is not limited by stomatal number or size during periods when conditions are favourable to photosynthesis, or constrained in their ability to respond to any future shifts in atmospheric composition via prior modification of stomatal density. Conversely, species with passive stomatal control do not have to invest in physiological mechanisms to sense short-term fluctuations in C a, signalling or control stomatal movements (Brodribb et al. 2009; Hu et al. 2010). Nevertheless, this may result in limitations to photosynthetic capacity during periods that may be favourable to photosynthesis or to a lower tolerance of water stress (Brodribb et al. 2009; McAdam et al. 2011). The two conifers with passive stomatal control in response to C a (A. australis and P. macrophyllus) both possess stomatal wax plugs, possibly suggesting that the wax plugs restrict stomatal closure (Feild et al. 1998). However, the stomatal complexes of the fern O. regalis that also exhibited passive stomatal control were not occluded by stomatal wax plugs. An inverse relationship between stomatal pore length or size and stomatal density has been observed in the gas exchange responses of plants in response to increased [CO2] (Franks et al. 2012; Roth-Nebelsick et al. 2012). However, a similar pattern was not observed in the species analysed in this study (Figs. 2, 3).

The co-ordination of stomatal C a sensitivity with morphological response to growth at elevated [CO2] suggests an apparent evolutionary trade-off in the control of G s (Fig. 4b, e) determined by the respective evolutionary costs of each stomatal control strategy (Haworth et al. 2011b). Selective pressures exerted by the respective costs of active and passive stomatal control may have induced the co-ordination of physiological and morphological stomatal responses to elevated [CO2] apparent in this dataset, suggesting an evolutionary trade-off between stomatal control strategies in the optimisation of water use efficiency. This relationship between stomatal C a sensitivity and the degree of stomatal density response to [CO2] may provide an explanation for the diversity of reported stomatal density responses to atmospheres enriched in [CO2] (e.g. Woodward 1987; Kürschner et al. 1997; Beerling et al. 1998a; Bettarini et al. 1998; Reid et al. 2003; Ainsworth and Rogers 2007; Haworth et al. 2010; Hirano et al. 2012). Shifts in stomatal function and morphological response to [CO2] are likely to affect transpiration rates at local to global scales and to influence ecological composition under rising atmospheric [CO2] (Lammertsma et al. 2011; Franks et al. 2012). Nevertheless, further work is required to establish any relationship between the degree of stomatal density response and stomatal C a sensitivity in plants grown across a range of [CO2] levels. Additionally, the stomatal density and/or index values of fossil plants are increasingly used as indicators of the palaeo-atmospheric level of [CO2] in which the leaf developed (e.g. Royer et al. 2001; Haworth et al. 2005; Kürschner et al. 2008; Passalia 2009; Smith et al. 2010; Grein et al. 2011; Stults et al. 2011). Future stomatal palaeo-[CO2] reconstructions should be undertaken within the context of the likely stomatal C a sensitivity of the fossil plant species based upon the stomatal control mechanisms employed by nearest living relative or analogue species.