The impact of growth at elevated [CO2] on stomatal anatomy and behavior differs between wheat species and cultivars

Elevated [CO2] affects stomatal characteristics differently in wheat elite cultivars and wild accessions, resulting in faster kinetics in wild relatives, a trait useful to enhanced photosynthesis and water use.


Introduction
Prior to the industrial revolution, the atmospheric CO 2 concentration (a[CO 2 ]) was maintained at a value close to 280 ppm for ~1000 preceding years (Tans and Keeling, 2016). Subsequently, anthropogenic CO 2 emissions, primarily through the burning of fossil fuels, have increased the present day atmospheric CO 2 concentration to 419 ppm (NOAA, 2022).
With the current increases in CO 2 emissions associated with modern day activities, the Intergovernmental Panel on Climate Change projections include scenarios of [CO 2 ] doubling from current levels by the end of the century (IPCC, 2021), which to date has resulted in a rise in global temperature (of ~1.1 °C, World Meteorological Organization, 2022) and is predicted to rise with further increases in [CO 2 ] (Stockwell et al., 2021). Elevated [CO 2 ] generally increases leaf photosynthetic rates in a range of C 3 crops from potatoes (Lawson et al., 2001) to soybean (Rogers et al., 2004), through increased substrate for Rubisco (the enzyme involved in the first major step of carbon fixation) and the suppression of photorespiration. A recent review of 18 C 3 crops grown using free-air CO 2 enrichment (FACE) technology with elevated CO 2 of 200 ppm above ambient [CO 2 ] reported that most species exhibited increased yields by ~18% (Ainsworth and Long, 2021). However, the same study also highlighted that yield increases were not consistent across species or cultivars, and yield benefits with elevated [CO 2 ] were correlated with sink strength (Ainsworth and . As stomatal conductance (g s ) regulates gas exchange between the leaf interior and the external environment, stomatal responses to changing climatic conditions are critical in determining CO 2 supply for photosynthesis (A) and water loss through transpiration (Lawson et al., 1998;Morison et al., 2008). Transpirational water loss also plays a key role in nutrient uptake from the plant roots as well as evaporative cooling of the leaf tissue and the maintenance of optimal leaf temperatures for photosynthesis (Raven, 1977(Raven, , 2002Hetherington and Woodward, 2003;Peterson et al., 2010;McAusland et al., 2016;Murray et al., 2016;Lawson and Vialet-Chabrand, 2019). Therefore, stomatal dynamics will have a pivotal role in determining C 3 crop productivity in future climates (Lawson et al., 2010(Lawson et al., , 2012. g s is determined by anatomical features as well as functional aspects of the guard cells, both of which are influenced by growth [CO 2 ] (and temperature) (Woodward, 1987;Matthews and Lawson, 2019;Stevens et al., 2021). Both stomatal anatomy and behavior are modified by elevated [CO 2 ], with most species responding by decreasing stomatal density (Woodward, 1987;Poole et al., 1996) and reducing aperture (see review by Stevens et al., 2021). Reducing stomatal aperture under elevated [CO 2 ] greatly increases intrinsic water use efficiency (WUEi; A/g s ) with potential benefits for plant growth (Leakey et al., 2009;Sreeharsha et al., 2015). On the other hand, reductions in g s can negatively impact on photosynthesis through diffusional constraints, as well increases in leaf temperature (Matthews and Lawson, 2019).
The number and size of stomata on the leaf determine the maximum potential stomatal conductance (g smax ; Lawson and Morison 2004;Lawson et al., 2010;McElwain et al., 2016), whilst pore aperture/behavior regulates the shorttime scale dynamics of g s and gas exchange (Dow et al., 2014;Takahashi et al., 2015). The majority of studies that have explored variation in g s or the influence of growth conditions on anatomy and function have examined steady-state conditions (Schlüter et al., 2003;Doheny-Adams et al., 2012;Tanaka et al., 2013); however, recent studies have illustrated the significant impact of dynamic g s responses on A (Sakoda et al., 2020(Sakoda et al., , 2022Vialet-Chabrand and Lawson, 2020) and WUEi (Papanatsiou et al., 2019;Acevedo-Siaca et al., 2021;Pignon et al., 2021). Generally, slow stomatal opening limits the CO 2 assimilation rate, reducing the speed of photosynthetic induction , whilst slow closure erodes WUEi (Lawson and Blatt, 2014;McAusland et al., 2016;Qu et al., 2016Qu et al., , 2020Lawson and Vialet-Chabrand, 2019). The rapidity of stomatal responses to changing climatic conditions is also critically important for maintaining optimal leaf temperature (Matthews and Lawson, 2019;Stevens et al. 2021) and for photosynthesis and plant productivity (Moore et al., 2021). Dynamic responses have been linked to both morphological and physiological variation in stomata (Drake et al., 2013;Lawson and Blatt, 2014;Zhang et al., 2019); however, few studies have explored the impact of changing climate conditions such as growth [CO 2 ] on morphophysiological characteristics. Consequently, natural variation in rapidity of stomatal conductance between cultivars and species, as well as the influence of changing climatic conditions on these traits, could provide currently unexploited targets for improving crop productivity in future climates (Lawson et al., 2012;Faralli et al., 2019;Faralli and Lawson, 2020).
Here we explored the impact of elevated CO 2 concentration (e[CO 2 ]) of 800 ppm, approximately double that of the current atmospheric [CO 2 ] (a[CO 2 ]), on the physiology and growth of 11 different wheat progenitor and elite cultivar accessions. Wheat (Triticum aestivum L.) is a principal global food grain source, grown on more land area than any other commercial crop. In addition, it is one of the largest traded primary crop commodities, along with maize and rice (FAO, 2014). Globally, wheat provides >20% of the calories consumed by the human population (Braun and Atlin, 2010;Lobell et al., 2011). Modern wheat is a hexaploid species containing three sets of chromosomes (A, B, and D subgenomes). These subgenomes originated from three different diploid grass species and combined during two hybridization events (Kerber and Rowland, 1974;Faris, 2014;Marcussen et al., 2014). Initially diploid wheat Triticum urartu (subgenome AA ancestor) hybridized with the B genome ancestor Aegilops speltoides ssp. ligustica (Huang et al., 2002;Dvorak and Akhunov, 2005;Peng et al., 2011) to produce wild emmer wheat Triticum turgidum ssp. dicoccoides (genome AABB). In the second event, T. turgidum ssp. dicoccoides hybridized with the wild goat grass Aegilops tauschii to produce the modern hexaploid Triticum aestivum ssp. aestivum (AABBDD; Huang et al., 2002;Charmet, 2011;Faris, 2014). In this study, phenotypic variation in morphological traits including stomatal size and density, as well as physiological responses and, critically, the effect of growth [CO 2 ] on these traits, was assessed in six wheat relative (WR) accessions (including the species Aegilops tauschii, T. turgidum ssp. dicoccoides, and T. turgidum ssp. dicoccon) and five elite wheat T. aestivum cultivars (Claire, Rialto, Robigus, Soissons, and Xi19) to identify possible beneficial traits that could be exploited for future breeding.

Plant growth conditions
Triticum and Aegilops species (listed in Table 1) were germinated in a greenhouse compartment (at BASF, Ghent, Belgium) with supplementary lighting (Master Greenpower CGT 400 W E40 HPS lights) to ensure a typical summer day length of 15.30 h. At 14 d post-emergence, plants were vernalized in a controlled environment (custom-made growth chamber, BASF, Ghent, Belgium) for 10 weeks at 4 °C, with 75 μmol m -2 s -1 PPFD, over a 10 h photoperiod using an in-house-produced 60/40 peat-based sowing and cutting soil (including NPK Compound Fertilizer 12-14-24 (0.8 kg m -3 ). Plants were then transferred into 4 liter pots using a peat-based, boron-free potting soil [including NPK Compound Fertilizer 12-14-24 (2 kg m -3 )] and grown in two separate growth environments, one at current (2018) atmospheric [CO 2 ] (408 ppm CO 2 ) and a second at an elevated [CO 2 ] of 800 ppm. Both growth chambers had a light intensity (at pot height) of 800 ± 20 µmol m -2 s -1 with a 2:1 high pressure sodium:metal halide lighting mix (Master Greenpower CGT 400 W E40 and Powerstar HQI-BT 400 W/D PRO 400 W Daylight E40, respectively) for a 15 h light/9 h dark photoperiod. With the exception of [CO 2 ], both growth environments were set to identical conditions: air temperature controlled to 20 °C and 18 °C (±1 °C) day and night, respectively, and relative humidity maintained at a constant 65%. Plants were well watered using a drip irrigation system to the roots. All wheat measurements were taken from the flag leaf, at Zadoks growth stage 49 (GS 49, first awns/scurs visible) to GS 59 (ear emergence complete) (Zadoks et al., 1974). Six repetitions of each measurement were completed per accession unless stated below.

Leaf anatomical measurements
Measurements of stomatal density and size Stomatal density (SD) was measured from impressions taken from both the adaxial (upper) and abaxial (lower) leaf surface using silicone impression material (Xantopren, Heraeus, Germany) following the methods of Weyers and Johansen (1985) using six leaves per species/cultivar, measured at the middle of the leaf lamina. SD, guard cell length (GCL; used as a proxy for stomatal size), and pore length (PL) were all measured via light microscopy (Olympus BX60, Essex, UK). Total magnification was 100fold for SD measurements and 400-fold for GCL and PL measurements.
Anatomical maximum stomatal conductance (g smax : mol m -2 s -1 ) was calculated from the measurements of SD and stomatal dimensions (Equation 1) following the equations of Franks and Farquhar (2001): Where d is the diffusivity of water in air (m 2 s -1 , at 22 °C), v is the molar volume of air (m 3 mol -1 , at 22 °C), and pore depth (l; μm) was equal to guard cell width at the centre of the stoma represented as half the GCL. The mean maximum stomatal pore area (a max ; μm 2 ) was calculated assuming stomatal pores were elliptical with the major axis equal to pore length and the minor axis equal to half pore length (see McElwain et al., 2015).

Leaf thickness
Leaf thickness (LT) measurements were taken using the MultispeQ v1.0 instrument (Michigan State University, MI, USA) (Kuhlgert et al., 2016). The device was calibrated using 0.18 mm thick filter paper (Whatman 1001-110, Maidstone, Kent, UK). A mean leaf thickness was calculated from three repeat measurements per leaf from three separate leaves per species/cultivar.

Dry weight and leaf area
Leaf area was measured using a bench-top area meter (LI-3100C, Li-Cor, Lincoln, NE, USA) where the mean leaf area was calculated from three repeat measurements per leaf from three separate leaves per species/cultivar. Leaves were then placed in paper bags and dried at 60 °C to constant weight and measured using a four-digit balance (Kern, Northamptonshire, UK).

Leaf gas exchange
Stomatal conductance to water vapor (g s ) and the rate of photosynthetic CO 2 assimilation (A) were measured using a portable gas exchange system (Li-Cor 6400XT, Li-Cor) with an integrated light source (Li6400-40, Li-Cor), consisting of blue and red light-emitting diodes. Leaf temperature and VPD were controlled to 22 °C and 1 ± 0.2 kPa, respectively, throughout the measurements. Gas exchange measurements had a constant flow rate set at 300 μmol s -1 , with cuvette conditions maintained at a CO 2 concentration of 400 μmol mol -1 (for both plant growth CO 2 treatments). Gas exchange analysis was completed within the first 7 h of the photoperiod, to minimize any diurnal effects on stomatal opening and photosynthetic activation. All measurements were conducted on the mid-point of fully expanded flag leaves, before anthesis (GS 49-59) (Zadoks et al., 1974). Intrinsic water use efficiency was calculated as WUEi=A/g s . Between five and seven repetitions of each measurement were completed per accession for gas exchange data. Species abbreviation is how the species is referred to in the text. All seeds were provided from the NIAB collection.

PPFD step measurements
To measure the response of A and g s to a single step increase in PPFD, leaves were equilibrated at a PPFD of 100 μmol m -2 s -1 until both A and g s were at steady state (defined as <2% change in rate over 5 min).
Measurements were made at 30 s intervals, for 10 min at 100 μmol m -2 s -1 , after which PPFD was increased in a single step to 1000 μmol m -2 s -1 and recorded for a further 60 min. Leaf temperature (Ti), VPD, and [CO 2 ] were all maintained at 22 °C, 1 ± 0.2 kPa, and 400 µmol mol -1 , respectively, throughout the measurement. These data were used to model the response of A, g s , and WUEi to changes in PPFD.

Modeling gas exchange parameters
The maximum velocity of Rubisco for carboxylation (V cmax ) and the maximum rate of electron transport demand for ribulose bisphosphate dehydrogenase (RuBP) regeneration (J max ) were calculated from the A/C i response using equations from von Caemmerer and Farquhar (1981), as described by Sharkey et al. (2007) using the Rubisco kinetic constants for wheat (Carmo-Silva et al., 2010). The response of g s to the step change in PPFD was analyzed following the method described in McAusland et al. (2016). In summary, the optimum function in R (www.r-project.org; version 3.5.3), a model representing g s as a function of time, was fitted on each observed response as shown in Equation 2: The model uses a sigmoidal equation rather than an exponential slope, with an initial time lag (the time before g s starts to increase, λ, min), a time constant (the time taken to reach 63% of the variation, k, min), an initial value (r 0 , mol m −2 s −1 ), and a steady-state target [the value when the plateau is reached (g smax , mol m −2 s −1 ]. The time was set to 0 when PPFD was increased from 100 μmol m −2 s −1 to 1000 μmol m −2 s −1 (Vialet-Chabrand et al., 2013).

Statistical analysis
All statistical analyses were conducted using R software (www.r-project.org; version 3.5.3). For SD, GCL, and g smax , a Shapiro-Wilk test was used to test for normality and a Levene's test of homogeneity was used to determine if samples had equal variance. A log transformation was applied when data were not normally distributed (P<0.05, Shapiro-Wilk test) to achieve normality and meet modeling assumptions of an ANOVA. Single factor differences were analyzed using t-tests with a Bonferroni-Hochberg end correction or a one-way ANOVA, as described in the figure legends. When more than one factor existed, a two-way ANOVA was applied with an interaction between the two factors, and, if a significant difference was found (P<0.05), a Tukey posthoc test was performed.

Stomatal anatomy
Stomatal anatomy including SD and GCL was measured in five elite T. aestivum cultivars (Claire, Rialto, Robigus, Soissons, and Xi19; all hexaploid) and six WRs (four diploid lines IG 48509, IG 48514, KU 2018, and KU 2036; and two tetraploid lines TRI 3432 and TRI 11502), grown at two [CO 2 ], atmospheric (a[CO 2 ]) at ~408 ppm and elevated (e[CO 2 ]) at ~800 ppm. Significant (P<0.05) variation in combined (adaxial+abaxial) leaf SD was found between species grown at a[CO 2 ] (Fig. 1A) with the hexaploid cultivars ranging from ~0 mm 2 to ~100 mm 2 and the wheat relatives showing a larger range of ~60 mm 2 to ~160 mm 2 , with a +60% difference between the lowest and the highest mean SDs. When grown at e[CO 2 ] (Fig. 1B), less variation between and within species was observed. The majority of WRs showed a decrease in SD, with the exception of the T. dicoccoides accession TRI 11502 in which SD increased. No consistent pattern of change was observed for the elite hexaploid cultivars, with two cultivars showing no change in SD, while Rialto decreased, and Claire increased SD (P<0.05). SD was higher on the adaxial (upper) leaf surface (Fig. 1C) compared with the abaxial (lower) (Fig. 1E) surface (P<0.05) and SD on the adaxial surface was influenced to a greater extent by e[CO 2 ] (Fig. 1D) and accounted for a greater proportion of changes in total leaf SD compared with the abaxial surface, and this was particularly evident in the WRs. Overall, there was no relationship between SD in plants grown a[CO 2 ] and e[CO 2 ] ( Supplementary Fig. S1A). However, those species showing a change in SD with e[CO 2 ] on the adaxial surface also had significantly altered SD on the abaxial surface, albeit of a smaller magnitude ( Fig 1F). These data suggest that the majority of the combined (adaxial+abaxial) SD is determined by adaxial density (Fig. 1). Although there was no consistent species response of GCL to growth at e[CO 2 ] ( Supplementary Fig. S2), a significant (P=0.0055) positive correlation was observed between GCL from plants grown at ambient and elevated [CO 2 ] ( Supplementary Fig. S1B). The smallest GCL was observed for Ae. tauschii accession KU 2036 at ~29 μm at a[CO 2 ] and ~35 μm at e[CO 2 ], while the largest GCL was found on the bread wheat cultivars Xi19 at ~47 μm at a[CO 2 ] and Robigus at ~45 μm at e[CO 2 ]. When species were separated by ploidy, diploid species tended to respond to e[CO 2 ] by increasing GCL; however, this was not always significant ( Supplementary Fig. S2). Tetraploids had a tendency to decrease in GCL, but no specific trends were observed for hexaploids. Unlike the case for SD, it appears that average leaf GCL was determined by both the adaxial and abaxial leaf surfaces, as similar responses to e[CO 2 ] were observed on both, and together reflected the observed differences in combined (adaxial+abaxial) leaf averages (Supplementary Fig. S2).
A negative correlation between SD and GCL was evident for wheat grown at ambient [CO 2 ] (P≤0.001; Fig. 2A), demonstrating a relationship between decreasing SD and increasing stomatal size, driven mostly by the change in the four diploid accessions. Aegilops tauschii accession KU 2036 had the highest SD and smallest GCL, and the accession with the lowest SD mean (cv. Claire) had one of the largest GCLs. Although a similar trend of decreasing GCL with increasing SD was observed for the species and cultivars when grown under e[CO 2 ], this relationship was not significant (Fig. 2B). This is most likely to be attributable to the reduced range in SD under e[CO 2 ], particularly for the four Ae. tauschii accessions (Fig. 1B).  (A and B), the adaxial leaf surface (C and D), and the abaxial leaf surface (E and F) for 11 wheat species grown at atmospheric CO2 (~408 ppm; A, C, and E) and elevated CO 2 (~800 ppm; B, D, and F). Different letters represent statistically significant differences (P<0.05) between species means using the results of a Tukey test following a two-way ANOVA. A dashed line represents mean stomatal density of all wheat lines for the specific CO 2 treatment and leaf surface. Dotted lines separate wheat by ploidy. To test the effect of growth at elevated [CO 2 ] on stomatal density, a t-test with a Bonferroni-Hochberg end correction (n=6) was used to compare stomatal density means of individual wheat lines, with gray asterisks indicating significant differences (P<0.05).
Anatomical potential maximum rate of g s SD and measurements of stomatal size (GCL) were used to calculate the maximum anatomical stomatal conductance (g smax ), assuming fully open pores. At a[CO 2 ], a considerable range of g smax values were determined between accessions (Fig. 3A), with cv. Soissons displaying the lowest and Ae. tauschii accession KU 2036 the highest values (an increase of ~41%, driven predominantly by the differences in SD; Fig. 1). Higher g smax values were observed on the adaxial leaf surface irrespective of growth [CO 2 ] (Fig. 3C, D), with typical values >1.0 mol m -2 s -1 , whereas g smax values on the abaxial surface (except e[CO 2 ]-grown Xi19) were <1.0 mol m -2 s -1 (Fig. 3E, F). There was considerable variation in the response of g smax when grown under e[CO 2 ] compared with a[CO 2 ] (Fig. 3B). These differences appear to be driven mostly by the changes in SD (Fig. 1), but not exclusively as the CO 2 response patterns between SD (Fig. 1) and g smax (Fig. 3) were not identical. Similar to the patterns described for SD, there was a tendency for reduction in g smax with e[CO 2 ] driven mostly by adaxial g smax . However, interestingly, not all of the changes in SD translated into changes in g smax , strongly indicating a role for changes in GCL with e[CO 2 ] to compensate for changes in density, maintaining a similar g smax (Lawson and Morison, 2004;Harrison et al., 2020;Wall et al., 2022); for example, the diploid and tetraploid species have similar g smax to that of the hexaploid species even though SD is much higher in the diploid species.

Leaf gas exchange
Response of gs and A to a step change in PPFD The effect of e[CO 2 ] on stomatal behavior/kinetics was assessed by measurements of g s and A following a step increase in PPFD (Fig. 4). As expected, all species and cultivars exhibited an increase in g s and A with increasing irradiance. In general, A rapidly increased compared with g s when light was increased ( Fig.   4A-D), and this resulted in the maximum WUEi value being reached within a few minutes of the change in PPFD (Fig. 4E, F). Further increases in g s with time drove a continuous decrease in WUEi, and this trend continued after A had reached a maximum steady state. Considerable variation in A, g s , and WUEi was observed in plants grown under both [CO 2 ] treatments, although the variation was more apparent in growth at e[CO 2 ], particularly for A and g s (Fig. 4A-D).
The time constant to reach 63% of the final value for g s (τ gs ) as an indicator of the rapidity (Fig. 5) was significantly greater (P<0.05) in the hexaploid wheat compared with other species, regardless of growth [CO 2 ] (Fig. 5A, B). Hexaploid lines averaged 20 min to reach maximum g s while the other species averaged 10 min. In general, e[CO 2 ] increased the time constant (indicating slower stomatal responses) in most species with the exception of cv. Claire and T. dicoccon accession TRI 3432 which showed no significant differences, and T. dicoccoides accession TRI 11502 and cv. Xi19 which were significantly faster with e[CO 2 ] (Fig. 5B). Interestingly, there was a significant positive correlation (P<0.05) between τ gs in plants grown at ambient [CO 2 ] and at e[CO 2 ], indicating that speed was inherent with limited influence of growth environment ( Supplementary Fig. S3). However, the speed of the g s response did not influence the overall final g s (g sF ) achieved ( Fig. 5C-F), with no correlation observed between the two. Growth at e[CO 2 ] did not influence g sF values, with no differences observed between most accessions, with the exception of cv. Soissons and KU 2018 (Fig. 5C, D). On the other hand, the magnitude of change in g s (Δg s ) decreased in almost all species and cultivars with growth at e[CO 2 ], with the exception of the hexaploid cv. Rialto and Robigus and the Ae. tauschii accession IG 48509, and this was related to the speed of response, with slow responding accessions (mostly the hexaploids) having a lower Δg s compared with the fast responders in which Δg s was greater (Fig. 5E), although this correlation was only significant (at P=0.0107) when plants were grown under Fig. 2. Correlation between total stomatal density (mm 2 ) and total guard cell length (µm) for each species, calculated for the average of both leaf surfaces, for 11 wheat species grown at atmospheric CO 2 (~408 ppm; A) and elevated CO 2 (~800 ppm; B). The black dotted line represents the trend in the data between the two variables. Atmospheric CO 2 correlation= -0.552 (P=6.79e-12) and elevated CO 2 correlation=0.0625 (P=0.483) using a Pearson's correlation test. Error bars represent the SE (n=12). Fig. S4A, B). The fact that there was no effect of e[CO 2 ] on g sF indicates that minimum g s must have been higher with growth at high [CO 2 ]. The more rapid g s responses did not, however, impact on τ A (Fig. 6), with no clear relationship between the two parameters. Δg s was positively correlated with g sF at both a[CO 2 ] and e[CO 2 ] ( Supplementary Fig. S4C, D). Under both ambient and elevated growth [CO 2 ] conditions, the greater the g sF , the higher the A F achieved ( Supplementary Fig. S4E, F), and under ambient but not elevated [CO 2 ] this was also correlated with a greater change in A (ΔA) (Supplementary Fig. S4G, H), suggesting diffusion constraints by g s on the kinetic responses of A. The final value of A (A F , Fig. 6) at 1000 μmol m -2 s -1 PPFD was similar across the different accessions, ~25 μmol m -2 Fig. 3. Mean (white dot) and variation (box and whisker plots displaying distribution of biological replicates) of flag leaf g smax (mol m -1 s -1 ), calculated from the average of both leaf surfaces (A and B), the adaxial leaf surface (C and D), and the abaxial leaf surface (E and F) for 11 wheat species grown at atmospheric CO 2 (~408 ppm; A, C, and E) and elevated CO 2 (~800 ppm; B, D, and F). Different letters within each graph represent statistically significant differences (P<0.05) between means using the results of a Tukey test following a two-way ANOVA. The dashed line represents the mean g smax of specific CO 2 treatment and leaf surface. Dotted lines separate wheat by ploidy. To test the effect of growth at elevated [CO 2 ] on g smax , a t-test with a Bonferroni-Hochberg end correction (n=6) was used to compare individual wheat line means, with gray asterisks indicating significant differences (P<0.05). s -1 for both CO 2 growth treatments, although values for two accessions were significantly (P<0.05) higher when grown at e[CO 2 ], hexaploid cv. Xi19, ~32 μmol m -2 s -1 and Ae. tauschii accession IG 48509, ~35 μmol m -2 s -1 , and, not unexpectedly, this was highly positively correlated with the ΔA (Supplementary Fig. S4I, J). The change in A (ΔA) was similar across the different species, showing a typical increase of 15 μmol m -2 s -1 with a few exceptions being higher at 20 μmol m -2 s -1 (Fig. 6E, F). However, there was no relationship between τ A and these values in plants grown under e[CO 2 ] (Fig. 6), but the speed of the A response was negatively correlated with g sF , suggesting possible differences in the induction of photosynthesis due to both stomatal and biochemical constraints ( Supplementary Fig.  S4K, L).

A/C i response analysis
In order to assess changes to photosynthetic capacity, the response of assimilation rate (A) as a function of internal [CO 2 ] (C i ; Supplementary Fig. S5) was determined on the flag leaf on plants grown in the two [CO 2 ] environments. All accessions exhibited the expected increase in A with increased C i before reaching a plateau. Accessions grown at ambient [CO 2 ] displayed significant variation in their responses ( Supplementary  Fig. S5A). In general, hexaploid accessions had the highest assimilation rates, and greater V cmax , J max , and A max values at both ambient and e[CO 2 ] (Supplementary Fig. S6) whilst those of the tetraploid and diploid species were lower, indicating a reduced photosynthetic capacity. Growth under e[CO 2 ] had no significant influence on photosynthetic capacity, in any of the species. Fig. 4. Temporal response of stomatal conductance (g s ; A and B), net CO 2 assimilation (A; C and D), and intrinsic water use efficiency (WUEi; E and F), to a step increase in light intensity (from 100 μmol m -1 s -1 to 1000 μmol m -1 s -1 PPFD for 60 min) for 11 wheat species grown at atmospheric CO 2 (~408 ppm) and elevated CO 2 (~800 ppm). Gas exchange parameters (g s and A) were recorded at 30 s intervals, and leaf temperature and VPD were maintained at 22 °C, and 1 ± 0.2 kPa, respectively. Error ribbons represent the mean ±SE (n=5-7).

Plant growth
Multiple leaf growth parameters were measured including flag leaf area (LA; Supplementary Fig. S7), DW ( Supplementary Fig.  S8), and leaf thickness (LT; Supplementary Fig. S8). In general, all hexaploid wheat accessions had a greater LA (Supplementary Fig. S7) than other species, except for the tetraploid T. dicoccon accession TRI 3432 when grown at a[CO 2 ]. A similar trend followed for e[CO 2 ]-grown wheat, although there was less variation between species. No significant differences were observed between accessions from the same species from a[CO 2 ] to e[CO 2 ] except for Ae. tauschii accession IG 48509 in which LA increased. DW (Supplementary Fig. S8) followed the same trends as LA. In general, there was a trend for hexaploid and tetraploid species having thicker leaves than the diploid species at both CO 2 growth treatments ( Supplementary Fig. S9), the exception being cv. Soissons in which LT was reduced at e[CO 2 ]. These data suggest that the diploid species had smaller thinner leaves compared with the hexaploid wheat species.

Discussion
The global human population is expected to reach >9.5 billion by 2050, putting increasing pressure on breeders and crop scientists to improve yields to ensure sufficient food (Asseng et al.,Fig. 5. Time constant for stomatal opening [τ gs (min); A and B], final stomatal conductance value [g sF (mol m -2 s -1 ); B and D] after a step increase in light intensity from 100 μmol m -2 s -1 to 1000 μmol m -2 s -1 PPFD, and the difference in g s [Δ gs (mol m -2 s -1 )] between 100 μmol m -2 s -1 and 1000 μmol m -2 s -1 PPFD (E and F). The 11 wheat species were grown at both at atmospheric [CO 2 ] (~408 ppm; A, C, and E) and elevated [CO 2 ] (~800 ppm; B, D, and F). Error bars represent 95% confidence intervals using the results of a Tukey test following a two-way ANOVA. To test the effect of growth at elevated [CO 2 ], a t-test with a Bonferroni-Hochberg end correction (n=5-7) was used to compare individual wheat line means, with gray asterisks indicating significant differences (P<0.05). 2020). However, with the continued increases in global [CO 2 ], along with predicted changes to climate, it is vital that crop improvement programs consider the impact of these changes on crop performance and identify valuable physiological resilience traits (and the underlying genetics) that maintain productivity in a diverse range of environmental conditions. Genetic engineering approaches have demonstrated that enhancing photosynthetic capacity and stomatal behavior can successfully deliver crops with greater yield and resource use efficiency (Ruiz-Vera et al., 2017;López-Calcagno et al., 2020;De Souza et al., 2022). However, another powerful approach is exploiting natural variation in various physiological traits including photosynthesis (Driever et al., 2014;Carmo-Silva et al., 2017;Faralli and Lawson, 2019) and stomatal dynamics (Faralli et al., 2019(Faralli et al., , 2022Sakoda et al., 2022). Exploiting variation in current elite bread wheat germplasm (e.g. Driever et al., 2014;Faralli et al., 2019) as well as crop relatives (McAusland et al., 2020;Sharwood et al., 2022) offers significant potential to identify novel allelic variation (Sakoda et al., 2022;Sharwood et al., 2022;Yin et al., 2022). Here we have explored the impact of growth [CO 2 ] on variation in photosynthesis, stomatal anatomy, and stomatal kinetics in several elite wheat cultivars and their tetraploid and diploid relatives.
It is well documented that significant variation in stomatal anatomy exists between and within species, spatially within Fig. 6. Time constant for light-saturated carbon assimilation [τ A (min); A and B), final light-saturated carbon assimilation rate [A F (µmol m -2 s -1 ); B and D] after a step increase in light intensity from 100 μmol m -2 s -1 to 1000 μmol m -2 s -1 PPFD, and the difference in A [ΔA (µmol m -2 s -1 )] between 100 μmol m -2 s -1 and 1000 μmol m -2 s -1 PPFD (E and F). The 11 wheat species were grown at both at atmospheric [CO 2 ] (~408 ppm; A, C, and E) and elevated [CO 2 ] (~800 ppm; B, D, and F). Error bars represent 95% confidence intervals using the results of a Tukey test following a two-way ANOVA. To test the effect of growth at elevated [CO 2 ], a t-test with a Bonferroni-Hochberg end correction (n=5-7) was used to compare individual wheat line means, with gray asterisks indicating significant differences (P<0.05).
leaves (Ticha, 1982;Smith et al., 1989;Willmer and Fricker, 1996; and on different leaf surfaces (Wall et al., 2022), all of which are influenced by the growth environment (Poole et al., 1996;Croxdale, 2000;Lawson et al., 2002). Stomatal density is one of the most plastic traits and is affected by a great number of environmental parameters (Matthews and Lawson, 2019;Stevens et al., 2021). Increasing growth [CO 2 ] most commonly decreases SD in the majority of plant species investigated (Woodward, 1987), but not all (Lodge et al., 2001), and the degree of change is not the same even within cultivars of the same species (Dusenge et al., 2019). Not unexpectedly, in this study we observed significant variation across and between species and cultivars. The highest SDs were observed in the tetraploid relatives, with some individuals having double that of some elite varieties. A possible explanation for the high SD in the WRs is the smaller leaf area in these species (Supplementary Fig. S7). Therefore, expansion or differentiation of the epidermal cells in the elite cultivars would reduce SD (Lawson et al., 2002). Interestingly, growth at e[CO 2 ] generally reduced SD in diploid species, but not in the elite cultivars (except Rialto), and therefore no relationship between SD in plants grown under the two [CO 2 ] was observed ( Supplementary Fig. S1). Furthermore, variation within and between cultivars was generally reduced at e[CO 2 ], although the underlying cause of the reduced variation is currently unknown. However, as these plants were grown in controlled environments, and the only changing variable was [CO 2 ] (with all other parameters kept constant), it is possible that plants grown at a[CO 2 ] were subjected to greater variation in [CO 2 ] (due to photosynthetic draw down), and that the plants were more sensitive and responsive to this variation. For example, the a[CO 2 ] growth chambers ware maintained at 400 ppm; however, photosynthetic CO 2 fixation would result in short-term dynamic draw down of [CO 2 ] to ~320 ppm, whilst the same draw down in e[CO 2 ] would result in variation only between 700 ppm and 800 ppm, and plants would be less sensitive to these changes  as these levels will saturate photosynthesis ( Supplementary Figs S2, S3).
The SD variation and response to [CO 2 ] were mainly the result of anatomical changes on the adaxial leaf surface, suggesting two important points. Firstly, the receptors or signaling pathways responsible for detecting and responding to growth at e[CO 2 ] which drive changes in stomatal patterning are complex and either they reside separately on the two surfaces (and are not mesophyll driven) or there is limited surface to surface communication. Secondly, stomata on the adaxial surface play a more prominent role in gaseous exchange than those on the abaxial surface. This agrees with the recent work by Wall et al. (2022) who demonstrated that adaxial stomata make the greatest contribution to leaf gas exchange in amphistomatous bread wheat. Tsutsumi et al. (2014) reported that elevated [CO 2 ] decreased leaf size in rice, and this was accompanied by a decrease in epidermal cell numbers on the adaxial surface, but a reduction in cell size on the abaxial surface, thus providing a possible explanation for the differences observed between surfaces in different cultivars. GCL (as an indicator of stomatal size) was generally lower in the WRs compared with the elite cultivars, and together with SD was used to determine the maximum potential g s (g smax ) for the accessions investigated. As above, the variation in g smax was driven mostly by SD (and at the leaf level due to differences on the adaxial surface), but not entirely, with GCL clearly having a secondary role, as has previously been shown (Lawson and Morison, 2004). These findings indicate that there are some compensatory processes between SD and GCL (or size) to maintain a level of g smax across species (Lawson and Morison, 2004;Bussis et al., 2006;McElwain et al., 2016). The strong negative correlation observed between SD and stomatal size in plants grown at ambient [CO 2 ] agrees with several reports that have shown that lowered SD results in increased size (Franks and Farquhar, 2007). What is particularly interesting is that this size-density relationship was lost in plants grown at e[CO 2 ], due in part to the decrease in SD variation with growth at e[CO 2 ], and the fact that no relationship between SD at the two growth [CO 2 ] were observed; however, GCL was positively correlated between plants grown in the two environments. The anatomical constraints of g smax can translate into species-specific differences in operational or functional g s , often with implications for carbon gain and water use efficiency-particularly in dynamic environments Lawson and Vialet Chabrand, 2019). Dynamic stomatal responses and the speeds of stomatal responses to changing environmental cues have recently received considerable attention for optimizing A relative to water loss and WUEi (Farquhar and Sharkey, 1982;Mansfield et al., 1990;Lawson et al., 2010;Buckley and Mott, 2013;Lawson and Blatt, 2014;Buckley, 2017;Vialet-Chabrand et al., 2017;Matthews et al., 2018;Papanatsiou et al., 2019;Yamori et al., 2020). Stomatal conductance, although closely correlated with A, is an order of magnitude slower to respond to these changes than photosynthetic responses and can therefore lead to a disconnect between A and g s , as slow stomatal opening can limit CO 2 uptake whilst slow closure can erode water use efficiency (Drake et al., 2013;Lawson and Vialet-Chabrand, 2019;Vialet-Chabrand and Lawson, 2019). Exploiting variation in kinetic stomatal responses has been proposed as a possible route to increase the speed of stomatal responses to be more in tune with photosynthetic demands for CO 2 . We know that stomatal kinetics depend on species (McAusland et al., 2016), cultivar (McAusland et al., 2020Stevens et al., 2021), environmental conditions Long, 2005, 2021;De Souza et al., 2020), and time of day . Here the kinetic responses of g s to increasing PPFD were up to 50% slower in the elite cultivars compared with the diploid and tetraploid WRs, and growth at e[CO 2 ] decreased the speed even further. This agrees with previous reports that g s responses are slower in species with a lower density of larger guard cells (e.g. Elliott-Kingston et al., 2016) as we have observed here in the WRs, and growth under elevated [CO 2 ] could amplify this, dampening the g s response (Knapp et al., 1994). Surprisingly, these differences did not directly translate into differences in final g s values at high PPFD, most probably due to greater variation in τ gs than g sF . However, slow g s responses did result in a lower Δg s under e[CO 2 ], implying that stomatal speed influences overall g s behavior at elevated but not a [CO 2 ]. This is most likely to be due to a greater variation in g s in plants grown under e[CO 2 ]. Δg s was positively correlated with g sF (Supplementary Fig. S4), further supporting the idea that stomatal kinetics influence overall g s behavior and final values achieved. Such a relationship has previously been shown for tobacco, with the greater the change the higher the g s value achieved (von Caemmerer et al., 2004). Together, these findings indicate that both anatomical and biochemical/physiological components determine the speed of g s responses (Lawson and Blatt, 2014) and that both the rapidity in stomatal responses and the magnitude of change influence g s values. Growth under e[CO 2 ] reduced the magnitude of change in g s following the step increase in PPFD; however, it is clear that this was driven by differences in minimum g s and not the maximum achieved (g sF ). This could be due to differences in SD with growth under e[CO 2 ] or that guard cell sensitivity to [CO 2 ] was reduced under these conditions, ultimately increasing g s at low light (Hetherington and Woodward, 2003;Chater et al., 2015). The final g s values (g sF ) positively correlated with A F at ambient and elevated [CO 2 ] ( Supplementary Fig. S4), clearly demonstrating a diffusional constraint on photosynthetic induction rates, and highlights the importance of stomatal behavior in carbon assimilation (Lawson et al., 2012;De Souza et al., 2020;Long et al., 2022). g sF also positively correlated with ΔA, providing further support for a diffusional constraint on A. The fact that a similar relationship was not observed in plants grown at e[CO 2 ] is most probably due to g sF not being influenced by growth at e[CO 2 ] and therefore decreased g s control on CO 2 diffusion and A. The negatively correlation between g sF and τ A (at both growth [CO 2 ]) illustrates the importance of g s in photosynthetic induction (Lawson et al., 2010;Long et al., 2022). Furthermore, the tight correlation between ΔA and A F (Supplementary Fig.  S4) suggests that photosynthetic capacity at low PPFD was less variable than at high PPFD, and the final A reached depends on the magnitude and kinetic changes in A, that are driven by both stomatal and biochemical traits (Lawson et al., 2012).
The kinetic responses revealed more variation between accessions in both A and g s at e[CO 2 ] compared with ambient; however, the two compensated for changes in one relative to the other to maintain a similar WUEi to plants grown in ambient conditions. This demonstrates the importance of measuring both physiological components that make up WUEi as well as the attributing anatomical features (Lawson et al., 2010). The strong correlation between the speed of g s at ambient [CO 2 ] and e[CO 2 ] indicates that the rapidity of g s depends on stomatal anatomy and biochemistry, and not only differences in photosynthetic biochemistry. This is also supported by the lack of any influence that growth [CO 2 ] had on V cmax , A max , and J max (Supplementary Fig. S6). Therefore, stomatal speed is an inherent trait within these species and cultivars, supporting the notion that such a phenotype could be a key trait which could be incorporated for future breeding programmes.
In conclusion, this study has demonstrated that there is significant variation between species and cultivars in stomatal anatomy and function as well as photosynthetic capacity, and that growth at e[CO 2 ] does not necessarily impact on all of them, or in the same way. Current hexaploid bread wheat has a number of desirable traits, such as larger leaves and higher photosynthetic capacity, lower SD with a small Δg s (and therefore potential water saving capacity) compared with their WRs. Furthermore, SD in these species was not influenced by growth at e[CO 2 ]. It is possible that these traits have been unintentionally selected for during the breeding process. However, the WRs have much faster stomatal kinetics compared with modern wheat species, and although here this did not directly translate into improved A (as in previous studies) it was directly related to g sF which correlated significantly with A max (Supplementary Fig. S4), suggesting some reduced stomatal diffusional constraints on A in cultivars with greater Δg s . Such phenotyping traits could also be beneficial for increased WUEi as well as maintaining optimal leaf temperatures, highlighting the potential to exploit natural variation in different species, WRs, and elite crop varieties to develop idiotypes to maintain productivity in future climates.

Supplementary data
The following supplementary data are available at JXB online. Fig. S1. Correlation between SD and GCL of 11 wheat species grown at a[CO 2 ] and e[CO 2 ].