In C3 leaves, the mesophyll conductance to CO2 diffusion, gm, determines the drawdown in CO2 concentration from intercellular airspace to the chloroplast stroma. Both gm and stomatal conductance limit photosynthetic rate and vary in response to the environment. We investigated the response of gm to changes in CO2 in two Arabidopsis genotypes (including a mutant with open stomata, ost1), tobacco and wheat. We combined measurements of gas exchange with carbon isotope discrimination using tunable diode laser absorption spectroscopy with a CO2 calibration system specially designed for a range of CO2 and O2 concentrations. CO2 was initially increased from 200 to 1000 ppm and then decreased stepwise to 200 ppm and increased stepwise back to 1000 ppm, or the sequence was reversed. In 2% O2 a step increase from 200 to 1000 ppm significantly decreased gm by 26–40% in all three species, whereas following a step decrease from 1000 to 200 ppm, the 26–38% increase in gm was not statistically significant. The response of gm to CO2 was less in 21% O2. Comparing wild type against the ost1 revealed that mesophyll and stomatal conductance varied independently in response to CO2. We discuss the effects of isotope fractionation factors on estimating gm.
For C3 photosynthesis to occur, CO2 has to diffuse from the atmosphere through stomata, intercellular air space and the liquid phase into the chloroplast for fixation by ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco). Both CO2 diffusion and the biochemical processes involved in CO2 fixation influence the carbon isotope discrimination during photosynthesis (Farquhar, Ehleringer & Hubick 1989). This affects isotopic signatures of atmospheric CO2, which have become an important tool for monitoring changes in global CO2 exchange processes (Yakir & Sternberg 2000). The conductance to CO2 diffusion from intercellular airspace to the chloroplast has been termed mesophyll conductance, gm, and has been shown to have a major influence on the carbon isotope discrimination, which depends on the ratio of intercellular to ambient CO2 (pi/pa) as well as CO2 assimilation rate (Evans et al. 1986, 2009; von Caemmerer & Evans 1991; Flexas et al. 2007, 2008).
It is not easy to measure the magnitude of gm. The two most common techniques rely on either combined measurements of gas exchange and chlorophyll fluorescence or combined measurements of gas exchange and the carbon isotope discrimination (Pons et al. 2009). There is an urgent need to assess to what extent gm is linked to leaf anatomical properties such as chloroplast surface area appressed to intercellular air space and cell wall thickness (Evans et al. 2009) and to what extent gm varies with short-term environmental perturbation (Flexas et al. 2007, 2008; Warren 2008). For example, there is controversy whether gm responds to perturbations in CO2 concentration (Flexas et al. 2007; Tazoe et al. 2009; Vrabl et al. 2009).
It is important that we understand the response of gm to CO2, for several reasons. Firstly, the widely used C3 photosynthesis model (Farquhar, von Caemmerer & Berry 1980) is increasingly being fitted to the partial pressure of CO2 at the sites of carboxylation and this generally assumes a constant value for gm (e.g. Ethier & Livingston 2004; Sharkey et al. 2007). Secondly, it alters the interpretation of atmospheric isotopic data that is used to infer CO2 fluxes between the biosphere and atmosphere (Suits et al. 2005). Thirdly, it may help in the discovery of ways to manipulate gm by helping quantify the roles of the cell wall, membranes and their protein pores and carbonic anhydrase (Evans et al. 2009).
Developments in tunable diode laser absorption spectroscopy (TDLAS) have improved our ability to make rapid measurements of the carbon isotope discrimination, Δ, concurrently with gas exchange (Flexas et al. 2006; Barbour et al. 2007; Uehlein et al. 2008; Bickford et al. 2009; Bickford, Hanson & McDowell 2010). Here we describe a system that allows Δ measurements to be made over a wide range of CO2 and O2 concentration using TDLAS. The aims of this study were to test the short-term response of gm to step changes in CO2 concentration and examine the reproducibility of estimates of gm through a day. We studied responses of gm to changes in CO2 in two Arabidopsis genotypes (including a mutant with open stomata; ost1), tobacco and wheat, at both 2% O2 where photorespiratory fractionations are minimized and 21% O2.
MATERIALS AND METHODS
Plant materials and growth conditions
Tobacco (Nicotiana tabacum cv Wisconsin 38) plants were grown in a growth cabinet, which was set to a 12 h photoperiod (25/20 °C day/night temperatures), 70% relative humidity and a photon flux density (PFD) at plant height of ca. 500 µmol quanta m−2 s−1 (Sunmaster, Warm Deluxe, Metal Halide 1000 W lamps, Streetsboro, OH, USA). Wheat (Triticum aestivum L. cv. Yecora 70) and Arabidopsis (Arabidopsis thaliana L. Heyhn.) Columbia-0 (Col-0), Landsberg erecta (Ler) genotypes and OPEN STOMATA 1 mutant (ost1), which are in the Ler background and have stomata that still respond to CO2 (Merlot et al. 2002; Mustilli et al. 2002), were grown in a greenhouse under full sunlight. Temperatures were controlled by an air conditioning system at 22/15 °C day/night. Plants were grown in a garden soil mix with a fertilizer (Osmocote, Scotts Australia Pty. Ltd., New South Wales, Australia) and watered daily.
Gas-exchange and carbon isotope measurements
Plants were transferred from the greenhouse or the growth cabinet to the laboratory (room temperature of 25 °C) in the early morning. One fully expanded leaf was placed across the 6 cm2 leaf chamber of the LI-6400 with a red-blue light-emitting diode (LED) light source (Li-Cor, Lincoln, NE, USA) and flow rate was set at 200 µmol s−1. For some measurements made with Arabidopsis Ler and ost1 mutant, a laboratory-constructed whole leaf chamber (115 × 110 × 25 mm depth) was used together with a red-green-blue LED light source (6400-18 RGB Light source, Li-Cor) and the LI-6400. The leaf chamber air was stirred with a tangential fan and boundary layer conductance to diffusion of water vapour was estimated to be 5 mol H2O m−2 s−1 (von Caemmerer & Evans 1991). Leaf temperature was controlled by a water jacket and it was measured with a chromel-constantan thermocouple (Omega Engineering, Manchester, UK) connected to the LI-6400.
Gas exchange was coupled to a tunable-diode laser absorption spectroscope (TDLAS, model TGA100A, Campbell Scientific, Inc., Logan, UT, USA) for on-line measurements of carbon isotope discrimination (Δ) (Bowling et al. 2003; Griffis et al. 2004). Input gases (N2 and O2) were mixed using mass flow controllers (OMEGA Engineering, Inc., Stamford, CT, USA) and split into various lines. Two lines supply the inlets of two LI-6400 consoles. Another line of the N2/O2 mixture was used to establish zero of the TDLAS throughout the day. 10% CO2 was added to the remaining fourth N2/O2 gas stream by a capillary gas mixing system to generate an air stream with approximately 1000 ppm CO2. This air stream was then used to generate six different CO2 concentrations (170, 330, 490, 660, 820, 970 ppm CO2) of the same isotopic composition (δ13CVPDB = −24.56 ± 0.06‰), which was used to calibrate the 13CO2 signal. (The absolute concentration of this gas mixed by capillaries varies slightly depending on the O2 concentration, but the resulting isotope calibration is independent of O2 concentration, see Supporting Information Fig. S1). Gas from a compressed air tank (380 ppm CO2) was measured to correct for gain drift of the TDLAS throughout the day. These calibration systems allow measurements to be made over a wide range of CO2 and O2 concentrations.
The carbon isotope composition (δ13C) was calculated as:
where RVPDB is 13CO2/12CO2 of the standard Vienna Pee Dee Belemnite (VPDB) and it is 0.0111797 (Griffis et al. 2004). The LI-6400 CO2 mixing system was used to generate different CO2 concentrations for gas exchange measurements. The δ13C of CO2 gas cylinders (δ13Ctank) used in the LI-6400 CO2 injector system was between −13 and −6‰. Reference and sample gases were collected from T junctions inserted into the reference gas tube and match valve outlet, respectively. These gases were dried by passing through a dryer assembly in the gas line, and then 12CO2 and 13CO2 concentrations were measured for each gas by the TDLAS. The sample cell of the TDLAS was maintained at 20 mbar and flushed with 150 mL min−1 by a gas stream, which changed every 20 s. Data were averaged over the last 10 s for the calculations.
where δ13Csam and δ13Cref are the carbon isotope compositions of the leaf chamber and reference gases of the LI-6400. ξ = Cref/(Cref − Csam), and Cref and Csam are the CO2 concentrations of dry air entering and exiting the leaf chamber, respectively, measured by the TDLAS.
An example of a typical measuring cycle is shown in Fig. 1. The initial staircase of six CO2 concentrations was generated from the 1000 ppm gas (Cal. #1). Uncorrected carbon isotope values declined slightly with increasing CO2 from −24.6 to −25.6‰ at 170 and 970 ppm CO2, respectively, and variability decreased with increasing CO2. Compressed air (Cal. #2; 380 ppm CO2) was used to adjust gain drift for the 12CO2 signal, and then the staircase of calibration CO2 was used to calibrate the 13CO2 signal in each measurement cycle. The calibration was applied to the reference and sample gases to calculate δ13Cref, δ13Csam and Δ for each LI-6400. Gases were measured every 4 min and the cycles were repeated eight times at a given CO2. The standard error of δ13Cref declined from 0.026‰ at 200 ppm to 0.012‰ at 1000 ppm whereas the standard error of Δ for tobacco averaged 0.15‰ and was independent of CO2 concentration (Supporting Information Fig. S2).
CO2 responses were measured using tobacco, Arabidopsis and wheat leaves. At first, photosynthesis was measured for approximately 40 min at irradiance of 1500 µmol quanta m−2 s−1, leaf temperature 25 °C, inlet CO2 of 400 ppm, and 21% O2. Then, O2 was changed to 2% to reduce the carbon isotope fractionation associated with photorespiration, and the oxygen effect on infrared analysis was corrected for using the LI-6400 prompt routine based on Bunce (2002). In some instances, CO2 responses were measured in 21% O2 rather than 2% O2 using tobacco and Arabidopsis. CO2 response measurements began at 200 ppm CO2, and then CO2 was increased to 1000 ppm, decreased stepwise until 200 ppm and increased stepwise back to 1000 ppm. Other leaves were treated to the reverse sequence. Measurements were made over 30 min at each CO2. After measuring the CO2 response, the light was turned off and dark respiration (Rdark) was measured at both 200 and 1000 ppm CO2.
We also compared the LI-6400 2 × 3 leaf chamber with a custom-built whole leaf chamber sealed with an O-ring and flexible putty (Terostat). As we could not discern any chamber effects on the relationships between gm and pi (Supporting Information Fig. S3), we have not applied any corrections for gasket leaks.
Calculation of mesophyll conductance
A full description of discrimination during C3 photosynthesis was given by Evans et al. (1986). Because of the high boundary layer conductance in the leaf chamber, the boundary layer term was ignored, such that:
where a is fractionation factor caused by diffusion through stomata (4.4‰), ai is fractionation factor for hydration and diffusion through water (1.8‰), and b is fractionation factor for the carboxylation reaction by Rubisco and phosphoenolpyruvate carboxylase (PEPC). In this study, b of 30‰ was used. The parameter e is associated with fractionation from day respiration and needs to account for differences between growth and measurement δ13C (Wingate et al. 2007). Following Tazoe et al. (2009), we assume no fractionation by day respiration and calculate e = δ13Ctank − δ13Catmosphere. In this study, δ13Ctank was between −13 and −6‰ and the carbon isotope composition of CO2 in glasshouse air, δ13Catmosphere, was assumed to be −8‰. The parameter f (11.6‰) is the fractionation factor for photorespiration (Lanigan et al. 2008). pa, pi and pc are the CO2 partial pressure of the ambient, intercellular airspaces and sites of carboxylation within chloroplasts, respectively. CO2 partial pressure equals CO2 concentration multiplied by atmospheric pressure, which in Canberra averages 953 mbar. Day respiration, Rd, is assumed to be the same as dark respiration (Rdark). Γ* is the CO2 compensation point in the absence of Rd and we have assumed the value previously measured for tobacco (von Caemmerer et al. 1994) for all species, 3.68 µbar in 2% O2 and 38.6 µbar in 21% O2 at 25 °C. The symbol k is the carboxylation efficiency of Rubisco and k=Vc/pc where Vc is RuBP carboxylase activity per unit leaf area and Vc = (A + Rd)/(1 − Γ*/pc) (von Caemmerer & Farquhar 1981).
Mesophyll conductance, gm is defined as gm = A/(pi − pc). Substituting for k and replacing pc by pi − A/gm, Eqn 3 can be solved for gm and
To incorporate the boundary layer effect, a can be weighted by the relative drawdowns in the boundary layer and through stomata. The CO2 partial pressure at the leaf surface, ps, is given by
where gb is the boundary layer conductance. Then
In the results presented here, we ignored the boundary layer effect. We have estimated that it reduces our estimate of gm by only 2%, irrespective of pa (data not shown).
Diurnal measurements of mesophyll conductance at different CO2
We established a new gas exchange protocol to examine the short-term response of mesophyll conductance, gm, to a large step change in CO2 and to a range of CO2, also whether gm varied over a day when measured under the same conditions. Gas exchange measurements were made over 6 h and examples of diurnal traces from single leaves of assimilation rate, A, stomatal conductance, gs, and mesophyll conductance, gm, at different CO2 are shown in Figs 2 and 3. Histograms show mean and standard errors for three or four independent leaves for gm at 200 and 1000 ppm CO2. Two protocols with CO2 staircases in both directions were used to check whether gm varied with time of day and pretreatment. In 2% O2, the initial step change in CO2 from 200 to 1000 ppm significantly decreased gm by 40% in tobacco, 26% in Arabidopsis (Col-0) and 36% in wheat (Table 1). For tobacco and Arabidopsis, gm then increased as CO2 was decreased stepwise back to 200 ppm (middle of the day) and decreased again as CO2 was increased stepwise back to 1000 ppm (late afternoon, Fig. 2). On the other hand, when the protocol was reversed, the responses of gm to changes in CO2 were not statistically significant for any of the species (Fig. 3), despite a 30% increase in gm following the first step down from 1000 to 200 ppm (Table 1). Wheat showed a variable response after midday, frequently having dramatic reductions in A, gs and gm irrespective of the direction of the CO2 change.
Table 1. Response of gm to a step change in CO2 in Arabidopsis genotypes, tobacco and wheat in 2 and 21% O2
200 to 1000 ppm
1000 to 200 ppm
gm (mol m−2 s−1 bar−1)
gm (mol m−2 s−1 bar−1)
Measurements were made at an irradiance of 1500 µmol quanta m−2 s−1, and leaf temperature 25 °C at the beginning of the day. Asterisks show statistically significant changes in gm. Data are means ± SE. Change in the value of gm expressed as a percentage increase or decrease from the initial value.
Arabidopsis, Ler (n = 3)
0.301 ± 0.013
0.195 ± 0.012
(n = 4)
0.214 ± 0.020
0.195 ± 0.022
Arabidopsis, ost1 (n = 3)
0.203 ± 0.031
0.132 ± 0.026
(n = 4)
0.195 ± 0.022
0.183 ± 0.038
Arabidopsis, Col-0 (n = 3)
0.175 ± 0.015
0.130 ± 0.009
0.161 ± 0.004
0.207 ± 0.016
Tobacco (n = 4)
0.567 ± 0.008
0.339 ± 0.050
0.413 ± 0.036
0.569 ± 0.035
(n = 3)
0.333 ± 0.031
0.350 ± 0.050
Wheat (n = 3)
0.682 ± 0.072
0.433 ± 0.043
0.516 ± 0.086
0.650 ± 0.052
Response of mesophyll conductance to CO2
We started the measurement series with the largest step change in CO2. When CO2 was switched from 200 to 1000 ppm, A in tobacco rapidly increased from 17 to 38 µmol m−2 s−1 (Fig. 4). gs was gradually increasing in 200 ppm, and then decreased gradually following the change to 1000 ppm. By contrast, gm rapidly decreased when CO2 was increased from 200 to 1000 ppm. Similar qualitative responses were evident in Arabidopsis and wheat following the step increase in CO2. When CO2 was decreased from 1000 to 200 ppm, A rapidly decreased and gs gradually increased. gm also gradually increased, but the change was not statistically significant (Fig. 4, Table 1). When measurements were made at 21% O2, there was no response of gm to a change in CO2 from 200 to 1000 ppm in either of two Arabidopsis genotypes or tobacco (Table 1).
The relationships between the CO2 partial pressure of the intercellular airspaces, pi, and A, gs, and gm for all three species are shown in Fig. 5. For wheat, the first staircase of CO2 change of the day was used because of the dramatic reductions in A, gs and gm in late afternoon (see Fig 2 & 3). In tobacco and Arabidopsis Ler genotype, the responses were also measured in 21% O2. The response of A to CO2 differed between 2 and 21% O2 as expected because of the suppression of photorespiration at low CO2, but gs showed similar declines with increasing pi. In 2% O2, gm decreased as pi increased whereas it was independent of pi in 21% O2 (Fig. 5).
We also measured the CO2 response in Arabidopis Ler genotype together with the ost1 (open stomata 1) mutant, which has more open stomata (Fig. 6). Measurements were made in both 2% and 21% O2 and the response of gm to CO2 was much less in 21% O2 compared with 2% O2 (Fig. 6), which was similar to the result in tobacco (Fig. 5). The ratio of intercellular CO2 to ambient CO2 (pi/pa) and gs were higher in ost1 than wild type. The absolute difference in gm between ost1 and Ler in 2% O2 reflects differences in A, which were caused by differences in leaf ages.
When measured under the same conditions (1500 µmol quanta m−2 s−1, 360 µbar CO2 and 2% O2), the drawdown of CO2 partial pressure from internal air spaces to chloroplasts (pi − pc) was less in tobacco and wheat compared with all Arabidopsis genotypes (Fig. 7).
Effect of fractionation factors, b, e and f on gm
The estimates of gm depend on the assumed fractionation factors for Rubisco (b) photorespiration (f) and respiration (e). Using two typical CO2 responses of gm in 21% and 2% O2, we assessed the sensitivity of gm to changes in values assumed for the various fractionation factors (Fig. 8). The value chosen for the Rubisco fractionation factor b had the largest effect on the both the absolute value of the calculated gm and the CO2 dependence of gm at 2% O2. This is probably tied to the fact that pi/pa increased with increasing CO2 in 2% O2 whereas pi/pa was almost constant under 21% O2. The effects of varying the fractionation factors e and f were small compared with the impact of b. The photorespiratory fractionation factor f affected the estimates of gm primarily at low CO2 and 21% O2. The value of e had only a small effect on gm in either 21% or 2% O2. A spreadsheet is provided in the Supporting Information to illustrate the impact of any combination of b, e, f and Rd on the estimation of gm.
Mesophyll conductance, gm, has garnered attention because gm limits photosynthesis at ambient CO2 concentration. Recently, some reports have shown that gm varies with CO2 concentration (Flexas et al. 2007; Vrabl et al. 2009). Variation in gm may be caused by gating of CO2 permeable aquaporins (cooporins, Terashima et al. 2006) located in the plasma membrane and inner envelope of chloroplasts (Uehlein et al. 2008). However, the effects of CO2 concentration on the cooporins are still unknown and there is still controversy about whether gm responds to CO2 concentration.
TDLAS enabled us to repeatedly estimate gm using the carbon isotope discrimination method. Careful analysis of the data is necessary because various parameters could affect the calculation of gm. We examined the effect of isotopic and photosynthetic parameters on gm in detail to establish the methodology of calculating gm using the TDLAS and reassessed the responses of gm to CO2 using tobacco, Arabidopsis and wheat.
The apparent increase in uncorrected δ13C of the calibration gas #1 with decreasing CO2 concentration (Fig. 1) is associated with a zero correction. The 13CO2 signal is deliberately set to have a small positive value at zero, which is determined and subtracted by the calibration routine. δ13C is independent of both CO2 and O2 concentration once the calibration routine has been applied (Supporting Information Fig. S1). The larger variance of δ13C at low CO2 is associated with calculating a ratio from smaller signals (Supporting Information Fig. S2a). This is offset by a decrease in ξ, which results in variance of Δ being independent of the CO2 concentration with the standard error averaging 0.16‰ for a tobacco leaf (Supporting Information Fig. S2b).
An example of the 13CO2/12CO2 in the air mixed by the LI-6400 entering the leaf chamber (Re) was 0.01105 and 0.01103 at 50 and 960 ppm CO2, respectively (data not shown, see also Supporting Information Fig. S1b). This contrasts with our previous data obtained using membrane inlet mass spectrometry, where Re appeared to decline at low CO2 concentration (see fig. 1a in Tazoe et al. 2009).
Effect of instantaneous change in CO2 on gm
The TDLAS enabled us to estimate gm more quickly than previously when CO2 had to be collected cryogenically. This has allowed us to measure gm following a step change in CO2 concentration. In response to an increase in CO2, both A and gm changed rapidly, whereas gs responded more slowly in all plant species (Figs 2 & 4). On the other hand, when CO2 was decreased from 1000 to 200 ppm, gm responded more slowly, taking 10 min to stabilize (Fig. 4). This slower response was similar to that of gs. We cannot explain why the speed of the CO2 response of gm differed depending on the direction of CO2 change. However, because this CO2 response of gm was observed in tobacco, Arabidopsis and wheat, it could be a common feature in plant leaves.
Response of gm to CO2
We are aware of only four published studies where gm has been compared in different CO2 concentrations using the carbon isotope method. Loreto et al. (1992) measured Quercus rubra and Xanthium strumarium with the isotope method and gm was 30% less at 750 µbar pi than at ambient CO2 in 21% O2, although this was only significant for X. strumarium. Flexas et al. (2007) reported that gm estimated by the isotope method was lower at high pi compared with ambient CO2 in 21% O2 by 54% for Nicotiana tabacum and 34% for Nicotiana sylvestris. The response to CO2 was larger in 2% O2, increasing to 49% for N. sylvestris. For Helianthus annuus, Vrabl et al. (2009) found that gm decreased by 74% as pi increased from 200 to 800 ppm, but was also lower at 50 ppm CO2 in 21% O2. For Triticum aestivum in 2% O2, we previously found that gm varied little with pi between 80 and 500 µbar (Tazoe et al. 2009). Given the variability in results, does the response of gm to CO2 differ between plant species?
To obtain unified information involving the effect of CO2 on gm, we measured gm over a wide range of CO2 and O2 concentrations using tobacco (Nicotiana tabacum), Arabidopsis (Arabidopsis thaliana) and wheat (Triticum aestivum). In 2% O2, as pi increased from low (92–117 µbar) to high (593–693 µbar), gm decreased by 32% for tobacco, by 28% for Arabidopsis Col-0 and by 29% for wheat (Fig. 5). In N. tabacum and A. thaliana (Col-0), the responses of gm to CO2 have also been measured by a chlorophyll fluorescence method in Flexas et al. (2007). Flexas et al. (2007) observed gm decreased by 35% for N. tabacum and by 83% for A. thaliana as pi increased from low (120 µbar) to high (560 or 660 µbar) in 21% O2. For N. tabacum, the response to CO2 was similar using either isotope discrimination or chlorophyll fluorescence to calculate gm (Flexas et al. 2007). Decreases in gm with increasing pi have also been found for 5 species of Banksia, again using chlorophyll fluorescence to calculate gm under 21% O2 (Hassiotou et al. 2009).
While wheat appears to be less sensitive to CO2 than other species, the results from Fig. 5 differed slightly from our previous study (Tazoe et al. 2009). There was a pronounced drop in photosynthetic properties, A and gs, later in the day (Figs 2 & 3). Such a pronounced drop was not observed in our previous study where instead, gs gradually declined with time irrespective of the direction of CO2 change (data not shown). We have been unable to explain why gs dropped later in the day. It does not appear to be associated with growing conditions or leaf/plant age but may reflect being measured in a gas exchange chamber for a longer time.
The response of gm to CO2 in 21% compared with 2% O2 is diminished either partially (Flexas et al. 2007) or to a large extent (Figs 5 & 6). It is hard to imagine that O2 concentration could directly affect gm. The effect of the photorespiratory fractionation factor f on Δ has been studied (Tcherkez 2006; Lanigan et al. 2008; Tcherkez et al. 2010). The contribution by photorespiration is greater in 21% than 2% O2 and increases at lower CO2 concentrations. We calculated gm using various reported values of f (Rooney 1988; Igamberdiev et al. 2004; Lanigan et al. 2008), but none could account for the effect of O2 on the CO2 dependence of gm (Fig. 8). Assuming a value of 0 for f (as in Vrabl et al. 2009) would result in lower estimates of gm. The impact on Δ from the substrate source and the time it takes for the photorespiratory pools to turn over needs to be considered (Tcherkez et al. 2010). Our experiments lasted the whole day and returned to both high and low CO2 at different times. If anything, there may have been a tendency for gm to decrease slightly over the day, but we used both increasing and decreasing CO2 concentrations for a given leaf and reversed the order these were applied through the day between leaves to correct for any bias this may have caused.
Effect of stomatal conductance on gm
To calculate gm, one needs to know pi, which raises the possibility that gm and gs are not independent. Water and salt stresses have been found to reduce both gs and gm (Flexas et al. 2007; Warren 2008). Warren (2008) showed that gs could be altered by changing VPD without affecting gm in Eucalyptus regnans, Phaseolus vulgaris and Solanum lycopersicum. Vrabl et al. (2009) found that treating Helianthus annuus leaves with ABA could reduce gs by 50% without affecting gm.
We used the ost1 mutant, which is insensitive to ABA and stomata do not close at high CO2 and in the dark (Merlot et al. 2002; Mustilli et al. 2002), to investigate the effect of gs on gm. In 21% O2, gs in ost1 was higher than wild type over all pi, which led to the higher pi/pa, but gm was identical (Fig. 6). The increase in gs at low pi was not associated with any increase in gm. In 2% O2, the ost1 leaves had similar gs but lower gm than wild type. The lower gm was associated with a lower mean photosynthetic capacity for the ost1 leaves that were measured in 2 versus 21% O2. Consequently, regardless of O2 concentration, we found that gm could vary independently of gs.
The growing number of experiments with a variety of species that have been able to vary gs using VPD, ABA and mutants without affecting gm, suggests that gs and gm can change independently from one another, both in the short-term and at steady state.
Effects of fractionation factors on gm
The most important factor in the calculation of gm using carbon isotope discrimination is that for Rubisco, b. Does it vary significantly between plant species? Estimates vary between 28 and 29.5‰ in Nicotiana tabacum (Evans et al. 1994; McNevin et al. 2007), 29‰ in Phaseolus vulgaris (von Caemmerer & Evans 1991), 30‰ in Spinacia oleracea (Roeske & O'Leary 1984) and 32‰ in Triticum aestivum (von Caemmerer & Evans 1991). We illustrated the effects of assuming various b-values ranging from 27 to 31‰ on the calculation of gm (Fig. 8, see also Supporting Information Table S1). The absolute value of the calculated gm increased as b decreased, regardless of O2 concentration, but in 2% O2, the response of gm to CO2 concentration also varied with b. Direct estimates of b for a range of species are needed.
The fractionation factor associated with day respiration, e, is also still uncertain. In our previous study (Tazoe et al. 2009), fractionation from day respiration was close to zero once we accounted for the difference between the isotopic composition under the growth and measurement conditions, as suggested by Wingate et al. (2007). Here, e was assumed to be δ13Ctank − δ13Catmosphere. By comparing various values of e, it was clear that when PFD is 1500 µmol quanta m−2 s−1, the effect of the factor e on gm was very small (Fig. 8). However, at low PFD (less than 150 µmol quanta m−2 s−1) when A is smaller, it was evident that Δ varied depending of the δ13C in the CO2 gas cylinders (δ13Ctank). The variation in Δ could be reduced by using e = δ13Ctank − δ13Catmosphere (Tazoe, von Caemmerer & Evans, unpublished data). The effects of e and f (described earlier) should be considered when A is small because of low CO2 or PFD.
The effect of CO2 on gm was re-examined in tobacco, Arabidopsis and wheat leaves using the TDLAS. gm tended to decrease with increasing CO2 in all plant species in 2% O2. While gm rapidly decreased when CO2 was stepped up from 200 to 1000 ppm, gm increased gradually when CO2 was stepped down from 1000 to 200 ppm. There was little change in gm when CO2 concentration was varied in 21% O2. The absolute value of gm calculated from carbon isotope discrimination depends on the values assumed for the fractionation by Rubisco, b. Photorespiration and day-respiration are expected to make only a slight contribution to the observed Δ, with their largest impact occurring at very low CO2 concentration.
We thank Steve Sargent and Joel Green for the design of the CO2 calibration system and installation of the TDL and Chin Wong and Peter Groeneveld for other construction, particularly the N2/O2/CO2 mixing system. We thank Stephanie McCaffery for cultivating Arabidopsis and wheat used in this study and Prof Ichiro Terashima for valuable discussions. Y.T. was supported by Research Fellowship of the Young Scientists of the Japan Society for Promotion of Science (JSPS). This work was funded by ARC Discovery grant DP0771413.