Mesophyll conductance to CO₂, assessed from online TDL-AS records of ¹³CO₂ discrimination, displays small but significant short-term responses to CO₂ and irradiance in Eucalyptus seedlings.

Mesophyll conductance (g(m)) is now recognized as an important limiting process for photosynthesis, as it results in a significant decrease of CO(2) diffusion from substomatal cavities where water evaporation occurs, to chloroplast stroma. Over the past decade, an increasing number of studies proposed that g(m) can vary in the short term (e.g. minutes), but these variations are still controversial, especially those potentially induced by changing CO(2) and irradiance. In this study, g(m) data estimated with online (13)C discrimination recorded with a tunable diode laser absorption spectrometer (TDL-AS) during leaf gas exchange measurements, and based on the single point method, are presented. The data were obtained with three Eucalyptus species. A 50% decrease in g(m) was observed when the CO(2) mole fraction was increased from 300 μmol mol(-1) to 900 μmol mol(-1), and a 60% increase when irradiance was increased from 200 μmol mol(-1) to 1100 μmol mol(-1) photosynthetic photon flux density (PPFD). The relative contribution of respiration and photorespiration to overall (13)C discrimination was also estimated. Not taking this contribution into account may lead to a 50% underestimation of g(m) but had little effect on the CO(2)- and irradiance-induced changes. In conclusion, (i) the observed responses of g(m) to CO(2) and irradiance were not artefactual; (ii) the respiratory term is important to assess absolute values of g(m) but has no impact on the responses to CO(2) and PPFD; and (iii) increasing irradiance and reducing the CO(2) mole fraction results in rapid increases in g(m) in Eucalyptus seedlings.

Introduction

During photosynthesis, CO2 diffuses from the atmosphere (at a mole fraction Ca) to the sites of carboxylation (Cc) inside the chloroplasts (Gaastra, 1959; Farquhar ). CO2 crosses the leaf boundary layer and traverses stomatal pores into the substomatal cavity. CO2 then diffuses through the gas phase between mesophyll cells before reaching cell walls, where it is solubilized. In the liquid phase, CO2 crosses the plasma membrane, the cytosol, and finally the chloroplast membranes to reach the sites of carboxylation in the chloroplast stroma. Most of the carboxylation is by RubisCO (ribulose-1,5-bisphosphate carboxylase/oxygenase), while a small fraction (estimated at 5%) is by phosphoenolpyruvate carboxylase (PEPc) in the cytosol. The mesophyll conductance to CO2 (gm) represents the conductance from the substomatal cavities at a mole fraction Ci, to the sites of carboxylation, and includes gas and liquid phase transfer.

In early leaf photosynthesis models, gm was considered to be infinite. This was an explicit assumption in the original formulation of Farquhar's model of C3 photosynthesis. However, many subsequent studies showed that there was a difference between Ci and Cc, suggesting that gm might be finite (Evans ; Lloyd ) and that it affects estimation and interpretation of photoynthetic parameters such as the maximal carboxylation activity of RubisCO (Vcmax) or the maximal light-driven electron flux (Jmax) (Epron ; Niinemets a). gm can limit the rate of photosynthesis by 25–30% (Epron ) and this limitation may be as large as the one due to stomatal conductance (Warren and Adams, 2006; Flexas c). The range of gm variation among species is similar to that of stomatal conductance, from 0.005 mol m−2 s−1 up to 0.5 mol m−2 s−1 (see the reviews by Evans and Von Caemmerer, 1996; Flexas ), with high gm occurring in herbaceous annuals and lower values in evergreen gymnosperms. gm is apparently influenced by leaf traits such as thickness, tissue density (and therefore leaf dry mass per unit area, LMA; for meta analyses, see Flexas ; Niinemets ), cell wall thickness, or the proportion of gaseous versus liquid mesophyll phases (Piel ; Terashima ; Evans ). Such morphological leaf properties would confer stable gm in the short term (e.g. a few days).

Since the early 1990s, gm was shown to be affected by environmental factors. At first, changes were thought to be rather long-term changes such as, for instance, the decline induced by leaf ageing (Scartazza ), by a gradual drought stress (Cornic ; Roupsard ; Warren, 2008), or by salinity (Bongi and Loreto, 1989). More recently, short-term changes were shown to occur in response to temperature (Bernacchi ; Pons and Welschen, 2003; Warren and Dreyer, 2006; Warren, 2008). Similarly rapid responses of gm (at the minutes scale) were reported to occur under varying CO2 mole fractions. A review by Flexas reported that gm decreases with increasing CO2 (data from Harley ; During, 2003; Flexas ). This pattern was also observed by Hassiotou among seven Banksia species, by Vrabl in Helianthus annus, and by Bunce (2010) in Glycine max and Phaseolus vulgaris. Tazoe recently observed the same decreasing pattern among three species with contrasting photosynthetic capacities. Only a few studies have tested the rapid response of gm to irradiance. Flexas reported an increase of gm with increasing irradiance (data from Gorton ; Flexas ).

Nevertheless, there is still no consensus about the reality of such rapid responses of gm to CO2 and irradiance. Several studies reported gm to be stable in response to changes in CO2 (Von Caemmerer and Evans, 1991; Loreto ; Tazoe ) and irradiance (Tazoe ; Yamori ). Some of the discrepancy may be due to measurement accuracy and/or artefacts. This can be true with the two main methods used to estimate gm: the fluorescence/gas exchange technique (Loreto ) and the isotopic discrimination method (Pons ). One of the complications with the isotopic method, which is used here, is that the contribution of 13C discrimination during respiration and photorespiration needs to be taken into account (see the model description below). Some studies have ignored discrimination during respiration and photorespiration (Flexas ; Vrabl ), or approximated respiration in the light by that in the dark (Tazoe , 2011). The fractionation factors associated with respiration and photorespiration can be taken into consideration using recent estimates (Lanigan ). An alternative approach is to limit the impact of fractionation during photorespiration by using low O2 in the measurement atmosphere (Tazoe ).

To date, there is no consensus regarding whether gm responds rapidly to irradiance and the CO2 mole fraction, despite the importance of this for interpreting responses of photosynthesis to environmental variables. The aim of this research was to assess whether short-term variations of gm occur in response to changes in irradiance and the CO2 mole fraction. Rapid response of gm was assessed by recording online 13CO2 discrimination during photosynthesis with a custom-built photosynthesis chamber adapted to a LI-COR 6400, and coupled to a TDL-AS (or tunable diode laser absorption spectrometer). To examine the ubiquity of responses, three species of Eucalyptus with differing rates of photosynthesis were used, and the respiratory term of online 13C discrimination was taken into consideration to reinforce the findings. Measurements were carried out on species with contrasting photosynthetic capacity to check whether gm variability is species dependent and how gm is related to A and gs within and among species.

Materials and methods

Plant material

Three species in the genus Eucalyptus were used: E. globulus Labill., E. saligna Sm., and E. sieberi L. A. S. Johnson. Plants were grown for 4 months in a naturally illuminated greenhouse at an average daily temperature of 25 °C in 8.0 l pots filled with compost-based substrate. They were watered with automatic drip irrigation. At the time of the experiment they were ∼60 cm high.

Gas exchange measurements

The response of the net CO2 assimilation rate (A, μmol m−2 s−1) to variations in substomatal CO2 mole fraction (Ci) and photosynthetic photon flux density (PPFD, μmol m−2 s−1) was measured in each of four replicate plants of the three species, for each treatment (e.g. 2 treatments×3 species×4 replicates=24 plants). All measurements were made on the youngest fully expanded leaves. Plants were transferred from the greenhouse to the laboratory at 25 °C. To assess photosynthesis and discrimination against 13CO2 simultaneously, a LI-6400 portable gas exchange system (LI-COR, Lincoln, NE, USA) equipped with a custom-built chamber of 18 cm2 and coupled to a TDL-AS (TGA100A, Campbell Scientific, Logan, UT, USA) was used.

Before measurements, leaves were exposed to 1000 μmol m−2 s−1 PPFD and 400 μmol CO2 mol−1 for 30 min to induce photosynthesis and stomatal opening. CO2 responses of photosynthesis were measured under a constant PPFD of 1000 μmol m−2 s−1 while the CO2 mole fraction for the reference gas (Ce) was sequentially set at 300, 500, 700, and 900 μmol mol−1. The PPFD responses at 200, 500, 800, and 1100 μmol m−2 s−1 were measured under a constant CO2 mole fraction of 400 μmol mol−1. During each step of the CO2 or PPFD responses, at least 20 min was allowed for A and gs to reach a steady state, and three individual points separated by 180 s were recorded. Each set of three individual data points was averaged to characterize the response at one step in the response curve. The standard error of the three individual points was used to assess the analytical variability.

Air flow into the chamber was set at 400 μmol s−1, and leaf temperature at 25 °C. Irradiance was provided by a LI-COR RGB light source (6400-18, LI-COR-0128), which covered the entire chamber surface. A subsample of air from the sample and reference gas lines of the LI-6400 was diverted to the TDL-AS. The TDL-AS measured sequentially gas from two calibration tanks, the LI-6400 reference gas, and finally the LI-6400 sample gas. Each intake was measured for 45 s, with the first 15 s ignored to minimize carryover and enable stabilization between intakes. The total time for a sequence of measurements was therefore 180 s.

The TDL-AS gas was connected through a ‘T’ tubing to the reference tube of the LI-6400 between the console and the IRGA. In the same way, the TDL-AS intake of sample gas was connected in between the LI-6400 chamber exhaust. The TDL-AS was set to continuously withdraw 150 ml min−1 (∼102 μmol s−1) from each of the sample and reference fluxes of the LI-6400. This withdrawal of air was much smaller than the flow through the LI-6400 chamber, which means that the TDL-AS could sample air from the LI-6400 while maintaining a positive pressure in the LI-6400 chamber. The existence of a positive pressure inside the LI-6400 chamber was checked through the curvature of the propafilm covering the top of the chamber. TDL-AS data were matched with LI-6400 data by taking into account the time lag of 37 s that was recorded between the chamber and the TDL-AS. The TDL-AS data were only used to record the isotopic composition of the reference and sample gas, while all photosynthesis parameters were estimated from the LI-6400 data.

Isotopic measurements and system testing

Discrimination by a leaf was assessed by measuring the isotopic composition in reference (δ13Ce, ‰) and sample gas (δ13Co, ‰). The isotopic composition (δ13C) was expressed as:where Rs is the isotopic ratio (R=13CO2/12CO2) of the sample and RVPDB is the isotopic ratio of Vienna Pee Dee Belemnite (VPDB, 0.0112372).

The TDL-AS was calibrated with two tanks (T1 and T2) with CO2 concentrations of 419±10 μmol mol−1 and 290±7 μmol mol−1 [mean ± the confidence interval (CI)], respectively, given by the provider and checked with a recently factory-calibrated LI-8100 IRGA (LI-COR). The isotopic composition of the each tank was measured by sampling air into 12 ml exetainers (Labco Limited, Buckinghamshire, UK) and analysed via the gas-bench inlet of an IRMS (Delta S, Finningan, Bremen, Germany) at INRA Nancy (n=10 exetainers per tank). For T1 and T2, δ13C was –36.6±0.08‰ and –36.9±0.2‰ [mean ± standard deviation (Sd), n=10], respectively. Absolute values of 12CO2 and 13CO2 were, respectively, 414.51 μmol mol−1 and 4.487 μmol mol−1 for T1 and 286.89 μmol mol−1 and 3.104 μmol mol−1 for T2, considering the CO2 mole fraction indicated by the provider took into consideration both isotopologues. A linear interpolation was used between these two points for each isotopologue. For further calibration, a range of CO2 mole fractions of 200, 300, 500, 700, 1000, 1500, and 2000 μmol mol−1 was generated using the CO2 injector of the LI-6400 fed with the same CO2 cartridge during the whole test. CO2 mole fractions above the calibration range led to a small deviation of +2.5‰. A second order polynom was fitted to describe the deviation of apparent δ13C from reference values measured at Ce=300 μmol mol−1 (because this is within the calibration range) along the extended CO2 range (δ13C=0.000007[CO2]2–0.0035[CO2]–5.2922; R2= 0.92, n=140). All TDL-AS values were corrected and a stable δ13C signal was obtained along the extended CO2 mole fraction range.

The noise of the system was assessed by observing the Sd of δ13C values within each CO2 step. An average Sd of 0.16 ‰ was observed for δ13C (n=21). These values were used for the computation of the standard deviation of the observed discrimination by the leaf (SdΔobs, see below).

Finally the empty photosynthesis chamber was used to test the absence of a δ13C difference between the inlet and outlet that might be caused by leaks and/or CO2 adsorption/desorption processes. It was observed that the δ13Ce–δ13Co difference was stable along the CO2 mole fraction range used during the test, and it was confirmed that the empty chamber did not affect the isotopic composition of the air (i.e. leaks and CO2 adsorption/desorption processes were negligible). The observed δ13Ce–δ13Co difference was much smaller than the δ13C difference recorded during actual measurements (Fig. 1), although under specific conditions (low photosynthesis) the two could overlap. A data filtering procedure was therefore implemented (see below).

Fig. 1.

Boxplots of the difference in δ13C between the outlet and inlet of the 18 cm2 chamber for (i) an empty chamber with inlet air CO2 varying from 200 μmol mol−1 to 1000 μmol mol−1 (n=10); (ii) the whole set of measurements under varying CO2 concentrations (300 μmol mol−1 to 900 μmol mol−1), n=40; and (iii) the whole set of measurements under varying PPFD (from 200 μmol m−2 s−1 up to 1100 μmol m−2 s−1), n=40. The middle line represents the median, the upper and lower box limit the 75% and 25% quartiles, respectively, and whiskers represent the extreme value.

Model description

The observed discrimination (Δobs) is usually calculated following Evans :where:

ξ is the ratio of CO2 entering the chamber over the CO2 drawdown induced by the leaf.

Δ is the result of discrimination by diffusion processes during CO2 movement from the atmosphere to the chloroplast, and biochemical fractionation during carboxylation processes. Each fractionation step is characterized by a fractionation factor (due to diffusion or biochemistry) weighted by the gradient of concentration. In the complete form, Δ is predicted by (Evans ):where:

• ab is the fractionation during CO2 diffusion in the boundary layer (2.9‰, Evans );

• a is the fractionation during CO2 diffusion in air through stomata into the leaf (4.4‰, O'Leary, 1981);

• es is the fractionation occurring when CO2 is dissolved in the cell solution (1.1‰ at 25 °C, O'Leary, 1981);

• ai is the fractionation during CO2 diffusion in the liquid phase (0.7‰, O'Leary, 1981);

• b is the discrimination during carboxylation, and is dependent on fractionation by both RubisCO (b3=30‰) and PEPc (b4= –5.7‰),

b is computed as:

where β (between 0.05 and 0.1) is the relative amount of carbon fixed by PEPc (Farquhar and Richards, 1984). In the present experiment a value of b=28‰ was used; that is, a β value of 0.055.

• f denotes overall discrimination during photorespiration. The value of f was set at 11‰, according to the theoretical approach of Tcherkez (2006), which was confirmed later by Lanigan ;

• e denotes overall fractionation during day respiration relative to photosynthetic products (Rd), and can vary between –10‰ and +10‰ (Ghashghaie ). e was set at 1‰ before correction following Wingate ;

• k is the carboxylation efficiency computed as (Farquhar );

• Γ* is the CO2 compensation point in the absence of day respiration (Brooks and Farquhar, 1985).

The estimation of g is based on the difference between the observed discrimination by the leaf (Δobs) and the discrimination predicted from the simplified form of the model (Δi) in which decarboxylation terms are ignored and gm is considered to be infinite:

With this ‘single point method’ first developed by Lloyd and recently described by Pons , gm can be estimated from a single value of Δobs:

Model parameters

Rd and Γ* were estimated with the ‘Laisk method’ (Viil ) for the three species (i.e. from the intersection of three A–Ci curves recorded at PPFDs of 100, 50, and 25 μmol photon m−2 s−1 and Ce of 125, 100, and 50 μmol mol−1). The ‘Laisk method’ provides Ci* or the ‘apparent’ CO2 compensation point in the absence of day respiration (Von Caemmerer ), and was used as a proxy of Γ*. Because these measurements are sensitive to errors due to CO2 leak diffusion (low A and low Ca compared with the atmosphere), the potential CO2 leaks due to diffusion though chamber gaskets were estimated (Flexas a; Rodeghiero ). A diffusion coefficient of the gaskets was computed with the procedure provided in the user manual of LI-COR, and a value of 0.938 μmol s−1 was found (while it usually is 0.46 for smaller 6 cm2 chambers). This correction was incorporated into all gas exchange computations used for Γ* and Rd estimations. The computed values of Γ* did not differ between species, so the mean (Γ*=38.7±0.51 μmol mol−1, n=13) was used as a common value for the three species. For E. globulus, E. saligna, and E. sieberi, Rd was 0.41±0.09, 0.31±0.09, and 0.68±0.07 μmol m−2 s−1, respectively.

As the isotopic signature of the reference gas provided by cartridges differed from that used by the leaves for earlier photosynthesis, e was replaced by e'=e+δ 13Ctank–δ13Catmosphere (Wingate ). δ13C in the cartridge was measured with the TDL-AS at the chamber inlet. In the present case, the LI-6400 was fed with compressed CO2 cartridges with δ13Ctank varying between –1‰ and –4‰ except for two cartridges with δ13Ctank= –19‰. e' therefore varied between +4‰ and +6‰, except for two plants where it was –11‰.

Cc, the CO2 concentration at the site of carboxylation, was calculated from Fick's Law as:

Finally, the total leaf conductance to CO2 was calculated following Ball (1988), assuming resistances in series as:

where the stomatal conductance to CO2 is g=gsw/1.6.

Propagation of uncertainty from measurement to Δ calculation and data filtering

The uncertainty (standard deviation) of Δobs due to the finite precision of δ13C measurements was estimated. This was achieved by propagating uncertainty (standard deviations) of δ13Ce and δ13Co through the equations estimating Δobs (see Appendix for details):

The propagated uncertainty in Δobs (i.e. SdΔobs) was used as the basis for a filter to remove unreliable estimates of Δobs. Computation of gm is based on the difference between Δi and Δobs (i.e. Δi–Δobs), thus it was reasoned that gm estimates would be unreliable if the difference Δi–Δobs was smaller than SdΔobs. Consequently all values where Δobs+SdΔobs >Δi were rejected. This filter was applied to the individual points in the data set, rejecting 33 among 238 points.

Statistical analyses

All statistical analyses were performed with R (R Development Core Team 2010, http://www.R-project.org). Mixed-effect linear models were run to assess species and treatment effects on A, gs, gm, Δi–Δ, Cc, and the Ci–Cc drawdown, as shown in Table 1. For the CO2 treatment, ‘species’ (as a factor) and ‘Ci’ (as a covariate) were incorporated into the model as fixed effects, and ‘individual within species’ as a random effect. For variations of Cc and the Ci–Cc drawdown, Ca was used as covariate. For PPFD treatment, species and PPFD were set as factors. Normality and heteroscedasticity were graphically checked with QQ-plots. In the case of heteroscedastic data, the mean was weighted as a function of the variance. In the case of non-normal distribution, variables were log-transformed. The species×treatment interaction was tested for each procedure, and was removed from the model when not significant. In the absence of interaction, comparison of the intercepts was performed to assess differences between species. In the case of interaction, slope comparisons were performed to test if species responses differed from each other. Significance was accepted at P <0.05. Mean least squares regression was used to assess the correlation between variables (R2 and P-value).

Table 1.

Mixed effects model for A, gs, gm, and Δi–Δ

		Ci				Ca		PPFD
		A	gs	gm	Δi–Δ	Cc	Ci–Cc	A	gs	gm	Δi–Δ	Cc	Ci–Cc
Variable	F	168.11	38.82	18.67	3.70	165.01	102.90	86.01	34.23	6.41	NS	NS	NS
	P	<0.001	<0.001	<0.001	(0.06)	<0.001	<0.001	0.001	<0.001	0.002	NS	NS	NS
Species	F	13.31	26.33	10.11	6.36	4.64	7.65	14.65	(2.91)	NS	NS	NS	NS
	P	0.002	<0.001	0.006	0.02	0.045	0.013	<0.01	(0.10)	NS	NS	NS	NS
Interaction	F	3.05	NS	3.44	NS	5.84	3.72	14.88	4.55	NS	NS	NS	NS
	P	(0.06)	NS	0.047	NS	0.008	0.038	<0.001	0.002	NS	NS	NS	NS

Species, Ca and Ci, or PPFD effects were incorporated into the model as fixed effects, and individual plant as a random effect. In the case of heteroscedastic data the mean was weighted as a function of the variance. For Ci and Ca, the degree of freedom (df) was 1, for PPFD df=3, and for species df=2. Significant values (P <0.05) are shown in bold.

Results

Variation of gm under changing Ce

The CO2 mole fraction was changed in the air entering the chamber in three steps from 300 μmol mol−1 to 900 μmol mol−1, inducing a range of Ci from 185 μmol mol−1 to 745 μmol mol−1. Net CO2 assimilation rate (A) was positively related to Ci and varied between 3 μmol mol−2 s−1 and 18 μmol mol−2 s−1, while stomatal conductance to water vapour (gs) was negatively related to Ci and varied between 0.02 mol m−2 s−1 and 0.8 mol m−2 s−1 (Fig. 2). Eucalyptus sieberi had significantly higher A and gs than E. globulus and E. saligna (t-test P <0.05, Fig. 2).

Fig. 2.

Relationships between the CO2 mole fraction in the substomatal cavities (Ci) and (A) net CO2 assimilation rate (A). (B) Stomatal conductance to water vapour (gs). (C) Difference between predicted and measured isotopic discrimination (Δi–Δ). (D) Mesophyll conductance (gm). Eucalyptus globulus is in black, E. saligna in grey, and E. sieberi in white. The SE is provided by the average of three measurements taken at 180 s intervals. Measurements were made at four levels of Ce on four plants per species, and were filtered against noisy values of δ13Ce–δ13Co.

The difference between δ13Ce and δ13Co decreased with increasing Ci (data not shown). Among all species, Δobs varied between 12‰ and 22‰ and was positively correlated to Ci/Ca (R2=0.79, P <0.001, data not shown). The difference between Δ calculated with infinite gm and no respiratory term (simple model) and observed Δ (Δi–Δobs) varied between 3‰ and 7‰ but showed no clear trend with Ci (Fig. 2, Table 1).

gm computed by taking into account the respiratory component of discrimination varied from 0.025 mol m−2 s−1 to 0.55 mol m−2 s−1 (Fig. 2). gm was larger in E. sieberi than in the other two species (t-test, P <0.05). gm was affected by Ci (Table 1), and decreased when Ci increased. Post-hoc tests revealed that E. globulus and E. saligna displayed gm–Ci slopes significantly different from zero. In E. sieberi, three individuals out of four showed a clear negative pattern when Ci increased, but not the fourth. The relationship between gm and gs was significant among all species (R2=0.54, P <0.001), and within E. globulus and E. saligna when treated separately (data not shown).

Cc and the Ci–Cc drawdown were, as expected, severely affected by Ca (Table 1). The Ci–Cc drawdown was ∼50 μmol mol−1 at Ca=200 μmol mol−1 and increased up to 200 μmol mol−1 at Ca= 900 μmol mol−1 (Fig. 3). When tested individually, all slopes of the responses of Cc and Ci–Cc to Ci were different from zero.

Fig. 3.

Left: CO2 mole fraction in the substomatal cavities (Ci, disks) and at carboxylation sites (Cc, triangles) as a function of Ca. The dashed grey line is the 1:1 relationship. Right: Ci–Cc drawdown as a function of Ca. Each point represents the mean of 1–3 analytical measurements ±SE, with E. globulus in black, E. saligna in grey, and E. sieberi in white.

Variation of gm with changing irradiance

A and gs increased significantly with irradiance (Table 1). A and gs were larger in E. sieberi than in the two other species (see Fig. 4, -test P <0.05). In each species, gs increased significantly from 200 μmol m−2 s−1 to 500 μmol m−2 s−1 PPFD, and then stabilized (Fig. 4). A and gs were positively correlated, for all species taken together (Fig. 5C, R2=0.76, P <0.001). Across the range of irradiance, Δobs varied between 11‰ and 20‰ and was positively correlated with Ci/Ca (R2=0.74, P <0.001, data not shown). Δi–Δobs varied between 3‰ and 9‰ but was not affected by irradiance or by species (Table 1, Fig. 4).

Fig. 4.

(A) Net CO2 assimilation rate (A). (B) Stomatal conductance to water vapour (gs). (C) Difference between predicted and measured isotopic discrimination (Δi–Δ). (D) Mesophyll conductance (gm). (E) CO2 mole fraction at carboxylation sites (Cc). (D) The Ci–Cc drawdown at four different levels of PPFD. Means ±SE (n=2–4 replicate plants), with E. globulus in black, E. saligna in grey, and E. sieberi in white.

Fig. 5.

Relationships between net CO2 assimilation rate (A) and (A) total leaf conductance to CO2 (gt, black squares), (B) mesophyll conductance to CO2 (gm, grey squares), and (C) stomatal conductance to CO2 (gsc, white squares) under varying PPFD.

gm varied between 0.04 mol mol−2 s−1 and 0.6 mol mol−2 s−1, and was positively related to PPFD. As for gs, the response consisted of a significant increase between 200 μmol m−2 s−1 and 500 μmol m−2 s−1 PPFD with a stabilization above this threshold. gm was positively correlated with gs (R2=0.36, P <0.001, data not shown) and A (R2=0.49, P <0.001, Fig. 5B), among all species. Significant A–gm and gm–gs relationships within each species were also detected (except for gs–gm in E. saligna).

Total leaf conductance to CO2 (gt) was strongly correlated to A (R2=0.83, P <0.001) as shown in Fig. 5A. Over the full set of irradiance values, Cc varied between 200 μmol mol−1 and 260 μmol mol−1, and the Ci–Cc drawdown between 40 μmol mol−1 and 60 μmol mol−1. None of these parameters displayed any variation with irradiance (Table 1), or with species.

Discussion

This study provides support to recent evidence that gm varies rapidly (within minutes) in response to environmental conditions. The rapid responses were observed under two different sources of variation: CO2 mole fraction and irradiance. In seedlings of three Eucalyptus species a modest but significant decrease of g with increasing CO2 mole fraction, and a significant increase with irradiance, was found. The effect was visible in the three species irrespective of the photosynthetic capacity.

Importance of the respiratory and photorespiratory terms in the estimation of gm

The isotopic method estimates gm from the 13CO2 discrimination during photosynthesis by comparing observed values with those derived from a model-based prediction of discrimination under infinite mesophyll conductance. This approach requires a high precision in discrimination records, which is now achieved by combining precise leaf gas exchange measurements with online TDL-AS records of changes in 13CO2/12CO2 in the atmosphere around the leaf (for a discussion of the technique, see Pons ; Tazoe ). One of the important problems with this method is the fact that several discrimination steps during photosynthesis, respiration, and photorespiration need be taken into account. In particular, a number of earlier studies omitted the respiratory and photorespiratory terms (Flexas ; Vrabl ). Moreover, the response of gm to CO2 was affected by the O2 mole fraction in the air; that is, by the occurrence of photorespiration during measurements: it was visible only under low O2 (Tazoe ).

In the present study, these two terms were incorporated into the gm estimates displayed in the results. Absolute values of gm were up to 50% larger when these terms were incorporated, and this enhancement was independent of the treatments applied (Fig. 6). Whether this potentially large change had an impact on the observed the effects of changing the CO2 mole fraction in the air or irradiance was tested: the response of gm to CO2 mole fraction and irradiance remained significant even when the respiratory term was omitted (see Table 2). the effects of substituting e (fractionation during day respiration) and f (fractionation during photorespiration) with extreme values (f=0‰, e= –10‰, e= +10‰) were similarly tested. Despite the fact that these changes resulted in significant differences of computed gm, they did not result in any loss of significance of the observed effects of CO2 or irradiance (data not shown). Vrabl compared gm estimated with the fluorescence method (which takes the respiratory terms implicitly into account) and with the isotopic method (without taking them into account) and found the same range of gm values and the same negative response to Ci. The present computations suggest that the respiratory terms can be important in the estimation of the absolute values of gm, but have only little influence on the observed CO2 or irradiance responses of gm. The respiratory terms of isotopic discrimination are unlikely to be responsible for the discrepancies among studies. Addressing the question of changing e and f during the CO2 and irradiance treatments was omitted in this discussion: up to now there is no reason to assume that these values are not stable across the whole range of environmental conditions.

Fig. 6.

Mesophyll conductance gm as computed by taking into account the contribution of respiration and photorespiration to 13CO2 discrimination versus without taking them into account (i.e. fractionation factors e and f set to 0 in Equation 7). Results obtained under changing CO2 (filled squares) or irradiance (open squares). The dashed line represents the 1:1 relationship. Each point represents the mean ±SE of a given measurement (n=3).

Table 2.

Impact of omitting the respiratory term in the discrimination model on the assessment of the impact of the CO2 mole fraction (Ci) or of irradiance on gm

		Ci		PPFD
		gm	gm (omitting the respiration terms)	gm	gm (omitting the respiration terms)
Variable	F	18.67	13.14	6.41	12.09
	P	<0.001	0.001	0.002	<0.001
Species	F	10.11	10.28	NS	4.89
	P	0.006	0.006	NS	0.03
Interaction	F	3.44	3.64	NS	NS
	P	0.047	0.04	NS	NS

Mixed effect model for mesophyll conductance computed including (as shown in Table 1) or omitting the respiration and photorespiration terms.

Response of gm to CO2 mole fraction

There was at maximum a 50% decrease in gm with increasing CO2 mole fraction (from 300 μmol mol−1 to 900 μmol mol−1). This response was general, as all three species displayed a similar response to CO2, with the exception of a single individual of E. sieberi. The results contrast with earlier studies reporting no response of gm to CO2. This was the case in Quercus ilex and Citrus aurantium (chlorophyll fluorescence method; Loreto ), Raphanus sativus (isotopic discrimination; Von Caemmerer and Evans, 1991), and Triticum aestivum (isotopic discrimination; Tazoe ). The present results confirm several studies that reported a negative relationship between gm and CO2 (Flexas ; Hassiotou ; Vrabl ; Bunce, 2010). The same magnitude of decrease of gm with Ci as in Flexas and Vrabl was found. Tazoe found an ∼30% decrease of gm when Ci increased under 1% O2 but no effect under 21% O2 in two Arabidopsis thaliana genotypes or in Nicotinia tabacum, while in the present study a significant effect was found even under 21% O2. The absence of a clear pattern of gm responses among studies could be interpreted as a species-dependent response of g to Ci, but no common trait seems to be shared by the ‘non-responsive’ versus the ‘responsive’ species.

There are a suite of methodological reasons and artefacts (e.g. choice or calculation of b, e, f, Rd, or Γ*) that might explain the discrepancy among studies. The case for e and f was discussed above. Measured values of Rd and Γ* (i.e. of its proxy Ci*) were used for each species rather than arbitrary values taken from the literature. Despite some uncertainties regarding the choice of b or f (Lanigan ; Pons ), it is concluded that the response of gm that was recorded here is unlikely to be an artefact.

Rapid responses of gm to CO2 could be mediated by aquaporins that might impact the permeability of plasma and chloroplast membranes to CO2 (Terashima and Ono, 2002), enhancing CO2 diffusion in the liquid phase. Flexas stated that the expression of NtAQP1 aquaporins can change gm values by 20–50% in N. tabacum. These variations of gm are of the same magnitude as those in the present study, but only direct measurements of aquaporin expression/activity could confirm the role of these proteins in the diffusion of CO2 through membranes. Establishing a parallel between short-term responses of gm and of the expression and activities of aquaporins is still an open area for research. Tholen also found that chloroplast movements can induce a variation of gm by 50% in A. thaliana. However, it is not known yet whether CO2 variations can directly mediate a displacement of chloroplasts.

A positive relationship was observed between gm and gs, as was also observed by Flexas . Interestingly, such a relationship was also found by Flexas when they compared plants overexpressing NtAQP1 aquaporin and controls. They suggested that the variation of gm primarily induced by manipulating NtAQP1 expression probably also led to an adjustment of gs and subsequently of A. This potentially indicates a physiological link between these two parameters (Flexas ; Vrabl ). Nevertheless, the precise signalling cascade that could cause a coordinated response of stomatal and of mesophyll conductance remains to be elucidated.

Response of gm to irradiance

Several studies observed an increase in gm with increasing irradiance. Flexas b) reported that gm increased in tobacco by ∼40% when irradiance increased from 250 μmol m−2 s−1 to 1000 μmol m−2 s−1. Data from Gorton , reanalysed by Flexas , also showed a positive effect of irradiance on gm. Hassiotou detected an ∼22% increase in six Banksia species as irradiance was switched from 500 μmol m−2 s−1 to 1500 μmol m−2 s−1. The present study corroborates this positive effect of irradiance, with an increase in gm by 60%, up to a plateau in irradiance reached between 500 μmol m−2 s−1 and 1100 μmol m−2 s−1, depending on the species. A slightly larger sensitivity of gm to irradiance than earlier studies was found. Sensitivity to irradiance could be (i) species dependent or (ii) due to different parameterization of the model enabling gm estimation. For instance, in the present case, removing the respiratory and photorespiratory terms in the estimation of gm led to a slightly smaller response of gm to irradiance (Table 2). A systematic analysis of the reported responses, with a standardized parameterization, would be very helpful.

As during the response to CO2, gm was positively correlated to gs. This relationship seems to be independent of the method used to vary net CO2 assimilation rates. A review by Flexas insisted that gs and gm responded in parallel to irradiance, CO2, temperature, and drought stress (Warren, 2008. On the other hand, Warren (2008 reported that gm was unaffected by increases in vapour pressure deficit (VPD) while gs decreased strongly, and Vrabl reported that gm was unaffected by feeding leaves with abscisic acid (ABA), whereas there was a clear decrease in gs. In the two latter cases the net CO2 assimilation rate remained unaffected by VPD and ABA despite the severe reduction of gs. The gs–gm relationship therefore may reflect a tight coordination between A and gm. Thus it seems that gm contributes to adjust the CO2 supply to the sites of carboxylation in response to photosynthetic limitations such as light availability, hydraulic constraints, or biochemical limitations (Warren ). In the present study, the coordinated variations of gs and gm apparently led to a very stable Cc (and Ci–Cc drawdown) across irradiance levels, despite large variation in A. Such a homeostasis of Cc was already observed across a range of leaf morphologies (Hassiotou ), during leaf ageing (Ethier ; Warren, 2006), and such an adjustment seems also to occur during short-term fluctuations of irradiance.

Conclusion

This study with three Eucalyptus species confirmed that gm estimated with the online 13C discrimination method declines in response to short-term increases of the CO2 mole fraction and increases with irradiance. The response to irradiance is saturated above 500 μmol m−2 s−1 PPFD. The respiratory term in the 13C discrimination equation was found to be important to estimate absolute values of gm but had little impact on the CO2 and irradiance responses. During the responses to CO2 and PPFD, gs and gm were tightly correlated and varied in parallel independently of the source of variation. Moreover, it was observed that coordinated adjustments of CO2 demand (A) and supply (gs and gm) led to a stability of Cc across irradiance variations. Cc homeostasis could be an advantage for the leaf to prevent large variation of the oxygenation/carboxylation ratio of the RubisCO, when the CO2 demand increases.