Leaf-level measurements have shown that mesophyll conductance (gm) can vary rapidly in response to CO2 and other environmental factors, but similar studies at the canopy-scale are missing. Here, we report the effect of short-term variation of CO2 concentration on canopy-scale gm and other CO2 exchange parameters of sunflower (Helianthus annuus L.) stands in the presence and absence of abscisic acid (ABA) in their nutrient solution. gm was estimated from gas exchange and on-line carbon isotope discrimination (Δobs) in a 13CO2/12CO2 gas exchange mesocosm. The isotopic contribution of (photo)respiration to stand-scale Δobs was determined with the experimental approach of Tcherkez et al. Without ABA, short-term exposures to different CO2 concentrations (Ca 100 to 900 µmol mol−1) had little effect on canopy-scale gm. But, addition of ABA strongly altered the CO2-response: gm was high (approx. 0.5 mol CO2 m−2 s−1) at Ca < 200 µmol mol−1 and decreased to <0.1 mol CO2 m−2 s−1 at Ca >400 µmol mol−1. In the absence of ABA, the contribution of (photo)respiration to stand-scale Δobs was high at low Ca (7.2‰) and decreased to <2‰ at Ca > 400 µmol mol−1. Treatment with ABA halved this effect at all Ca.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Mesophyll conductance (gm) limits the diffusive flux of CO2 from the substomatal air space to the sites of carboxylations (Lloyd et al. 1992; Epron et al. 1995; Evans & von Caemmerer 1996; Flexas et al. 2008). This limitation is complex, as it involves diffusion through intercellular airspaces and in the liquid phase through cell wall, plasma membrane and cytosol, and the envelope and stroma of the chloroplast (Evans et al. 2009). Recent research has revealed strong short-term, reversible and acclimation responses of gm to several environmental factors, such as light, temperature, drought, salinity and CO2 (see Flexas et al. 2008 for a compilation of gm responses). This reversible or adaptive component appears to be associated with the conductance of membranes, which may be modified by the expression of cooporins (Hanba et al. 2004; Flexas et al. 2007), which are aquaporins capable of transporting CO2 across plasma membranes (Terashima et al. 2006).
Interestingly, adaptive responses of gm to single environmental drivers do not always agree in direction and magnitude in different experiments with different species. In most cases, gm increases with decreasing atmospheric CO2 (e.g. Flexas et al. 2007). However, gm did not respond to CO2 in Triticum aestivum (Tazoe et al. 2009) and Raphanus sativus (von Caemmerer & Evans 1991). Finally, also a biphasic response of gm has been found in Helianthus annuus (Vrabl et al. 2009) with gm increasing with increasing CO2 below 200 µmol mol−1 and decreasing when CO2 was increased beyond.
The negative gm response to water stress and salinity is fairly consistent among species (see compilation of Flexas et al. 2008). Also, the majority of studies agree that gm decreases with leaf age, decreasing nitrogen content and shading. However, these factors often co-vary, making it difficult to unravel the underlying mechanism. Piel et al. (2002) showed a marked difference in gm between sun and shade leaves of Juglans sp., but these leaves also differed in nitrogen content. A similar co-variation was observed by Niinemets et al. (2005) in Mediterranean evergreen species, where the decreasing nitrogen content with leaf age was accompanied by decreasing gm. In herbaceous plants older leaves become successively more shaded during canopy development when nitrogen is re-allocated to younger leaves formed near the top of the canopy (Connor, Sadras & Hall 1995). Both factors may affect gm.
The gm response to a particular environmental driver – for example, CO2– may well depend on interactions with other factors. However, such studies are rare. No interaction between irradiance and CO2 was observed as either both parameters did not affect gm (Triticum aestivum; Tazoe et al. 2009) or lowering irradiance decreased gm but had no effect on the general response to CO2 (Banksia species; Hassiotou et al. 2009). When studying the interaction of CO2 and drought on gm, care must be taken for co-varying parameters as explained earlier. Drought induced changes in relative water content might affect gm via anatomical or morphological changes without altering the physiology. In this respect, exogenous application of abscisic acid (ABA) is a useful tool to mimic drought effects on stomatal conductance without affecting the morphology of leaves.
Thus, in this work, we addressed the following main questions: (1) does canopy-scale gm respond dynamically to short-term variation of CO2 similar to the response of leaves? and (2) does ABA influence the gm response to CO2?
Plants of sunflower (Helianthus annuus L., cv. Optisol, CARG, BSA-Nr. 241) were sown individually in plastic pots (35 cm deep, 5 cm diameter), filled with 0.8 kg of washed quartz sand (0.4–0.8 mm particle size), transferred to four growth cabinets (Conviron E15, Conviron, Winnipeg, Canada) and arranged at a density of 78 plants m−2. Throughout the experiment an automatic drip irrigation system supplied nutrient solution [modified half-strength Hoagland solution with 7.5 mM N L−1; composition: 2.5 mM KNO3, 2.5 mM Ca(NO3)2, 1 mM MgSO4, 0.5 mM KH2PO4, 0.1 mM Fe–ethylenediaminetetraacetic acid (EDTA), 23 µM H3BO3, 4.6 µM MnCl2, 0.4 µM ZnSO4, 0.16 µM CuSO4, 0.26 µM H2Mo8O4] to every pot every 4 h. The supply rate was adjusted weekly to satisfy the water demand of plants. Light was supplied by cool white fluorescent tubes and incandescent lamps. Irradiance at the top of each canopy was maintained at 600 µmol m−2 s−1 photosynthetic photon flux density (PPFD) during the 16 h photoperiod. Temperature was controlled at 20/16 °C and relative humidity at 75/85% during the light and dark periods of the day. At 28 d after sowing ABA (Sigma-Aldrich, Seelze, Germany) dissolved in methanol was added to the nutrient solution of two growth cabinets. The final ABA concentration in the nutrient solution was 20 µM L−1.
Gas exchange measurements and isotope analysis
The four growth cabinets formed part of the 13CO2/12CO2 gas exchange and labeling system described by Schnyder et al. (2003). In short, air supply to the growth cabinets was generated by mixing CO2- and H2O-free air with CO2 of known carbon isotope composition (δ13C, where δ13C = [(Rsample/Rstandard) − 1], and R the 13C/12C ratio in the sample or Vienna Peedee belemnite (VPDB) standard. The CO2-free dry air was produced by passing compressed ambient air through an absorption dryer (KEN 3100, Zander, Essen, Germany) filled with a molecular sieve (activated aluminium oxide F200, Alcoa, Houston, TX, USA). The dryer was fed by a screw compressor (S40, Boge, Bielefeld, Germany), which compressed ambient air to ∼0.7 MPa at a rate of ≤300 m3 h−1. After passage through the dryer, the air had a residual CO2 concentration of <0.5 µmol mol−1 and its dew point was ≤70 °C. The rate of air supply and the CO2 concentration in air injected into the growth cabinets was controlled by four independently operated gas mixing systems. Each served one cabinet and comprised two computer-operated mass-flow controllers (FC-2925V for air and FC-2900 4S for CO2, Tylan General, San Diego, CA, USA). A computer-controlled gas sampling system (DMP, Hegnau-Volketswil, Switzerland) sequentially sampled air entering and leaving each growth cabinet. Each air sample was split: CO2- and H2O-concentration was analysed with an IRGA (Licor 6262, Li-Cor, Lincoln, NE, USA) and the δ13C of CO2 by an online gas isotope ratio mass spectrometer (IRMS, Delta plus, Finnigan MAT, Bremen, Germany). Air for mass spectrometric analysis was pumped continuously via a steel capillary from the gas sampling system to a 300 µL sample loop which was attached to a six-port Valco valve (Valco Instruments Co. Inc., Houston, TX, USA) mounted in a GC interface (GP-GC Interface, Finnigan MAT, Bremen, Germany). Air samples were fed to the mass spectrometer by periodically flushing the sample loop with He carrier gas and flushing the content through a Nafion® water trap and a GC column (25 m × 0.32 mm Poraplot Q, Chrompack, Middelburg, Netherlands). Then, the CO2 was introduced into the ion source of the IRMS via a glass capillary (0.1 mm i.d.) connected to the interface via an open split.
The system allowed a high frequency of sampling (approx. one sample per 2 min), so that each inlet and outlet of the four growth cabinets was sampled every 25 min. Each sample was compared with a VPDB-gauged working standard reference CO2 gas. The external precision (standard deviation) at 300 µmol mol−1 CO2 was <0.2‰ for δ13C.
During the entire experiment, including the establishment phase of the stands, two of the growth cabinets were supplied with CO2 from a fossil-organic source (δ13C −47.1‰) and two with CO2 from a mineral source (δ13C −5.4‰; both CO2: Messer Griesheim, Frankfurt a.M., Germany).
Rates of net CO2 exchange in light (A, µmol CO2 m−2 s−1) were calculated as the difference between the CO2 fluxes entering (Fin, µmol CO2 s−1) and leaving (Fout, µmol CO2 s−1) the cabinets divided by the cabinets' ground area (s, 1.4 m2):
δA, the δ13C of net CO2 exchange of canopies during the light was obtained by isotopic mass balance:
with subscript in and out denoting the inlet and outlet of the growth cabinet. δRn, the δ13C of CO2 respired in the dark was obtained in the same way using measurements taken during the dark period.
with and C the CO2 concentration at standard humidity.
Cout was maintained at 310 µmol mol−1 (SE = 0.6 µmol mol−1) throughout the experiment. Cin and the flux of air into the chamber were adjusted at intervals of 2 to 3 d to maintain ξ between 5 and 8. Thus, with constant Δobs during canopy development, δ13C of the CO2 in the cabinet atmosphere (δout) was constant (SE = 0.04‰ for both CO2) throughout the experiment.
Transpiration rate (E, mmol H2O m−2 s−1) was calculated analogous to A (Eqn 1). Thus, E = (FH2Oout − FH2O in)/s, with out designating the flux of water vapour leaving the cabinet (mmol H2O s−1) and in that added to cabinet air by the individual cabinets' humidification system (note that air entering the chambers was dry; see earlier discussion). The humidification system controlled the water vapour concentration in cabinet air by adding just enough vapour to match the difference between that generated by E and the nominal water vapour concentration of the chamber. Water addition by the humidification system occurred by activation (opening) of spray nozzles whenever the measured vapour concentration decreased below the nominal concentration. The rate of vapour addition of each sprayer was obtained from a calibration of vapour production versus nozzle activation times. E was recorded for each step during CO2 response measurements for approximately 15 min when the CO2 concentration at the chamber outlet had reached the new steady state. Evaporation from the sand surface (15% of chamber area) was a negligible component of (evapo)transpiration of the closed sunflower canopy. This was true, except for a small evaporation signal during a short period during and after irrigation events when the sand surface of pots was wetted and the relative humidity of chamber increased briefly. Therefore, data were not taken for a period of 20 min after irrigation. The CO2 response of A and E was measured in duplicate in two growth cabinets.
Calculation of gm
According to Fick's first law, gm is expressed as,
with A the net photosynthesis rate of the stand and Ci and Cc the CO2 concentrations in the intercellular airspace and in the chloroplast. Canopy-scale on-line carbon isotope discrimination (Δobs) is (Tcherkez et al. 2010):
with Ca and Cs denoting the CO2 concentration in the atmosphere and at the leaf surface, ab the fractionation during diffusion through the leaf boundary layer; a that during diffusion in air, am the combined fractionation during dissolution of CO2 in water and fractionation during diffusion of dissolved CO2 in water; b the fractionation associated with carboxylations, Rd the rate of leaf dark respiration in light; Rh the dark respiration of heterotrophic plant parts (root and stem), k the carboxylation efficiency, Γ* the CO2 compensation point in the absence of dark respiration, f the fractionation associated with photorespiration, e the fractionation associated with leaf dark respiration in light and eh the fractionation associated with respiration of heterotrophic plant parts. d*/Ca, the last term on the right-hand side of Eqn 5a, represents the effect of total mesocosm respiration in light on Δobs. This includes photorespiration and daytime dark respiration of autotrophic (leaf) and heterotrophic plant parts and soil, and was termed (photo)respiration (in accordance with Tcherkez et al. 2010). This term is a potentially large component of Δobs as respiration is a much bigger fraction of net CO2 exchange at mesocosm than at leaf scale, and carbon isotope fractionation in photorespiration and dark respiration of different plant parts is significant (Ghashghaie et al. 2003; Klumpp et al. 2005). The contribution of (photo)respiration to Δ, d*/Ca was obtained as:
(cf. Eqn 3 in Tcherkez et al. 2010), where Δ0 denotes Δ at the CO2 compensation point, Γ, of the mesocosm. Δ0 was estimated as the intercept of a linear regression of Δobs versus A (‘second method’ in Tcherkez et al. 2010) using the data obtained during the CO2 response measurements (as presented in Fig. 2a,c). This method has the advantage, that it does not require assumptions on the rates of (photo)respiration of leaves, and respiration of roots and stems and their specific fractionation factors (see discussion in Tcherkez et al. 2010).
Mesophyll conductance was calculated using the difference between Δobs (Eqn 3) and an estimate of Δ assuming infinite gm (Δi):
Leaf boundary layer conductance used to calculate Cs was not measured but assumed to be 750 mmol m−2 s−1 for water vapour (within the range found for broad-leafed species in greenhouses; Morrison & Gifford 1984; Aphalo & Jarvis 1993; Gan et al. 2002). Ci was calculated from stomatal conductance, which was based on measurements of transpiration (E) and leaf temperature. Leaf temperature was measured using six thermocouples evenly distributed within the top 20 cm of the canopy.
The last term in Eqn 7[i.e. the (photo)respiratory contribution] is the same as the last term in Eqn 5a. Thus, any difference between Δi and Δobs was caused by the difference between Ci and Cc:
Equation 8 could be solved for Cc as the only unknown and used to calculate gm using Eqn 1.
Canopy gas exchange and 13C discrimination under growth conditions
A and Δobs of sunflower canopies were measured in four different mesocosms under growth conditions. Measurements started after canopy closure when A, and Δobs were near-constant on a day-by-day basis (Fig. 1 and Table 1). A averaged 15.6 µmol CO2 m−2 s−1 in the four chambers and varied little between them (Table 1). Similarly, A was quite stable during the course of individual light periods. In general, it decreased slightly between 1 and 15 h in light (Fig. 1a). Δobs was high in all growth chambers (average 24.1‰). But it was virtually the same in growth chambers with different δ13CCO2 (Table 1).
Table 1. Net mesocosm CO2 exchange in light (A) and associated on-line carbon isotope discrimination (Δobs) of sunflower stands in four growth chambers (mesocosms)
Before ABA application
After ABA application
µmol CO2 m−2 s−1
µmol CO2 m−2 s−1
Growth conditions were identical in the chambers except for the δ13C of CO2 (δ13CO2). Data report the mean A and Δobs during 3 d prior to addition of ABA to the nutrient solution, and the mean rate on the third day after abscisic acid (ABA) addition (chambers 3 and 4).
Effect of ABA on gas exchange and 13C discrimination under growth conditions
ABA had strong effects on A and Δobs (Fig. 1). A and Δobs started to decrease immediately after ABA addition to the nutrient solution (data not shown). The effect of ABA on A and Δobs saturated after about 2 to 3 d when A and Δobs stabilized at near-constant values. At that time A was 11.8 µmol CO2 m−2 s−1, or 25% less than the rate just prior to ABA addition. Again, there was a slight decrease in A throughout the light period. Δobs was 17.2‰, i.e. 6.9‰ less than before ABA addition (Table 1).
ABA also strongly affected the relationship between δ13C of net CO2 exchanged (δA) and the CO2 respired in the night (δRn) or the (photo)respired CO2 (δPR). In control canopies, δRn was approximately 5‰ enriched relative to δA (Table 2). Conversely, δPR was depleted by 2.6‰ relative to δA. After ABA application, δRn was still slightly enriched relative to δA (approx. 1‰; Table 2). ABA had a particularly strong effect on δPR, which was 3‰ enriched relative to δA after ABA addition, but was significantly depleted before addition.
Table 2. δ13C of net CO2 exchanged in the light (δA), δ13C of CO2 respired in the dark (δRn) and δ13C of (photo)respiratory CO2 in the light (δPR) of sunflower canopies grown and measured at 350 µmol mol−1
Before ABA addition
After ABA addition
Measurements were performed during 3 d prior to addition of abscisic acid (ABA) to the nutrient solution, and on the third day after ABA addition. Data from chambers 3 and 4.
Effect of ABA on the CO2 response of A, Δobs and transpiration
The CO2 response of A, Δobs and canopy conductance was measured on the day before first ABA addition, and after 3 d of ABA addition, when A and Δobs had decreased to a stable level. All measurements were performed between 2 h of ‘lights on’ and 6 h before the end of the light period, when A and Δobs were near constant. In the absence of ABA, A followed the characteristic response of photosynthesis to changes in CO2 partial pressures (Ca) as described by leaf- or cell-based models of Farquhar, von Caemmerer & Berry (1980) and von Caemmerer and Farquhar (1981) (Fig. 2a). Transpiration also showed a strong response to increasing Ca (Fig. 2b), with the rate decreasing by 30% between 100 and 900 µmol mol−1 CO2. Furthermore, Δobs increased greatly with increasing Ca (Fig. 2c). At 900 µmol CO2 mol−1Δobs was 28.3‰[±0.1 standard deviation (SD)], close to the 13C discrimination value of Rubisco.
ABA strongly modified the short-term CO2 responses of A, Δobs and transpiration. ABA had no effect on A at Ca < 150 µmol mol−1, but at higher CaA was significantly reduced relative to the control (Fig. 2a). At a Ca of 900 µmol mol−1A was reduced by 25% relative to the control.
Transpiration (E) was affected more strongly by ABA addition (Fig. 2b) and the reduction, relative to the control, was similar (near 40%) over the whole range of Ca. Overall, ABA caused a substantial improvement of water use efficiency (A/E) at all Ca (Supporting Information Fig. S1). ABA did not affect the sensitivity of E to Ca when this sensitivity was expressed as the change in E produced by a small change in Ca (dE/dCa). However, ABA did affect the response of E to Ca: an increase from 120 to 860 µmol mol−1 CO2 caused a reduction in E by 20% in the absence of ABA, while the same change caused a reduction of 40% in the presence of ABA.
ABA and Ca had strong interactive effects on Δobs. Δobs was unaffected by ABA when Ca was low. But, the responses of Δobs to Ca diverged strongly when Ca was increased above 150 µmol mol−1: Δobs decreased in the presence of ABA, but increased in its absence. At a Ca of 500 µmol mol−1, Δobs was 15‰ in the presence of ABA, 12‰ less than in its absence. In both treatments, changes of Ca above 500 µmol mol−1 had little effect on Δobs.
Effect of ABA on the components of Δobs
Using d*/Ca (Eqn 6), the (photo)respiratory contribution to Δobs could be calculated for each measurement taken during the CO2 response (Fig. 3). This term described the total effect of all respiratory activities of the mesocosm (including photorespiration) on Δobs measurements in light. In control canopies (photo)respiratory discrimination was a very significant component of Δobs at low Ca (7.2‰ at Ca of 120 µmol mol−1) decreasing strongly with increasing Ca (1‰ at the highest Ca). After ABA application (photo)respiratory discrimination showed a similar shape of the response to a change in Ca. However, at each CO2 concentration, the contribution to Δobs was only half of that in control canopies (3.8 and 0.6‰ at the lowest and highest Ca, respectively).
Effect of ABA on the CO2 response of mesophyll conductance
Canopy-scale mesophyll conductance was calculated from estimates of Ci and Cc, and the net rate of CO2 fixation (A), using Fick's first law (Eqn 1). Estimates of Ci were obtained from E, leaf temperature and the vapour pressure deficit of chamber air. Canopy conductance was very high and was ignored in the estimation of Ci. Estimates of Cc were obtained with Eqn 8, using parameters as given in the Materials and Methods section.
In control canopies, the difference Δi − Δobs was highest at low Ca and decreased nearly linearly with increasing CO2 concentration (Fig. 4a). Again, there was virtually no difference in Δi − Δobs between ABA-treated and control canopies at low Ca. But Δi − Δobs increased with ABA application up to 400 µmol mol−1Ca, but did not change with higher CO2 concentration.
Accordingly, gm in control canopies averaged 0.45 mol CO2 m−2 s−1 and hardly responded to Ca (Fig. 4b). In contrast, gm in ABA-treated canopies strongly responded to Ca: at low Ca, gm was in the range observed in control canopies, but it decreased exponentially with Ca and attained <0.1 mol CO2 m−2 s−1 at the highest CO2 concentration.
Taken together, the CO2 responses of gs (not shown, but see Fig. 2b) and gm implied that the Cc to Ci ratio of ABA-treated canopies decreased with Ca, while it increased in controls (Fig. 5a). In ABA-treated canopies the ratio was ∼0.8 at a Ca of <200 µmol mol−1, decreased to 0.6 up to 400 µmol mol−1 and remained at ∼0.6 beyond 400 µmol mol−1. Conversely, in the control treatment, the ratio increased from ∼0.8 at a Ca of <200 µmol mol−1 to ∼0.95 at 850 µmol mol−1. Nevertheless, Cc (and Ci) increased continuously with Ca in both treatments (although to different extents), meaning that the degree of CO2 limitation of Rubisco decreased with Ca (Fig. 5b,c). Still, because of the very low gm, Cc barely exceeded 400 µmol mol−1 at the highest Ca (870 µmol mol−1) in the ABA-treated canopies (Fig. 5c).
This work revealed a significant co-limitation of canopy photosynthesis by gm in sunflower under ambient CO2. Short-term variation of CO2 had little effect on gm in control conditions, but addition of ABA led to dynamic CO2-responses of gm. In the presence of ABA, gm decreased very strongly and rapidly with increasing CO2, significantly decreasing Cc below Ca at ambient to elevated Ca. Decreasing CO2 to low Ca reversed this effect.
As discussed by Warren (2006), the estimation of gm with the isotopic method depends on the values of the parameters used for the estimation of Δi. These include the rate of (photo)respiration, the fractionation associated with photorespiration and dark respiration in light and the fractionation by carboxylases (cf. Eqn 7), which are all associated with uncertainties. Therefore we elected to use the approach of Tcherkez et al. (2010) to estimate discrimination associated with total mesocosm respiration in light (termed ‘(photo)respiration’ and including all photorespiration, and dark respiration by autotrophic and heterotrophic components of the system in the light). This approach avoided assumptions about individual respiratory fractionation factors and rates of photo- and dark respiration in light. The integral contribution of these respiratory activities to mesoscosm-scale Δobs was especially important at low Ca, where the (photo)respiratory signal accounted for up to 35 % of Δobs in the control mesocosm (Fig. 3). This derived from the fact that the ratio of (photo)respiration to gross photosynthesis is highest at low Ca and decreases with increasing Ca.
Importantly, the ranking of the treatments and the shape of the response functions of gm to Ca were also not altered when we modified the fractionation factors affecting d* (b and Δ0, see Eqn 6) in a wide range. Increasing (decreasing) Δ0 led to decreasing (increasing) gm in both treatments without affecting their ranking. In ABA-treated canopies, for example, an increase of Δ0 from 21.4‰ (i.e. the measured value) to 25‰ caused a decrease in gm of 13%. Increasing (decreasing) b, led to lower (higher) estimates of gm in both treatments, again with no effect on treatment ranking. The same was true for the shape of the gm response to Ca. Also, consideration of canopy-scale heterogeneity in stomatal conductance of leaves (which may vary as a function of leaf position/irradiance, age and ABA effects) would not alter our conclusions. Lloyd et al. (1992) have shown that even in situations of very high heterogeneity of stomatal conductance the error in the difference between Δi and Δobs would not exceed 1‰. This error is only a small fraction of the ABA effect on Δi − Δobs observed in our study. Finally, we are very confident that artefacts in Δobs measurements were insignificant: we observed the same Δobs for canopies growing in the presence of CO2 with different δ13C (Table 1), meaning that there were no leaks in the chambers, which could have influenced mesocosm 13CO2/12CO2 exchange and, hence, Δobs measurements (Schnyder 1992). Moreover, the external precision of 13CO2/12CO2 measurements was very high (SD < 0.2‰). This measure of precision included all errors associated with CO2-free air generation, source CO2 composition, gas mixing, sampling and determination of 13CO2/12CO2 ratios (see Schnyder et al. 2003). Lastly, all 13CO2/12CO2 exchange measurements were performed in the same isotopic environment as stand growth.
The data also demonstrate that daytime (photo)respired CO2 was 13C-depleted by a few permil relative to CO2 net fixed, whereas CO2 respired in darkness was enriched by several permil (Table 2). This variation was also observed in previous work at mesocosm-scale with sunflower growing at 200 and 1000 µmol mol−1 CO2 (Tcherkez et al. 2010) and in leaf-scale studies (Ghashghaie et al. 2003; Tcherkez et al. 2004) and is probably related (at least in part) to diurnal variations in the isotopic composition of sucrose (Gessler et al. 2008), the main substrate for respiratory metabolism. These variations result from an isotope effect of aldolases (Gleixner & Schmidt 1997) that catalyse the production of fructose-1,6-bisphosphate from triose phosphates. This causes a circadian variation in the 13C content of sucrose, depleting sucrose formed in light, and enriching sucrose formed from transitory starch at night (Tcherkez et al. 2010). Perhaps, the divergent effect ABA on δPR is also related to an ABA effect on C metabolism. Water stress can change C partitioning between starch and soluble sugars (sucrose and hexoses) synthesis in favour of sugars (Vassey & Sharkey 1989; Kanechi et al. 1998), to assist osmotic adjustment (Villadsen, Rung & Nielsen 2005). External addition of ABA can also cause changes in the activity of enzymes associated with primary carbohydrate metabolism (Yang et al. 2004). In this context, it is interesting to note that after ABA application, the (photo)respiratory flux was slightly 13C-enriched relative to that net fixed (Table 2). This effect could be related to an inhibition of starch biosynthesis by ABA, which would lead to less negative δ13C of triose phosphates exported from chloroplasts during the light period. This effect would also cause a less negative δ13C of sucrose synthesized in the cytoplasm and exported to other plant parts.
The gm responses observed here were highly dynamic and fully reversible, and were unrelated to anatomical features. This corroborates the view that gm has a protein- or enzyme-facilitated component (Bernacchi et al. 2002; Warren & Dreyer 2006; Flexas et al. 2007) with aquaporins as likely candidates. In tobacco leaves, the highest concentration was found in cells adjoining to the substomatal cavities (Otto & Kaldenhoff 2000). Aquaporins can mediate CO2 transport across membranes (Maurel et al. 2008), and in tobacco were shown to enhance membrane permeability for CO2 (Uehlein et al. 2003). Overexpression in rice (Hanba et al. 2004) and tobacco (Flexas et al. 2006) significantly increased gm. Flexas et al. (2007) hypothesized that maintaining high gm could be an energy consuming process. Thus, at low Ca, when CO2 availability is limiting photosynthesis, the excess energy could be used to increase gm. At high Ca, diverting energy to maintain a high gm would be less efficient, as the extra CO2 at the site of carboxylation would result in only little additional CO2 fixation. In our study, the modulation of gm seemed strictly dependent on the presence of ABA. The mechanism underlying this effect is presently unclear. However, if aquaporins played a role, then it should be interesting to investigate ABA effects on their CO2 transport function. Interestingly, the CO2 response of gm observed at the mesocosm scale, and the effect of ABA on this response, differed from that observed at leaf-scale by Vrabl et al. (2009). In the presence of ABA, they found a similar CO2 response of gm in the reliable range of their data. However, in controls, they also observed a decrease of gm with increasing CO2, while we found no effect. Absence of a CO2 effect on gm has also been observed by others (see compilation in Flexas et al. 2008). It is well possible, that the difference between the leaf- (Vrabl et al. 2009) and the present mesocosm-scale CO2 response of gm was related to interactions with other external factors, such as light and CO2 exposure of plants. The leaf measurements of Vrabl et al. occurred in high light (irradiance was >2 times higher than during growth), whereas our measurements were performed at the same irradiance as during growth (600 µmol m−2 s−1 PPFD). Interestingly, Flexas et al. (2007) showed that in tobacco the response of gm to a change in CO2 concentration was far less pronounced at low compared with high light intensity during measurements. Furthermore, Vrabl et al. (2009) measured the CO2 response on a portion of a leaf, whereas the remaining part and all other leaves experienced a different CO2 concentration. This contrasts with the mesocosm-scale response, which was obtained by exposing all leaves to the same CO2. Lastly, Vrabl et al. (2009) measured only young leaves, whereas all leaf-age categories of plants contributed to the mesocosm-scale measurement. It is known that physiological responses to ABA addition can depend on interactions with particular environmental scenarios (Tardieu, Parent & Simonneau 2010), and availability of light could be an important factor.
Guillaume Tcherkez is thanked for helpful discussions and Thomas Gebbing and Wolfgang Feneis for fine help with running and maintaining the mesocosm facility. This research was partially supported by DFG (SCHN 557/4-1) and the European Community's Human Potential Program under contract HPRN-CT-1999-00059, NETCARB, coordinated by Jaleh Ghashghaie. J.S. was supported by GA CR grant 206/08-0787 and MEYS grants 6007665801 and AV0Z50510513.