Project Topics
Get Complete Project Material File(s) Now! »
Discrimination during photorespiration
Photorespiration is a complex cycle that includes the fixation of a molecule of O2 by RubisCO, the resulting production of glycolate which enters the peroxysome to produce glycine (Figure 2). Glycine then enters the mitochondria to be decarboxylated into serine which releases the photorespired CO2. Serine comes back to the peroxisome, produces glycerate which re-enters the Calvin cycle as 3-phosphoglycerate. The ratio of oxygenation to carboxylation (φ) by RubisCO can be estimated from Γ*, the CO2 compensation point in absence of Rd. Γ* is thus incorporated in the model of discrimination to account for photorespiration. Evans and Loreto (2000) provided a list of Γ* estimated in 11 species, with extremes values varying between 33 and 47 µmol mol-1, with average Γ* lies around 40 µmol mol-1.
It has been shown that photorespiratory decarboxylation fractionates against 13C glycine decarboxylation, (f in the discrimination model, see Ghashghaie et al., 2003 for a review). In vitro estimations of f conducted on glycine decarboxylase extracted from eight plants species showed that the photorespired CO2 could be 13C depleted (-8‰) or enriched (+16‰) with respect to the substrate (Ivlev et al., 1996). The direction of fractionation was species-dependent, but also influenced by the pH and the enzyme co-factors used for the reaction. f was estimated from a theoretical approach which predicted that photorespired CO2 should be 13C depleted by 11‰ compared to glycine (Tcherkez, 2006). A statistical approach provided close value with f=11.6‰ (Lanigan et al., 2008). This approach consisted to set b, f and gm as unknowns, and fix known values for Γ*, e and Rd. The three unknowns were then estimated simultaneously to fit a modelled and a measured ∆13C in three Senecio species. These latter values bring support for a positive f value, which implies emission of depleted CO2, and a fractionation around 11‰. We can note that the enriched serine comes back to the Calvin cycle via glycerate and 3-phosphoglycerate. This can cause enrichment of the carbohydrate pool formed via the Calvin cycle. This influence is not accounted in the ∆13C model, and was not estimated yet.
Sensitivity analysis of∆13C estimates
To our knowledge, there was no sensitivity analysis performed on the overall ∆13C model yet. Indeed, quantifying at the same time the relative importance of all the parameters of a model can be difficult. We chose here two different methods. The first method consisted to generate a dataset with fixed values of gm=0.2, 0.5 and 0.8 µmol m-2 s-1 (corresponding to a realistic range of Vcmax between 103 and 74 µmol m-2 s-1) ; b=26, 28, 30‰ ; f=0, 5, 15‰ ; e=-15, 1 , +15‰ ; Rd=0, 1, 2 µmol m-2 s-1 ; and Γ*=35, 42.5, 50 µmol mol-1, the range of each parameter being chosen following values found in the literature (Table 2). For each of the 35 possible combinations we calculated ∆13C with a fixed value of A=16.9 µmol m-2 s-1, Ci=269.1 µmol mol-1 and Ca=317.1 µmol mol-1, corresponding to measured data on Eucalyptus sieberii (Douthe et. al, in press). Then, a multiple linear regression was performed to explain the variations of the computed ∆13C with gm, b, e, f , Rd and Γ* as covariates and partial R2 for each parameter was compared to the total R2 of the model. gm appeared to be largely the most influent parameter in the discrimination model, with a relative R2 of 70% (Figure 4, lower left panel). When gm is fixed thus not accounted for in the analysis (Figure 4, lower right panel), b is the most important parameter (partial R2 of 65%), followed by f (30%) and e (10%). Rd and Γ* appeared to have a small effect on the model.
Another approach was tested to confirm these results. We used the Sobol’s index, which compares the effect of a variation of the input parameters on the variance of the output variable (here ∆13C) (see R Development Core Team 2010, package “Sensitivity” version 1.4.0 and Saltelli 2002). Sobol’s index is comprised between 0 and 1; higher values mean a higher relative influence on the given parameter. We used the same range of variations for each parameter than for the first method. Again, gm is largely the most influent parameter on ∆13C (Figure 4, upper left panel), and b when gm is fixed (Figure 4, upper right panel). e and f have approximately the same influence, and Rd and Γ* are less influential parameters in the model.
Could variations of b with irradiance or O2 cancel short-term response of gm?
There is an increasing number of studies testing and observing a rapid (within minutes) response of gm to changes of environmental conditions (CO2 and irradiance, in particular, see Flexas et al., 2008 for review). In these studies it has been considered that other fractionation factors used in the gm equation remain stable with time. This is a crucial hypothesis to validate the rapid response of gm, but has not yet been tested. Here we focus on possible variations of b during changes of irradiance to test if b could vary enough to fully account for the observed changes of ∆13C. We based this analysis on irradiance response curves from 200 to 1000 µmol m-2 s-1 PPFD performed on Eucalyptus sieberii with a calibrated TDLAS, under 21% and 1% O2 to vary the influence of photorespiration, with all parameters constant with O2, except Γ*=38.7 and 1.85 µmol mol-1 respectively (Figure 7). Considering that gm was stable with irradiance and equal to the maximum value measured at an irradiance of 1000 µmol m-2 s-1 (Figure 7, gm= 0.37 µmol m-2 s-1 under 21% O2), b should vary from 28‰ at high irradiance to 26‰ at low irradiance. The same pattern was observed under 1% O2, according to a constant gm= 0.49 µmol m-2 s-1, with b should be 27‰ at low irradiance. Under 21% O2, a decrease of b from 28 to 26‰ could be explained by an increase of β (relative amount of carbon fixed by PEPc) from 0.055 to 0.11 (i.e. 5.5% to 11% carbon fixed by PEPc), for b3= 30‰ and b4=-5.7‰ and constants with irradiance. There are however several possible combinations of b and β to explain this variation (see Figure 8). These values of b and β are comprised in a realistic range of variation, according to the literature (Table 2), suggesting that b variations with irradiance are plausible. Such phenomena could occur since RubisCO carbon fixation is directly dependent of irradiance (via electron transport chain) but not that of PEPc. Thus, β could be higher at low compared to high irradiance. However, to the best of our knowledge, such hypothesis was not directly tested yet. This possibility was mentioned by Von Caemmerer and Evans (1991) and Lloyd et al. (1992), but the authors concluded in favour of a constant β. Von Caemmerer and Evans (1991) cited estimation of β at low light being of the same range as in high light and Lloyd et al. (1992) concluded that the variations of b needed was very unlikely because too large (from 20 to 35‰, for Citrus) to fully explain the variations of ∆13C attributed to gm. Nevertheless, there is a real need to estimate β with independent methods. At the moment, only estimations via RubisCO and PEPc activities are used to assess the relative part of carbon fixed by each enzyme, but we don’t know if maximum activities can be systematically related to the effective amount of carbon fixed.
Importance of non-photorespiratory decarboxylation: e and Rd
Fractionation during respiration other than from photorespiration (Rd), so-called “day respiration”, has only a small influence on the estimation of gm, as highlighted by the sensitivity analysis. Using the standard parameterization (Table 3) and changing Rd from 0 to 2 µmol m-2 s-1 affected gm by only 1% (see Figure 6). Even with extreme values of b, f or Γ*, Rd has a small influence. An exception was when Rd and e varied together. With e=-15‰, increasing Rd decreased gm by 0.1 mol m-2 s-1 (20%) and when e=+15‰, enhanced Rd increased gm by 0.15 mol m-2 s-1 (35%). This close relationship is obviously induced by the e*Rd factor in the equation. The e parameter has a larger effect on gm estimation than Rd. Variation between extreme values (e=-15 and +15‰) changed gm by almost 0.1 mol m-2 s-1 (i.e 30% variation). This effect is larger when f=15‰ or b=26‰ is used (60% or 100% increase compared to the standard parameterization, respectively). It has been reported that δ13C of respired CO2 could change with leaf temperature and by maintaining the leaf under prolonged darkness (Tcherkez et al., 2003). These variations are mainly caused by a change of respiratory substrate, with decarboxylation of sucrose resulting in enriched respired CO2 and that of lipids or proteins resulting in depleted respired CO2. Such changes could impact ∆13C recorded during photosynthesis, and should be included in the discrimination model via the e parameter. This is the same rationale as (Wingate et al., 2007) who considered that if the δ13C of source CO2 during the experiment is different from that during growth, then the respired CO2 would be affected because of a mix between freshly synthesised and older carbon pool. We tested the possibility that such changes in isotopic signature of respired CO2 during an experiment could give rise to artefactual short-term variations gm. We estimated that e would have to change from ~3‰ to 25-30‰ under 21% O2, and from ~3‰ to 15‰ under 1% O2 with decreasing irradiance to negate the computed variation of gm. Values of e for 21% O2 are clearly out of the range found in the literature for darkened leaves. Under 1% O2, the values of e are within the range found in the literature (for darkened leaves), except if we consider estimations based on sucrose as respiratory substrate (then e=2-5‰). If we now compare to recent estimations of e during the light period (e<1‰, Tcherkez et al., 2011; Tcherkez et al., 2010), these variations of e are even more unlikely. However, such rapid variations would need large change of the respiratory substrate during the experiment but this phenomenon remains very unlikely due to the constant temperature used in our experiment, and the fact that carbohydrate starvation probably did not occur. This clearly shows that potential changes of e cannot lead to the short-term response of gm to irradiance. We then considered that using values of e close to 0 for estimating mesophyll conductance is probably adapted. It is the same story for Rd. To fully explain the change in gm, Rd would have to increase up to 20 µmol m-2 s-1 under 21% O2 which is not possible. Under 1% O2, Rd would have to increase to 2.2 and 4.5 µmol m-2 s-1 at 600 and 200 µmol m-2 s-1, respectively. These values are far higher than values reported in the literature. Moreover, there is no study observing increasing Rd in such proportions, with decreasing irradiance (between 1000 and 200 µmol m-2 s-1). We conclude that Rd can’t explain short-term variations of gm under either 21% or 1% O2.
Importance of photorespiration: f and Γ*
The fractionation during photorespiration, noted f, is the second most influential parameter in the estimation of gm, according to the sensitivity analysis (Figure 4). Enhanced f increases the value of gm, with gm lowered by 20% when f=0‰ and increased by 30% when f=15‰ (Figure 7). In terms of absolute values this represents a change of gm by 0.3 mol m-2 s-1 with f varying from 0 to 15‰ for b=30‰, but for b=26‰ gm changed by ≈0.5 mol m-2 s-1 (higher sensitivity of gm to f with low b). Comparatively, Γ* has only a small effect on the estimation of gm, with variations remaining 0.05 mol m-2 s-1 when Γ* varies between 35 and 50 µmol mol-1. There is a larger effect with f=15‰, with gm varied by 0.1 µmol mol-1.
Changers in f are unlikely to explain gm variations with irradiance, with respect to the large range of values needed (Figure 6). f should switch from 11‰ under high irradiance up to 26‰ under low irradiance (21% O2), which is double recent estimates (Lanigan et al., 2008; Tcherkez, 2006). Moreover, it has been never suggested that f could vary with irradiance. We found a very large value, with f=150‰ at 1% O2, clearly a computation consequence, regarding of the low Γ* (1.85 µmol mol-1) associated with f. To account for variation in gm at low irradiance, Γ* would have to increase to 60 µmol mol-1 (compared to 38.7) at 21% O2 and 25 µmol mol-1 (compared to 1.85) at 1% O2 . This represents very large increases compared to the value under high irradiance, and since Γ* reflects RubisCO affinity for CO2/O2, it is not likely to vary with O2. This evidence that neither variations of Γ* or f could explain gm variations with irradiance.
Table of contents :
CHAPTER I Relationship between 13C discrimination and leaf gas exchange: Analysis of the model, influence of the parameters and implications for estimating mesophyll conductance to CO2.
INTRODUCTION
Diversity of the Wi-Δ13C relationships found in the literature
Using the complete form of the discrimination model
On the importance of the b parameter
Importance of non-photorespiratory decarboxylation: e and Rd
Importance of photorespiration: f and Γ*
Concluding remarks
CHAPTER II Mesophyll conductance to CO2, assessed from on-line TDL-AS records of 13CO2 discrimination, displays small but significant short term responses to CO2 and irradiance in Eucalypt seedlings.
ABSTRACT
INTRODUCTION
MATERIAL & METHODS
RESULTS
DISCUSSION
Importance of the respiratory and photorespiratory terms in the estimation of mesophyll
conductance
Response of gm to CO2 mole fraction.
Response of gm to irradiance
CONCLUSION
CHAPTER III Is mesophyll conductance to CO2 in leaves of three Eucalyptus species sensitive to shortterm changes of irradiance under ambient as well as low O2?
ABSTRACT
INTRODUCTION
MATERIAL AND METHODS
RESULTS
DISCUSSION
Rapid response of mesophyll conductance to irradiance under 21 and 1% O2
Sensitivity of gm estimates to changes of model parameter values
Are the observed variations of gm with O2 realistic?
CONCLUSION
CHAPTER IV
Mesophyll conductance of poplar leaves varies rapidly with changes in CO2 and irradiance: an assessment from on line 13CO2 discrimination records with TDL-AS.
ABSTRACT
INTRODUCTION
MATERIAL & METHODS
RESULTS
DISCUSSION
Rapid variations of gm with PPFD and CO2
Impact of a stable gm on net CO2 assimilation rate under PPFD variations
CONCLUSION
CONCLUSIONS
and PERSPECTIVES
ANNEX I
REFERENCES
