Leaf photosynthesis and respiration of three bioenergy crops in relation to temperature and leaf nitrogen: how conserved are biochemical model parameters among crop species?

Given the need for parallel increases in food and energy production from crops in the context of global change, crop simulation models and data sets to feed these models with photosynthesis and respiration parameters are increasingly important. This study provides information on photosynthesis and respiration for three energy crops (sunflower, kenaf, and cynara), reviews relevant information for five other crops (wheat, barley, cotton, tobacco, and grape), and assesses how conserved photosynthesis parameters are among crops. Using large data sets and optimization techniques, the C(3) leaf photosynthesis model of Farquhar, von Caemmerer, and Berry (FvCB) and an empirical night respiration model for tested energy crops accounting for effects of temperature and leaf nitrogen were parameterized. Instead of the common approach of using information on net photosynthesis response to CO(2) at the stomatal cavity (A(n)-C(i)), the model was parameterized by analysing the photosynthesis response to incident light intensity (A(n)-I(inc)). Convincing evidence is provided that the maximum Rubisco carboxylation rate or the maximum electron transport rate was very similar whether derived from A(n)-C(i) or from A(n)-I(inc) data sets. Parameters characterizing Rubisco limitation, electron transport limitation, the degree to which light inhibits leaf respiration, night respiration, and the minimum leaf nitrogen required for photosynthesis were then determined. Model predictions were validated against independent sets. Only a few FvCB parameters were conserved among crop species, thus species-specific FvCB model parameters are needed for crop modelling. Therefore, information from readily available but underexplored A(n)-I(inc) data should be re-analysed, thereby expanding the potential of combining classical photosynthetic data and the biochemical model.

Introduction

In conventional crop modelling leaf photosynthesis is calculated from net photosynthesis light response curves (An–Iinc; see symbols explanation in Table 1) at ambient atmospheric CO2 level using empirical functions (e.g. SUCROS; Goudriaan and van Laar, 1994). In the context of better understanding biological processes and exploring the impact of climate change, recent crop models (e.g. GECROS; Yin and van Laar, 2005), 3D models (e.g. Evers ), or terrestrial ecosystem models (e.g. LPJmL; Beringer ) calculate photosynthesis based on the mechanistic model of Farquhar, von Caemmerer, and Berry (Farquhar ; the FvCB model hereafter).

Table 1.

List of main symbols used in this study with their definitions and units

Symbol	Definition	Unit
Ac	Rubisco-limited net photosynthetic rate	μmol CO2 m−2 s−1
An	Net assimilation rate	μmol CO2 m−2 s−1
Aj	Electron transport-limited net photosynthetic rate	μmol CO2 m−2 s−1
An,max	Light-saturated An	μmol CO2 m−2 s−1
aR	x-axis intercept in Equation 12	μmol CO2 m−2 s−1
bR	Slope parameter in Equation 12	–
Cc	CO2 chloroplast partial pressure	μbar
Ci	Intercellular CO2 partial pressure	μbar
Dj, Dv	Deactivation energy of Jmax and Vcmax (Equation 6)	J mol−1
EKmc, EKmo	Activation energy for Kmc and for Kmo	J mol−1
Ej, ERn, Ev	Activation energy of Jmax, Rn, and Vcmax (Equations 5–6)	J mol−1
ERn(a)	Constant parameter (Equation 11)	J mol−1
ERn(b)	Slope parameter in Equation 11	J m−2 mol−1 g−1 N
gm	Mesophyll conductance for CO2 diffusion	mol m−2 s−1
gs	Stomatal conductance for H2O	mol m−2 s−1
Iinc	Incident light on leaf surface	μmol photons m−2 s−1
J	Photosystem II electron transport rate	μmol e− m−2 s−1
Jmax	Maximum electron transport rate	μmol e− m−2 s−1
Jmax25	Value of Jmax at 25 °C	μmol e− m−2 s−1
Kmc	Michaelis–Menten constant for CO2	μbar
Kmo	Michaelis–Menten constant for O2	mbar
Na	Leaf nitrogen per unit area	g N m−2 leaf
Nb	Minimum Na required for photosynthesis	g N m−2 leaf
O	Oxygen partial pressure of the air (=210)	mbar
R	Universal gas constant (=8.314)	J K−1 mol−1
Rd	Day respiration rate	μmol CO2 m−2 s−1
Rn	Night respiration rate	μmol CO2 m−2 s−1
Rn25	Value of Rn at 25 °C	μmol CO2 m−2 s−1
Sj, Sv	Entropy term for Jmax and Vcmax (Equation 6)	J K−1 mol−1
Vcmax	Maximum carboxylation rate	μmol CO2 m−2 s−1
Vcmax25	Value of Vcmax at 25 °C	μmol CO2 m−2 s−1
Γ*	Ci-based CO2 compensation point in the absence of Rd	μbar
θ	Convexity factor for the response of J to Iinc	–
κ2LL	Conversion efficiency of Iinc into J at low light	mol e− mol−1 photons
ΦCO2LL	Apparent quantum yield of An at low Iinc	mol CO2 mol−1 photons
χj	Slope of the Jmax25 and Na relationship (Equation 10)	μmol e− g−1 N s−1
χR	Slope of the Rn25 and Na relationship (Equation 8)	μmol CO2 g−1 N s−1
χv	Slope of the Vcmax25 and Na relationship (Equation 9)	μmol CO2 g−1 N s−1

The FvCB model describes photosynthesis as the minimum of the Rubisco-limited rate and the electron transport-limited rate. The key parameters of the model are the maximum Rubisco carboxylation rate (Vcmax), the maximum electron transport rate (Jmax), and the mitochondrial day respiration (Rd). These biochemical parameters are influenced both by the physiological status of a leaf such as the amount of leaf nitrogen per unit area (Na) (e.g. Harley ) and by short- and long-term changes of environmental variables such as temperature, light (e.g. Hikosaka, 2005), CO2 (e.g. Makino ), and drought (e.g. Galmes ).

Usually, the FvCB parameters are obtained by analysis of net photosynthesis response to CO2 at the stomatal cavity (An–Ci) (e.g. Sharkey ) or by combining An–Ci and An–Iinc curves (e.g. Braune ) or by combining these curves with chlorophyll fluorescence measurements (Yin ). Obviously, to parameterize the FvCB model, information on An–Ci is predominantly considered to be essential, and an ongoing discussion is mainly focused on improving the methods of analysing these An–Ci curves (Ethier ; Sharkey ; Gu ).

In the context of forward crop modelling typically for predictions at the ambient CO2 level, the FvCB model is used to project leaf photosynthetic rates in response to both temporal (diurnal and seasonal) and spatial (within a crop canopy) variation in light intensity. This implies that in the context of inverse modelling, important FvCB model parameters Jmax and Vcmax should and can be estimated from An responses to Iinc. This would reflect better the tradition whereby crop modellers describe leaf photosynthesis from its response to light intensity (e.g. Goudriaan, 1979), in contrast to the tradition that photosynthesis physiologists study gas exchange measurements mainly across various levels of CO2 (e.g. von Caemmerer and Farquhar, 1981). In fact, the FvCB model can be parameterized from analysis of An–Iinc data alone (Niinemets and Tenhunen, 1997; Kosugi ), but so far there is no information about the accuracy of Jmax and Vcmax parameters derived from such an analysis. If Jmax and Vcmax estimates derived from analysis of An–Iinc are similar to those obtained from the common An–Ci analysis or combined analysis of An–Ci and An–Iinc curves, it may generate an opportunity to reduce empiricism in crop models by using readily available An–Iinc data. Therefore, the first objective of this study is to explore this opportunity by parameterizing the FvCB model using An–Iinc data.

In the light of current trends for a parallel increase in food and energy production from crop species in the context of climate change, the use of the FvCB-based simulation models together with an urgent need to feed these models with photosynthetic and respiration parameters has been increased (e.g. Beringer ). Compared with the rich information found for trees in the literature, there are only a few reports on Jmax, Vcmax, Rd, and night respiration (Rn) parameters in relation to environmental and management factors for economically important crop species (e.g. Müller ; Braune ; Yin ) and these are virtually lacking for new bioenergy species. Therefore, the second objective of this study is 3-fold: (i) to provide new information on photosynthesis and respiration for three Mediterranean energy crops (Helianthus annuus, sunflower; Hibiscus cannabinus, kenaf; and Cynara cardunculus, cynara); (ii) to summarize existing information for five major cash crops (wheat, barley, cotton, tobacco, and grape); and (iii) to assess how conserved FvCB parameters are among crop species to better assist modellers in this exploitation.

Sunflower, kenaf, and cynara crops were chosen because these crops have great potential to increase bioenergy production in the Mediterranean region (Archontoulis , ; Danalatos and Archontoulis, 2010). In addition, the chosen crops cover a wide range of bioindustrial applications (biodiesel, bioethanol, heat, and electricity) and fit into different cropping strategies (short or long growing period, cultivation with or without irrigation, etc.). Sunflower is widely grown in the Mediterranean region, but kenaf and cynara cultivation is still in the experimental phase; relevant information for crop modelling is currently being accumulated for these crops, including vertical distribution of light and nitrogen within crop canopies (Archontoulis ). Photosynthetic gas exchange studies for sunflower have been reported (e.g. Connor ), but there are only a few for kenaf (Muchow, 1990; Cosentino ) and none for cynara.

The present analysis focuses on the FvCB parameters in response to temperature and Na for these bioenergy crops. This is because earlier studies on Vcmax and Jmax temperature dependencies showed great species-to-species variability (Leuning, 2002; Medlyn ), and because Na is linearly related to Rubisco content that drives CO2 fixation (Makino ), reflects leaf dynamics well (leaf age, rank; Archontoulis ), and comprises a reference index for scaling photosynthetic CO2 assimilation from leaf to canopy levels (de Pury and Farquhar, 1997). Among bioenergy crops, the perennial cynara has long annual growth cycles (∼10 months each; Archontoulis ). Given the numerous reports together with their diverse findings on photosynthetic and respiratory acclimation to growth environment (Atkin ; Ow ; Yamori , 2010; Silim ), seasonal acclimation effects on photosynthesis and respiration for the cynara crop are also investigated.

Materials and methods

Literature data for An–Ci versus An–Iinc curves

The first objective of this study was to compare Vcmax and Jmax estimates derived either from An–Ci or from An–Iinc curves. For this, published data from Yin for Triticum aestivum (cv. Minaret) were used. All relevant parameter values required to fit the FvCB model to the An–Ci or An–Iinc data set were available, therefore avoiding any statistical artefact in Vcmax and Jmax estimation. Wheat measurements (four replicates; all at 25 °C) were conducted on leaves with different Na status (15 sets of An–Ci and 15 of An–Iinc curves), allowing the comparison of Jmax and Vcmax estimates to be made over a wide range of their values. For more information about the measurements, see Yin .

Energy crop species and study site

Sunflower (cv. Panter), kenaf (cv. Everglades 41), and cynara (cv. Biango avorio) crops were grown in different sections of the same field (for details, see Archontoulis ) in central Greece (39°25'43.4'' N, 22°05'09.7'' E, 105 m asl) for 3 years (2007–2009). The site has a Mediterranean climate with cold/wet winters and warm/dry summers (Supplementary Fig. S1 available at JXB online). The soil was loamy, classified as Aquic Xerofluvent, with a shallow groundwater table (1.8–2.8 m below the surface during May). In general, crops grown at that site produce much higher biomass yields than crops grown on dry soils (e.g. Archontoulis ). During summer, sunflower and kenaf crops were frequently irrigated at intervals of 4–6 d according to potential evapotranspiration (for site-specific calculations, see Danalatos and Archontoulis, 2010) while cynara was irrigated only a few times, when necessary during May–June but not during November–April (see precipitation in Supplementary Fig. S1).

Gas exchange measurements and experimental protocol

Leaf gas exchange (GE) measurements were implemented in situ in fully expanded leaves using a portable open gas exchange system with a 6.25 cm2 clamp-on leaf chamber (ADC, LCi/LCpro+, Bioscientific Ltd, Hoddesdon, UK). CO2/H2O exchanged by the leaf was measured using an infrared gas analyser in a differential mode. The system allowed for an automated microclimate control in the leaf chamber. Before each measurement, attached leaves were adapted for 10–45 min to chamber conditions, depending on leaf age, time of the day, and season. Daytime GE measurements were taken within 1–2 d after irrigation and during morning hours to ensure no water stress and to avoid midday depression of photosynthesis. Night-time GE measurements were initiated 30–45 min after sunset and lasted for 4–5 h each time.

To parameterize the model, a common experimental protocol was applied per species, including four different sets of GE measurements. In all sets, CO2 concentration was kept at 380±5 μmol mol−1. The first set aimed to determine the response of net photosynthesis (An) to incident light (Iinc). Accordingly, at fixed leaf temperature and measured Na, An was determined in 11 Iinc steps (2000, 1500, 1000, 500, 250, 200, 150, 100, 50, 20, and 0 μmol photons m−2 s−1); in total, 76 curves were constructed. Adaptation time to each Iinc level was ∼5 min, except for Iinc=0 where it was >10 min; 3–5 replicated An measurements were taken at each Iinc step to ensure stability and precision of measurements. Given that the examination of steady-state photosynthesis takes considerable time and that GE measurements should be done within a limited time frame in order to avoid stress conditions (see above), the response of An to leaf temperature (set II) was determined at three Iinc levels: 450, 900, and 1800 μmol photons m−2 s−1. Na was also determined. At each Iinc, leaf temperature was increased or decreased up to 10 °C from the ambient temperature in steps of 2–4 °C and replicated An measurements were recorded every 5 min. To establish the relationship between net photosynthesis and Na (set III), it was necessary to evaluate leaves with as wide an Na range as possible. So, in addition to earlier sets, An measurements were done at saturated Iinc (1600–1800 μmol photons m−2 s−1) on leaves from different insertion heights in the canopy, from different growth stages, and from plots with different N status. Per leaf (∼180 leaves assessed), 5–10 measurements were taken at leaf temperature close to the ambient temperature.

To obtain direct measurements of the mitochondrial respiration occurring in the night (Rn), the response of Rn to temperature was investigated (set IV). Leaf temperature increased or decreased up to 10 °C from the ambient temperature in small steps of 1–2 °C, and replicated Rn measurements were recorded every 4 min. Measurements were done on leaves with (as much as possible) variable Na.

To validate the models, GE measurements obtained from the same genotypes growing in the same site during summer 2005 and 2006 (set V) were used. Sunflower and kenaf GE measurements were collected using similar techniques and time frames to those described for sets I–IV. In cynara, a different protocol was followed. The external unit that controls chamber microclimate was removed to obtain measurements under real ambient conditions. Measurements were recorded every 4–8 min, while climatic variables were continuously changing following 24 diurnal trends, thereby providing a data set to assess whether the FvCB model can predict An under real fluctuating field conditions.

The wide range of measuring temperature used (15–40 °C) unavoidably resulted in variation in vapour pressure difference (VPD). An effort was made to reduce that variation by keeping humidity high at high temperature. In most cases, VPD was maintained below 3 kPa to prevent stomatal closure (Bernacchi ). Although VPD was sometimes above 3 kPa at the highest temperatures, the stomatal conductance for H2O vapour was not less than 0.30 mol m−2 s−1 (as in Yamori ).

All measured An data were corrected for the CO2 respired under the gasket surface (total 4 mm width; R. Newman, personal communication) following the common approach of Pons and Welschen (2002). All GE characteristics were re-calculated according to von Caemmerer and Farquhar (1981), for example to provide the Ci values that are required as input to the FvCB model (see below). In addition, the number of replications and observations were increased to reduce the measurement noise, especially when low CO2 exchange rates were measured (e.g. respiration).

The portion of the leaf used for measurements was cut and its area was measured with a Li-Cor area meter. The leaf material was then weighed after drying at 70 °C to constant weight and its total nitrogen concentration was measured using the Kjeldahl method. From these measurements, the leaf nitrogen content Na (g N m−2) was calculated.

Model and its parameterization

The FvCB model predicts An (μmol CO2 m−2 s−1) as the minimum of two processes (see Fig. 1), the Rubisco carboxylation-limited rate (Ac) and the RuBP regeneration- or electron transport-limited rate (Aj):Rubisco-limited photosynthesis is calculated as a function of maximum carboxylation capacity (Vcmax, μmol CO2 m−2 s−1):where Ci (μbar) and O (mbar) are the intercellular partial pressures of CO2 and O2, respectively, Kmc (μbar) and Kmo (mbar) are the Michaelis–Menten coefficients of Rubisco for CO2 and O2, respectively, and Γ* (μbar) is the CO2 compensation point in the absence of Rd (day respiration in μmol CO2 m−2 s−1, which comprises mitochondrial CO2 release occurring in the light other than photorespiration; von Caemmerer ).

Fig. 1.

Main panel: typical net photosynthesis light response curve (An–Iinc) at ambient CO2 concentration. Curve regions for the Rubisco carboxylation-limited rate (Ac-limited, Equation 2; solid line) and the electron transport-limited rate (Aj-limited, Equation 3; dotted line) are indicated. Usually, Ac-limitation occurs above 1500 μmol photons m−2 s−1; however, it is also possible that the entire An–Iinc curve is described as Aj-limited. Inset panel: representative portion of the An–Iinc curve used in calculations of the day respiration (Rd), night respiration (Rn), and apparent quantum yield (ΦCO2LL). Rd and ΦCO2LL were calculated from linear regression analysis to open circles while the filled circle represents the value of the Rn. For details, see the Materials and methods.

There are various equations to describe the rate of photosynthesis when RuBP regeneration is limiting (Farquhar and von Caemmerer, 1982; Yin ). The most widely used form is given by:where J (μmol e− m−2 s−1) is the photosystem II electron transport rate that is used for CO2 fixation and photorespiration. J is related to the amount of incident photosynthetically active irradiance (Iinc; μmol photons m−2 s−1) by:where Jmax (μmol e− m−2 s−1) is the maximum electron transport rate at saturating light levels, θ is a dimensionless convexity factor for the response of J to Iinc, and κ2LL (mol e− mol−1 photons) is the conversion efficiency of Iinc into J at limiting light levels (Yin and Struik, 2009; Yin ). The formulation of Equations 2 and 3 assumes infinitive mesophyll conductance (gm) for CO2 transfer to chloroplasts, so that Ci is used as the proxy for the chloroplast CO2 level (Cc). There is increasing evidence that gm might be low enough to allow a significant drawdown of Cc from Ci in most species (Warren, 2004; Flexas ). However based on the available GE data, it was risky to evaluate gm (Pons ; von Caemmerer ; Yin and Struik, 2009), hence the forms of Equations 2 and 3 had to be used, as in most earlier studies (e.g. Medlyn ; Kosugi ). Omitting gm in the analysis means that an appropriate consideration is needed in choosing values of the Rubisco kinetic constants (see below).

The temperature responses of respiration and of Rubisco kinetic properties (Kmc and Kmo) are described using an Arrhenius function (Equation 5) while the temperature responses of Vcmax and Jmax were explored using a peaked Arrhenius function (Equation 6); both functions were normalized with respect to their values at 25 °C:where T is the leaf temperature (°C); X25 is the value of each parameter at 25 °C (Rn25, Kmc25, Kmo25, Vcmax25, and Jmax25); Ex is the activation energy of each parameter (ER, EK, EK, Ev, and Ej; in J mol−1); Dx is the deactivation energy for Jmax and Vcmax (Dj and Dv in J mol−1); Sx is the entropy term for Jmax and Vcmax (Sj, Sv in J K−1 mol−1), and R is the universal gas constant (=8.314 J K−1 mol−1). Given that Equation 5 is a special case of Equation 6, F-tests were performed to determine whether Equation 6 described temperature responses of Vcmax and Jmax significantly better than did Equation 5. When Equation 6 was overparameterized, as often observed in the literature (Dreyer ; Medlyn ), then Sx was fixed at 650 J K−1 mol−1 (Harley ).

Rubisco kinetic properties are generally assumed constant among C3 species (von Caemmerer ). However, values of these constants and their temperature dependency reported in the literature vary appreciably, so the choice of Rubisco parameters is a matter of considerable uncertainty (Dreyer ). In this work, similar to many other reports (e.g. Medlyn et al., 2002a; Müller ), Rubisco parameters reported by Bernacchi were selected because these values (i) were estimated from in vivo measurements without disturbance of the leaf; and (ii) were derived using the Ci-based FvCB model and hence are compatible with the present analysis assuming an infinite gm (see above). The parameter values are: Kmc25=404.9 μbar; Kmo25=278.4 mbar; EKmc=79 430 J mol−1; and EKmo=36 380 J mol−1 (Table 1). Furthermore, using these values. the temperature dependence of Γ* was calculated as (Yin ):where the factor 0.5 is mol CO2 released when Rubisco catalyses the reaction with 1 mol O2 in photorespiration. The term in the brackets was derived using Bernacchi parameters for temperature dependence of maximum carboxylation and oxgenation rates of Rubisco.

The basal capacity of Rn25, Vcmax25, and Jmax25 is linearly related to Na (Harley ; Hirose ; Müller ; Braune ):where χR (μmol CO2 g−1 N s−1), χv (μmol CO2 g−1 N s−1), and χj (μmol e− g−1 N s−1) are the slopes for Rn25, Vcmax25, and Jmax25, respectively, and Nb (g N m−2) is the minimum value of Na at or below which An is zero. In principle, this Nb is practically impossible to measure and its estimation depends on the statistical methods used and on the available data sets. For instance, different Nb estimates were found when different data sets were examined (An or Vcmax, or Jmax; e.g. Harley ; Müller ; Supplementary Table S1 at JXB online) or when Nb was estimated simultaneously with other parameters in optimization procedures or when different equations (linear or non-linear) were applied to the same data set (Niinemets and Tenhunen, 1997). Given the simplicity required in modelling and the lack of biological interpretation of different Nb values for the same species, a unique Nb value (per species) was determined beforehand from direct assessments of An–Na plots. Then this estimate was used as input parameter.

There is some evidence that the activation energy for respiration (ER) depends on the position of the leaf in the canopy (Bolstad ; Griffin ) and perhaps ER is also associated with Na since a close relationship between leaf canopy position and Na usually exists (Archontoulis ). This was tested by assuming a linear relationship between ER and Na:and it was checked whether the slope parameter ER differed significantly from zero.

So far, temperature and nitrogen relationships for Rn have been described, as extensive GE measurements during the night period were available. However, the FvCB requires estimates for Rd, which is much more difficult to measure. To estimate Rd, regression analysis was applied to the linear sections of the An–Iinc curves for each species (Fig. 1, inset; Kok method; Sharp ). From this analysis, Rd was calculated as the y-axis intercept of the linear regression and the corresponding Rn was estimated as the mean of the An values at 0 μmol photons m−2 s−1. Additionally, the apparent quantum efficiency at limiting light (ΦCO2LL, mol CO2 mol−1 photons) on the incident light basis was calculated from the slope of the regression. The Iinc range for this regression analysis was typically 20–150 μmol m−2 s−1 (Fig. 1, inset), while in a few cases the Iinc range was slightly different, especially for data sets obtained at high temperatures. The estimated Rd was then related to Rn as:where bR and aR are the slope and the x-axis intercept of the linear model, respectively. By assuming that activation energies for Rd and Rn are similar and taking into account the precise quantification of Rn based on a large data set, the temperature and nitrogen dependencies of Rd can be calculated from combining Equations 5, 8, 11, and 12. This approach allows Rd values to be estimated for sets II and III (see above) where Iinc exceeds 350 μmol m−2 s−1, for which it was not possible to use the Kok method for estimating Rd.

Summary of parameters and statistics

The basic equations of the FvCB model, Equations 1–4, capture the response of An to Ci and to Iinc. Coupled with auxiliary temperature (Equations 5–7) and nitrogen (Equations 8–12) equations, the model also quantifies leaf photosynthesis and respiration (Rd and Rn) in response to these environmental variables. Data from sets I–IV were analysed using step-wise optimization procedures. Per crop, 16 parameters were estimated following the order: step 1, Nb; step 2, χR, ER, ER; step 3, bR, aR; step 4, χv, Ev, Dv, Sv; step 5, κ2LL; step 6, χj, Ej, Dj, Sj, and θ (see the Results). Inputs to the model are: Ci, Iinc, leaf temperature, and Na. So, just like using An–Ci curves, using An–Iinc data to calculate FvCB model parameters (e.g. Vcmax) also requires Ci as an input to the model, meaning that any (short-term) change in stomatal aperture during the An–Iinc measurements will have been reflected in the values of Ci and thus have little effect on the calculation of the FvCB parameters. For a similar reason, the direction of changing Iinc levels for measuring A–Iinc curves will also have little impact on parameter estimation (see Yin ).

For each step, regression fitting was carried out using the GAUSS method in PROC NLIN of SAS (SAS Institute Inc.). To investigate seasonal effects of acclimation on photosynthesis and respiration rates of cynara, data sets were split into two periods: a cold period with low light from November to April and a warm period with high light from May to June (Supplementary Fig. S1 at JXB online). Then, dummy variables (Z1=1 and Z2=0 for warm and Z1=0 and Z2=1 for cold periods, respectively) were introduced into the regression analysis to separate for the effects. A dummy variable was also used to best estimate the Nb parameter (see the Results).

The goodness of model fit was assessed by calculating r2 and the relative mean root square error (rRMSE). A sensitivity analysis was also performed. Model predictions were validated against independent data sets (set V).

Results

Vcmax and Jmax estimates from An–Ci and/or An–Iinc curves

Vcmax and Jmax were estimated for wheat, from either An–Iinc or An–Ci curves alone or from the combined data of the two curves. The following parameters were set as inputs to the model (see Equations 1–4 and 7): Kmc25 and Kmo25 from Bernacchi ; and Rd25, Rn25, κ2LL, and θ per set of data from Yin . Vcmax and Jmax were successfully estimated simultaneously in 40 out of the 45 cases (15 sets×3 methodologies). In five cases, it was not possible to estimate Vcmax from An–Iinc curves because in these cases the entire curve was Aj limited (Fig. 1). Then we first calculated Vcmax directly from Equation 2 with observed Ci as input for simple substitution using data points where Iinc >1500 μmol photons m−2 s−1, and secondly by setting Vcmax as an input to the model, the Jmax parameter was estimated again. To be consistent, results for all An–Iinc curves were presented following the two-step approach, because estimates from both approaches were very close.

Figure 2 illustrates Vcmax and Jmax estimates from An–Ci and from An–Iinc curves versus the combination of those curves. As expected, Vcmax and Jmax estimates obtained from An–Ci curves were almost identical to the estimates based on the combined data (r2=0.97–0.99). However, it was found that An–Iinc curves alone also provided sufficient estimates (r2=0.91–0.93) and thus can be considered as an alternative to predominant An–Ci curves to parameterize the FvCB model. In fact, regression lines in Fig. 2 were matching across a very wide range of Vcmax and Jmax values. Even in cases where photosynthetic responses to light were entirely Aj limited (Fig. 1), Vcmax estimates obtained from either An–Iinc or An–Ci data were close (Fig. 2). The slight discrepancy of the estimates at high Vcmax and Jmax values (Fig. 2) caused a lower r2 for the An–Iinc compared with the An–Ci estimates.

Fig. 2.

Relationships between Vcmax (μmol CO2 m−2 s−1) and Jmax (μmol e− m−2 s−1) estimated from photosynthetic light response curves at ambient CO2 concentration (open circles; An–Iinc) or from photosynthetic CO2 response curves at saturated light (filled circles; An–Ci) versus estimates obtained from an analysis of combined An–Iinc and An–Ci curves. Data for An–Ci and An–Iinc measurements are from Yin for Triticum aestivum (n=15).

Step-wise estimation of model parameters for bioenergy crops

Step 1: Nb estimation

Measured light-saturated An (An,max) responded non-linearly to increasing Na in all tested crops (Fig. 3; r2 > 0.81; P < 0.001). An effect of temperature was detected in this relationship only at high Na (Fig. 3). To estimate the Nb value properly from these plots a dummy variables approach was used, in order to obtain a unique Nb estimate per crop, while allowing the equation to vary with different temperatures (optimum versus non-optimum temperature ranges; Fig. 3). Derived parameters are listed in Table 2. Nb values for all crops were close to 0.4 g N m−2, while the lack of Na data below 0.7 g m−2 caused a high standard error of the Nb estimate (Table 2).

Fig. 3.

Relationships between light-saturated net photosynthesis, An (Iinc >1500 μmol m−2s−1; CO2=380 μmol mol−1), and leaf nitrogen content, Na. Filled symbols refer to data obtained at temperatures near the optimum temperature for photosynthesis per species (sunflower, 26–34 °C; kenaf, 27–35 °C; cynara, 23–31 °C) and open symbols refer to data obtained at sub- (open squares) or supra- (open triangles) optimum temperature ranges. Each point is an average of 4–10 measurements. Lines are fits from a three-parameter non-linear equation: An=An,max {2/[1+exp(–c(Na–Nb))]–1}, (see Sinclair and Horie 1989), where An,max is the asymptote (maximum value) of the dependent variable; c is the parameter determining the steepness of the curve; and Nb is the intercept of the x-axis denoting a threshold leaf nitrogen value at or below An equals zero. Estimates of parameters are given in Table 2. Cynara’s data points were mostly collected during May–June.

Table 2.

Estimates (SE in parentheses) of the non-linear equation used to describe data illustrated in Fig. 3

Species	Symbola	An,max	c	Nbb
Sunflower	Filled (26–34 °C)	36.6 (2.48)	1.19 (0.195)	0.387 (0.078)
	Open	26.4 (1.45)	1.65 (0.313)
Kenaf	Filled (27–35 °C)	35.8 (2.18)	1.29 (0.269)	0.390 (0.126)
	Open	29.2 (2.12)	1.45 (0.334)
Cynara	Filled (22–31 °C)	36.4 (2.41)	1.08 (0.191)	0.416 (0.097)
	Open	23.9 (2.16)	1.22 (0.725)

An,max is the maximum net assimilation rate (μmol CO2 m−2 s−1) at saturated light, maximal leaf nitrogen content, ambient CO2 concentration, and at optimum (filled symbols) and non-optimum (open symbols) temperature ranges; c is a dimensionless factor determining the steepness of the non-linear model; and Nb is the minimum leaf nitrogen content (g N m−2) required for photosynthesis.

Symbols in Fig. 3.

The confidence limits for Nb are: 0.231–0.541, 0.139–0.640, and 0.225–0.608 for sunflower, kenaf, and cynara, respectively (P=0.05).

Step 2: Rn in relation to temperature and Na

By combining Equations 5, 8, and 11, Rn parameters were estimated (Table 3). In cynara, an additional seasonal effect was found, with significantly higher Rn rates for the winter/cold- compared with the summer/warm-growing leaves (Fig. 4). Incorporation of this effect into the model improved r2 from 0.68 to 0.72. Of the two Rn parameters, temperature sensitivity (ERn) was significantly (P < 0.01) affected by season, but the slope of the Rn–Na relationship (χR) was not (P=0.263); thus, a common χR value was calculated (Table 3). The Rn models’ goodness of fit was satisfactory (r2 > 0.72; rRMSE < 0.28 across species).

Table 3.

Estimates (SE in parentheses) of parameters used to describe temperature and nitrogen sensitivities of photosynthesis and respiration rates in three bioenergy crops For cynara, when significant differences between warm and cold seasons were found, two estimates are given. For units see Table 1.

	Parameter	Sunflower	Kenaf	Cynara-warma	Cynara-colda
Rn	χR	0.609 (0.006)	0.954 (0.015)	0.775 (0.009)
	ERn(a)	117 912 (1814)	100 740 (3250)	–10 900 (5617)	146 956 (4281)
	ERn(b)	–23 346 (770)	–15 743 (1455)	33 040 (2490)	–26 640 (1858)
	n (night)b	2492	1403	3212
	r2	0.799	0.793	0.724
Rd/Rn	bR	0.843 (0.040)
	aR	0.390 (0.107)
Vcmax	χv	73.8 (0.94)	66.7 (0.92)	65.2 (0.62)
	Ev	53 688 (1631)	61 812 (1402)	190 831 (33 853)
	Dv	205 638 (355)	0	158 486 (30 907)
	Sv	650c	0	550 (108.2)
J	χj	144.2 (3.4)	122.1 (1.88)	100 (0.91)	92.2 (0.88)
	Ej	43 295 (5122)	28 584d (1131)	23 111 (971)
	Dj	125 324 (12 653)	0d	204 489 (218)
	Sj	405 (38.47)	0d	650c
	κ2LL	0.255 (0.018)	0.278 (0.013)	0.314 (0.014)	0.419 (0.011)
	θ	0.607 (0.027)	0.627 (0.023)	0.847 (0.011)
	n (day)b	1366	2042	2334
	r2	0.928	0.909	0.916
Ratio	Jmax/Vcmaxe	1.95	1.83	1.53	1.41
	Rd/Vcmaxe	0.0057	0.0103	0.0085

Warm period=from early May to end of June; cold period=from November to mid-April; see supplementary Fig. S1 at JXB online.

Number of data used in the analysis.

Fixed value (see the Materials and methods).

Alternatively the following parameters: Ej=28 149, Dj=474 614, and Sj=1482 (with a temperature optimum of 41.7 °C) gave equal temperature sensitivities but values were rejected due to a high standard error of the estimate.

Normalized to 25 °C.

Fig. 4.

Cynara’s night respiration rates (Rn) in relation to leaf temperature. Data are presented per growth season and include leaves with various Na. (a) The predicted Rn from a simple temperature-sensitive model (Equation 5; parameter values used are shown). (b) The predicted Rn from a combined nitrogen-, temperature-, and acclimation-sensitive model (see parameter values in Table 3).

Step 3: relationship between Rd and Rn

Plotting Rd versus Rn gave a good linear relationship with no significant differences among species (P=0.225; Fig. 5; Table 3). Analysis showed that mitochondrial respiration was inhibited by ∼28% in the light. The observed x-axis intercept (aR=0.39) differed significantly from zero (P=0.0039), indicating that Rn and Rd were not entirely proportional (Fig. 5). Additionally, no effect of Na (r2=0.01; P=0.67) but a significant effect of temperature (r2=0.18; P=0.008) was found on the Rd/Rn ratio, showing that the ratio approached unity at high temperatures. Similarly, the Rn/An,max ratio—ranging from 7% to 11% across bioenergy species—was insensitive to changes in Na (P > 0.05), but increased significantly with increasing temperature (r2=0.62; P < 0.01; data not shown).

Fig. 5.

Relationship between day (Rd) and night (Rn) respiration rates (see also Table 3 and Equation 12).

Step 4: Vcmax in relation to temperature and Na

The relationships of Vcmax to temperature and Na were quantified by fitting Equations 2 and 5–12 to data obtained at high light levels (Iinc ≥1500 μmol m−2 s−1) to ensure that An is limited only by Rubisco. All required parameters (χv, Ev, Dv, and Sv) were well estimated. Across species, there were small differences in χv (<12%; Table 3), and large differences in temperature sensitivities >30 °C (Fig. 6a; including other crops). Sunflower temperature sensitivity was best described by the peaked Arrhenius equation (r2=0.736; P < 0.001; Table 3), showing an optimum temperature for Vcmax at 38.7 °C (calculated from Equation A1 in the Appendix). For kenaf and cynara no optimum temperature was observed within the measurement range tested (18–41 °C; Fig. 6a). To explore any acclimation of Vcmax to growth environments in cynara, the model was allowed to estimate different parameters for two contrasting seasons. No significant effect of the growing season on χv (65.8 versus 64.3; P=0.094) or on Ev, Dv, and Sv parameters (P=0.247) was found, meaning little seasonal Vcmax acclimation.

Fig. 6.

Temperature sensitivities for Vcmax (a) and Jmax (b). Values normalized to 1 at 25 °C. Filled symbols refer to bioenergy crops while open symbols refer to wheat (de Pury and Farquhar, 1997), barley (Braune ), cotton (Harley ), grapevine (Schultz, 2003), tobacco (Bernacchi ; 2003), and perennial Plantago asiatica (Ishikawa ). For kenaf, both observed Jmax temperature sensitivities are plotted (see Table 3; note that for kenaf the measurement range was up to 41 °C).

Step 5: κ2LL in relation to temperature and Na

κ2LL was estimated indirectly from ΦCO2LL information (see Equation A2). Correlations of κ2LL with temperature, light, and nitrogen were investigated afterwards. The results indicated poor correlations with Na (r2=0.26, P=0.025), leaf temperature (r2=0.19, P=0.104), and the combination of the above (r2=0.44, P < 0.01; data not shown). However, better relationships were obtained when κ2LL was regressed against seasonal temperature (r2=0.40, P=0.004) and radiation data (r2=0.34, P=0.003), showing a long-term κ2LL acclimation. This became clearer when average κ2LL values per crop and per growth environments were considered (Fig. 7). These findings were supported fairly well by literature data (Fig. 7). Based on this analysis, average κ2LL values per species were considered in further analyses (including acclimation effect for cynara, Table 3).

Fig. 7.

Conversion efficiency of incident light into linear electron flux (κ2LL) in relation to (a) seasonal growth temperature, (b) short-term changes in leaf temperature, (c) seasonal irradiance, and (d) Na, for three bioenergy crops (filled symbols: see key in Fig. 5) and four major field crops (open symbols; see panels for details). Growth temperatures and irradiances were calculated from Supplementary Fig. S1 at JXB online. For sunflower and kenaf one average κ2LL value (± vertical standard error) was calculated because measurements (set I) were conducted during the July–August period when temperature and radiation do no change much (Supplementary Fig. S1 at JXB online). In contrast, for cynara (cold and warm) four average κ2LL values were calculated, reflecting the months November, April, May, and June, respectively. Horizontal bars (when larger than symbols) indicate the mean standard error of the explanatory variable. Information on growth temperature and irradiance could not be retrieved from from the studies Müller and Schultz (2003).

Step 6: Jmax in relation to temperature and Na

All Jmax temperature sensitivities (except kenaf; Table 3) were best described using Equation 6. Across species, Jmax temperature sensitivity was highly variable (Fig. 6b including other crops), while the maximum Jmax was obtained at lower temperature than the maximum Vcmax (temperature optimum of 32, 42, and 33 °C for sunflower, kenaf, and cynara, respectively; Fig. 6). As a result, there was a decreasing trend of the Jmax/Vcmax ratio with increasing temperature (Fig. 8). For cynara, a significant (P < 0.05) temporal change was found for the χj parameter (Table 3). χj showed a larger variability (36% change) than χv (12% change) among species and growth environments studied (Table 3). The parameter θ was lower for sunflower (0.60) and higher for cynara (0.84), but close to the commonly used value of 0.75 in all cases. All these differences (including temperature and nitrogen sensitivities) among species and growth environments became smaller when the Jmax/Vcmax ratio was plotted against leaf temperature (Fig. 8).

Fig. 8.

Jmax/Vcmax ratio versus leaf temperature. Closed symbols refer to bioenergy crops, open symbols to major field crops. Note that for cynara two lines were plotted because the parameter Jmax25 differs between seasons (see Equation 10 and Table 3).

Sensitivity and validation analysis

To investigate the uncertainty introduced into the estimates by the chosen Rubisco kinetic parameters, the initial values of Bernacchi were increased or decreased by 20% and optimization procedures were repeated. Not surprisingly, a maximum change was obtained in the estimated Vcmax25, whereas the remaining parameters were less affected (<5%; data not shown). Given that even the maximum change in Vcmax was ∼11% in response to a 20% change, the parameter estimates were quite stable despite the uncertainties in values of Rubisco kinetic constants. A further analysis showed that the predicted An was sensitive to a 20% decrease in χv and χj, whereas its sensitivity to other changes was weak (Fig. 9).

Fig. 9.

Sensitivity analysis of the predicted An in response to a ±20% change in input parameter values for the photosynthesis model. The relative change in predicted value was calculated as: 100×(An, predicted–An, predicted, original)/An, predicted, original. When input parameters were part of a linear or polynomial equation (e.g. Ej, Dj, Sj; Equation 6) and strongly intercorrelated, a combined change was implemented.

Lastly the models were validated against independent data sets (Fig. 10). Predictions versus observations for sunflower and kenaf were satisfactory (rRMSE < 0.15; Fig. 10a, b). For cynara the FvCB model was tested using measurements from a series of 24 h diurnal cycles (Fig. 10c), where stress conditions were unavoidably present (data sets outside the calibration range). In general, predictions were close to actual measurements, except for those data obtained from 14:00 h to 18:00 h, where a systematic overestimation was detected (Fig. 10c). The FvCB model responded to lowering temperature in late afternoon by increasing An; however, actual measurements indicated that the photosynthetic apparatus could not recover so quickly from the ‘photosynthesis midday depression’. The failure in predicting the depression and its after-effect during the recovery hours (Fig. 10c) might be attributed to the ‘steady-state’ character of the FvCB model. These results suggest that prediction of diurnal photosynthesis for species grown in the Mediterranean region requires more detailed approaches in which gm, recovery functions for An (midday depression), and the effects of leaf water potential should be included (see Tuzet ; Vico and Porporato, 2008; Yin and Struik, 2009).

Fig. 10.

Measured versus predicted photosynthesis (all panels) and measured versus predicted night respiration (only in the lower panel c). In c, canopy CO2 varied from 350–380 μmol mol−1 during day time to 450–600 μmol mol−1 during night-time; VPD followed temperature variations, and stomatal conductance ranged from 0.05 mol m−2 s−1 during the night up to 0.48 mol m−2 s−1 during the day time. The model predicted diurnal trends moderately (r2=0.814, rRMSE=0.553, n=720). When midday measurements were excluded (14:00–16:00 h), the model fit was improved (r2=0.930, rRMSE=0.335, n=543).

Discussion

Use of An–Iinc curves to parameterize the FvCB model

The FvCB model parameters, Jmax and Vcmax in particular, have been predominantly estimated from An–Ci data sets (Harley ; Medlyn ). The value of An–Ci curves for parameterizing the FvCB model is confirmed (Fig. 2). It was also shown that Vcmax and Jmax can be estimated sufficiently well by an appropriate analysis of An–Iinc data alone (r2=0.91–0.93; Fig. 2). Unlike Jmax, Vcmax cannot always be estimated from An–Iinc curves; that is when the entire curve is Aj limited (Fig. 1). This is often observed in field crops (e.g. cotton; Wise ). Actually, Boote and Pickering (1994) used only the Aj equation of the FvCB model to calculate leaf photosynthesis in their canopy photosynthesis model. For the purpose of using the complete FvCB model, the two-step approach is proposed to estimate both Vcmax and Jmax from An–Iinc data. This is in line with the approach of Niinemets and Tenhunen (1997), but in contrast to that of Kosugi and Müller who assumed a fixed Jmax/Vcmax ratio of 2.1 at 25 °C (based on Wullschleger, 1993) in their analyses. This assumption does not allow for the flexibility of the ratio as observed for different species or for the same species when grown under different environments, thereby introducing many uncertainties in parameter values (see Fig. 8 and discussion below).

The present results indicated that information from An–Iinc curves has been underexplored. Use of An–Iinc curves has an additional advantage in that data of An–Ci curves may be uncertain due to CO2 leakage during gas exchange measurements when CO2 set point values are either below or above the ambient air CO2 level (Flexas ). Crop modellers used to measure An–Iinc curves under an ambient CO2 condition, upon which an empirical model for light–response curves is parameterized. Provided that values of Ci across Iinc levels are properly monitored, re-analysing readily available An–Iinc data to parameterize the FvCB model will strengthen photosynthesis calculations in crop models. This would expand the potential of combining classical photosynthetic data and the biochemical FvCB model to assess the impact of climate change on crop production and to examine options of bioenergy production under a changing climate.

On the other hand, caution should be exercised that use of An–Iinc data sets does not allow the model to account for the TPU (triose phosphate utilization)-limited rate, the third limitation added by Sharkey (1985) to the FvCB model. TPU limitation sets an upper limit to the maximum photosynthetic capacity and is usually observed at high CO2 or/and low O2 levels (e.g. Wise ), although many studies still ignore this limitation (e.g. Wohlfahrt ). The limitation, if it occurs, can be easily identified, at the high end of An–Ci curves, versus the Rubisco limitation that can be identified at the low end of An–Ci curves. In the present study, where essentially An–Iinc curves were used, it was not possible to detect this limitation, because both Rubisco and TPU limitations, if any, will occur at the high end of An–Iinc curves. This is certainly the disadvantage of using An–Iinc curves to parameterize the FvCB model. Fortunately, the present light response curves were obtained under ambient CO2 conditions, so any TPU limitation, if it exists, can be assumed to be negligible under these measurement conditions. In the future climate where the ambient CO2 level is expected to increase, the TPU limitation will be more likely to occur. Therefore, use of An–Iinc curves to estimate FvCB model parameters needs to be tested across high CO2 levels and a broad range of other environmental variables in order to decide how conserved these parameters are.

Below the effects of temperature, Na, and season on photosynthesis and respiration parameters, all derived from the current An–Iinc data for three bioenergy crops, are discussed. The present findings will be compared with those reported for the crops wheat, barley, cotton, tobacco, and grapevine based on An–Ci or combined An–Ci and An–Iinc data sets, with attention to any conserved nature in these parameters among species.

Night and day respiration parameters: χR, ER, ER, bR, and aR

This study is among few in the literature providing direct Rn measurements, underlining the great importance of respiration in carbon budgets (Valentini ). The present estimates for χR (range: 0.61–0.95 μmol CO2 g−1 N s−1; Table 3) agree well with previous reports for crops (Hirose ; Reich ; Müller ; Braune ), but current values are almost double compared with those for trees (Bolstad ; Griffin ). The temperature sensitivity for respiration (ERn) was significantly correlated with Na in all species (Equation 11; Table 3), indicating that respiration in leaves with high Na values (young/sun leaves) was less sensitive to changes in temperature, while leaves with lower Na values were more sensitive (senescence/shade leaves). Griffin and Bolstad working with tree leaves that were positioned in different canopy layers—also having different Na values—found temperature sensitivities similar to those in the present study, while Turnbull reported the opposite. However, in none of these studies was ERn significantly correlated with Na.

For cotton, Harley reported a simple temperature-sensitive Rn model for leaves with variable Na. The present analysis indicated that it is useful to calculate both Rn components as a function of Na (e.g. Fig. 4; across all species, r2 scaled from 0.53 to 0.77). The component Rn25 accounted for 27% and ERn for the other 5% of this improvement in r2. However, the remaining unexplained variability in night data sets (see r2 in Table 3; Fig. 4) means that apart from Na, other factors should be explored.

Unlike for Rn, it is difficult to measure Rd directly as such measurements require sophisticated methodologies (e.g. Haupt-Herting ; Pinelli and Loreto, 2003; Pärnik and Keerberg, 2007). Its value is empirically estimated indirectly using various methods (for a comparison see Yin ), or is commonly fixed as 1% of Vcmax or as 50% of Rn (de Purry and Farquhar, 1997; Wohlfahrt ; Medlyn ; Kosugi ; Braune ). Here, application of the Kok method (Sharp ) indicated a 28% reduction in Rd compared with Rn, an estimate which is positioned at the lowest reported range (light inhibition range: 24–90%; Buckley and Adams, 2011, and references therein).

Rubisco and electron transport parameters: Nb, χv, χj, κ2LL, θ, Ev, Ej, Dv, and Dj

The present findings for Nb (Fig. 3) along with published data support the idea that this threshold value for photosynthesis is not affected by temperature (Sage and Pearcy, 1987; Makino ; Niinemets and Tenhunen, 1997), CO2 (Harley ; Hirose ), or irradiance levels (Makino ). Excluding the statistical bias that usually exists in Nb estimations (see the Materials and methods) it is believed that a common Nb is 0.3–0.4 g N m−2 for C3 crop species (excluding legume crops; Supplementary Table S1 at JXB online). For use in modelling, it was shown that a ±20% change in the Nb value resulted in a <5% change in the predicted An (Fig. 9).

The relationships between An,max and Na at near-optimum temperature ranges for sunflower, kenaf, and cynara (Fig. 3) agreed well with several non-legume C3 species (Supplementary Fig. S2 at JXB online). The observed decline in An,max at high temperature (Fig. 3; Na >2 g N m−2) is associated with gm (Bernacchi ) and/or Vcmax and Jmax limitations of photosynthesis (Fig. 6). Na and leaf temperature explained >81% of the temporal (seasonal) and spatial (within a crop canopy) variation in An,max values (Fig. 3). The remaining unexplained variability might be due to leaf adaptation to different microenvironments created by CO2 and light gradients within crop stands (Buchmann and Ehlinger, 1998; Archontoulis ). This may have an additional impact on Jmax and Vcmax estimates and their ratio.

Nevertheless, the observed consistency among An,max–Na plots (Supplementary Fig. S2 at JXB online) along with the similar χv estimates for sunflower, kenaf, cynara, cotton, wheat, and barley (range: 60–82 μmol CO2 g−1 N s−1; Table 3; Harley ; de Pury and Farquhar, 1997; Müller , 2008; Braune ) suggests that χv is very conserved for this plant group (An,max=30–35 μmol CO2 m−2 s−1; Supplementary Fig. S2).

Unlike χv, χj for the same group was highly variable (90–165 μmol e− g−1 N s−1). However, the parameter χj (which determines Jmax25; Equation 10) is not independent of, but interrelated to, the values of κ2LL and θ (see Equation 4). This means that use of constant κ2LL and θ values across species and environments will bias Jmax estimates and therefore the Jmax/Vcmax ratio. Among sunflower, kenaf, and cynara, χj varied by 36%, κ2LL by 39%, and θ by 28%, but in different directions (Table 3). When κ2LL was fixed to 0.3 and θ to 0.7 (commonly assumed values; de Pury and Farquhar, 1997; Medlyn ), the χj variation among crops and growing environments became smaller (15%), and the Jmax/Vcmax ratio less variable.

The present analysis showed that variation in the electron transport rate among bioenergy crops followed changes in environmental conditions during growth (Supplementary Fig. S1 at JXB online), with higher J rates for cynara in low light (<700 μmol m−2 s−1; winter period) and higher J rates for sunflower and kenaf in high light conditions (>700 μmol m−2 s−1; summer period; Table 3, Equation 4). This is consistent with recent findings for tobacco (Yamori ) where plants grown under low light enhanced the efficiency of light acquisition while those grown under high light enhanced the capacity of light utilization, through changes in chlorophyll contents, the chlorophyll a/b ratio, and cytochrome f and Rubisco contents.

In studies of Wullschleger (1993), Dreyer , and Medlyn the κ2LL was fixed as a constant at 0.18, 0.24, and 0.30, respectively, across species, crop stages, and environments. However, Yin directly demonstrated a positive relationship between κ2LL and Na, which was confirmed by the results of a model curve-fitting procedure (Müller ; Braune ; Yamori ).

In Fig. 7, κ2LL information for eight crops is summarized and this large variation is interpreted in the light of long- or short-term response to temperature or irradiance. Across species, the highest κ2LL values were found in crops grown under long-term low irradiance and temperature conditions (Fig. 7a, c). To understand this, it is necessary to underline the components of the κ2LL parameter (see Equation A3 derived by Yin , 2009; Yin and Struik, 2009; also see equation 6 in Niinemets and Tenhunen, 1997). The fraction of Iinc absorbed by the leaf photosynthetic pigments (parameter β in Equation A3) is affected by long-term changes in light and temperature through its changes in leaf morphology. Leaves grown at high temperature are generally thinner, with a lower ability to absorb light (Poorter and Evans, 1998; Yamori ), therefore providing a reasonable explanation for the observed κ2LL reduction with increasing temperature. On the other hand, leaves grown at high irradiance are thicker (Niinemets and Tenhunen, 1997), indicating that κ2LL variation is much more complex and still not fully understood. Nonetheless, caution should be exercised when modelling canopy photosynthesis based on the sun/shade approach (de Pury and Farquhar, 1997; Yin and van Laar, 2005) because κ2LL increases with increasing Na (Fig. 7d), while κ2LL also increases with decreasing light (Schultz, 2003; shade leaves which generally have low Na values; Fig. 7b).

The normalized temperature functions of Vcmax and Jmax were variable across crops (Fig. 6), particularly above 30 °C, in line with Leuning (2002). This mean that the assumption used in crop modelling, a unique An response to temperature across crop species, is inappropriate when photosynthesis is calculated by the FvCB model. In the case of no available data, it is suggested that researchers as a first approximation use Vcmax and Jmax temperature parameters from species that belong to the same family (see Fig. 6; cotton and kenaf belong to Malvaceae; sunflower and cynara to Asteraceae).

The Jmax/Vcmax ratio provides an estimate of the relative activities of RuBP regeneration and Rubisco carboxylation, and incorporates both temperature and Na effects. This study confirms (Table 3) the generally reported Jmax/Vcmax value of 2.0±0.5 (Wullschleger, 1993; Poorter and Evans, 1998; Bunce, 2000; Leuning, 2002; Medlyn ). However, this ratio should not be considered constant in absolute terms. Vcmax is dependent on the Rubisco parameters used (up to 11% change; see also Medlyn ) and Jmax is affected by the assumed κ2LL and θ values used (see earlier discussion). For instance, grape showed a much higher Jmax/Vcmax ratio compared with other crops (Fig. 8). Apart from the effect of species, there are two possible artefacts causing this: the different Rubisco parameters used in that study (Schultz, 2003) and the lower grape κ2LL values compared with the other crops (Fig. 7b). Also use of Ci instead of Cc affects this ratio. Thus approaches (e.g. Kosugi ; Müller ) that fix the Jmax/Vcmax ratio at a constant value to parameterize the FvCB model should receive critical reservation.

Seasonal effects on photosynthesis and respiration in cynara

Direct interpretation of the seasonal effects on An and Rn for cynara is difficult because both the climate (Supplementary Fig. S1 at JXB online) and the plant stage are different, with new and old leaves being present (Archontoulis ; Searle ). Rn acclimated to cold and warm environments to a larger extent than did An (Table 3; Fig. 4). This is consistent with previous studies (Yamori ; Ow ; Silim ).

The nature of Rn acclimation is variable within and among plant species, and it is usually related to changes in ERn and/or to changes in Rn25 (Atkin ; Searle ). Given that χR did not change between seasons (P=0.269; Table 3) and that the measured winter leaves had higher Na values than the summer leaves (on average 2.48 versus 1.53 g N m−2; see also Fig. 7d), this indicates that basal capacity, Rn25, plays an important role in this acclimation. Secondly, ERn was also higher during winter periods. Apparently, cynara follows an ‘acclimation type II’ (Atkin ) where the overall elevation of the Rn–temperature response was affected by season and growth stage (Fig. 4).

Among FvCB parameters analysed, seasonal effects were found on two electron transport parameters, χj and κ2LL (Table 3 and earlier discussion), and none related to Vcmax. Literature information on An acclimation is diverse among studies (Wilson ; Medlyn ; Bernacchi ; Hikosaka, 2005; Yamori ; Braune ; Silim ). As far as is known, only Wilson reported both χj and χv seasonal changes in trees, while Braune found only χj variation for barley as in the present study. For cynara, the normalized Vcmax and Jmax temperature functions were slightly changed between seasons, in line with other field studies (Medlyn ; Schultz, 2003), but in contrast to growth chamber studies (Bernacchi ; Yamori ; Ishikawa ; Braune ) where plants were grown only at different temperatures. The fact that this study assessed leaves with different Na status may be a reason, but an inconsistency between actual field and controlled chamber studies is obvious.

The Jmax/Vcmax ratio has been reported to be either sensitive or insensitive to growth temperature (see discussion by Hikosaka ), growth irradiance (Poorter and Evans, 1998; Yamori ), and seasonal changes (Bunce, 2000; Medlyn ). The present results suggest that cynara regulates the balance between RuBP regeneration and Rubisco carboxylation to maintain the Jmax/Vcmax ratio almost constant (change <8%; Table 3) across seasons and growth stages.

Conclusions

This study provides new information on photosynthesis and respiration rates for three bioenergy crops, sunflower, kenaf, and cynara. It provides an alternative way to parameterize the FvCB model from An–Iinc data, instead of using An–Ci data that are more expensive to obtain. It was shown that major FvCB model parameters, Vcmax and Jmax, derived from either An–Ci or An–Iinc analysis, are very close (r2=0.92). Present models can predict photosynthesis under varying levels of Ci, Iinc, temperature, and leaf nitrogen, and can estimate night respiration under varying levels of temperature and leaf nitrogen, for the three bioenergy crops. Comparisons of FvCB model parameters among sunflower, kenaf, cynara, cotton, wheat, barley, tobacco, and grapevine indicated that only a few parameters were conserved. This means that in order to feed crop models properly, species-specific FvCB model parameters are needed. In this context, readily available An–Iinc data—that have been underexplored—can assist in that respect. By combining classical photosynthetic data and the biochemical model, the potential of crop growth models to assess the impact of climate change on crop production and to examine options of bioenergy production under a changing climate is enlarged. Further research is needed to quantify reliably the effects of photosynthetic acclimation and diurnal midday depression identified in this study.

Supplementary data

Supplementary data are available at JXB online.

Average monthly temperatures, radiation, and precipitation at the experimental site (period: 2007–2009). Sunflower measurements were taken from July to August; kenaf measurements from July to September, and cynara measurements from November to June.

Reported relationships between light-saturated net assimilation rate at ambient CO2 concentration and at near-optimum temperature (An,max in μmol CO2 m−2 s−1) and leaf nitrogen content (g N m−2) for C3 crops (a), C3 legume crops and trees (b), and C4 crops (c). (d) An average relationship for C3 and C4 crops.

Reported Nb values (minimum leaf nitrogen for photosynthesis, in g N m−2) for various species.