Global decadal variability of plant carbon isotope discrimination and its link to gross primary production

Abstract Carbon isotope discrimination (Δ13C) in C3 woody plants is a key variable for the study of photosynthesis. Yet how Δ13C varies at decadal scales, and across regions, and how it is related to gross primary production (GPP), are still incompletely understood. Here we address these questions by implementing a new Δ13C modelling capability in the land‐surface model JULES incorporating both photorespiratory and mesophyll‐conductance fractionations. We test the ability of four leaf‐internal CO2 concentration models embedded in JULES to reproduce leaf and tree‐ring (TR) carbon isotopic data. We show that all the tested models tend to overestimate average Δ13C values, and to underestimate interannual variability in Δ13C. This is likely because they ignore the effects of soil water stress on stomatal behavior. Variations in post‐photosynthetic isotopic fractionations across species, sites and years, may also partly explain the discrepancies between predicted and TR‐derived Δ13C values. Nonetheless, the “least‐cost” (Prentice) model shows the lowest biases with the isotopic measurements, and lead to improved predictions of canopy‐level carbon and water fluxes. Overall, modelled Δ13C trends vary strongly between regions during the recent (1979–2016) historical period but stay nearly constant when averaged over the globe. Photorespiratory and mesophyll effects modulate the simulated global Δ13C trend by 0.0015 ± 0.005‰ and –0.0006 ± 0.001‰ ppm−1, respectively. These predictions contrast with previous findings based on atmospheric carbon isotope measurements. Predicted Δ13C and GPP tend to be negatively correlated in wet‐humid and cold regions, and in tropical African forests, but positively related elsewhere. The negative correlation between Δ13C and GPP is partly due to the strong dominant influences of temperature on GPP and vapor pressure deficit on Δ13C in those forests. Our results demonstrate that the combined analysis of Δ13C and GPP can help understand the drivers of photosynthesis changes in different climatic regions.

Under RuBisCO limitation, AC is defined as: is the ratio of the effective Michaelis constant (K) to ca (µmol mol -1 ), and 5234 is the maximum rate of carboxylation (µmol m -2 s -1 ). 5234 and 234 depend on the leaf temperature (Tk) in K following an Arrhenius function and a peaked function as in Medlyn et al. (2002): where Ha and Hd are the rates of exponential increase and decrease of the function below and above the optimum, respectively (or activation and deactivation energies, kJ mol -1 ). DS is the entropy factor (J mol -1 K -1 ). The values of Ha, Hd and DS for 5234 and 234 depend on the plant functional type (PFT, see Table S1). R is the universal constant (8.314, J mol -1 K -1 ).
5234 and 234 are then calculated from their respective values at 25ºC (µmol m -2 s -1 ), for each PFT as: The ratio of 234,)> to 5234,)> is set constant and depends on the PFT considered. 5234,)> is calculated as in Medlyn et al. (1999): 3 is the leaf nitrogen content per unit area (kgN m -2 ), and @ (µmol m -2 s -1 ) and @ (µmol gN -1 s -1 ) are the intercept and slope of the relationship between 5234,)> and 3 , respectively, which depend on the PFT considered (Kattge et al., 2009). 3 is calculated as the product of PFT-dependent leaf traits following Harper et al. (2016): 23AA is the top-leaf nitrogen mass per leaf mass (kgN kg -1 ), LMA is the leaf mass per unit area (kg m -2 ) and LAI is leaf area index (m 2 m -2 ).
All the PFT-dependent parameters in JULES are reported in Table S1.
Text S2 Similar influence of and on D 13 C and on the limiting rates of photosynthesis (AC and AJ) Assuming infinite mesophyll conductance (gm), D 13 C can be written as: where a, @ and f are the isotope fractionations due to CO2 diffusion across the stomata (4.4‰), effective RuBisCO carboxylation (28‰) and photorespiration (12‰). and are as defined Assuming finite gm, D 13 C is: where am is the isotope fractionations due to CO2 diffusion across the mesophyll cell (1.8‰), b is equivalent to @ assuming finite gm (30‰) and 5 is the ratio of chloroplastic (cc) to ambient (ca) partial pressure of CO2 (with cc < ci).
The derivatives of D 13 C with and when assuming infinite gm (Eqn S9a) are: The derivatives of D 13 C with , and 5 when assuming finite gm (Eqn S9b) are: Thus, D 13 C increases with rising (or 5 ) but decreases with rising .
The derivatives of log-transformed AJ and AC (Equations 1 and 2) with and are: Thus, AJ and AC increase with rising but decrease with rising .
The derivatives of D 13 C and of log-transformed AJ and AC are always positive with rising and negative with rising , so D 13 C and A are expected to vary in similar directions with changing and .     Only groups with more than 20 grid-points are considered.