Declining global sensitivity of stomatal conductance to photosynthesis

Abstract

Terrestrial ecosystems play a pivotal role in mitigating climate change through photosynthesis and transpiration regulated by plant stomata. Terrestrial biosphere models (TBMs) commonly couple a biochemical photosynthesis model, i.e., the Farquhar model, and a semi-empirical stomatal conductance model, i.e., the Ball-Berry model, to estimate gross primary productivity (GPP) and evapotranspiration (ET). However, the key ecosystem parameters (i.e., maximum carboxylation rate Vcmax and the Ball-Berry slope m) are commonly prescribed as constant in TBMs, hindering accurate quantification of global GPP and ET. m depicts ‘compromise between the costs and benefits of stomatal conductance relative to the photosynthetic activity of the leaf’, namely the sensitivity of stomatal conductance to photosynthesis. However, although recent work has retrieved continuous seasonal V_cmax over the globe, there is a lack of an effective method to derive spatially and temporally explicit m in TBMs. In this study, we estimated monthly m through a Bayesian parameter optimization approach at 136 eddy covariance flux sites (809 site years) and upscaled site-level m to global gridded m during 2001-2020 via a machine learning model. We found a significant declining trend of global m (R2 = 0.72, P < 0.001) in 2001-2020. We found the largest decline of m in grassland (0.04 per year), moderate declines in deciduous broadleaf forests, mixed forests, and cropland (0.02 per year), the smallest decline in evergreen needleleaf forests and savannas (0.01 per year). We found the air temperature plays the predominant role in driving the declining trend of m. This study generates and analyzes a global timeseries of m by ‘handshaking’ the measured fluxes, terrestrial biosphere models, and remote sensing. Our results highlight the changing relationship between photosynthesis and stomatal conductance under climate change and identify the importance of incorporating m as a dynamic parameter in future GPP and ET simulations with terrestrial biosphere models.