Environmental responses of gross primary production: emerging knowledge gaps

Accurate simulations of gross primary production (GPP) are vital in our efforts to model the global carbon cycle. The instantaneous controls of leaf-level photosynthesis, which can be studied in manipulative experiments, are well established; but there is no consensus on how canopy-level GPP depends on spatial and temporal variation in the environment. Models make a variety of assumptions when ‘scaling-up’ the standard model of photosynthesis. These assumptions are consequential, leading to large differences in the apparent environmental dependencies of modelled GPP.We have attempted to understand and resolve these inconsistencies using both theoretical analysis of the processes involved in scaling-up from photosynthesis to GPP, and empirical analysis by generalized linear modelling of GPP inferred from eddy-covariance flux measurements. Theoretical analysis has explained why ‘light-use efficiency’ (LUE) models work – and has led to the ‘P model’, a notably parsimonious model that coordinates capacities for CO2 fixation, water- and electron-transport to simulate GPP. For empirical analysis we used eddy-covariance data from over 100 sites worldwide. We combined these flux data with in situ radiation measurements and the MODIS FPAR product. Soil moisture data were estimated using the SPLASH model, with appropriate meteorological inputs, and soil water-holding capacity derived using SoilGrids.In arriving at a preferred statistical model, we showed that daytime air temperature and vapour pressure deficit, and soil moisture content are salient predictors of LUE. Despite taking LUE (GPP normalised for absorbed light) as our response variable, we found that the diffuse fraction of solar radiation has a strong influence on production: second only to VPD in predictive power. That finding challenges the idea, dating back 50 years to studies of crop yield, that time-averaged carbon assimilation is simply proportional to the amount of absorbed light. Our empirical analysis of GPP data has led us to seek ways to improve the performance of the P model without sacrificing its simplicity and transparency. Differential canopy penetration by diffuse and direct radiation is one line of development. Also needed is an improved representation of the temperature dependency of GPP. The empirical analysis suggested a generally increasing (asymptotic) trend over the observed range in growth temperature rather than the temperature optimum of 15°C displayed by the current P model simulations.Our analysis suggests it is feasible to predict GPP using a single model structure, common across vegetation categories. But the goal of a model design that is at once simple, theoretically well-founded and robust continues to generate scientific challenges.