Extending a first-principles primary production model to predict wheat yields

Climate exerts a major influence on crop development and yield. Despite extensive modelling efforts, there is still considerable uncertainty about the consequences of a changing climate for the yields of major crops. Existing crop models are complex and rely on many assumptions and parameters, motivating a quest for more parsimonious models with stronger theoretical and empirical foundations. This paper presents a prototype of such a model for wheat, informed by measurements of gross primary production (GPP), biomass and yield at research sites across the wheat-growing regions of China. First, GPP was predicted using a recently developed first-principles model driven only by climate, carbon dioxide (CO2) concentration, and light absorbed by leaves. Modelled GPP was shown to agree well with eddy-covariance measurements. Second, the data were used to show that above-ground biomass (AB) is proportional to time-integrated GPP, and that grain yield shows a saturating relationship with AB. Simple empirical equations based on these findings were combined with modelled GPP to predict yield, including propagated errors due to parameter uncertainty in both the GPP model and the empirical equations. The resulting 'hybrid' model, applied in a variety of climates, successfully predicted measured interannual variations in AB and yield. Third, the model was extended to include a phenology scheme, a massbalance equation relating mean leaf area index to accumulated GPP over the growth phase, and an independently observed response of leaf mass-per-area to CO2. Sensitivity analyses and scenario runs with this extended model showed a positive but saturating (at similar to 600 ppm) response of yield to rising CO2, consistent with experimental evidence. This positive effect was partially counteracted by a net negative response of yield to increasing temperature, caused by increasing photorespiration and an accelerated growth cycle.