Will Interannual Climatic Variability Prevent Any CO2-driven Stimulation of Grassland Biomass?

Abstract:

The future productivity of this world depends upon how ecosystems respond to the changing environment. It is generally expected that the increasing concentration of CO₂ in the atmosphere should stimulation biomass production, especially in semi-arid or seasonally dry environments. However, many experiments in grassland systems have shown clearly that the effect of elevated CO₂ (eCO₂) on biomass production varies substantially among years, with the eCO₂ effect varying from strongly positive to strongly negative across years. Recently, we demonstrated that the eCO₂ effect on biomass in one such highly-variable system (the TasFACE experiment) can be explained very accurately by the seasonal distribution of rainfall, with more rainfall in summer leading to eCO₂ stimulation of biomass and more rainfall in spring or autumn reducing the eCO₂ effect. Thus, the overall impact of eCO₂ on net primary productivity in such systems depends upon the seasonal rainfall balance, not the total amount that falls. Here, we use the relationship derived from the TasFACE experiment to simulate biomass production and the eCO₂ effect using both historical and projected future rainfall patterns. We also compare the projections from this simple, experimentally-derived empirical relationship with those from a more complex, mechanistic ecosystem model that accurately predicts current biomass production in this system. Both sets of projections show that both historical and future rainfall variability substantially influence the simulated eCO₂ effect, but the mechanistic model fails to predict the negative effects of eCO₂ on biomass production. It is probable that linkages between rainfall and soil nitrogen availability are not sufficiently captured in the mechanistic model and that this influences the projections of ecosystem productivity. Such effects need to be captured in mechanistic models if we are to improve our ability to predict and model future land-atmosphere feedbacks.