The role of non-structural carbohydrates in predictions of ecosystem carbon fluxes

Abstract

Land surface models (LSM) represent a significant source of uncertainty in predictions of future climate. Many LSMs are unable to account for differences between plant carbon assimilation through photosynthesis, and plant carbon expenditure through autotrophic respiration and growth, as they do not comprehensively represent labile non-structural carbohydrate (NSC) stores that allow asynchrony between assimilation and expenditure to occur. This limits the ability of LSMs to accurately capture seasonal and inter-annual variation of ecosystem carbon fluxes in particular during periods of environmental stress. This thesis discusses the current empirical understanding of NSC, and examines previous representations of NSC storage and utilisation within LSMs. A simple model of NSC designed to decouple plant carbon assimilation and expenditure in LSMs and improve predictions of ecosystem carbon fluxes, is presented. The model is tested at three scales and under varying climatic conditions. First, in simulations across the Amazon rainforest, the model decouples respiration and growth from photosynthesis, resulting in shifts in the seasonal cycle of total carbon expenditure. Then at a tropical drought experiment in Caxiuan˜ a, Brazil, the model allows more accurate predictions of carbon fluxes relative to a LSM that does not represent NSC. Finally, at a global scale, the model is used to highlight the potential importance of NSC in predictions of global terrestrial carbon uptake. The thesis concludes by outlining possible developments for future work.