Aim Using a Dynamic Global Vegetation Model (DGVM), we assessed the independent and co-varying effects of temperature, precipitation and atmospheric CO2 on nonlinear (threshold) responses in carbon-based processes, and evaluated whether these underlying process thresholds translate to the ecosystem-scale.
Location Amazon Basin, South America.
Methods The Lund-Potsdam-Jena model (LPJ) was employed to determine responses in net primary production (NPP), heterotrophic respiration (RH), vegetation carbon (CV), soil carbon (CS), and plant functional type (PFT) composition to variations in temperature (± 9 °C relative to the control), precipitation (up to 80% reduction in rainfall relative to the control) and atmospheric CO2 (± 100 p.p.m.v. relative to the control).
Results Our modelling experiments show that increases in temperature result in lower and steeper NPP and RH curves, indicating a thermal threshold at current temperature conditions. Under a combination of temperature and precipitation change, CV responds more to precipitation, while CS closely follows temperature gradients. Ecosystem thresholds, measured in terms of PFT composition stability, are surprisingly few. Simulations indicate an ecosystem threshold occurring at 80% reduction in rainfall; however, due to modelling limitations, this threshold is likely to occur at earlier drought stress conditions. Further empirical research on abiotic stress tolerance levels in tropical ecosystems must be performed in order to refine PFT descriptions used in DGVMs.
Main conclusion In evaluating simulation scenarios that promote major changes in PFT assemblage, we conclude that the ‘natural’ Amazonian rain forest is resilient to environmental change, particularly to decreases in temperature and precipitation. Determining to what extent anthropogenic pressures have altered this resiliency is of utmost importance in predicting the future fate of the Amazon Basin.