Dynamic global vegetation models (DGVMs) can disentangle causes and effects in the control of vegetation and fire. We used a DGVM to analyse climate, CO2 and fire influences on biome distribution and net primary production (NPP) in last glacial maximum (LGM) and pre-industrial (PI) times. The Land surface Processes and eXchanges (LPX) DGVM was run in a factorial design with fire 'off' or 'on', CO2 at LGM (185 ppm) or PI (280 ppm) concentrations, and LGM (modelled) or recent climates. Results were analysed by Stein-Alpert decomposition to separate primary effects from synergies. Fire removal causes forests to expand and global NPP to increase slightly. Low CO2 greatly reduces forest area (dramatically in a PI climate; realistically under an LGM climate) and global NPP. NPP under an LGM climate was reduced by a quarter as a result of low CO2 . The reduction in global NPP was smaller at low temperatures, but greater in the presence of fire. Global NPP is controlled by climate and CO2 directly through photosynthesis, but also through biome distribution, which is strongly influenced by fire. Future vegetation simulations will need to consider the coupled responses of vegetation and fire to CO2 and climate.
Keywords: CO2 fertilization; biomass burning; climate change; fire; fire-climate-CO2 interactions; fuel; primary production; vegetation distribution.
© 2015 The Authors. New Phytologist © 2015 New Phytologist Trust.