UCLA

Abstract Recent modeling studies of future vegetation change suggest the potential for large-scale forest die-off in the tropics. Taken together with observational evidence of increasing tree mortality in numerous ecosystem types, there is clearly a need for projections of vegetation change. To that end, the authors have performed an ensemble of climate–vegetation experiments with the National Science Foundation–DOE Community Atmosphere Model (CAM) coupled to the Community Land Model (CAM–CLM-CN) with its dynamic vegetation model enabled (CAM–CLM-CNDV). To overcome the limitations of using a single model, the authors employ the sea surface temperature (SST) warming patterns simulated by eight different models from the Coupled Model Intercomparison Program phase 3 (CMIP3) as boundary conditions. Since the SST warming pattern in part dictates how precipitation may change in the future, in this way a range of future vegetation–climate trajectories can be produced. On an annual average basis, this study’s CAM–CLM-CN simulations do not produce as large a spread in projected precipitation as the original CMIP3 archive. These differences are due to the tendency of CAM–CLM-CN to increase tropical precipitation under a global warming scenario, although this response is modulated by the SST warming patterns imposed. However, the CAM–CLM-CN simulations reproduce the enhanced dry season in the tropics simulated by CMIP3. These simulations show longer fire seasons and increases in fractional area burned. In one ensemble member, extreme droughts over tropical South America lead to fires that remove vegetation cover in the eastern Amazon, suggesting that large-scale die-offs are an unlikely but still possible event.