Tree mortality, climate change and dynamic global vegetation models

Background/Question/Methods

The fate of tropical forests may depend, at least in part, on their response to potential future changes in their rainfall regime.  We present results from an in-depth model-data comparison study examining the ability of four leading terrestrial biosphere models (CLM-DGVM, ED2, IBIS, JULES), to simulate observed reductions in biomass in two throughfall exclusion experiments (TFEs) in the Brazilian Amazon. Models were forced with local meteorological drivers and parameterised with local soil data, so as to represent field conditions as closely as possible. 

Results/Conclusions

Observations show a 20-25% loss of aboveground biomass in the two TFEs over 5-7 years.  The models, however, are generally unable to reproduce this result.  In fact, biomass stocks in most of the models appear highly insensitive to the large (~50%) reductions in rainfall imposed upon them in the simulations.   We decompose the simulated changes in biomass into impacts on growth and mortality and show that mortality processes are particularly poorly represented in the models.  We find that the insensitivity in the biomass response can lead to unrealistic lags relative to the response of other variables to drought.   Background mortality in JULES and IBIS depend on fixed turnover rates, so that any short-term changes in biomass in these models are dominated by reductions in NPP.  Over longer timescales, biomass losses driven by changes in fractional cover may ensue in these models, but the lack of explicit mortality functions precludes their ability to adequately simulate biomass responses to extreme drought events.  In contrast to JULES and IBIS, CLM-DGVM and ED2 both have explicit mortality functions and we explore the relative magnitudes of different modes of mortality in these models under progressively more severe drought scenarios.  We conclude by identifying mortality patterns that DGVMs should ultimately aim to reproduce.  These include the greater vulnerability of large canopy trees to drought relative to smaller trees but also include mortality patterns which may not be related to drought, such as the East-West tree mortality gradient observed in Amazonia and the increases in mortality observed over recent decades in tropical forests.