Many globally applied dynamic vegetation models used by the IPCC climate assessment process predict widespread ecosystem die-back in the next 100 year as the result of climate change. However, these models are all of minimum functional diversity, in particular with regard to the mechanisms which control tree mortality in extreme heating or drying events, which are the most usual causes of modelled dieback. We used the Ecosystem Demography model, coupled to the JULES land surface scheme and defined ten plant functional for each of seven biome types along a single trade-off surface, from fast growing intolerant plants (high assimilation and stomatal conductance, low wood density and carbohydrate storage) to slow growing, intolerant plants, with the opposite trait properties. The model was globally applied and spun-up for 500 years, before being subjected to climate anomalies for the 21st century derived from the HADCM3 climate model.

Results/Conclusions

Results indicate that a key parameter controlling dieback in a dynamic vegetation model with relatively high biodiversity is the degree to which fast growing trees are able to competitively exclude slow growing trees, as this influences the liklihood of climate resilient vegetation types being present at the end of the spin-up period. The majority of DGVMs are unable to simulate co-existance of multiple vegetation types. therefore conclude that increasing the biodiversity represented DGVMs will not necessarily increase ecosystem resilience to climate change, unless mechanisms controlling co-existence are both understood and simulated in ecosystem models.