The savanna ecosystems are characterized by a coexistence of trees and grasses. In this biome, tree abundance strongly influences ecosystem dynamics and exhibits large spatial variability. Tree abundance in tropical savannas is known to be strongly limited by annual rainfall input. However, it is still unclear how intra-annual rainfall variability (e.g. rain intensity and seasonality) constrains tree abundance. This knowledge gap may hamper future predictive ability for tropical savannas, especially considering intra-annual rainfall variability is projected to change under global warming. Here, we first examined the relation between tree abundance and intra-annual rainfall variability at global scale by combining the multi-year average tree cover from MODIS and daily rainfall from TRMM. We calculated the potential tree cover (without disturbance) at a given mean annual rainfall (MAR) and further investigated its dependence on rainfall intensity and rainy season length. Second, we tested the hypothesis that the observed relation is a result of the differentiated water use strategies between trees and grasses. To do this, we coupled a biophysical tree-grass competition model and a stochastic rainfall generator. After calibrated with field observations, we conducted numerical experiments under various rainfall variability scenarios with and without tree-grass competitions and analyzed the modelled tree abundance.
Results/Conclusions
(1) We find that potential tree cover at given MAR increases with rain season length and decreases with rainfall intensity globally. At continental scale, the effect of rain season length is weaker in American savannas compared with Australian and African savannas. The effect of rainfall intensity is largest in Australian savannas. However, the magnitude of the effects must be interpreted cautiously as rainfall intensity and length are not independent and are negatively correlated, especially in Australia.
(2) Numerical simulations show that the observed negative relation between maximum tree abundance and rainfall intensity can be explained only when tree-grass competition is included. More specifically, differentiated water transport and water use strategies between trees and grasses are essential to reproduce the observed pattern.
(3) Our results emphasize the importance of vegetation physiology and tree-grass competition in determining the responses of tree abundance to climate variations in tropical savannas and suggest that more extreme rainfall patterns in the future may lead to an increase in grass in tropical savannas.