Forest cover–rainfall relationships in a biodiversity hotspot: the atlantic forest of brazil

It is now generally accepted that the relationship between vegetation and climate is dynamic: vegetation is influenced by climate, but feedbacks between terrestrial ecosystems and the atmosphere mean that vegetation also affects climate. From this it follows that land‐use changes may have climatic consequences. Specifically, it is widely believed that forest clearance may inhibit rainfall. Although models often support this view, this is not universally the case, and empirical evidence is scarce. We have compiled a database of forest cover and precipitation for the state of São Paulo, which lies within the diverse and highly endangered Atlantic forest region of Brazil. We do not find a strong relationship between forest cover and total rainfall, which appears to be influenced primarily by factors such as distance to the coast; but significant positive relationships between tree cover and the number of rain days consistently emerge. The degree of forest fragmentation seems to influence this relationship, with patchier forests associated with fewer rain days; and tree cover also predicts interannual variability in rainfall. The directionality of these relationships is inferred by considering only sites which are known to have been forested in the recent past. Given that across São Paulo state there is no overall trend for a reduction in rainfall over the study period (1962–1992), it would appear that important local‐scale processes may be hidden by regional trends. We discuss these results with reference to conservation strategies in fragmented landscapes. For instance, the spatial scale over which tree cover–precipitation relationships occur appears to be somewhat larger than that over which studies of tropical forest fragmentation typically operate, but similar to the area of habitat deemed necessary to maintain viable populations of several Atlantic forest species.