Humans are transforming the upland landscapes of Amazonia from nearly continuous expanses of high forest to mosaics of pasture, secondary forests, logged forests and primary forest remnants. The large-scale ecological changes that accompany this radical alteration of Earth's biosphere are poorly understood, but probably very important. For example, the best predictions of regional climatic impacts of this transformation suggest that Amazonia will be drier and warmer in the years to come and that climatic shifts may follow in the grain belt of south-central Brazil. Our poor ability to predict the consequences of Amazonian deforestation can be traced to a surprising dearth of comparative information for primary forests and the anthropogenic ecosystems that are replacing them, particularly in the major zone of forest conversion extending from Belem south to Goiania and west to Acre. For example, there are no published studies comparing the hydrology of primary and secondary ecosystems of this 1.2 M km2 region. The only estimates of evapotranspiration and water balance for Amazonian ecosystems based on empirical measurements were conducted on primary forests in central and northern Amazonia, far from the advancing agricultural frontier. It is hypothesized that the deep soils and pronounced seasonal drought that characterize the "arc of forest conversion" support deep-rooted primary forests that use water and mineral nutrients in ways that are not predicted by the existing literature on moist tropical forest ecology. Deep-soil functions of these forests, including carbon storage, dry season water uptake, wet season water storage, fine root production, and nutrient uptake are the least studied aspects of moist tropical forest and yet must be understood to predict the ecological changes that will accompany forest conversion. This research on water use by deep-rooted, Amazon forests and the ecosystems that are replacing these forests will assist in testing the hypothesis above. Results from initial studies suggest that the new ecosystems of Amazonia are relatively shallow-rooted and functionally distinct. This study will provide the basis for development of a model relating seasonal patterns of rainfall to root and canopy dynamics in deep-rooted, moist tropical forests.
