Regional effects of deforestation on soil biogeochemistry in the southwestern Amazon (Brazil)

Ecosystem changes due to deforestation in the Amazon have received much attention because of potential regional and global impacts to biogeochemical cycles, particularly related to increases in greenhouse gas emissions. Soil is an integral part of all terrestrial ecosystems, and well-constrained field studies have shown significant changes in soil properties (e.g. pH, exchangeable base cations, organic carbon content) following forest clearing. However, most field investigations have documented changes over very small areas, and use coarse methods of extrapolation to understand the implications for ecosystem changes over large areas. This dissertation presents regional quantitative soil analyses over Rondônia (240,000 km2), Brazil, conducted using a large soil database (3000 profiles) collected by the state for agricultural zoning purposes.; Non-spatial analysis revealed that the temporal soil response to land-cover change was not consistent over this large, geographically diverse area, and suggested different processes may control patterns in soil properties over broader scales. Geostatistical methods were applied to both quantitatively characterize individual soil properties, as is often preferred for input to ecological and hydrological models, and to provide a regional context in which to interpret the process-level studies produced by field investigations. The geostatistical approach captured soil property variation much more effectively than traditional classified soil maps, and better represented the continuously varying nature of most properties. Chapters 3 and 4 illustrate the usefulness of these methods for clarifying patterns at resolutions more detailed than the original sampling design, due to the incorporation of covariance models. Spatial relationships among 11 soil properties were quantified to define common scales at which correlations among the properties changed in magnitude or direction. Four major scales were determined which best represent significant changes in integrated soil biogeochemistry: <3 km, <10 km, <68 km, and >68 km. Land-cover change was a significant predictor at every spatial scale, however. Without a clear understanding of the environmental context in which changes due to land-cover change are occurring, tightly controlled studies field studies may not contribute to understanding regional impacts.