RESUMO
Sustainable biodiversity and land management require a cost-effective means of forecasting landscape response to environmental change. Conventional species-based, regional biodiversity assessments are rarely adequate for policy planning and decision making. We show how new ground and remotely-sensed survey methods can be coordinated to help elucidate and predict relationships between biodiversity, land use and soil properties along complex biophysical gradients that typify many similar landscapes worldwide. In the lower Zambezi valley, Mozambique we used environmental, gradient-directed transects (gradsects) to sample vascular plant species, plant functional types, vegetation structure, soil properties and land-use characteristics. Soil fertility indices were derived using novel multidimensional scaling of soil properties. To facilitate spatial analysis, we applied a probabilistic remote sensing approach, analyzing Landsat 7 satellite imagery to map photosynthetically active and inactive vegetation and bare soil along each gradsect. Despite the relatively low sample number, we found highly significant correlations between single and combined sets of specific plant, soil and remotely sensed variables that permitted testable spatial projections of biodiversity and soil fertility across the regional land-use mosaic. This integrative and rapid approach provides a low-cost, high-return and readily transferable methodology that permits the ready identification of testable biodiversity indicators for adaptive management of biodiversity and potential agricultural productivity.