Predictors of Deforestation in the Brazilian Amazon

摘要: Aim and Location We assessed the effects of biophysical anthropogenic predictors on deforestation in Brazilian Amazonia. This region has world's highest absolute rates forest destruction fragmentation. Methods Using a GIS, spatial data coverages were developed for three types potential predictors: (1) human-demographic factors (rural-population density, urban-population size); (2) that affect physical accessibility to forests (linear distances nearest paved highway, unpaved road navigable river), (3) may land-use suitability human occupation agriculture (annual rainfall, dry-season severity, soil fertility, waterlogging, depth). To reduce autocorrelation among variables, basin was subdivided into >1900 quadrats 50 × 50 km, random subset 120 selected stratified intensity. A robust ordination analysis (non-metric multidimensional scaling) then used identify key orthogonal gradients ten original predictor variables. Results The revealed two major environmental study area. Axis 1 discriminated areas with relatively dense populations highways, sparse no highways; whereas axis 2 described gradient between wet sites having low many rivers few roads, those opposite values. multiple regression both highly significant predictors, collectively explaining nearly 60% total variation intensity (F2,117=85.46, P < 0.0001). Simple correlations variables concordant model suggested highway density rural-population size most important correlates deforestation. Mainconclusions These trends suggest Amazon is being largely determined by proximate factors: population highways all which increase deforestation. At least at scale this analysis, fertility waterlogging had little influence activity, depth only marginally significant. Our findings current policy initiatives designed immigration dramatically expand infrastructure networks are likely have impacts activity. Deforestation will be greatest seasonal, south-easterly basin, accessible centres where large-scale cattle ranching slash-and-burn farming easily implemented.