Determinants of the spatial distribution of cultivated land in the North Argentine Dry Chaco in a multi-decadal study

Deforestation in the Northern Argentine Dry Chaco (NADC) has been mainly driven by soybean expansion and, more recently, by the expansion of implanted pastures. In areas with fast land use transformations, it is important to identify variables that determine the spatial distribution of land use change. The kinds of exploratory analyses that do so contribute to understanding the logic of deforestation agents and to identifying more probable sectors for land use change. We produced maps of cultivated land in NADC for different years (1972, 1991, 1997, 2002, 2007). Based on these maps, we evaluated the importance of environmental and accessibility variables over the spatial distribution of cultivated land in NADC using multiple and simple logistic regressions. Environmental variables (soil suitability for agriculture, rainfall, topography) and accessibility variables (distance to main roads, distance to main towns) were used to fit logistic regressions to the occurrence of cultivated land at different years as dependent variable. Goodness-of-fit was evaluated by the Relative Operating Characteristic (ROC) and Pseudo R2 indices. Results indicate that the main variables explaining the spatial distribution of cultivated land in 2007 are distance to main towns (ROC = 0.76; Pseudo R2 0.12) and soil suitability (ROC = 0.72; Pseudo R2 0.11). The capacity of environmental and accessibility variables to explain the spatial distribution of cultivated land decreased through time (ROC in 1972 = 0.91 and ROC in 2007 = 0.77). Results also suggest that rainfall has not been a major restriction to cropland expansion, and that the main limitations are imposed by infrastructure and services provided from main towns. A decreasing goodness-Of-fit over time suggests that initial limitations have been overcome by cropland expansion and the consolidation of productive areas. Based on these results, we suggest that cropland expansion may generate positive feedbacks in infrastructure and services (i.e. agglomeration economies) that could explain why initial limiting factors related with distance to roads and towns have been gradually overcome. (C) 2015 Elsevier Ltd. All rights reserved.