A behavioral model of landscape change in the Amazon basin: The colonist case

Abstract

This paper presents the prototype of a predictive model capable of describing both magnitudes of deforestation and its spatial articulation into patterns of forest fragmentation. In a departure from other landscape models, it establishes an explicit behavioral foundation for algorithm development, predicated on notions of the peasant economy and on household production theory. It takes a "bottom-up" approach, generating the process of land-cover change occurring at lot level together with the geography of a transportation system to describe regional landscape change. In other words, it translates the decentralized decisions of individual households into a collective, spatial impact. In so doing, the model unites the richness of survey research on farm households with the analytical rigor of spatial analysis enabled by geographic information systems (GIS). The paper describes earlier efforts at spatial modeling, provides a critique of the so-called spatially explicit model, and elaborates a behavioral foundation by considering farm practices of colonists in the Amazon basin. It then uses insight from the behavioral statement to motivate a GIS-based model architecture. The model is implemented for a long-standing colonization frontier in the eastern sector of the basin, along the Trans-Amazon Highway in the State of Pará, Brazil. Results are subjected to both sensitivity analysis and error assessment, and suggestions are made about how the model could be improved.