Simulating the response of land-cover changes to road paving and governance along a major Amazon highway: the Santarem-Cuiaba corridor

摘要

The spatial distribution of human activities in forest frontier regions is strongly influenced by transportation infrastructure. With the planned paving of 6000 km of highway in the Amazon Basin, agricultural frontier expansion will follow, triggering potentially large changes in the location and rate of deforestation. We developed a land-cover change simulation model that is responsive to road paving and policy intervention scenarios for the BR-163 highway in central Amazonia. This corridor links the cities of Cuiaba, in central Brazil, and Santarem, on the southern margin of the Amazon River. It connects important soybean production regions and burgeoning population centers in Mato Grosso State with the international port of Santarem, but 1000 km of this road are still not paved. It is within this context that the Brazilian government has prioritized the paving of this road to turn it into a major soybean exportation facility. The model assesses the impacts of this road paving within four scenarios: two population scenarios (high and moderate growth) and two policy intervention scenarios. In the 'business-as-usual' policy scenario, the responses of deforestation and land abandonment to road paving are estimated based on historical rates of Amazon regions that had a major road paved. In the 'governance' scenario, several plausible improvements in the enforcement of environmental regulations, support for sustainable land-use systems, and local institutional capacity are invoked to modify the historical rates. Model inputs include data collected during expeditions and through participatory mapping exercises conducted with agents from four major frontier types along the road. The model has two components. A scenario-generating submodel is coupled to a landscape dynamics simulator, 'DINAMICA', which spatially allocates the land-cover transitions using a GIS database. The model was run for 30 years, divided into annual time steps. It predicted more than twice as much deforestation along the corridor in business-as-usual vs. governance scenarios. The model demonstrates how field data gathered along a 1000 km corridor can be used to develop plausible scenarios of future land-cover change trajectories that are relevant to both global change science and the decision-making process of governments and civil society in an important rainforest region.