Spatial complexity in fragmenting Amazonian rainforests: Do feedbacks from edge effects push forests towards an ecological threshold?

Deforestation and resulting landscape fragmentation are important concerns in many tropical areas. Deforestation is a complex process with many potential feedback loops, many of which are ignored in models that attempt to interpolate forest loss based on past deforestation rates. In addition, most ecological studies of the impacts of deforestation have focused on landscapes that are already fragmented. These studies ignore the fact that edge effects, such as anthropogenic fire, reach their maximum well before habitat connectivity is lost and may create positive feedbacks that result in further fragmentation. We developed a simple model to explore the potential influence of edge effects on fragmentation rates and used remotely sensed data from the MAP (Madre de Dios, Acre, and Pando) region of the Brazilian Amazon to parameterize the relationships of interest. Under reasonable real-world parameter combinations, edge effects can have a significant impact on deforestation rates, supporting the hypothesis that the true tipping point in a forest to pasture regime shift occurs earlier (i.e., ∼50% forest loss) than analysis of a loss in connectivity would suggest (i.e., ∼60% forest loss). Our results have important implications for understanding deforestation, edge-driven processes, regime shifts, and the management of complex pattern-process relationships.

► We first quantified empirical relationships between forest area, edge, patch size, and connectivity for the MAP region of Amazonia.
► We then developed a simple deforestation model that included a feedback from edge creation to forest loss.
► Comparing models with and without feedbacks suggested that even a small edge effect can greatly increase deforestation rates.
► Edge reaches its maximum well before connectivity is lost; a focus on connectivity obscures the underlying dynamics of forest loss.
► System-wide movement towards a deforestation threshold starts earlier and gains momentum faster than loss of connectivity implies.