Fire and edge effects in a fragmented tropical forest landscape in the southwestern Amazon

Abstract Amazonian forests are subject to both anthropogenic and climate impacts despite the recent decline of annual deforestation rates. Forests in the southwestern Amazon have been intensively fragmented due to deforestation and recently intense drought events have caused extensive forest fires in this region. However, the impacts of fires and fragmentation on forests and forest responses to disturbances in the southwestern region are still poorly studied. We examined impacts of forest fire and fragmentation on forests in a highly fragmented landscape in Acre state to characterize post-fire forest recovery from wildfires, in 2005 and 2010, and analyze edge effects on forest aboveground biomass (AGB) and stem density. Forest inventories were conducted in five areas that included forests burned in 2005 and 2010 and unburned forests. Thirty-two 500 m transects were established with diameter at breast height (DBH) of stems and species identifications collected for subplots every 100m from forest edge to forest interiors. AGB and stem density in different DBH classes were estimated for trees, with pioneer species (Cecropia sp.) used as a disturbance indicator in our analysis. We used generalized liner mixed models to examine the importance of forest fragmentation variables including edge age, edge distance, forest cover and edge density, in addition to elevation and slope as environmental variables. Compared with unburned forests, the impact of forest fire on AGB and stem density is still significant four years after the event. However, rapid recovery within nine years after a fire event made AGB levels statistically indistinguishable from unburned areas. Pioneer species, AGB and densities of small stems of Cecropia sp. in recently burned forests and larger stems in the 2005 burned forest plots were significantly higher than unburned forests. As for forest fragmentation, high spatial variability of AGB and stem density was observed across the studied sites and limited edge effects were observed on AGB and tree density. Among forest fragmentation variables, edge age was the most important predictor of AGB and structural changes in forests followed by edge density. Although AGB changes were poorly correlated with edge distance, relative changes of forest AGB at different edge distances to AGB at 500m, the deepest edge distance, indicates a substantial reduction of AGB near forest edges.