O papel relativo da configuração da paisagem, fatores naturais e manejo da terra na estrutura e diversidade de florestas secundárias no leste da Amazônia

Secondary forests cover about 23% of deforested areas of the Brazilian Amazon. Although they have clear differences with primary forests, forests in succession have undeniably ecological, social and economic importance. The eastern region of the Amazon, specifically the area known as arc of deforestation, presents a fragmented landscape, characterized by several land uses and large extent of secondary forests. This study aimed to evaluate the relative role of landscape variables, natural variables and use and soil management on the structure and floristic diversity of secondary forests in the southeast of Pará. We surveyed 20 secondary forests along an age gradient from 5 to 20 years. In each fragment, were established: i) transects of 10 x 250m (0.25 ha) to study the plants with diameter at breast height (DBH) ≥ 10cm (DBH ≥10cm) in plots of 10 x 10m (total 25 ) ii) 5 subplots of 5 x 20m nested within the transects to study the plants (DBH) <10cm and ≥2cm. (DBH <10cm). Structural variables were measured (plant density, height, basal area, number of stems) and species diversity variables (richness, and Shannon and Simpson diversity index) considering trees, palm trees and lianas. A set of landscape configuration variables for each fragment were considered (eg. Size, perimeter, distance and proportion of primary forest and total forest in a radius of 500 m, 1 km and 3 km); natural variables (altitude and slope) and the historical use of each transect: land use intensity index (LUI) and number of agricultural cycles. The history of land use was assessed by a combination of satellite images time series and interviews with landowners. We undertook analyses with RandomForest package in the R software environment. Structural and diversity parameters had great variation among the 20 secondary forests studied. Variables such as height and basal area have not grown consistently over the age gradient. The percentage of variation of the response variable explained by the models ranged from 0 to 38.75%. In general, all ...