Species diversity, habitat use and mobility of bats in a fragmented landscape in central Amazonia, Brazil

As a step towards achieving a better understanding of the distribution of bats species and their response to forest fragmentation in the Brazilian Amazonia, I conducted a study in a fragmented landscape composed by primary forests and forest fragments surrounded by savanna-like vegetation, close to the village of Alter do Chao, in the west of Para State. After 102 night of sampling, I captured 3978 bats representing 70 species, 40 genera and 7 families, representing the highest bat species diversity for a single locality in Brazil. Fifty species were recorded in savannas, 44 in forest sites and 41 in forest fragments. My species list includes new records for Brazil and extends the knowledge of the distribution of others species in hundreds of kilometres.; I used species accumulation curves and species-richness estimators to evaluate my bat sampling and they indicated that 75--100 species may occur in the studied area, suggesting that my inventory recorded approximately 67--89% of the bat fauna there. Using cluster analysis, I compared the bat fauna in Alter do Chao with the faunas from 17 other sites in the Neotropics and observed 65% of similarity with the fauna from Manaus (Brazil), 60% with that from Iwokrama (Guyana) and 57% with that from Paracou (French Guyana).; I performed ordinations using multidimensional scaling (MDS) based both on frequencies of captures (FC) and on presencelabsence of species (PA) to investigate the effects of forest fragmentation on the local bat communities. MDS indicated separation between habitats, both for presence/absence of species and frequency of captures, but this separation was not sharp and savannas were more distinct than forest and fragments. MANOVA indicated significant separation between the bat faunas in the three habitats based on FC and PA (Pillai-Trace = 0; Pillai-Trace = 0.003, respectively). However, when savanna samples were removed from the analysis, there was a significant separation between forests and fragments based on PA (Pillai-Trace = 0.029) but not based on FC (Pillai-Trace = 0.124). There was significant correlation between the number of captures and the shape of the area sampled (Pearson Correlation, p = 0.031), but not with the size of the area sampled (p = 0.118) or the tree density (p = 0.118). My results indicate that the savannas in Alter do Chao presented a distinct bat fauna when compared with forests and forests fragments and there was a difference between the bat fauna in forests and fragments but it was not as evident as the difference between the savannas and those two habitats. There was an erect of forest fragmentation on the local bat fauna, but this effect was not strong or clearly evident. (Abstract shortened by UMI.)...