Vertebrate community disassembly in human-induced Amazonian forest islands

The extent to which land-use change impairs tropical biodiversity is controversial. This is because while some species may thrive in human-modified landscapes, others are locally extirpated – the so-called “few winners and many losers” paradigm. Furthermore, reliable environmental impact assessments imply that the correct drivers of biodiversity change are recognised, and suitable reference conditions are available. Herein, I examine vertebrate responses to habitat fragmentation induced by two mega hydroelectric reservoirs in Brazilian Amazonia. In the Tucuruí Hydroelectric Reservoir, bird point count surveys were carried out on 36 forest islands in 2006 and 2007. In the Balbina Hydroelectric Reservoir, I carried out bird mist-net surveys on 33 forest islands and five continuous forest sites in 2015 and 2016; I also used autonomous recording units to survey anuran species on 74 forest islands and four continuous forest sites in 2015. At Tucuruí, bird species traits associated with vulnerability to forest fragmentation included rarity and forest dependency. At Balbina, species richness of understorey insectivorous birds was more affected by island size than the amount of habitat surrounding mist-net lines, indicating that a patch-centric approach is most appropriate to measure species responses to habitat loss if species exhibit low dispersal ability and the surrounding habitat matrix is hostile. Regarding environmental impact assessments, the higher the intactness of the reference condition, the greatest the estimated impact of habitat fragmentation on bird species richness. In other words, studies of biodiversity responses to land-use change will likely yield over-optimistic results if they are masked by the low conservation status of either degraded or insufficiently large habitat patches regarded as the reference condition. Finally, island size per se played a decisive role in explaining anuran species richness on Amazonian forest islands, yet the inferential power of island species-area relationships is only reliable if derived from an appropriate study design