As tropical rainforests are cleared, forest remnants are increasingly isolated within agricultural landscapes. Understanding how forest loss impacts on species diversity can, therefore, contribute to identifying the minimum amount of habitat required for biodiversity maintenance in human-modified landscapes. Here, we evaluate how the amount of forest cover, at the landscape scale, affects patterns of species richness, abundance, key functional traits and common taxonomic families of adult trees in twenty Brazilian Atlantic rainforest landscapes. We found that as forest cover decreases, both tree community richness and abundance decline, without exhibiting a threshold. At the family-level, species richness and abundance of the Myrtaceae and Sapotaceae were also negatively impacted by the percent forest remaining at the landscape scale. For functional traits, we found a reduction in shade-tolerant, animal-dispersed and small-seeded species following a decrease in the amount of forest retained in landscapes. These results suggest that the amount of forest in a landscape is driving non-random losses in phylogenetic and functional tree diversity in Brazil’s remaining Atlantic rainforests. Our study highlights potential restraints on the conservation value of Atlantic rainforest remnants in deforested landscapes in the future.

Habitat loss and the illegal exploitation of natural resources are among the main drivers of species extinction around the world. These disturbances act at different scales, once changes in the landscape composition and configuration operate at large scales and exploitation of natural resources at local scales. Evidence suggests that both scales are capable of triggering genetic erosion in the remaining populations. However, most of the studies so far did not evaluate simultaneously the effects of these disturbances on genetic diversity and structure of plants. In this study, we used a multiple scale approach to empirically evaluate the impacts caused by local and landscape scale disturbances in the genetic diversity and structure of an endangered palm tree, Euterpe edulis. We sampled and genotyped with microsatellite markers 544 juveniles of E. edulis in 17 fragments of Atlantic Forest in Brazil. In addition, we estimated the local logging rate and the forest cover and isolation at landscape scale. We found that the palm populations have not undergone any recent bottleneck events and that only logging intensification had affected the fixation index and the number of private alleles. Additionally, we did not detect any evidence of spatial genetic structure or genetic divergence associated with environmental disturbance variables at different scales. However, we identified distinct genetic clusters, which may indicate a reduction of gene flow between fragments that were previously a continuous habitat. Our results show that local disturbances, which act directly on population size reduction, such as logging, modified the genetic diversity more rapidly, whereas genetic structure is probably more influenced by large-scale modifications. In this way, to maximize the conservation efforts of economically exploited species, we recommend to increase the inspection to reduce the illegal exploitation, and reforestation of degraded areas, in order to increase the gene flow in Atlantic Forest fragments.

Anthropogenic landscape modification may lead to the proliferation of a few species and the loss of many. Here we investigate mechanisms and functional consequences of this winner–loser replacement in six human-modified Amazonian and Atlantic Forest regions in Brazil using a causal inference framework. Combining floristic and functional trait data for 1,207 tree species across 271 forest plots, we find that forest loss consistently caused an increased dominance of low-density woods and small seeds dispersed by endozoochory (winner traits) and the loss of distinctive traits, such as extremely dense woods and large seeds dispersed by synzoochory (loser traits). Effects on leaf traits and maximum tree height were rare or inconsistent. The independent causal effects of landscape configuration were rare, but local degradation remained important in multivariate trait-disturbance relationships and exceeded the effects of forest loss in one Amazonian region. Our findings highlight that tropical forest loss and local degradation drive predictable functional changes to remaining tree assemblages and that certain traits are consistently associated with winners and losers across different regional contexts.