Forest loss and isolation impacts on mutualistic bat-plant interaction networks in a Brazilian Cerrado

Abstract

In recent decades, human land use has led tropical forests to fragmentation and isolation, disrupting biological communities and changing species interactions, like the plant-animal interactions with implication for pollination and seed-dispersal. Altered landscapes affect environmental conditions and resource quality and availability, impacting the roles of nectarivorous and frugivorous animals in sustaining ecosystem functions. Against this backdrop, my research investigates two interconnected questions: (1) how forest loss and isolation of fragments from large continuous forests influence the abundance and richness of interacting bat and plant species, as well as the structure of the local bat-plant interaction networks; and (2) how the modularity of the regional metanetwork correlates with bat body mass, forest cover, and isolation from large continuous forest. I used a seven-year dataset of interacting bats and plants (pollen and seeds surveyed from the bats’ faeces) across 20 sites at variable distances to the Serra da Bodoquena National Park, a continuous deciduous-semideciduous forest in southwestern Brazil. Forest cover was measured at two spatial scales (1 and 2.5 km radius buffers around each site), and isolation was calculated as the linear distance between each site and the nearest border of the national park. The results revealed that forest loss and increased isolation significantly affect bat and plant species richness and abundance, as well as bat-plant network connectance, modularity, and nestedness. Along continuous gradients of forest cover loss and isolation in the landscapes, bat and plant species richness and abundance followed a hump-shaped pattern, peaking at intermediate values (around 50% forest cover). Local network modularity was highest and connectance lowest in landscapes with approximately 30% forest cover. In addition, the modularity of the Serra da Bodoquena metanetwork varied along environmental gradients, suggesting that bat body mass was the strongest predictor of modular structure, followed by forest cover and isolation. In conclusion, local bat-plant networks tended to be smaller, more connected, and less modular in sites embedded in highly deforested and isolated, while the emergence of modules of the metanetwork is greatly associated with bat size and forest loss and isolation. Overall results emphasize that conservation of interacting bats and plants should prioritize the maintenance of at least 30 to 50% forest cover in areas designed for land use, along with the preservation of large continuous forests in the region.