The effects of habitat quality and matrix quality on tropical mammal biodiversity in human-modified landscapes

Background/Question/Methods

There is a vast body of research dedicated to understanding how the amount of habitat in fragmented landscapes determines biodiversity rates. Yet, there is limited understanding of how the quality of the habitat and the matrix (surrounding landcover not considered primary habitat) affects biodiversity, especially for tropical forest-dependent mammals. What are the interacting effects of habitat quality and matrix quality, and how do they influence the mammalian biodiversity in fragmented landscapes? To address this knowledge gap, we surveyed the biodiversity of forest-dependent mammals in 16 fragmented human-modified landscapes in Brazil within a wide gradient of habitat and matrix quality. We identified habitat as patches of forest and savanna using land-use/landcover maps from Mapbiomas. We then measured quality using remote sensing proxies for ecological processes and resources, such as vegetation productivity, vegetation height, and water availability. With these variables, we created a quality index to scale the relative habitat and matrix quality of each landscape. Finally, we deployed camera traps in each landscape for at least 45 days (1080 camera trap hours) from November 2021 to January 2022 to record the species. For analysis, we focused on terrestrial medium-large mammals ( >1kg) and removed domesticated and invasive species from the dataset.

Results/Conclusions

Qualitative interpretation of the preliminary data suggests landscapes with low habitat and low matrix quality are more likely to have greater species richness, contrary to our hypothesis. Landscapes with a high percentage of soy or mosaic (mix of agriculture and pasture) in the matrix also showed greater species richness compared to pasture dominated landscapes. This suggests pasture is the least favorable land-use in this region for mammal species, unless it is within a mosaic. To understand which landscape variables affect the species richness at each site, we fit the data to generalized linear models with a Poisson distribution for alpha diversity counts. Preliminary results suggest shorter vegetation height of the habitat (F=-0.295, p=0.008) is associated with higher species richness at a 1km radius landscape scale (12.56km2). We will require additional species data and further analysis of landscape variables to understand how habitat and matrix quality interact to affect species richness and other valuable measures of biodiversity, such as functional diversity and abundance. Our results show there is still much to understand about how landscape changes affect mammal biodiversity in fragmented human-modified landscapes. Future studies should also consider temporal trends, such as seasonal changes in the environment and historical land-use change.