RESUMEN
Several viruses from the genus Orthohantavirus are known to cause lethal disease in humans. Sigmodontinae rodents are the main hosts responsible for hantavirus transmission in the tropical forests, savannas, and wetlands of South America. These rodents can shed different hantaviruses, such as the lethal and emerging Araraquara orthohantavirus. Factors that drive variation in host populations may influence hantavirus transmission dynamics within and between populations. Landscape structure, and particularly areas with a predominance of agricultural land and forest remnants, is expected to influence the proportion of hantavirus rodent hosts in the Atlantic Forest rodent community. Here, we tested this using 283 Atlantic Forest rodent capture records and geographically weighted models that allow us to test if predictors vary spatially. We also assessed the correspondence between proportions of hantavirus hosts in rodent communities and a human vulnerability to hantavirus infection index across the entire Atlantic Forest biome. We found that hantavirus host proportions were more positively influenced by landscape diversity than by a particular habitat or agricultural matrix type. Local small mammal diversity also positively influenced known pathogenic hantavirus host proportions, indicating that a plasticity to habitat quality may be more important for these hosts than competition with native forest dwelling species. We found a consistent positive effect of sugarcane and tree plantation on the proportion of rodent hosts, whereas defaunation intensity did not correlate with the proportion of hosts of potentially pathogenic hantavirus genotypes in the community, indicating that non-defaunated areas can also be hotspots for hantavirus disease outbreaks. The spatial match between host hotspots and human disease vulnerability was 17%, while coldspots matched 20%. Overall, we discovered strong spatial and land use change influences on hantavirus hosts at the landscape level across the Atlantic Forest. Our findings suggest disease surveillance must be reinforced in the southern and southeastern regions of the biome where the highest predicted hantavirus host proportion and levels of vulnerability spatially match. Importantly, our analyses suggest there may be more complex rodent community dynamics and interactions with human disease than currently hypothesized.