Thu, Aug 18, 2022: 5:00 PM-6:30 PM
ESA Exhibit Hall
Background/Question/Methods: Bats are a very diverse group of mammals known to host many pathogens of increasing domestic animal and human health concern. It is crucial to identify factors that may explain variation in pathogen diversity within this order. Previous studies have identified IUCN status, population structure, longer lifespan, larger group size, and geographic distribution in the eastern hemisphere as factors that drive the richness and diversity of viruses in bats. However, bat roosting ecology is often not included in such models owing to a sparsity of standardized data. This omission could have important consequences for predicting virus spillover risk, particularly as the propensity of bat species to roost in anthropogenic structures (ex. buildings, bridges, homes, and tunnels, etc.) directly relates to human exposure. This study combines novel roosting ecology data manually aggregated from existing literature, host-virus associations extracted from The Global Virome in One Network (VIRION) database, and the COalesced Mammal dataBase of INtrinsic and Extrinsic traits (COMBINE). At the species level, a machine learning approach is used to investigate if anthropogenic roosting (1) is important in predicting viral richness and (2) improves identification of undetected but likely bat reservoir species.
Results/Conclusions: In a preliminary analysis of bats species with known viral associations, 45.5% of species were found to not roost in human-made structures whereas 54.4% were found to roost in human-made structures. These anthropogenic structures included bridges, attics, mines, and overhangs of buildings. Descriptive analyses suggest negligible differences in viral richness between natural and anthropogenic roosting bats when controlling for the number of citations per species, a proxy for surveillance sampling effort. However, relationships between viral richness and roosting status are distinct across bat families, such that members of the Phyllostomidae, Emballonuridae, and Hipposideridae do not show pronounced differences whereas anthropogenic roosting species in the Vespertilionidae and Molossidae have greater virus richness than their natural-roosting counterparts. In subsequent analyses, boosted regression trees will be used to derive the relative importance of bat roosting ecology for classifying viral hosts (and zoonotic virus hosts) as well as how this trait affects model predictions. This work emphasizes the need to characterize and synthesize species traits that relate directly to the wildlife-human interface, such as the use of anthropogenic structures as roosts, where spillover occurs.
Results/Conclusions: In a preliminary analysis of bats species with known viral associations, 45.5% of species were found to not roost in human-made structures whereas 54.4% were found to roost in human-made structures. These anthropogenic structures included bridges, attics, mines, and overhangs of buildings. Descriptive analyses suggest negligible differences in viral richness between natural and anthropogenic roosting bats when controlling for the number of citations per species, a proxy for surveillance sampling effort. However, relationships between viral richness and roosting status are distinct across bat families, such that members of the Phyllostomidae, Emballonuridae, and Hipposideridae do not show pronounced differences whereas anthropogenic roosting species in the Vespertilionidae and Molossidae have greater virus richness than their natural-roosting counterparts. In subsequent analyses, boosted regression trees will be used to derive the relative importance of bat roosting ecology for classifying viral hosts (and zoonotic virus hosts) as well as how this trait affects model predictions. This work emphasizes the need to characterize and synthesize species traits that relate directly to the wildlife-human interface, such as the use of anthropogenic structures as roosts, where spillover occurs.