COS 95-8
Predicting bat reservoirs of zoonotic infectious diseases
Bats have long been known to harbor infectious diseases harmful to humans, but have been increasingly linked with novel emerging infections such as those caused by the SARS coronavirus, Nipah, and Ebola virus. Recent work has postulated that feeding and ranging habits may contribute to high probability of human transmission (e.g., rabies), and that aspects of bat social behavior and physiology may make this group ideally suited to be zoonotic reservoirs. In this analysis, we apply machine learning methods to examine all 1116 species of extant bats and over 50 variables describing their life history, physiology, and ecology to investigate whether there are common traits shared among species that carry the greatest diversity of zoonotic diseases. We also identify which particular species of bats have the highest probability of carrying a zoonotic pathogen.
Results/Conclusions
Results suggest that larger size at birth size and larger size at weaning are the most important predictors of bat species carrying many zoonotic infections. These traits are most common in bats that are long lived and have multiple litters per year. Our results also suggest that environmental factors such as mean AET that are characteristic of highly vegetated habitats are important predictors of zoonotic bat reservoirs. We highlight several bat species that show greater than 70% probability of carrying an undiscovered zoonotic pathogen that should be targeted for surveillance, and identify multiple geographic hotspots where the geographic ranges of these novel reservoir species overlap.