The macroecology of viral coinfection

Abstract

Coinfection is common in wild animals, and can profoundly influence disease outcomes and transmission. However, most coinfection research is based on laboratory experiments or a few well-studied wildlife systems. Here, we use data from the PREDICT project - the largest standardized wildlife disease surveillance project ever conducted - to describe patterns of viral coinfection and evaluate their association with host and virus factors. Within the viruses prioritized for testing, we find that coinfection is rare (detected in just 223 of 65,662 animals), but still more common than expected by chance - especially coinfections of coronaviruses, paramyxoviruses, and influenza A virus. We further find that coinfection is associated with host age, but bats and rodents exhibit opposing relationships. We find that captive wild animals (specifically rats and mallards) exhibit higher coinfection than free-ranging wild animals, highlighting potential risks to wildlife and human health from the wildlife trade. Finally, we find that biases in sampling and testing can shape observed patterns of coinfection. Our results characterize associations among viruses in the largest relevant dataset currently available. Factors associated with coinfection in this dataset such as host-level variation, virus-virus interactions, and human interference should be further tested via surveillance and laboratory approaches to better resolve the role of coinfection as an important and likely overlooked driver of viral dynamics in nature.