Accounting for imperfect detection reveals role of host traits in structuring viral diversity of a wild bat community

Abstract

<jats:title>Abstract</jats:title><jats:p>Understanding how multi-scale host heterogeneity affects viral community assembly can illuminate ecological drivers of infection and host-switching. Yet, such studies are hindered by imperfect viral detection. To address this issue, we used a community occupancy model – refashioned for the hierarchical nature of molecular-detection methods – to account for failed detection when examining how individual-level host traits affect herpesvirus richness in eight species of wild bats. We then used model predictions to examine the role of host sex and species identity on viral diversity at the levels of host individual, population, and community. Results demonstrate that cPCR and viral sequencing failed to perfectly detect viral presence. Nevertheless, model estimates correcting for imperfect detection show that reproductively active bats, especially reproductively active females, have significantly higher viral richness, and host sex and species identity interact to affect viral richness. Further, host sex significantly affects viral turnover across host populations, as females host more heterogeneous viral communities than do males. Results suggest models of viral ecology benefit from integration of multi-scale host factors, with implications for bat conservation and epidemiology. Furthermore, by accounting for imperfect detection in laboratory assays, we demonstrate how statistical models developed for other purposes hold promising possibilities for molecular and epidemiological applications.</jats:p>

Mammal virus diversity estimates are unstable due to accelerating discovery effort

“…Multiple studies have found that correcting for undersampling undermines widespread assumptions about zoonotic risk [13,14], and we suggest that future studies should similarly attempt to reject the null hypothesis that downstream patterns of zoonotic risk are a neutral consequence of total observed viral diversity. Given that present-day data are a tiny subset of the latent 'true' host-virus network, there will also be value in employing network-or measurement error-based methods that explicitly account for observation biases in analyses [25]. Overall, because current patterns of host-level viral richness represent an unstable and biased snapshot of the mammal virome, we suggest that inference from host-virus association data needs to be carefully qualified and may not by itself be a comprehensive foundation for setting future agendas on viral zoonosis research or One Health policy.…”

Mammal virus diversity estimates are unstable due to accelerating discovery effort

“…Multiple studies have found that correcting for undersampling undermines widespread assumptions about zoonotic risk (13,14), and we suggest that future studies should similarly attempt to reject the null hypothesis that downstream patterns of zoonotic risk are a neutral consequence of total observed viral diversity. Given that present-day data are a tiny observed subset of the latent "true" host-virus network, there will also likely be value in employing network-or measurement error-based methods that explicitly account for observation biases in analyses of the wildlife virome (24). Overall, since current patterns of host-level viral richness represent an unstable and biased snapshot of the mammal virome, we suggest that inference from host-virus association data needs to be carefully qualified, and may not be a comprehensive foundation for setting future agendas on viral zoonosis research or One Health policy.…”

SARS-CoV-2: Cross-scale Insights from Ecology and Evolution

“…State-space models account for imperfect observations in time series data by separating the dynamics of the biological process (e.g., infection dynamics) from noise or bias in the observation process (e.g., false negatives) [91]. Extensions of these two methods can incorporate multiple infection states [93], estimate transmission and recovery rates [86], and include multiple host or virus species [94]. entities (e.g., cell, tissue, organ, person, or population) that are classified by their infection state: susceptible (S, green), infectious (I, purple), and recovered (R, blue).…”