Heterogeneity in social network connections is density-dependent: implications for disease dynamics in a gregarious ungulate

Incorporating host behavioral variation into epidemiological models is important for predicting host-pathogen dynamics. Animals living at high densities or with many strong social connections are predicted to have greater risk of acquiring pathogens. Using social network analysis, we tested the hypothesis that variation in the strength of social connections would influence simulated elk (Cervus canadensis) pathogen dynamics. We quantified fine-scale social connections for captive elk at three experimentally manipulated densities and wild elk at two natural densities. We applied susceptible-infected epidemiological models to networks to infer the relationship between fine-scale host sociality and simulated pathogen dynamics. Networks were filtered based on four association thresholds to determine how variation in the strength of social connections influenced pathogen dynamics. Our simulations suggest that social behavior interacts with population density to predict pathogen dynamics, but this effect was sex-specific. For both males and females at higher density, elk had strong social connections, resulting in higher number of infected individuals. We observed differences in social connections across density, and these results translated to our simulations, which predicted density-dependent pathogen dynamics for captive and wild elk networks. Our results highlight host social behavior as a potential mechanism driving variation in the relationship between population density and pathogen dynamics. Elk are reservoir hosts for numerous emerging infectious diseases, and our models suggest that density-dependent host social behavior could influence pathogen dynamics in elk social networks.