UC Davis

Understanding the disease transmission at the wildlife and livestock interface is critical to better prevent and manage emergent infectious diseases. This dissertation aims to better understand disease transmission risks at the wildlife-livestock interface through direct animal contacts by use of animal relocation data obtained through GPS collars. The objectives are to 1) characterize contacts occurring between white-tail deer, feral pigs and cattle, 2) identify factors that drive new animal contacts and drivers that motivate animals to return to contact partners over time, and 3) model the disease spread process and potential spillover through direct animal contacts using Foot and Mouth Disease (FMD) as an example. A combination of network analysis, separable temporal exponential random graph modeling (STERGM), and generalized epidemic mean-field (GEMF) diffusion modeling techniques were used to achieve these goals.

Chapter one static and dynamic network analyses provided rich descriptive insights of contact frequencies across multiple species over a larger time frame (~3 years) than previous studies. Less than 1% of contacts observed occurred between wildlife and livestock species. Descriptive findings characterized cattle as the most densely, interconnected, and network centric species. White-tailed deer and feral pigs, on the other hand, were more socially fragmented, existing in smaller network subgroups and likely to play a smaller role in facilitating disease transmission across the network compared to cattle. In Chapter two, a STERGM was built focused on identifying drivers of new and persistence of contact formations. Results suggest that neither temperature nor precipitation conditions were predictive of new contact formations, but woody land use (i.e., presence of trees and shrub landscapes) significantly drive new contact formations. When evaluating species-specific odds forming new contacts, feral pigs (OR: 0.15, p < 0.001) and white-tailed deer (OR: 0.073, p < 0.001) both estimated significantly lower odds forming new contacts compared to cattle. Social network topographic attributes were significantly predictive of both, new contact formations and their subsequent persistence. Chapter three took the STERGM built in chapter 2 to simulate three new animal contact networks for better generalizability. A susceptible-exposed-infected-removed (SEIR) model within the GEMF framework allowed to better understand disease dynamics at the wild-livestock interface and could be used to implement risk-based mitigation strategies.