Estimating Environmental Transmission Risk From Host Movement Data

Environmentally mediated transmission is a critical pathway for disease spread, particularly for pathogens with environmental persistence. However, distinguishing which overlapping host movements result in transmission remains difficult, as most mobility models assume homogeneous exposure and overlook key spatial and behavioral factors. We extend previous models of indirect contact to account for variation in host movement behavior that could affect the probability of indirect transmission and examine these effects on transmission dynamics. We considered four models with different assumptions about the influence of movement behavior on indirect contact and transmission probability. The pathogen decay model includes all overlaps occurring within the pathogen's viability window. The high-use area model restricts transmission to frequently visited locations, assuming greater pathogen deposition where hosts repeatedly congregate. The behavioral model filters contacts based on host behaviors relevant to deposition or acquisition. The integrated model incorporates both spatial and behavioral constraints. We applied these models to GPS telemetry data from wild pigs in Florida and simulated disease dynamics using SEIR models with environmental transmission for Influenza A and Brucella suis, representing pathogens with short and long environmental persistence, respectively. Compared to pathogen decay model, total contact numbers decreased by 0.3% in high-use model, 70.3% in behavioral model, and 70.4% in integrated model under the short-lived condition. Edge density declined from 0.60 to 0.50, transitivity from 0.81 to 0.77, and assortativity from 0.43 to 0.17. R ₀ declined from 2.42 ± 0.47 to 2.06 ± 0.39 (high-use), 1.56 ± 0.42 (behavioral), and 1.35 ± 0.40 (integrated), with a lag of approximately 21 days in peak incidence. The long-lived scenarios, were quantitively similar to those of short-lived scenarios. These results underscore that assumptions about how indirect contact from movement data maps to pathogen transmission significantly influence network configuration and epidemic projections. Our findings highlight the importance of mechanistically defining indirect transmission in disease modeling.