Date of this Version
Snow, N. P., W. F. Porter, and D. M. Williams. 2015. Underreporting of wildlife-vehicle collisions does not hinder predictive models for large ungulates. Biological Conservation 181:44–53
Conflicts from wildlife–vehicle collisions (WVCs) pose serious challenges for managing and conserving large ungulates throughout the world. However, underreporting of large proportions of WVCs (i.e., two-thirds of WVCs in some cases) creates concern for relying on governmental databases to inform management strategies of WVCs. Our objective was to test the sensitivity of WVC studies to underreporting using 2 species of large ungulates that experience substantial incidences of collisions but exist in different environmental settings: white-tailed deer (Odocoileus virginianus) in agricultural-dominated central Illinois and moose (Alces alces) in forest-dominated western Maine, USA. We estimated baseline relationships between the landscape, traffic, and abundance of wildlife on the probabilities of WVCs using the total number of reported WVCs. Then, we simulated underreporting by randomly excluding reports of WVCs and evaluated for relative changes in precision, parameter estimates, and prediction. Point estimates of the relationships between environmental influences and WVCs for both species were reliable until high rates of underreporting occurred (≥70%). When underreporting occurred with spatial bias, shifts in point estimates were detected only for variables that spatially-corresponded with the rate of reporting. Prediction estimates for both species were also reliable until high rates of underreporting occurred (≥75%). These findings suggest that predictive models generate reliable estimates about WVCs with large ungulates unless underreporting is severe; possibly because they occur in non-random patterns (i.e., hotspots) and variability in their environment influences is low. We recommend that concern about underreporting not impede research with existing databases, such as those in this study, for analyzing predictive models and developing management strategies for reducing WVCs.