COS 114-3
Major issues in wildlife mortality-based and connectivity model-based prioritization of transportation mitigation

Thursday, August 14, 2014: 2:10 PM
Regency Blrm C, Hyatt Regency Hotel
Fraser Shilling, Department of Environmental Science and Policy, University of California, Davis, Davis, CA
David Waetjen, Environmental Science & Policy, University of California, Davis, CA

Wildlife-vehicle collisions (WVC) are a source of conflict between traffic and natural systems and a major source of mortality for certain animal species. State wildlife agencies and Departments of Transportation (DOTs) are increasingly looking for solutions to improve connectivity by prioritizing locations along roadways for mitigation actions. In some states, agencies and conservation scientists have treated connectivity models in GIS as a source of prioritization information, but outputs from these models have not been appropriately treated and tested as hypotheses . Other states use WVC information to locate sites for mitigation. Many DOTs collect data about locations of collisions with animals. These data are seldom freely available, but could be useful for understanding and resolving collisions with deer, a safety concern in most states, and other wildlife.

We developed a high-resolution connectivity model that calculates least-cost paths among randomly placed points (representing moving animals). We tested whether this model and the current “linkage” model for California could predict wildlife movement and conflict with traffic. We developed state-scale, crowd-source WVC  reporting systems and tested whether there were differences between priority mitigation locations  found using these systems (n>30,000) and priorities identified using DOT-collected data (n>20,000) along >2,000 km of state highway.


We found that existing and novel connectivity model outputs failed to consistently predict locations of WVC (Getis-Ord cluster analysis). We propose that conservation scientists and state agencies should treat “linkage” and “corridor” model outputs as hypothetical until proven otherwise using wildlife-occupancy or movement observations. We demonstrate that DOT data vary widely in their accessibility, from inaccessible (CA) to accessible online (ID), which may reflect varying concerns about personal-injury based liability of the relevant DOTs. We also found that DOT-collected WVC observations usually lack the spatial precision to be useful for locating mitigation to reduce collisions, but when they are precisely located, result in different priority locations  along roadways when compared to precisely-located, crowd-sourced observations. This demonstrates that a combination of accessible DOT-collected and crowd-sourced observations may be critical for prioritizing conservation actions for wildlife and connectivity.