Prioritizing road sections for wildlife fencing: Considering scales, thresholds, and trade-offs

Roads have many negative effects on wildlife populations, the most visible of which is wildlife mortality due to vehicle collisions. Fences and wildlife passages have been applied to reduce roadkill. However, wildlife passages without fencing, in general, have been shown to not reduce roadkill. Therefore, fencing is the most important component for mitigating roadkill. Understanding where and why wildlife-vehicle collisions occur can inform planners about where mitigation measures would be most effectively placed. However, it has not been discussed how the choice of scales and confidence levels influence the results and how the locations of the warm- and coldspots should be included in the decision-making. We used roadkill data of reptiles and medium-sized mammals from three roads and applied multiple scales of analysis and several confidence levels to answer two questions: (1) Are there thresholds in the effect of the extent of fencing (total fence length) on the expected reduction in road mortality? (2) What are the effects of varying scales and varying confidence levels on the road section prioritization results for fencing? We used the software Siriema to identify hotspots, warmspots, and coldspots of road mortality at multiple scales. Our results show that the choice of confidence intervals and scales affects the amount of hot-, warm-, and coldspots identified. At lower confidence levels, there are more hotspots and coldspots than at higher confidence levels. When roadkill data are analyzed at a smaller scale (e.g. 100 m), there are more hotspots identified, but combined they cover a shorter overall length of the road than hotspots at larger scales. Our study shows how identifying hotspots, warmspots, and coldspots at multiple scales allows for a more comprehensive approach for locating and prioritizing road sections for wildlife fencing. We discuss the existence of thresholds in the amount of total fencing needed, the importance of considering the fence-end effect when defining the length of the fences to be installed, and the FLOMS trade-off: "Few-Long-Or-Many-Short fences". Based on these results, we propose an Adaptive Fence Implementation Plan with steps to prioritize road sections for wildlife fencing. The steps of this plan consist of collecting roadkill data to maintaining the installed wildlife fences, integrating hot-, warm-, and coldspots as well as multiple scales and confidence intervals. --
Comment: J. Jaeger would present this talk via the internet (Skype or Zoom) to avoid CO2 emissions (as discussed with Tony Clevenger). If this is not possible, then Fernanda Teixeira can present it (who will attend the conference in person).