Collisions between wildlife and vehicles are detrimental to both wildlife and human safety. A variety of mitigation methods have been deployed with the intent of increasing ungulate awareness of approaching vehicles. Wildlife warning reflectors are one such method. Roadside reflectors are designed to reflect headlights into the right-of-way and alert ungulates to approaching vehicles. Studies of the effectiveness of these reflectors have yielded mixed results. We conducted a robust test of reflectors in central Wyoming, USA, during 2013 and 2014, and unexpectedly discovered a potentially more effective method than the reflectors for reducing collisions between vehicles and mule deer (Odocoileus hemionus). In our initial experiment, we manipulated 10 1.6-km segments of highway by leaving their reflectors exposed or covering them with a white canvas bag with the intention of neutralizing the reflector. The treatment configuration was swapped monthly. We counted deer carcasses under each treatment for 1 year and observed deer roadcrossing behavior using thermal video cameras. Carcass rates were 33% less in the white canvas treatment relative to uncovered reflectors. Deer in the white canvas treatment also stopped before entering the road 20% more often, ran into the road from the right-of-way 11% less often, and fled from the road 12% more often than when reflectors were uncovered. In a follow-up experiment, we found that deer carcass rates were 32% less when reflectors were exposed versus covered with black canvas. We further found that deer road-crossing behavior was least risky in a white canvas treatment, intermediate in a reflectors treatment, and most risky when reflectors were removed from posts. Taken together, these results indicate that, although reflectors were moderately effective, white canvas was substantially more effective in reducing deer-vehicle collisions. This unexpected finding suggests that new vigilance-enhancing mitigation methods should be explored as a way to reduce wildlife-vehicle collisions. Ó

Collisions with vehicles are a major threat to wildlife populations and often occur in identifiable patterns. To reduce wildlife road mortalities, mitigation structures including exclusionary fencing and wildlife crossings are constructed. Openings in fencing at road intersections may lead to concentration of road mortality hot spots at openings leading to a belief that these gaps concentrate road mortalities. However, it is also possible that hot spots existed at these locations before construction indicating that road mortality patterns have not changed with mitigation structure construction. Therefore, to assess mitigation structure effectiveness, it is important to examine both road mortality numbers and road mortality spatial distribution. Wildlife road mortality data was collected on a 15-km section of rural highway in Texas, USA before, during, and after the construction of wildlife mitigation structures. We expected that the number of road mortalities would decrease after construction compared to before construction and that road mortalities would become more concentrated around openings in the fence. We used ANOVA to compare numbers of road mortalities and emerging hot spot analysis and generalized linear modelling to assess changes in road mortality spatial distribution. Road mortalities were not significantly different in the before and after construction periods (p = 0.092). While there were no significant changes in road mortality patterns with construction, cluster intensity was greater when nearer to fence openings in all three time periods. Emerging hot spot analysis provides an effective and easy way to visualize road mortality patterns through time, however, due to low numbers of mortalities in many road mortality studies, including this one, the power of this analysis to detect significant changes in road mortality may be limited. This technique can provide both ecologists and transportation planners an effective tool for identifying patterns that may warrant further investigation using traditional statistical techniques.

This is an open access article under the terms of the Creat ive Commo ns Attri bution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.