Identification and prioritization of road sections with a relatively high concentration of large wild mammal-vehicle collisions in Gallatin County, Montana, USA

Abstract

The primary objective of this project is to identify and prioritize the road sections in Gallatin County that have a relatively high concentration of collisions involving large wild mammals. These road sections may then later be evaluated for potential future mitigation measures aimed at 1. Reducing collisions with large wild mammals, and 2. Providing safe passage across roads for large wild mammals, as well as other wildlife species in the area. We acquired the 3 datasets related to large wild mammal-vehicle collisions in Gallatin County: 1. Wildlife-vehicle crash data collected by law enforcement personnel, 2. Carcass removal data collected by road maintenance personnel; and 3. Grizzly bear road mortality data by the U.S. Geological Survey. The carcass removal data and grizzly bear road mortality data were merged into one carcass database. We conducted separate analyses for the crash data and the carcass data. We conducted two different types of analyses to identify and prioritize road sections with the highest number of wildlife-vehicle crashes and carcasses: 1. Kernel Density Estimation (KDE) analysis that identifies road sections with the highest concentration of collisions, and 2. Getis-Ord Gi* analysis identifies road sections that have statistically significant spatial clusters of collisions. There was great similarity between the hotspots identified through the Kernel Density Estimation analyses for 2008-2022 and 2018-2022 for both the crash and carcass removal data. The same was true for the Getis-Ord Gi* analyses. Especially sections of I-90 and US Hwy 191 between I-90 through Four Corners to the mouth of Gallatin Canyon had the highest concentration of wild animal crashes and large wild animal carcasses. Based on the Getis-Ord Gi* analyses, these road sections generally had concentrations of crashes and carcasses that were significantly higher than expected should the crashes and carcasses have been randomly distributed. In other words, these road sections do not only have the highest concentration of crashes and carcasses, but the identification of these road sections is not based on coincidence. These road sections have a concentration of crashes and carcasses that is beyond random.