An Analysis of Deer and Elk-Vehicle Collision Sites along State Highways in Washington State

We analyzed records of deer and elk carcasses removed from Washington State highways between 2000-2004 where 14,969 deer (6,135 mule deer, 4,543 white-tailed deer, 4,014 black-tailed deer, and 277 unidentified) and 415 elk (249 in western Washington and 166 in eastern Washington) were removed following collisions with vehicles. Our goals were to model the frequency of ungulate-vehicle collisions (UVCs) on state highways and identify key factors associated with collision sites, describe temporal and spatial characteristics associated with deer and elk carcass removal sites, and identify sites of potential conflict or hazard where highways were bisected by deer and elk movement corridors.

A key factor influencing the potential for UVCs on Washington State Highways was the level of deer concentration in the surrounding area. Improved knowledge of deer concentration areas and migration corridors may be an important tool for mitigating the likelihood of UVCs in Washington.

The presence of roadside cover and forage, both important deer habitat components, was associated with higher collision counts in both our white-tailed and mule deer models. In eastern Washington, most deer cover occurred in the form of forested upland, which is a commonly preferred habitat type of white-tailed deer. Habitat covariates usually associated with lower quality deer habitats showed negative associations with collision counts. Watercourses and associated riparian areas contain important components of quality deer habitat including forage, cover, and travel corridors. The covariate �"total water” was important only in the mule deer model and suggests that riparian areas may have use in predicting UVCs in more arid environments.

Scale may be an important consideration when attempting to interpret factors potentially influencing UVCs. Important variables which could affect UVCs such as roadway, adjacent roadside features, deer movement patterns, and habitat quality can change considerably within a mile of distance which may result in misinterpretations of covariate influences. Because of the potential effects of spatial scale, there are likely a number of factors that may contribute to where and when UVCs occur on Washington’s state highways that should be measured at a finer scale.

Increasing levels of average annual daily traffic (AADT) were associated with higher numbers of collisions in most eastern Washington models. Conversely, AADT was negatively associated with collision counts in our overall and urban western Washington models; this negative association may have been a result of confounding effects due to the highly positive correlation between AADT and total developed area in western Washington. Highly developed areas generally provide little, if any, deer habitat, so a negative association with collision counts could be expected.

Speed limit and 2 arterial road types, rural interstate and rural principal arterial, were found to be positively associated with collision counts Higher vehicle speed limits are generally associated with arterial road types, and the results suggest a confounding influence between increased speed limits and some road types, making differentiation of independent effects difficult.

The seasonality we observed in UVCs in Washington likely was related to changes in ungulate behavior and environmental factors. Most vehicle collisions involving mule deer occurred during October â€" January; most white-tailed deer were killed during the months of December, October, and January. Similarly, we also found more black-tailed deer were killed by vehicles during October and November. Autumn and early winter over-lap with deer hunting seasons, a time of increased disturbance to deer. Deer may increase their movement to avoid hunters, increasing the likelihood of their being near or crossing highways. Fall (particularly November) is also the breeding season for most deer populations; during this time deer increase movements in search of mates. In addition, day length is declining and precipitation is increasing during this period, lowering driver visibility and causing peak drive times (early morning and early evening) to coincide with dawn and dusk, periods of high deer activity.

We identified sites or aggregates of sites that incurred very high numbers of vehicle collisions with deer and elk in Washington. The number of collisions at these sites demonstrated their uniqueness and required additional study and attention. All sites in eastern Washington were within deer winter ranges and 2 were located at the intersection of an active migration corridor and state highway. Winter ranges are traditional use areas, usually at lower elevations, where forage is relatively more available and deer concentrate to spend the winter season. During migratory periods, relatively high numbers of deer moving between seasonal use ranges tend to use traditional movement corridors and where migration corridors intersect highways, seasonally high incidents of deer-vehicle collisions will occur. Further investigation into these high-kill areas, including site visits, may help identify unique characteristics that can be used in management.

The WDOT dataset and the accidents they represent are most likely minimum estimates; documented removals of deer and elk carcasses from Washington state highways probably represent only a portion of an unknown number of road kills that actually occur.

Our modeling demonstrated those parameters indicative of higher quality deer habitat (modest slopes, near water/watercourses, southern exposure, forage, cover), as well as deer concentrations, were associated with higher collision counts. Similarly, highways bisecting areas of high deer use such as winter ranges or migration corridors experienced higher numbers of UVCs. These results suggest that providing passageways for deer to cross over or under highways, constructing barriers that prohibit entry onto roadways, and discouraging deer use near highways indirectly by affecting the quality of adjacent habitats or directly reducing deer densities through harassment or lethal removal may reduce UVC rates on state highways. Similarly, when new highways are being designed, evaluating potential UVC rates should be an integral part of the planning process. To achieve the lowest potential level of UVCs, new routes should avoid deer concentration areas and known migration corridors; habitat and geographic features shown by our models to have significant relationships with high UVC rates also should be avoided. Improved delineation of deer concentration areas and migration corridors should be a priority for guiding placement of future roadways.

Our analysis provided unique insights into characteristics associated with sites of collisions between deer or elk and vehicles on Washington highways. While this information will be valuable to traffic planners and wildlife managers, we view these efforts to date to be introductory. Additional research to accurately identify, predict, and mitigate ungulate-vehicle collision sites is needed to help reduce and prevent future accidents, personal injuries, property damage, and loss of deer and elk resources. Future work should focus on: 1) review of existing telemetry data sets of deer and elk locations collected from animals wearing GPS collars to assess movement patterns near and across state highways; 2) field inspection and mapping of high level deer and elk collision sites identified in this study to document road, vegetation, and terrain features at a local scale, and to identify site-specific options for mitigation; 3) implementation and evaluation of mitigation techniques at test locations; 4) field studies of deer movement patterns and mortality factors in relation to highway crossing patterns and habitat use adjacent to state highways to identify key factors associated with deer-vehicle collisions and accurately estimate number of deer killed by collisions with vehicles; 5) improved UVC data base including more accurate kill locations and, where possible, descriptors of drivers, vehicles involved, and extent of vehicle damage; 6) driver surveys to assess driver attitudes and collision involvement; and 7) experimental hunting seasons to reduce deer densities and gauge its affects on deer-vehicle collisions.