Landscape influences on deer-vehicle accident areas in an urban environment

Deer-vehicle accidents (DVA) have become an important human safety concern in the United States, and few studies have focused specifically on urban areas. We used remotely sensed data, multivariate statistics, and a geographic information system (GIS) to quantify landscape factors influencing DVA in 2 suburbs of Minneapolis, Minnesota, USA, during 1993-2000. We determined DVA areas (n = 80) containing greater than or equal to2 roadkills and control areas (n = 80) containing 0 or 1 roadkill based on numbers of DVA within 0.13-km(2) buffered road segments. The most important variables (based on Akaike weights) that differentiated between DVA areas and control areas were number of buildings and number of public land patches. A logistic regression model containing these variables best fit the data and correctly classified 31 of 40 (77.5%) areas not used for model building. Local wildlife biologists and urban planners can use this information in managing deer habitat to minimize DVA by reducing forest cover and shrubby areas on public land near roads. We suggest that community officials consider using the potential for DVA as justification for proactive population and habitat management programs to maintain deer populations at levels that minimize deer-human conflicts.