Half a million deer-vehicle accidents (DVAs) occur annually on North Americ
an roads, yet the possible factors associated with these accidents are not
well understood. We examined the influence of landuse patterns and highway
characteristics on Iowa DVAs from 1990-97. Over 25% of DVAs occurred at 3.4
% of all mileposts in Iowa. Annually, the greatest numbers of DVAs occurred
in May and November. We examined 2.59-km(2) areas around a set of randomly
selected milepost locations (n = 1,284) to identify factors that may have
influenced DVAs. Stepwise logistic regression produced a B-variable model (
chi(6)(2) = 331.543, P = 0.001) that included 4 landscape variables, the nu
mber of bridges, and the number of lanes of traffic. A factor classificatio
n tree constructed from these variables revealed that 97% of mileposts asso
ciated with greater than or equal to 4 bridges were "high" DVA sites (great
er than or equal to 14 DVAs). The logistic model correctly classified 63.3%
of 245 sites in a validation data set. Our results suggest that mitigation
of DVAs should be focused on areas with a high number of bridges.