Figure 1.
Distribution of deer–vehicle collisions.
Number of DVCs for different types of roads in 2006 and 2009. Each dot represents one municipality crossed by at least one road of the corresponding type. The expected numbers of DVCs were derived from the parametric model (1).
Figure 2.
Goodness of fit for the parametric model (1) and the nonparametric model (2) illustrated by a scatterplot of the observed number of DVCs (ordinate) plotted against the fitted number of DVCs (abscissa).
Figure 3.
Collision risk, based on DVC index of climate, land use, browsing intensity, and spatial heterogeneity for 2006. The map shows seven risk classes based on the -means classification method.
Figure 4.
Multiplicative decomposition of the DVC index into contributions of climate, land use, browsing intensity, and spatio-temporal terms.
Figure 5.
Selected partial contributions of climate variables. The ordinate values can be interpreted as multiplicative changes in the expected number of DVCs per kilometer when all other variables remain constant.
Figure 6.
Effects of land use variables.
Selected partial contributions of land use variables. The ordinate values can be interpreted as multiplicative changes in the expected number of DVCs per kilometer when all other variables remain constant.
Figure 7.
Effects of browsing variables.
Selected partial contribution of browsing variables. The ordinate values can be interpreted as multiplicative changes in the expected number of DVCs per kilometer when all other variables remain constant.
Figure 8.
DVC index and harvest numbers.
Comparison of DVC index (for 2006) and roe deer harvest numbers for 2004–2009 for each of the Bavarian game management districts. Mean regression was fitted by a LOWESS smoother.