RESULTS
The average δ15N values for female and male deer were 3.1‰ (range −1.1‰ to 7.3‰) and 2.4‰ (range 0.5‰ to 4.1‰), respectively (Table S1). The average δ15N values for female deer killed by trapping and shooting were 3.3‰ (range −0.5‰ to 7.3‰) and 2.7‰ (range −1.1 to 5.2‰), respectively (Table S1).
In the female model, accessibility to agricultural crop fields was positively related to the δ15N values given that the 95% confidence interval did not include zero (coefficient = 0.55; 95% CI = 0.29 to 0.80; Fig. 2). The log-transformed scaling factor ln(α) was 2.30 (95% CI = 0.83 to 3.77). The α determines the distance-dependence in accessibility to a crop field illustrated by exp (−αd ), which provides a measure of the relative crop consumption of female deer in relation to the distance to agricultural crop fields (d ; Fig. 3). Figure 3 shows that where d is zero, the likelihood of crop consumption completely depends on the crop field area in each 1 × 1 km mesh. The likelihood of crop consumption decreased by half when the distance to agricultural crop fields (d ) was 10 km and fell to essentially zero at a distance of approximately 40 km (Fig. 3). The method of culling did not affect the δ15N values (coefficient = −0.37; 95% CI = −1.00 to 0.26). In the model for male deer, accessibility to agricultural crop fields did not affect the δ15N values (coefficient = 0.31; 95% CI = −0.10 to 0.72; Fig. 2). ln(α) was −0.66 (95% CI = −2.94 to 1.62).