Results of the linear mixed effects model.

<p>Fixed effect size estimates of (i) four distance classes (200, 1000, 2500 and 4000 m), (ii) the inclusion of topological variables (TV) and (iii) the choice of the environmental dataset (measured MV vs. assessed AV). Effect sizes are estimates how the cross-validated AUC changes compared to the base model for measured variables without TV at distance class 0.The linear mixed model structure follows: <i>arcsin(√ AUC) ~ α+β</i>_<i>D</i><i>+ β</i>_<i>TV</i><i>+ β</i>_<i>AV</i><i>+ a</i>_{<i>Species</i>}, where <i>α</i> = intercept, <i>β</i> = single effect sizes and <i>a</i>_{<i>Species</i>} = within species as random effect. 95%-Confidence intervals (CI) are based on parametric bootstrapping. Significant effects are highlighted in bold.