Results of Boosted Regression Trees models showing the influence of each predictor on the respective response variable, threshold estimates in non-linear relationships, model parameters and model performance.

PC1-2 = principal components 1–2 (accounting for > 60% of the variance) from a Principal Component Analysis on vegetation variables (see Methods); PA = protected area; Mammals = perceived mammal densities along road transects; tc = tree complexity; lr = learning rate; bf = bag fraction; CV-deviance = Cross-Validated deviance; Performance = 1 –(residual deviance / total deviance).