FIGURE 4.

Importance of variables to train the model. Regularized training gain represents how much better the distribution fits the presence data compared with uniform distribution. “With only variable” represents result when only the particular variable is run, “Without variable” represents effect of removing particular variable from model and “With all variables” represent results of model when all variables are run. The figure shows jackknife regularized training gain of four carnivore groups (a) common leopard (b), small felid (c), Herpestidae, and (d) Viverridae