Asset Details

Description:
Maxent models of the abiotically suitable area of the Caribbean spiny pocket mouse (Heteromys anomalus), showing the effect of changes in the regularization multiplier. Examples appear here for the randomly partitioned approach, one iteration of the cross‐validation experiment and selected regularization multipliers: (a) 0.25, (b) 1.00, (c) 2.00 and (d) 6.00. The predictions show a suitability gradient from low (0, blue) to high (1, red). Squares correspond to calibration (white) and evaluation (purple) localities. In these examples, note signs of overfitting at the two lower values of the regularization multiplier and a loss of discriminatory ability at the highest regularization value (see Fig. ).

Credit:
Radosavljevic, A., Anderson, R. P., & Araújo, M. (2014). Making better Maxent models of species distributions: complexity, overfitting and evaluation. Journal of Biogeography, 41(4), 629-643..