Model Evaluation
The Area under the receiver operating characteristic curve (AUC) was
used to assess the models’ predictive performance of the models ( Hanley
and McNeil, 1982) and the True Skill Statistic (TSS; . The sensitivity,
or the percentage of known presences predicted as presences, is plotted
on the ROC curve against specificity, or the percentage of
pseudo-absences predicted as absences. These validation techniques are
well-known and perform very well . Models with an AUC of less than 0.7
were disqualified. It has been demonstrated that weighing the individual
model projections according to their AUC scores is a particularly
trustworthy technique . Additionally, the ensemble model’s relative
importance of the variables was calculated from the ‘biomod2 ’
package devoted functionality . The ensemble models’ projections from
the two-stage sampling mentioned above were averaged to produce the
final potential distribution. To evaluate the spatiotemporal habitat
dynamics, the ensemble models were projected at roughly 1 km resolution.