Decide biomod2 models nmodels and number of run to use
So far using c('GLM','GBM','RF')
('GAM lead to some convergence error and Maxent is not working with presence/absence data)
Ensemble model with 10 runs per model with calibration on 70% of the data.