To elaborate our model we removed uncertain localities such as 'Bogota skins' and all records with coordinates laying down in sites with estimated elevations above 1,811 m. Also a record in BioMap from Buenavista (Meta), which clearly represents corrupted data, was removed previously to modelling.
The habitat suitability model generated in Maxent showed areas suitable in climatic terms for this species in a continuous belt along the eastern slope of the Eastern Anden from Arauca to the south, and a very few areas in the far east of Colombia. These areas are not known to be occupied by the species and were excluded from the potential distribution map.
Several specimens in collections from the area near the border with Panama have been identified as ssp subrufescens and not saturatus. Apparently saturatus has been assumed to be in Colombia because of an old specimen placed in the British Natural History Museum with locality ‘Colombia’ collected by R.J. Balston. This specimen well may come from Panama and not Colombia. The evidence suggests possibly ssp saturatus is not in Colombia, at most in the extreme NW Colombia may be an intergrading area between both subspecies, but so far there is no evidence pointing to this.
Assuming that the distribution of the species may have filled the complete climatic model generated, its distribution today in remnants of forest is about 116,342 km2, which corresponds to a loss of 60 % of its potential original distribution due to deforestation in Colombia.
Regularized training gain is 1.375, training AUC is 0.946, unregularized training gain is 1.860.
Algorithm terminated after 2000 iterations (67 seconds).
The follow settings were used during the run:
121 presence records used for training.
10120 points used to determine the Maxent distribution (background points and presence points).
Environmental layers used (all continuous): bio10co bio11co bio12co bio13co bio14co bio15co bio16co bio17co bio18co bio19co bio1co bio2co bio3co bio4co bio5co bio6co bio7co bio8co bio9co
Regularization values: linear/quadratic/product: 0.050, categorical: 0.250, threshold: 1.000, hinge: 0.500
Feature types used: hinge product linear threshold quadratic
'Equal Training Sensitivity and Specificity' and 'Equate Entropy of Thresholded and Original Distributions' thresholds and omission rates:
0.132-0.256-Fractional predicted area
0.132-0.008-Training omission rate