To elaborate our model we removed uncertain localities such as those localities with no coordinates and all records with coordinates laying down in sites with estimated elevations below 861 m and above 2,803 m.
The habitat suitability model generated in Maxent showed very good fit not showing suitable areas out of the main Andean Ranges. A very few pixels were predicted in serrania de San Lucas and the northern spurs of the Western Andes, those were left in the final potential map to inform, although must be considered areas where the species most likely does not occur.
Distribution of specimens according to BioMap suggest two zones of intergradation between subspecies rubinoides and aequatorialis. Both of them of at least 17,533 km2 in central Antioquia and from Quindio and Valle del Cauca towards the south to Nariño.
The literature points to the nominate subspecies as the one present in the Central Andes (HBW Alive 2016). Nonetheless, distribution of BioMap specimens shows that this is not certain at all. Most of the Western Andes and the Central Andes need collections to clarify better the boundaries between subspecies rubinoides and aequatorialis. Some authors have suggested that subspecies cervinigularis enters Colombia in the extreme southeast Andes (McMullan & Donegan 2014). Nonetheless, this is not likely since there are some collections confirming the presence of subspecies rubinoides.
Assuming that the distribution of the species may have filled the complete climatic model generated, its distribution today in remnants of forest is about 39,107 km2, which corresponds to a loss of 61 % of its potential original distribution due to deforestation.
Regularized training gain is 2.405, training AUC is 0.979, unregularized training gain is 2.782.
Algorithm terminated after 2000 iterations (84 seconds).
The follow settings were used during the run:
192 presence records used for training.
10192 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:
0.062-0.090-Fractional predicted area
0.062-0.005-Training omission rate