To elaborate our model we removed uncertain localities such as 'Bogota skins', those localities with no coordinates and all records with coordinates laying down in sites with estimated elevations above 2,301 m.
The habitat suitability model generated in Maxent showed areas suitable in climatic terms for this species in the middle section of the Western and Central Andes, Sierra Nevada de Santa Marta, serrania de San Lucas, and some areas in the far south Amazon. These areas are not known to be occupied by the species and were excluded from the potential distribution map of this hummingbird. Also is important to note that in our final maps we presented some extended areas (predicted as suitable) in the southern portion of the western slope of the Eastern Andes, the extreme northwest and the southwest. Also serrania del perija, from were there are no known records in Colombia but from where apparently
there are records in the Venezuelan side.
Assuming that the distribution of the species may have filled the complete climatic model generated, its distribution today in remnants of forest is about 83,586 km2, which corresponds to a loss of 52 % of its potential original distribution due to deforestation.
Presence of subspecies merrittii in extreme northwest Colombia is hypothesised here similarly to other authors, such as McMullan & Donegan (2014), since there are no collected specimens identified as such currently.
Regularized training gain is 1.519, training AUC is 0.921, unregularized training gain is 1.925.
Algorithm converged after 540 iterations (15 seconds).
The follow settings were used during the run:
48 presence records used for training.
10048 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: 0.199, categorical: 0.250, threshold: 1.520, hinge: 0.500
Feature types used: hinge linear quadratic
'Equal Training Sensitivity and Specificity' and 'Equate Entropy of Thresholded and Original Distributions' thresholds and omission rates:
0.167-0.219-Fractional predicted area
0.167-0.104-Training omission rate