To elaborate our model we removed uncertain localities such as 'Bogota skins' and all records with no coordinates.
The habitat suitability model generated in Maxent showed areas suitable in climatic terms for this species west of the Andes. These areas are not known to be occupied by the species and were excluded from the potential distribution map of this hummingbird.
Distribution of specimens, according to BioMap, suggests that nominate subspecies guainumbi is in Colombia but this is apparently an error and all must be reclassified to subspecies andinus. Nevertheless, is not clear why in BioMap most specimens were classified as the nominate subspecies while in the online collection of the Instituto de Ciencias Naturales all are classified as andinus. The literature suggests that subspecies andinus was described from Bogota skins by Simon in 1921. In the original document 'Historie Naturelle des Trochilidae', Simon states in page 316 the distribution as 'Colombie: savane de Bogota'. Later by 1945, Peters in the volume V of the 'Check-list of the birds of the world' states in page 58 the distribution as 'Colombia; distribution not worked out'. The most likely is that Bogota specimens might have been collected nearby in the 'Llanos' of Colombia. From this point it seems that until recently it has been assumed that specimens from Colombia are all andinus. Although, it is not clear the steps taken to assume this in the current literature such as in the HBW Alive (2016). Will be of interest to make a revision of the species in Colombia to define clearly the validity and distribution limits of andinus.
Regularized training gain is 1.445, training AUC is 0.957, unregularized training gain is 2.063.
Algorithm converged after 600 iterations (17 seconds).
The follow settings were used during the run:
19 presence records used for training.
10018 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: 0.462, categorical: 0.250, threshold: 1.810, 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:
0.105-0.235-Fractional predicted area
0.105-0.000-Training omission rate