To elaborate our model we removed uncertain localities such as 'Bogota skins' and all records with coordinates laying down in sites with estimated elevations below 240 m and above 2,455 m. Also two records in BioMap and DatAves from Magdalena and Cesar, which represent corrupted data, were deleted previous to model the distribution. A further record in DatAves from La Planada (Nariño) that seemed too far south in the Pacific slope was also deleted.
The habitat suitability model generated in Maxent showed areas suitable in climatic terms for this species in Sierra Nevada de Santa Marta and Perija. Also a very few areas far east of the Andes in eastern Guainia. 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 at least two possible areas (≈ 9,028 km2) of intergradation between subspecies emiliae and apicalis, located between south Norte de Santander, west Cesar and north Santander and in the south in southern Huila. Furthermore, it is interesting to note that there are 20 specimens in BioMap from the West and Central Andes catalogued as coruscus. These specimens might represent intergradation with subspecies emiliae or they can be not correctly identified; this needs further revision. Peters (1945) in his hummingbirds volume has a footnote confirming these observation '1Specimens from western Colombia are certainly nearer to P. g. coruscus than P. g. apicalis.'
Assuming that the distribution of the species may have filled the complete climatic model generated, its distribution today in remnants of forest is about 86,496 km2, which corresponds to a loss of 58 % of its potential original distribution due to deforestation. Nonetheless, this species favours edges and secondary vegetation and therefore possibly deforestation has not negatively affected greatly its populations.
Regularized training gain is 1.733, training AUC is 0.954, unregularized training gain is 2.136.
Algorithm terminated after 2000 iterations (66 seconds).
The follow settings were used during the run:
134 presence records used for training.
10133 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:
0.100-0.179-Fractional predicted area
0.097-0.045-Training omission rate