To elaborate the model of this species we removed in BioMap two records related to skins allegedly collected by brother Niceforo in Medellin in 1915, which most likely represent 'Medellin skins' collected elsewhere in east or north Antioquia and three more records in DatAves, two of them from golfo de Morrosquillo that might be slightly north to be correct, and one from bajo Calima that seems placed some 400 m above the upper altitudinal limit for this species (i.e. 1,000 m). Additionally, we moved about 8.4 km southeast the coordinates of two records in BioMap of skins collected by K. v. Sneidern in Volador (Bolivar), to a site where coordinates lay down at an elevation closer to that specified in the records (≈ 900 m).
The habitat suitability model generated in Maxent predicted a few areas in the east bank of the Magdalena valley, mid-high Magdalena valley and in extreme southwestern and southeastern Amazonia. These areas are not known to be occupied by this quail and were trimmed off from our final potential distribution map. Equally, we limited its distribution to areas below 1,000 m.
Assuming that the distribution of the species may have filled the areas predicted as suitable (i.e. marginally suitable, suitable and highly suitable) and those not predicted but thought to be likely suitable, its potential distribution today in remnants of forest is about 82,572 km2, which corresponds to a loss of ≈ 50 % of its potential original distribution due to deforestation.
Regularized training gain is 1.888, training AUC is 0.968, unregularized training gain is 2.327.
Algorithm converged after 660 iterations (19 seconds).
The follow settings were used during the run:
28 presence records used for training.
10028 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: 0.288, categorical: 0.250, threshold: 1.720, hinge: 0.500
Feature types used: linear quadratic hinge
'Equal Training Sensitivity and Specificity' and 'Equate Entropy of Thresholded and Original Distributions' thresholds and omission rates:
0.083-0.151-Fractional predicted area
0.071-0-Training omission rate