To elaborate the habitat suitability model we removed two corrupted records from Magdalena department (one in BioMap and one in DatAves), which are clearly erroneous and may belong to O. Garrula. Additionally, during modelling we added two records from EBIRD: one made in Las Tangaras reserve from ProAves and one made 1 km south of the urban area of Quibdo. There are seven records in BioMap identified as O. garrula chocoensis that must be transferred to this species. These records we believe would not have a major effect if added to modelling the habitat of this chachalaca since they add just one more nearby locality in Katios National Park. Both species are considered currently monotypic (HBW Alive, 2015). Additionally, we moved slightly (centecimals of a degree) several coordinates that lay down out of the mask of Colombia, bringing them inside of it and making use of them during the modelling process.
The habitat suitability model generated in Maxent showed extensively areas that are suitable in climatic terms for this species in the southern Pacific region, the lowlands adjacent north of the Western and Central Andes and around entering into the west side of the mid Magdalena valley, the low passes on the Western Andes in the area above the high Atrato and San Juan valleys, the high Patia valley, some sections of the foothills of the Central Andes and even in the eastern slope of the Eastern Andes. These areas are not known to be occupied by the species and were excluded from our potential distribution model for this chachalaca; except areas from the low Atrato to the Sinu valley and in the southern Pacific region, which might be possibly occupied by this species. Usually, the limit of the distribution of O. cinereiceps has been thought to be the rio Atrato (Hilty & Brown, 1986). However, there are a couple of records from Katios National Park in DatAves, with coordinates that lay down on the east side of the Atrato valley, and there is no apparent reason to consider them erroneous.
It is important to note that the record from Las Tangaras (at ≈ 1,600 m) is much higher than the usual elevations at which this species has been recorded in Colombia previously (≤ 300 m). Although not outside it elevational known range in its whole distribution from Honduras to Colombia (BirdLife International, 2015). Interestingly, our model did not predict that particular locality, despite it extended predictions closer to the foothills of the Western Andes. This suggests that our proposed distribution for this chachalaca may possibly extend further into the Western Andes. Thus, it is still necessary to determine if it occurs in other localities at similar elevations in this cordillera.
Assuming that the distribution of the species may have filled the complete climatic model proposed, its potential distribution today in remnants of forest is about 52,977 km2, which corresponds to a loss of 45 % of its potential original distribution due to deforestation. If we restrict this model to the most likely historical distribution according to records, the species can be in about 29,847 km2 of forested areas, being lost from 50 % of what originally it may have occupied. Additionally, if we extend the potential distribution to the slopes of the Western Andes below ≈ 1,600 m, the suitable areas for this chachalaca in remnants of forest increase in about 12,772 km2 in the whole Pacific region and in 7, 226 km2 in the restricted version of its distribution. Taking these into account, the loss of potential habitat increased to 46% for the whole region and decreased to 49% for the restricted distribution.
Regularized training gain is 2.045, training AUC is 0.983, unregularized training gain is 2.944.
Algorithm converged after 680 iterations (19 seconds).
The follow settings were used during the run:
15 presence records used for training.
10015 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: 0.586, categorical: 0.321, threshold: 1.850, 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:
0.067-0.129-Fractional predicted area
0.067-0.067-Training omission rate