NOTES
To elaborate a historical habitat suitability model we removed one corrupted record from Meta department and one 'Bogota' specimen. Additionally, seven records in BioMap identified as O. garrula chocoensis were transferred to O. cinereiceps following the HBW Alive taxonomy (2015) and were not used here for O. garrula. Both species are considered currently monotypic (HBW Alive, 2015). It is important to add that in this first exercise we just used records that represented to our knowledge the best known historical distribution of this species in Colombia. Otherwise, we conducted a second modelling experiment to elaborate a habitat suitability model of the potential areas of range expansion for this species following deforestation; in this last case, we used 26 accessions representing 10 localities in the departments of Choco, Antioquia and Caldas. After modelling we noticed that the accession 306220 in DatAves made in Cimitarra (Santander) was moved 1 degree to the west. Because it is just one locality and the results seemed to be relatively robust we decided not to repeat the experiment.
The 'historical' habitat suitability model generated in Maxent showed a few areas that are suitable in climatic terms for this species in the high Magdalena valley, central west Colombia and in the eastern slope of the Eastern Andes and to the east of the Andes. These areas are not known to be occupied by the species and were excluded from our potential distribution model for this chachalaca. Additionally, the model predicted an area in the Alta Guajira and another in extreme northwest Colombia, west of the low Atrato. This two areas are unlikely to be occupied by this species but we left them because they represent extensions of the climatic envelope that are of special interest since other species fill them; O. cinereiceps in the northwest and O. ruficauda in the northeast.
The 'possible areas of expansion' habitat suitability model generated in Maxent showed areas that are suitable in climatic terms for this species in southwest Colombia, the eastern slope of the Eastern Andes and to the east of the Andes. These areas are unlikely to be occupied by the species and were excluded from our potential distribution model. It is interesting to note that this model predict relatively well the historical distribution of the species. Additionally, it predicts extensive areas in the inter–Andean valleys and in the Pacific slopes of the northern West Andes. This model suggest that possibly this species still may continue expanding its range into the Cauca and Magdallena valleys.
Finally, both models were refined excluding areas above 1,800 m of elevation.
This species occupies deciduous forest, arid scrub, mangrove forest, riparian forest, borders and second growth (Hilty & Brown 1986) and may appear almost anywhere in its delimited potential area of distribution. Therefore, we did not elaborate a map overlapping its distribution with forested areas.
MODEL METADATA
'Historical' model
Regularized training gain is 2.073, training AUC is 0.969, unregularized training gain is 2.468.
Algorithm converged after 1000 iterations (28 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: linear quadratic hinge
responsecurves: true
jackknife: true
maximumiterations: 2000
'Equal Training Sensitivity and Specificity' and 'Equate Entropy of Thresholded and Original Distributions' thresholds and omission rates:
ETSS-EETOD-Description
16.069-10.404-Cumulative threshold
0.217-0.135-Logistic threshold
0.091-0.126-Fractional predicted area
0.083-0.021-Training omission rate
'Potential areas of expansion' model
Regularized training gain is 0.913, training AUC is 0.910, unregularized training gain is 1.443.
Algorithm converged after 160 iterations (1 seconds).
The follow settings were used during the run:
10 presence records used for training.
10010 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: 0.800, categorical: 0.500, threshold: 1.900, hinge: 0.500
Feature types used: linear quadratic
responsecurves: true
jackknife: true
maximumiterations: 2000
'Equal Training Sensitivity and Specificity' and 'Equate Entropy of Thresholded and Original Distributions' thresholds and omission rates:
ETSS-EETOD-Description
33.798-7.971-Cumulative threshold
0.465-0.151-Logistic threshold
0.172-0.401-Fractional predicted area
0.2-0-Training omission rate