Odontophorus gujanensis

NOTES

We elaborated a first habitat suitability model in Maxent for the whole species, the predictions of this model performed fairly well west of the Andes, whilst missed extensive areas in the Amazon region. Because of this, we decided to elaborate individual models for each one of the three subspecies confirmed in the country. It is important to highlight that subspecies medius has not been previously reported in the literature (HBW Alive, 2015) and specimens in the ICN are the first confirming the presence of this subspecies in Colombia. Additionally, to improve predictions for the subspecies buckleyi, we added during the elaboration of its model one point locality derived from three specimens (Catalogue No. 2455, 2456, 2457) collected in Tarapaca, Amazonas and deposited in the IAvH ornithological collection. Subspecies marmoratus was modelled twice, separating populations west and east of the Andes and finally joining individual trimmed models.

There are a few records in BioMap and ProAves datasets from Valle del Cauca, Cauca and west Nariño that are corrupted data and were removed during modelling.

Most of the original subspecies models elaborated predicted areas where other subspecies are present and therefore they were trimmed to fit better the known distributions. Thus, for the model elaborated for buckleyi it was trimmed off areas predicted in the slopes of all three cordilleras west of the Andes and in serrania del Baudo; for the model elaborated for marmoratus (west of the Andes) it was trimmed off areas in the Pacific and the Caribbean, except those adjacent to the central north Andean region; for the model elaborated for marmoratus (east of the Andes) it was trimmed off areas in the Pacific, Caribbean and central north Andean region, leaving just those in central northeastern Colombia (i.e. Norte de Santander-Arauca.);finally, for the model elaborated for medius it was trimmed off a few areas predicted out of Vaupes, in Putumayo, central northern Amazonas, central northern Guainia and west of the Andes.

Distribution of specimens, according to BioMap, does not suggests any intergradation between subspecies medius and buckleyi. It is important to highlight that the boundary of the distribution between these two subspecies it is not clear and needs further research. At the moment we assigned most of possible areas in central and eastern Guaviare and central and northern Guainia to subspecies buckleyi, but this still needs clarification.

MODEL METADATA

O. g. buckleyi

Regularized training gain is 0.966, training AUC is 0.913, unregularized training gain is 1.711.

Algorithm converged after 480 iterations (16 seconds).

The follow settings were used during the run:

23 presence records used for training.

10023 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: 0.385, categorical: 0.250, threshold: 1.770, hinge: 0.500

Feature types used: linear quadratic hinge

responsecurves: true

jackknife: true

maximumiterations: 2000

'Fixed Cumulative Value 5', 'Equal Training Sensitivity and Specificity' and 'Equate Entropy of Thresholded and Original Distributions' thresholds and omission rates:

FCV5-ETSS-EETOD-Description

5-34.412-11.17-Cumulative threshold

0.152-0.385-0.232-Logistic threshold

0.477-0.175-0.381-Fractional predicted area

0.043-0.174-0.087-Training omission rate

O. g. medius

Regularized training gain is 3.328, training AUC is 0.999, unregularized training gain is 5.416.

Algorithm converged after 1500 iterations (8 seconds).

The follow settings were used during the run:

3 presence records used for training.

10003 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: 1.000, categorical: 0.605, threshold: 1.970, hinge: 0.500

Feature types used: linear

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

74.292-14.298-Cumulative threshold

0.723-0.113-Logistic threshold

0.001-0.035-Fractional predicted area

0-0-Training omission rate

O. g. marmoratus (west of the Andes)

Regularized training gain is 1.404, training AUC is 0.930, unregularized training gain is 1.669.

Algorithm converged after 560 iterations (16 seconds).

The follow settings were used during the run:

40 presence records used for training.

10039 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: 0.221, categorical: 0.250, threshold: 1.600, 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

18.812-9.442-Cumulative threshold

0.279-0.128-Logistic threshold

0.15-0.246-Fractional predicted area

0.15-0.05-Training omission rate

O. g. marmoratus (east of the Andes)

Regularized training gain is 1.074, training AUC is 0.923, unregularized training gain is 2.002.

Algorithm converged after 140 iterations (1 seconds).

The follow settings were used during the run:

7 presence records used for training.

10007 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: 1.000, categorical: 0.545, threshold: 1.930, hinge: 0.500

Feature types used: linear

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

34.502-17.225-Cumulative threshold

0.321-0.163-Logistic threshold

0.143-0.342-Fractional predicted area

0.143-0.143-Training omission rate