Evaluating Habitat Suitability Models for Nesting White-Headed Woodpeckers in Unbumed Forest

• Published in: The Journal of wildlife management 2015-02, Vol.79 (2), p.263-273
• Main Authors: LATIF, QURESH S., SAAB, VICTORIA A., MELLEN-MCLEAN, KIM, DUDLEY, JONATHAN G.
• Format: Article
• Language: English
• Subjects: Animal nesting; Bird nesting; Ecological modeling; Forest canopy; Forest habitats; Habitat conservation; Habitat Relations; Habitat selection; Modeling; Pixels; Woodpeckers
• ISSN: 0022-541X; 1937-2817

Habitat suitability models can provide guidelines for species conservation by predicting where species of interest are likely to occur. Presence-only models are widely used but typically provide only relative indices of habitat suitability (HSIs), necessitating rigorous evaluation often using independently collected presence-absence data. We refined and evaluated presence-only habitat suitability models for nesting whiteheaded woodpeckers (Picoides albolarvatus), a narrowly distributed species of conservation concern that occupies dry conifer forests of the Inland Pacific Northwest, USA. We developed 2 models using Mahalanobis D² and Maxent techniques from nest location datasets collected on the eastside of the Cascade Mountain Range, Oregon (1 dataset in 1997-2004 and another, sampling a broader spatial extent, in 2010-2011). Consistent with known ecology of white-headed woodpeckers, both HSI models related positively with percent ponderosa pine, moderate levels of canopy cover (approx. 40%), and moderate-to-high levels of heterogeneity in forest structure. Unlike Mahalanobis HSIs, however, Maxent HSIs were consistently and positively related with nest prevalence and positively related with habitat use estimated with independent point count data. Locations with high Maxent HSIs were characterized by canopy openings adjacent to closed canopy forests. The fact that this habitat feature was described by Maxent HSIs but not by Mahalanobis HSIs possibly explains why Maxent HSIs better predicted white-headed woodpecker occurrence. Additionally, we used presence-absence data for model evaluation that sampled a broader spatial extent than nest surveys and therefore allowed us to demonstrate the generality of Maxent HSIs. Additional nest location data collected across a broader portion of the species range would be valuable for further model improvement and evaluation, but until such data are available, we recommend use of Maxent HSIs to guide habitat conservation and restoration efforts in unburned dry forests of Oregon.