Predictive habitat suitability models for nesting woodpeckers following wildfire in the Sierra Nevada and Southern Cascades of California

Abstract Woodpeckers are often focal species for informing management of recently burned forests. Snags generated by wildfire provide key nesting and foraging resources for woodpeckers, and nest cavities excavated by woodpeckers are subsequently used by many other species. Habitat suitability models applicable in newly burned forest are important management tools for identifying areas likely to be used by nesting woodpeckers. Here we present and test predictive models for mapping woodpecker nest-site habitat across wildfire locations that can be used to inform post-fire planning and salvage logging decisions. From 2009 to 2016, we monitored 313 nest sites of 4 species—Black-backed Woodpecker (Picoides arcticus), Hairy Woodpecker (Dryobates villosus), White-headed Woodpecker (D. albolarvatus), and Northern Flicker (Colaptes auratus)—from 3 wildfires in the Northern Sierra Nevada and Southern Cascades 1–5 yr after fire. Using these data, we developed habitat suitability index models that compared nest vs. non-nest sites for each species using (1) exclusively remotely sensed covariates, and (2) combinations of remotely sensed and field-collected covariates. We emphasized predictive performance across wildfire locations when selecting models to retain generalizable habitat relationships useful for informing management in newly burned locations. We identified models for all 4 species with strong predictive performance across wildfire locations despite notable variation in conditions among locations, suggesting broad applicability to guide post-fire management in the Sierra Nevada region. Top models for nest-site selection underscored the importance of high burn severity at the local scale, lower burn severity at the 1-km scale, mid-sized nest-tree diameters, and nest trees with broken tops. Models restricted to remotely sensed covariates exhibited similar predictive performance as combination models and are valuable for mapping habitat across entire wildfire locations to help delineate project areas or habitat reserves. Combination models are especially relevant for design of silvicultural prescriptions.