Fire severity is hypothesized as an important driver of bird responses to wildfire. For those species that typically respond negatively to increasing severity, Accessibility of high-severity burned forest may be dependent on the proximity of unburned or low-severity burned forest to meet all of the needs of breeding bird species that typically respond negatively to increasing fire severity. We used dynamic occupancy modeling and intensive surveys to test the hypothesis that the presence of live trees explained distribution of birds in high-severity burned forest in the first three years following the Biscuit Complex Fire in southwest Oregon.
Proximity of low-moderate severity burned forest appeared to be important for some foliage- and bark-gleaning species. However, probability of occurrence in high-severity patches for several canopy-associates was not associated with distance from low-moderate severity burned forest. We found strong evidence that large snags (>52-cm) and large live conifers provided important nest substrates for several of these open-cup nesting species. Sprouts from top-killed hardwood species provided primary nesting substrates for early-seral associates and some species that nest in live trees or mature shrubs in unburned forest. Large areas of high-severity burned forest appeared to provide important breeding habitat for edge-sensitive cavity-nesting species. Most cavity nests were in large fire-killed or legacy snags with significant decay.
In all, bird species that were likely to nest in high-severity forest also were likely to persist in the early post-fire period. Identifying the structures used for nesting by these species provides a mechanistic understanding of the distributions of many species of breeding birds following wildfire, and underscores the importance of local ecological context in interpretation of the effects of disturbance.
We used intensive surveys to evaluate the efficacy of point count surveys to detect breeding birds. Most species had significantly higher detection probabilities at sites with evidence of breeding activity than at sites without such evidence. Among species, however, detection probability did not predict prevalence of breeding. Most species for which detection probabilities during point counts were low or moderate were those predicted to decline following high-severity fire. This finding highlights the importance of evaluating detection probabilities to identify effective sampling methods for species expected to respond negatively to habitat alteration.