Uncovering Breeding Habitat Use of an Uncommon Songbird in Pennsylvania Using Large Scale Acoustic Data

ABSTRACT The Canada Warbler ( Cardellina canadensis ) population has declined by over 60% across its range since 1966. Knowledge of breeding habitat use is crucial to the conservation of songbird species but is missing for significant portions of the breeding range of the Canada Warbler's genetically distinct eastern population. We used a large‐scale deployment of autonomous recording units (ARUs) to record Canada Warbler song across north‐central and southwestern Pennsylvania. We deployed ARUs at 664 locations in forested ecosystems, setting them to record for 2 h each morning from mid‐May to mid‐June. We used classifier‐assisted listening to detect Canada Warbler song and generate detection histories. Our occupancy models revealed six significant habitat associations. We highlight forest conditions that can be targeted for the conservation of steeply declining eastern Canada Warbler populations. Our study is an example of the gains in statistical power allowed by ARUs and machine learning methods.