Biases in Bird-Window Collisions: A Focus on Scavengers and Detection Rates by Observers

Abstract

Bird–window collisions (BWCs) represent a source of mortality for both resident and migratory birds that researchers aim to quantify. Factors that limit carcass detection, including removal of carcasses by scavengers and inherent error in observer detection, can complicate these efforts. We combined 2 studies to examine what proportion of carcasses were scavenged and removed and how successful observers were at detecting bird decoys of multiple types. In fall 2019 and spring 2020, we deployed 40 bird carcasses at Radford University campus buildings and monitored visitation events with game cameras for up to 4 nights/5 days. Scavengers visited 31 of the carcasses and scavenged (disturbed but not removed) 14/31. Scavengers included Felis catus (Domestic Cat, n = 6 events), Mephitis mephitis (Striped Skunk, n = 3), and Procyon lotor (Raccoon, n = 3). All mammalian scavenging events were nocturnal. Cats visited carcasses without disturbing them (n = 22 events), whereas Raccoons (3/3 events) and Striped Skunks (3/4) scavenged carcasses on discovery. Mammals scavenged or removed 3 carcasses in the first night after deployment (e.g., in the first 14 h), suggesting that scavenging can cause daily surveyors to miss ∼7.5% of carcasses. Across spring and fall 2020, we deployed 60 decoys of 4 unique types (varying in color or shape), partially obscuring them under shrubs and along landscaped herbaceous vegetation. We found that observers detected 48% of the decoys on their first attempts (min–max: 8.7–100%). Decoy type did not affect observer detection. By our estimates, 2 observers per day could have collectively missed 27% of carcasses. Although BWC projects typically complete surveys once or twice daily, we suggest twice-daily surveys when monitoring BWC during migratory seasons and once-daily surveys at other times to minimize carcass loss due to scavenging.