Abstracts

Presented by Nick Masto - Email: nmasto@g.clemson.edu

Aerial surveys are efficient and effective strategies to estimate occurrence and abundance of wildlife populations especially in large and inaccessible landscapes. However, a fundamental concern is that animal detection is imperfect. Thus, accounting for imperfect detection is critical for accurate inference of abundance, population trends, and spatiotemporal distributions. We used two simultaneous aerial observers and analyzed data as replicated counts using N-mixture models to estimate detection, abundance, and species-habitat relationships. Specifically, we desired accurate (i.e., adjusted by detection probability) and precise (i.e., CV 1520%) estimates of abundance of dabbling and diving ducks in South Carolina, based on data in January and February 2018. Model-based inference using habitat covariates generally was precise for dabbling and diving ducks (11% CV 27%) compared to single-observer, design-based estimation (17% CV 36%). However, both observers exhibited low detection probabilities for dabbling and diving ducks (p = 11% and 8%, respectively) and detection probability varied among habitat types for duck taxa. Thus, we suggest using methods such as double or repeated sampling to minimize detection bias during aerial surveys for wintering waterfowl. Additionally, habitats influencing duck abundance varied between January and February 2018. Managed and non-managed historic rice fields influenced dabbling and diving duck abundances positively in January 2018 but not in February 2018. Complete drawdowns of managed tidal impoundments following waterfowl hunting in late January may have influenced duck-habitat relationships in February 2018. Additional analyses are in progress, but we provide first estimates of wintering waterfowl using N-mixture models with data collected from aerial surveys in South Carolina. Similar methods can be adopted to improve monitoring and estimation of wintering and breeding waterfowl populations.