Hierarchical model detects decadal changes in calibration relationships of single-pass electrofishing indices of abundance of Atlantic salmon in two large Canadian catchments

Abstract

Electrofishing is a commonly used technique to assess freshwater fish population abundance, and in many programs, there has been a shift in the sampling methodologies towards less laborious techniques. These new techniques usually only provide an index of abundance and require calibration with other sampling methods such as successive removal to be used for absolute abundance estimation. Using data for juvenile Atlantic salmon (Salmo salar) collected in 400 sites sampled over 21 years in two large Canadian river catchments with a single sampling protocol, we developed a hierarchical Bayesian model to account for effort, day of sampling, area of the site, and catchment effects on the relationship between the single-pass index of abundance and the fish densities, thereby illustrating the importance of carrying out a calibration exercise on a regular basis. Our work indicates that calibration relationships can change over time even with standardized sampling protocols and that these directional changes in important components of the sampling procedure can bias the estimate of population abundance and misinform the understanding of population dynamics.