W-108-7
A Hierarchical Bayesian Multistate Mark-Recapture Model to Estimate Fish Collection Efficiency and Route Selection Probability for Multiple Release Groups of PIT-Tagged Fish at the Cowlitz Falls Dam
A Hierarchical Bayesian Multistate Mark-Recapture Model to Estimate Fish Collection Efficiency and Route Selection Probability for Multiple Release Groups of PIT-Tagged Fish at the Cowlitz Falls Dam
Passive integrated transponder (PIT) technology was used to evaluate performance of juvenile salmonids in response to modifications aimed at improving smolt collection efficiency at the first of three dams on the Cowlitz River, Washington, USA. Marked fish were detected in one of two fish passage flumes in the Cowlitz Falls Dam and subsequently at the Cowlitz Falls Fish Facility. We used a Bayesian multistate mark-recapture model to analyze fish collection efficiency and route selection across a range of environmental conditions and project operations for steelhead, coho salmon, and Chinook salmon. Results from 2013 and 2014 indicate that all species preferentially selected the route along the historic river course and that fish collection efficiency was highly variable throughout the migration period. Hierarchical Bayesian analysis allowed for simultaneous use of all the relevant data to provide estimates for all parameters, even when the number of fish recaptured was small. This model framework, combined with a robust database of project operations and environmental covariates, provides the grounds to identifying limiting factors and improve fish collection.