Estimation of seabird bycatch for North Pacific longline vessels using design- and model-based methods

Bycatch estimation for sensitive species is becoming increasingly important with the shift toward an ecosystem-based approach to fisheries management. Incidental mortalities for various seabird species occur on longline vessels throughout the world, including those in the North Pacific groundfish fleet. We present an approach to seabird bycatch estimation for North Pacific longline vessels using observer-collected data. Observers collect enormous amounts of data through a complex sampling design, but some information deficiencies preclude bycatch estimation using only probability sampling. Our approach combines probability sampling with model-dependent techniques to overcome these information deficiencies. The resulting bycatch estimator reflects the observer sampling design as closely as possible and minimizes reliance on untested model assumptions. We apply our estimator to black-footed albatross (Phoebastria nigripes) bycatch as an example and compare yearly estimates to those previously published. We also suggest changes in data collection that would further reduce dependence on model assumptions.