When is a Ratio Estimator not a Ratio Estimator? Evaluation of Catch Estimators for Alaska Groundfish Fisheries.*Jennifer Cahalan, Pacific States Marine Fisheries Commission & NOAA Fisheries
Jason Gasper, NOAA Fisheries Alaska Regional Office
Jennifer Mondragon, NOAA Fisheries Alaska Regional Office
Keywords: Alaska, Catch Estimation, Fisheries, Ratio Estimators, Simulation
The federal commercial groundfish fisheries operating in Alaska are some of the largest in the world, with annual catches approaching 1.43 million tons worth an estimated $640 million in 2009. Fishery management decisions and decisions by industry to control harvest are made using timely electronic data feeds to monitor catches. Therefore, catch estimation needs to be transparent and easily understood by all stakeholders. Deterministic nearest-neighbor imputation is one method used in Alaska to estimate the species-specific catches for fishing events that are not monitored. Evaluating the statistical properties of imputation provides important analytical advice to customers using the estimates, including fishery managers, stakeholders, and scientists. While, ratio estimators are often used in catch estimation, the underlying model assumptions are not always realistic. We conducted a simulation study comparing properties of the imputation-based estimator with the simple mean and ratio estimators. Three ratio estimator variances are evaluated. No single estimator outperformed the others, making the choice of estimator difficult. Implications of estimation options will be discussed.