Model-based estimation of average fish weights from recreational fisheries

Catch estimates from recreational fisheries are an important component of many fishery management plans. Estimates of recreational catch (in weight) on the U.S. West Coast are often derived as the product of catch in numbers of fish and average fish weights. When estimates of average fish weight are imprecise (e.g., due to small sample sizes), the resulting estimates of catch in weight can fluctuate and unnecessarily trigger or delay management actions. This and other challenges associated with average weight estimation are currently addressed through replication of data based on deterministic algorithms ('borrowing rules'). These methods differ among states and do not present a viable method for variance estimation. In this study, we describe a model-based framework for estimation of average fish weights, with an application to the recreational groundfish fishery off Washington, U.S.A. The model-based framework identifies important sources of variability in mean weight, quantifies uncertainty in estimates, pools information to better inform strata with small sample sizes, predicts average weight for unsampled strata, and does not require data replication. We examine the effect of sample size on model-based estimates, and recommend propagation of uncertainty in average catch into estimates of recreational catch in weight.

Automated summary

Estimates of recreational catch are crucial for fishery management plans, especially when average fish weight estimates are imprecise. Currently, challenges with average weight estimation on the U.S. West Coast are addressed through data replication algorithms, which differ among states and are not suitable for variance estimation. This study presents a model-based framework for more accurate average fish weight estimation and highlights the importance of incorporating uncertainty into catch estimates.