Evaluating the Orcs Working Group Approach for Data-Poor Catch Estimation Using the RAM Legacy Stock Assessment Database

In response to the 2006 reauthorization of the Magnuson-Stevens Act, the National Marine Fisheries Service established a requirement to set acceptable biological catch limits for all stocks, including those for which reliable catch data are the only information available. The “Only Reliable Catch Stocks” (ORCS) Working Group approach for the data-poor estimation of acceptable biological catch is being increasingly used but has yet to be properly evaluated. We evaluated the ORCS method using data-rich stocks from the RAM Legacy Stock Assessment Database. First, we identified stock assessments considered reliable by knowledgeable experts and scored a subset of these assessments according to the “Table of Attributes” (TOA) criterion defined in the ORCS method. We used this training dataset to assess the ability of each TOA criterion to predict stock status using boosted regression trees, which identified threshold values that most accurately classify stocks into vulnerability categories. We applied this refined approach to the subset of data-rich stocks reserved for testing classification accuracy and compared the estimated status to the known status for each stock. We will also evaluate the consistency of TOA scores from different stakeholders to identify criteria that will be widely seen as objective.