Eliciting expert knowledge to inform stock status for data-limited stock assessments

Abstract

Data-limited fisheries are a major challenge for stock assessment analysts, as many traditional data-rich models cannot be implemented. Approaches based on stock reduction analysis offer simple ways to handle low data availability, but are particularly sensitive to assumptions on relative stock status (i.e., current biomass compared to unperturbed biomass). For the vast majority of data-limited stocks, stock status is unmeasured. The present study presents a method to elicit expert knowledge to inform stock status and a novel, user-friendly on-line application for expert elicitation. Expert opinions are compared to stock status derived from data-rich models. Here, it is evaluated how experts with different levels of experience in stock assessment performed relative to each other and with different qualities of data. Both “true” stock status and expert experience level were identified as significant factors accounting for the error in stock status elicitation. Relative stock status was the main driver of imprecision in the stock status prior (e.g., lower stock status had more imprecision in elicited stock status). Data availability and life-history information were not identified to be significant variables explaining imprecision in elicited stock status. All experts, regardless of their experience level, appeared to be risk neutral in the central tendency of stock status. Given the sensitivity to stock status misspecification for some popular data-limited methods, stock status can be usefully elicited from experts, but expert subjectivity and experience should be taken under consideration when applying those values.