Stock assessment framework for Scallop Fishing Areas 25, 26, and 27B : stock assessment models for SFAs 25A and 26A

Abstract

This research document focuses on the development of new stock assessment methodologies for the stocks in scallop fishing area (SFA) 25A (Sable Bank) and SFA 26A (Browns Bank North). A Bayesian state-space delay difference model (BSSM) has been used to assess the status of SFA 26A for over a decade, but no analytical models have ever been implemented for SFA 25A. A modified version of the BSSM currently used for the SFA 26A assessments is used to model the population dynamics for these two stocks. In addition, the population dynamics of these two stocks are modelled using a spatially explicit assessment model (SEAM); this model takes advantage of advances in computing power and the development of new statistical methods while retaining the same conceptual delay-difference population dynamics framework. Overall the two modelling frameworks provided broadly similar results, but there were differences in the productivity parameters between the models; these differences were more notable in SFA 25A. The exploitation rate in SFA 25A has been low throughout the timeframe covered by the model and the fishery appears to have had little impact on stock dynamics. In SFA 26A, there was evidence that the fishery impacted the stock dynamics, for this stock the fully-recruited biomass generally declined when the exploitation rate exceeded approximately 10%. For both stocks, the retrospective patterns indicated that fully-recruited biomass predictions were not strongly influenced by the inclusion of additional years of data. The analyses of the process error, residuals, and, for SEAM, the random fields, indicated that the process component of these models tended to overestimate fully-recruited biomass in recent years, but the models were largely able to compensate for this through their error structures. The prediction evaluation results indicated that default parameterizations for the one-year ahead fully-recruited biomass projections tended to over-estimate the realized biomass, but this bias could be substantially reduced by using methods which made different assumptions about the productivity of the stocks. Overall, the results indicated that, for both stocks, BSSM and SEAM provided reasonable approaches for modelling the stock dynamics given the available data and either of these approaches could be used to provide science advice. In SFA 25A, we recommend using BSSM to provide science advice with the no fully-recruited growth method used for projections. In SFA 26A, we recommend using SEAM to provide science advice with the previous year method used for projections.