The evolution, current practices, and future opportunities of retrospective analysis in stock assessment

Retrospective patterns are consistent directional changes in terminal year estimates as new data are added and remain one of the most persistent challenges in fishery stock assessment. This paper provides a “retrospective of retrospective analysis,” tracing the evolution of the practice from its early descriptions in the 1980s to its current status as a standard global diagnostic. Using the Georges Bank yellowtail flounder as a case study, I illustrate how strong retrospective patterns can lead to systematic reductions in the perception of biomass and the setting of unsustainable catch quotas, ultimately undermining stakeholder trust in the scientific process. This paper explores the technical shift from virtual population analysis to modern state-space models, such as the Woods Hole Assessment Model, with this technical shift reducing, though not eliminating, retrospective patterns through the use of random effects and environmental covariates. I evaluate current quantification metrics, specifically the widely adopted Mohn’s rho, while discussing the potential advantages of alternative measures. Finally, the paper identifies remaining challenges, including the difficulty of isolating specific causal mechanisms (e.g. missing catch versus changes in natural mortality) and the risks of incorrect model fixes. I conclude by proposing future research directions, emphasizing the role of closed-loop simulations in developing management strategies that are robust to retrospective instability.