Rose vs. Rho: a comparison of two approaches to address retrospective patterns in stock assessments

Abstract

Abstract Two approaches to address retrospective patterns in stock assessments are compared. The Rose approach is an ensemble of models that all remove the retrospective pattern through changes in data, parameter values, or model assumptions. It is time intensive and can result in a wide range of historical abundance trends. The Rho approach modifies the terminal year estimates of a single model that exhibits a retrospective pattern. It is fast and easy to apply but results in a discontinuous time series. Neither approach identifies the source of the retrospective pattern. The pros and cons of these two approaches are compared in terms of catch advice and stock status using four examples with varying strength and direction of retrospective patterns. The choice of which approach to use could be based on time and expertise available to conduct and maintain an assessment, with Rose preferred if a lot of both are available while Rho preferred otherwise. If the Rho approach is used, managers should consider adjusting their control rule or risk buffer to account for the difference between Rose and Rho results shown here.