Estimation of growth within Stock Synthesis models: Management implications when using length-composition data

Abstract

Growth modeling is an important component of contemporary fisheries stock assessment, and is typically conducted external to assessment models. However, direct growth estimates may be problematic for some species because of difficulty in aging older individuals using hard parts. Stock assessment results and the resulting management advice can be sensitive to growth specification, particularly when fitting to length-composition data. This study evaluates the influences of mean length-at-age (mean length) and variation in length-at-age (standard deviation; SD) in relation to length composition on management advice and determines if mean length and SD can be estimated inside stock assessment models. The Stock Synthesis assessment for bigeye tuna (Thunnus obesus) in the eastern Pacific Ocean, demonstrates that management quantities were sensitive to the assumptions of mean length and the SD of old fish. Using simulation analysis, we found that mean lengths of both old fish and young fish, and the SD of young fish, can be reliably estimated, and that the estimates were robust to the misspecification of length-based longline selectivity curve (asymptotic or dome-shaped). The SD of old fish and the growth coefficient were less reliably estimated. This study also demonstrates that equilibrium yield is robust to uncertainty in growth parameters when management is based on setting fishing mortality equal to FMSY.