A comparison of 4 primary age-structured stock assessment models used in the United States


Bai Li, Kyle W. Shertzer, Patrick D. Lynch, James N. Ianelli, Christopher M. Legault, Erik H. Williams, Richard D. Methot Jr., Elizabeth N. Brooks, Jonathan J. Deroba, Aaron M. Berger, Skyler R. Sagarese, Jon K. T. Brodziak
Ian G. Taylor, Melissa A. Karp, Chantel R. Wetzel, and Matthew Supernaw
Cover date: 
Supplementary material: 
Supplementary material
Supplementary table 1
Supplementary table 2
Supplementary table 3
Supplementary figures 1 and 2
Supplementary figures 3-8
Published online 30 August 2021

The National Marine Fisheries Service conducts fishery stock assessments to provide the best scientific information available for the U.S. regional fishery management councils. The assessment models applied in the United States are often region specific, although the models share similar mathematical and statistical attributes. However, comprehensive comparison studies identifying similarities and differences among these assessment models remain scarce. We developed a multi-model comparison framework to evaluate the reliability of 4 age-structured assessment models that are commonly used in the United States: the Assessment Model for Alaska, the Age Structured Assessment Program, the Beaufort Assessment Model, and Stock Synthesis. When applied to simulated data, all 4 models produced reliable estimates of assessment quantities of interest, such as fishing mortality, spawning biomass, recruitment, and biological reference points. Although there were differences among models in the calculation of the initial population numbers at age and in the bias adjustment of recruitment, their effects on model outputs were minor when estimation models were configured similarly. In addition, we provide guidelines for converting unfished recruitment and steepness between 2 methods of bias adjustment. We recommend that next-generation stock assessment models include recruitment bias adjustment and that more research be conducted to provide guidelines for which methods might be preferred under which situations.