Comparative performance of data-poor CMSY and data-moderate SPiCT stock assessment methods when applied to data-rich, real-world stocks

Paul Bouch, Coílín Minto, Dave G. Reid

Research output: Contribution to journalArticlepeer-review

35 Citations (Scopus)

Abstract

All fish stocks should be managed sustainably, yet for the majority of stocks, data are often limited and different stock assessment methods are required. Two popular and widely used methods are Catch-MSY (CMSY) and Surplus Production Model in Continuous Time (SPiCT). We apply these methods to 17 data-rich stocks and compare the status estimates to the accepted International Council for the Exploration of the Sea (ICES) age-based assessments. Comparison statistics and receiver operator analysis showed that both methods often differed considerably from the ICES assessment, with CMSY showing a tendency to overestimate relative fishing mortality and underestimate relative stock biomass, whilst SPiCT showed the opposite. CMSY assessments were poor when the default depletion prior ranges differed from the ICES assessments, particularly towards the end of the time series, where some stocks showed signs of recovery. SPiCT assessments showed better correlation with the ICES assessment but often failed to correctly estimate the scale of either F/FMSY of B/BMSY, with the indices lacking the contrast to be informative about catchability and either the intrinsic growth rate or carrying capacity. Results highlight the importance of understanding model tendencies relative to data-rich approaches and warrant caution when adopting these models.

Original languageEnglish
Pages (from-to)264-276
Number of pages13
JournalICES Journal of Marine Science
Volume78
Issue number1
DOIs
Publication statusPublished - 1 Jan 2021

Keywords

  • benchmarking
  • data-limited stock assessment
  • management advice
  • model bias
  • surplus production models

Fingerprint

Dive into the research topics of 'Comparative performance of data-poor CMSY and data-moderate SPiCT stock assessment methods when applied to data-rich, real-world stocks'. Together they form a unique fingerprint.

Cite this