A fish stock assessment model using survey data when estimates of catch are unreliable

Abstract

Listen

Translate article

Methods of assessment that depend upon commercial catch data can be undermined by misreporting or where parts of the catch, such as discards, are not accounted for. An age-structured model that makes use of survey data alone, and avoids this problem, is developed within a Bayesian framework so that routine stock summary statistics such as fishing mortality, recruitment and spawning stock biomass can be estimated with associated levels of uncertainty. It is also possible to estimate catch on a relative scale which can be compared to reported catches. The model is applied to West of Scotland haddock (Melanogrammus aeglefinus), a stock with suspected high catch misreporting. Stock trends derived from the model are consistent with conventional assessments that use catch data during periods of low misreporting. Estimated proportions of fish at each age in the catch correspond closely with observed values. Model estimates of total catches suggest substantial misreporting in some years, though the precision of the estimates is very low. Revised estimates of natural mortality are obtained from the model that are higher than conventional values used for this stock. These new values are generally consistent with those obtained from multispecies predation modelling for the adjacent North Sea stock. The model provides many of the basic quantities used for management advice. It should not be regarded as a replacement for more comprehensive analyses, but an additional tool to explore available data when catch information is unreliable.