Bayesian integrated survey-based assessments: an example applied to North Sea herring (Clupea harengus) survey data

Abstract

Stock assessments that use only fisheries independent data have been developed from a number of different standpoints over the last 15 years. While the ability of such stock assessments to avoid the use of potentially compromised or hard to interpret commercial data or making certain assumptions that are common to traditional assessments has been established, little has been made of their potential for detecting and estimating complex mortality trends over time or their potential utility in survey-based management procedures. Using North Sea herring ( Clupea harengus ) data as an example, a Bayesian survey-based assessment method that is able to estimate all the key population variables is detailed. However, survival probability, and not fishing mortality that is conditional on natural mortality, is the key parameter. Reversible jump Markov chain Monte Carlo routines were developed to explore the range of ages over which survival separates into year and age effects (a common assumption in many stock assessments). Post hoc estimates of natural mortality suggest that changes over years and ages may have occurred in relativity to historic levels. The derivation of reference points based on survival probability and surplus biomass production are detailed as proxies for more common F-based reference points. The potential role for the outputs of such assessments in a management procedure sense is discussed.