Bayesian state-space modelling of the De Lury depletion model: strengths and limitations of the method, and application to the Moroccan octopus fishery

The strengths and limitations of a Bayesian state-space modelling framework are investigated for a De Lury depletion model that accommodates two recruitment pulses per year. The framework was applied to the Moroccan fishery for common octopus (Octopus vulgaris) between 1982 and 2002. To allow identifiability, natural mortality (M ) and the recruitment rhythm were fixed, and the variance of both process and observation errors were assumed to be equal. A simulation–estimation (SE) approach was derived to test the performance of the method. If the data showed responses to harvest, the estimates of the most important figures, i.e. the initial abundance and the second recruitment pulse, were accurate, with relatively small bias. Results confirm that greater depletion yields smaller bias and uncertainty and that inferences are sensitive to the mis-specification of M. The 21 depletion series in the Moroccan dataset were jointly treated in a hierarchical model including random walk to capture the systematic fluctuations in estimates of catchability and initial abundance. The model provides estimates of the annual recruitment and monthly octopus population size. The recruitment estimates could be used to investigate the link between recruitment variability and the coastal North African upwelling regime to improve understanding of the dynamics and management of octopus stocks.