Application of Bayesian biomass dynamic models to neritic tuna species in the Indian Ocean

Stock assessment has been conducted for three neritic species, Kawakawa, longtail tuna and narrow-barred
Spanish mackerel, in the Indian Ocean based on the biomass dynamic models. Two different approaches were
applied; 1) state-space biomass dynamics models using both of catch series and standardized abundance
index (here, the Iranian coastal gillnet CPUE, annually averaged) and 2) Catch only analyses using the Cmsy
method. In the second analyses, we focused on the sensitivity/robustness of the results to i) the assumption
of prior distributions for r, K, initial and final depletions and ii) the assumption of production functions. For all
the analyses, we employed Bayesian methods to estimate parameters and evaluate associated estimation
uncertainty. Non-informative priors were used, and posterior samples were generated using a Markov chain
Monte Carlo (MCMC) method or acceptance/rejection sampling. Our overall conclusions were a) analyses
with CPUE series looked too optimistic for all the species, which was driven by a recent increasing trend of
CPUE; b) Cmsy method provided with robust results to some extent even when the prior assumptions were
moderately changed; c) however the result of the Cmsy method seemed sensitive to the production functions,
and therefore there should be careful diagnostic examinations using retrospective analyses and hindcasting
approaches.