ESTIMATION OF FISH MORTALITY-RATES FROM TAG RECOVERIES USING PARTIAL LIKELIHOOD

This paper considers the partial likelihood procedure for estimating f ishing and natural mortality rates from tag recoveries. We consider th e likelihood function of the multinomial model. and two models of mort ality rates with a change-point against constant mortality are introdu ced. Explicit estimators of fishing mortality rates are derived. The a symptotic variances of total mortality rates and fishing mortality rat es are also obtained explicitly, though a simple numerical solution is needed for estimating the total mortality rates. The procedure was ap plied to recovery data of red seabream Pagrus major in the Seto Inland Sea. The partial likelihood worked in the same way as the full one an d was effective. It took about only four minutes to obtain estimates u sing partial likelihood, but it took about eight hours from the approa ch of full likelihood. Partial likelihood is powerful for analysis und er the multinomial model.