抄録
To estimate parameters of a variety of DeLury's models by maximum likelihood method, the probability distribution of catch should be specified. But this is not a straightforward procedure because it is difficult to determine which distribution is a true representation of the catch variation. Taylor's Power Law covers a wide range of distributions from regular through random to aggregated. A generalized DeLury's method based on the law was proposed and applied to the catch-effort data of the purse seine fishery for a pelagic fish species in southern Lake Tanganyika. Dynamics models with variable catchability and natural mortality were given for the fish population. Binomial and negative binomial distributions were also considered for comparison purposes. Estimates of parameters indicated an aggregated distribution of the fish. Among the three models, Taylor's Power Law model was the optimal with the lowest AIC and the best fit to the data, which suggested its wide applicability.
