Comparison of Statistical Models for Expanded DeLury's Method

Abstract

Nine statistical models with different probability distributions are introduced for the expanded DeLury's method with variable catchability coefficients according to such environmental factors as water temperature, lunar cycle, and the intensity of ocean waves. After being applied to the catch-effort data of spiny lobster gillnet fishery, statistical models are compared and discussed with AIC and the confidence intervals of the estimators. The AIC values indicate that the negative binomial model is the optimum model. This result adequately reflects the distributional nature of spiny lobster in the field which shows some aggregation, coupled with other fishing characteristics. Though the confidence intervals of the estimated parameters were rather broad in the negative binomial model, those calculated through simultaneous analysis using three-year data were narrower than those calculated using single-year data. © 1994, The Japanese Society of Fisheries Science. All rights reserved.