Use of generalized linear models in analyzing the catch and effort data on the western king prawn Penaeus latisulcatus Kishinouye in the Gulf St. Vincent, Australia

摘要

The western king prawn Penaeus latisulcatus occurs in coastal waters off Western Australia, the Northern Territory, Queensland, New South Wales, and South Australia. In South Australia, it supports a commercial trawl fishery with an annual landed value of about AU$ 35 million. In this paper, I derive generalized linear models from an observational model of a sex-, age- and time-dependent model for a single prawn population, provide them with a firm theoretical basis, and make explicit some of their assumptions. Following an exploratory analysis by use of a generalized additive model, I use generalized linear models to analyze the commercial catch and effort data for the changes in its stock size with time, in fishing efficiency among vessels, in catch with time of day, fishing effort and area, and the effects of searching on catch. The results obtained by assuming that the catch follows independent log-gamma, log-normal, log-negative binomial, or log-Poisson distribution were similar. The relative exploitable biomass in the population immediately before the fishery was closed in 1991 was 31% of that when it was resumed in December 1993, remained at about the same level of December 1993 until 30 June 1995, slightly declined from 1 July 1995 to 30 June 1997, and was 81% of the level of December 1993 from I January 1998 onward. The relative instantaneous rate of fishing mortality of prawns of a particular sex and age was largely unchanged, but became less variable with time. Fishing efficiency was similar among all vessels. Normal fishing shots (using three nets) produced about three times the catch of searching shots (using only one net). Fishers learned during a sequence of shots, for prawn catch initially increased with shot number and then leveled off. A significant variation in catch with the time of day was detected, with the maximum catch occurring at 22 h. Finally, catch did not change proportionally with fishing effort; a 1% of change in effort led only to a nearly 0.9% of change in catch, so that raw catch per unit of effort decreased with fishing effort and is a biased index of the size of a prawn population. This work suggests that the current level of fishing effort cannot sustain the exploitable biomass of the population at the level of December 1993.