Analysis of Catch/Effort Data: A Comparison of Three Methods

Abstract

The models of Schnute and Deriso are compared with a simple autoregressive model (SA model). Comparisons are made using both stepwise and omnibus prediction. In the latter case predictions are made for at least 3 or 10 yr ahead. The SA model is shown to be consistently better at predicting both catch and catch-per-unit effort than either of the other two models. Furthermore, the percentage error is frequently sufficiently small to make the predictions useful for stock assessment. It is shown that in practice the Schnute and the Deriso models are variants of the SA model. It is suggested that catch and effort data cannot be used to establish equilibrium catches. The reasons for this are that unique equilibria probably do not exist and the catch/effort data are generally too unreliable to detect anything except major fluctuations in population size. It is suggested that catch-and-effort data can be used by adopting strategies designed to track the population fluctuations.