Estimation of gillnet and hook selectivity using log-linear models

Abstract

The log-linear modelling capabilities of existing statistical software permit a generalization of Holt's indirect estimation method by allowing several different selection curve models to be fitted to catch data from an arbitrary number of mesh sizes. This also facilitates the use of formal inferential procedures such as model selection, assessment of relative fishing powers, estimation of standard errors, and permits inclusion of knowledge regarding the shape of the population length distribution. This methodology is equally applicable to estimation of hook selectivity.