Estimation of gillnet selectivity curve by maximum likelihood method

ABSTRACT: A maximum likelihood method of estimating gillnet selectivity when data are obtained by gillnet fleets consisting of several nets of differing mesh size is presented in this paper. The SELECT model is expanded by application of the relative length (i.e. the ratio of fish length to mesh size) to obtain a master curve of gillnet selectivity. Four kinds of functional model, normal, lognormal, skew-normal and bi-normal are fitted to the data. In addition, two cases in which the relative fishing intensity is either estimated or fixed by catch effort are compared. The bi-normal model has the lower model deviance regardless of whether the relative fishing intensity is estimated or not. The estimation of relative fishing intensity by catch effort is also examined in which the estimates of the parameter of the SELECT model are compared with the catch effort as determined by the number of nets of each mesh size used. For the bi-normal model these quantities compare well. Thus, it is concluded that this method gives reliable estimates even if data for each mesh size is obtained with different catch efforts.