Analysis of trawl selectivity studies with an application to trouser trawls

Abstract

Data gathered from trouser trawl, split trawl, twin trawl, or alternate haul surveys do not conform to the assumptions required by conventional statistical methodology for analysing count data. Consequently, the application of conventional statistical methods may lead to incorrect results and conclusion. This paper presents a new statistical model that is appropriate for these types of selectivity data. This model was used to fit logistic selection curves to American plalce data obtained from trouser trawl studies using either a 155 mm diamond or square large-mesh codend. The model also estimated the split of fish into the two legs of the trawl (large-mesh codend and small-mesh codend) and these were found to be statistically significantly different from 50:50 in both the diamond and square mesh studies. The fitted selection curves remain valid because the unequal split was included in the model. The standard errors of the selection curve parameter estimates, the 25, 50 and 75% retention lengths, and the estimated split proportions are given.