A method for calculating numerical estimates of gear selectivity curve

Abstract

The present paper proposes a method for estimating the numerical curve of gear selectivity without the assumption of function of selectivity curve and estimating confidence intervals, using data of catch per unit effort at length for several nets of different mesh sizes obtained from fishing experiments. The paper assumes such geometric similarity as the selectivity is the same if the rate of fish length to mesh size is the same. The basic idea is due to a technique of numerical integration that uses the derivatives at various points. The natural logarithm of selectivity is expressed by an approximation equation using its cumulated derivatives. The values of derivatives at various lengths are calculated from the data. The confidence intervals are made from the bootstrap samples. The method was applied to the data of catch at length by mesh size of gill net for pink salmon Oncorhynchus gorbuscha. Data collection, modification and issues of the method are discussed.