Thomas E. Helser, Wildlife and Fisheries, College of Agriculture and Forestry, West Virginia University, PO Box 6125, Morgantown, WV 26506-6125; James P. Geaghan, Department of Experimental Statistics, Louisiana State University, Baton Rouge, LA 70803; and Richard E. Condrey, Coastal Fisheries Institute, Wetland Resources Building, Louisiana State University, Baton Rouge, LA 70803

Abstract. Gill nets are widely used as research tools to sample fish populations and as commercial fishing gears. Size selectivity of gill nets must be estimated to correct for sampling bias and to manage the commercial fishery. While myriads of methods have been proposed to estimate the selectivity of sill nets, few have been developed within a statistical framework. We present a method of estimating the selectivity of experimental gill nets in which type A and type B curves are solved simultaneously as a response surface using nonlinear regression. The modeling approach provides a general statistical framework for estimating selectivity parameters, evaluating different functional forms of the selectivity model, and testing for differences between models. We applied this approach to the gill net catches of Louisiana spotted seatrout in a five-panel experimental net from data collected from 1988-1995. The selectivity of the experimental gill nets for female and male spotted seatrout could be described by a common response surface that was based on a 4-parameter normal probability density function (r² = 0.95). By estimating type A and type B curves simultaneously, the model explained 74% more variation in the data than compared to methods which estimate type B curves for each size-class individually. Statistical comparisons of annual response surfaces were not significantly different (p>0.05) and suggest that the estimation approach was insensitive to annual variation in population size composition.