A peaked logistic-based selection curve

Abstract

Length-based selection curves define the relative catchability of fish to specific types of fishing gear, with catchability often highest at intermediate fish lengths. Distributions such as the normal, lognormal, or gamma are often used to define “peaked” selection curves, but these have limited capabilities to describe strongly asymmetric selection relationships, such as those sometimes observed for hooks or gillnets. Another, more flexible, peaked selection curve is proposed, which is derived by combining multiple logistic distributions. While the logistic distribution is frequently used to describe monotonic selection curves, incorporating multiple logistic equations (that describe either the increasing or decreasing catchability) can define a large range of asymmetric peaked selection curves. This “peak-logistic” curve also allows nonzero asymptotic selection for the largest size classes, which may be the selection occurring in some hook-and-line fisheries. We demonstrate examples of selection in hook, haul net, and mixed hook fisheries, for which the peak-logistic curve is more appropriate than comparative lognormal and binormal selection curves. We also promote an alternative to the peak-logistic: the two-sided normal curve.