The functional form of hook and gillnet selection curves cannot be determined from comparative catch data alone

Catch data from simultaneous fishing with two sizes of hook were analysed to determine plausible shapes for hook selection curves. The estimation problem suffers from ambiguity of fit because the lack of nonselective catch data allows an infinite number of selection curve shapes to give precisely the same fit to the catch data. Restricting consideration to just a few possible functional forms (e.g., normal, gamma, lognormal, and logistic) for the selection curve does not eliminate the problem and it is shown that two very different selection models, one using normal selection curves and the other using gamma selection curves, give precisely the same fit. Moreover, allowing for unequal fishing efficiencies of the two hooks permits identical fits to arise from vastly different selection curves of the same functional form. It is seen that the estimated length of maximum retention (of unimodal selection curves) is quite insensitive to the shape of the curve, but is extremely sensitive to relative fishing efficiencies. The same phenomena exist in many gillnet selectivity studies because, although several different mesh sizes are typically deployed, the selection curves are often fitted using the catch data from pairs of gillnets of adjacent size. Further research is required to determine whether it is preferable to use methods of analysis that fit simultaneously to all sizes of gear deployed.