Gillnet selectivity for North Sea Atlantic cod (Gadus morhua): model ambiguity and data quality are related

Gillnet selectivity curves for North Sea Atlantic cod (Gadus morhua) were f itted to catch data obtained with six different mesh sizes. The selectivity curves investigated included frequently used selectivity models following the normal, lognormal, and gamma distributions. Another group of selectivit y models that take the method of capture (gilled, maxillae, or "randomly" e nmeshed) into consideration was also included. The best description of the selection data was found for the latter models. Therefore, the capture proc esses and girth measurements should be recorded as a matter of routine and such data used when constructing and evaluating gillnet selectivity models. The shape of the selectivity curve for those size intervals where there we re satisfactory catch information was well defined, while the selection cur ve was ambiguous outside the interval with adequate data. Ambiguities in th e shape of the selectivity curve can be diminished by choosing an appropria te range in mesh sizes and ensuring that the mesh sizes match the size dist ribution of the population fished. It is furthermore suggested that the est imated length distribution of the fish encountering the nets be robust to m isspecification of the selectivity model.