An Atlantic herring (Clupea harengus) size selection model for experimental gill nets used in the Sound (ICES Subdivision 23)

Size selection investigations were performed in the Sound with experimental , multipanel gill nets equipped with a broad range of mesh sizes targeting Atlantic herring (Clupea harengus). Each of the 20 experimental fishery sur veys covered the central Sound from Helsingor-Helsingborg (north) to Drogde n-Klagshamn (south) during the autumn, winter, and spring periods from 1993 to 1998. There was high variability in catch sizes both between and within surveys. The overall best fits, in terms of minimum total deviance across all sets, were achieved using a unimodal normal scale model. Almost all sur veys showed high between-set variation, which was accounted for when estima ting the mean selectivity curves using a random effects model. The selectiv ity curves fitted to individual sets were compared with those fitted to the combined data sets. It was demonstrated that analysing pooled data sets re sulted in overdispersion and bias in the selection parameters. Irregular ti me series of the estimates of the selectivity parameters in the investigati on period indicated negative autocorrelation with a seasonal pattern.