Background. Roach, Rutilus rutilus, comprise a large proportion of the total catch in the Szczecin Lagoon, and are caught mainly as bycatch in gillnets targeting perch, Perca fluviatilis. The minimum landing length for roach and perch is 17 cm TL and the minimum mesh size is 30 mm bar length. However, little is known about the selectivity of roach, which may have implications for implementing effective regulations. Consequently, the aims of the study were to estimate roach selectivity curves and to compare the results with perch selectivity. Materials and methods. The study was based on the length frequency of catch from standardized fishing practice with commercial gillnets of mesh sizes ranging from 25 to 35 mm (bar length) and modelled using various generalized linear models. The best model was selected using deviance criteria. Results. The log-normal selectivity model provided the best fit, with spread relevant to mesh size under the assumption of equal fishing power. The estimated parameters were 18.7 cm and 1.5 (mode and spread of selectivity curve for the smallest mesh size). Conclusion. Although the most efficient mesh size for roach was 26 mm, its use in the Szczecin Lagoon is limited by impact on perch. All the studied mesh sizes prevented catches of undersized roach, while mesh sizes from 28 mm prevented catches of undersized perch (MLL = 17 cm TL). The authors concluded that the current 30 mm minimum mesh size is likely effective at preventing the harvest of undersized perch and roach.

roach, gillnet, selectivity, log-linear modelling, SELECT method