Gill-net selectivity off the Portuguese western coast

Fishing with experimental gill-nets was carried out during 1994 and 1995 off the western coast of Portugal, using mesh sizes between 40 and 90 mm. Most sets were concentrated in the region between Setúbal and Lisbon in shallow waters up to 100 m, with about 1/4 being carried out at greater depths. A total of 88 species were captured over the 2 years, of which the majority had small or no commercial value. The most important commercial species (hake, Merluccius merluccius, pouting, Trisopterus luscus, axillary seabream, Pagellus acarne, red mullet, Mullus surmuletus, and horse mackerel, Trachurus trachurus) accounted for about 12% of the catch weight in the 40 mm mesh panels, between 32 and 44% for mesh panels ranging between 60 and 80 mm, and 70% in the 90 mm mesh panels. The estimation of selectivity was carried out for the latter five species and also for the dogfish, Scyliorhinus canicula, wedge sole, Dicologlossa cuneata, and spotted flounder, Citharus linguatula, applying the SELECT method to the data pooled over both years. In the case of hake, pouting and dogfish, sufficient data were available to estimate selectivity on a yearly basis. Four different uni-modal and a bi-modal (bi-normal) models were fitted, with the bi-normal always providing the best fit as given by the smallest value of the ratio deviance/degrees-of-freedom. The exception was the wedge sole where the bi-normal model did not converge and the gamma was the unimodal model providing the best fit. The 40 mm mesh panels mostly retained species of low or no commercial value and undersized hake and pouting. Mesh size panels of 60, 70 and 80 mm proved to be the most efficient in catching the commercially valuable species over the length range available, with virtually no catch of undersized fish. The 90 mm panels retained mainly valuable species, but the overall catches were poor, with the exception of the axillary seabream. These results highlight the difficulties of managing multispecies fisheries based only on mesh size, since the optimal mesh varies considerably among the target species.