Abstract
Selectivity results for cod, using a codend with Exit Windows panels and T90 codend, are in line with previous experiments conducted with pelagic trawls and bottom trawls, and considered to be very good from the management point of view. The results showed that both codends effectively selected small fish at very high density of fish, and both provided a stable selection regardless catch size. Sorting grid selection (Sort-V) did not give as good results at high densities of fish as the codend with Exit Windows or the T90 codend did, and retained much more small fish. This is mainly because the grid had a much smaller sorting area than the other two selection devices, but also because the sorting grid systems provoked a reduction of the water flow inside the grid section. Consequently, a lot of fish accumulates in front and behind of the sorting grid and did not fall back to the codend. In addition, the sorting grid influenced the operation of the catch sensors, and these did not provide an accurate picture of the real catch during trawling. This presentation also includes work done in a flume tank in which we assessed the reasons of malfunctioning of sorting grids. The main findings are that the lifting the panel is the element that causes the greatest water flow reduction and that this should be removed or modified considerably. A new 4-panel grid sections has several advantages over the existing two-panel grid sections: it has a better water flow, it is geometrically more stable, and it have larger cross-sectional area.