Testing the effect of square mesh window position on Atlantic cod (Gadus morhua) size selectivity in a trawl codend

Atlantic cod ( Gadus morhua ) is one of the most economically important species for commercial fisheries in the Northern Atlantic, but the stocks have been declining in many areas for decades. Few size selectivity studies on cod in directed trawl fisheries for gadoid species have been conducted in the Greater North Sea ecoregion, where cod is highly economically important. Improvements in the size selectivity of cod trawls are essential to improve the exploitation pattern because there is a wide range of challenges. Square mesh net panels are the most used selective device to improve cod selectivity, henceforth referred to as windows (inserted square mesh net panels). The positioning of the window is a main controllable influential factor on cod escape. This study tests the effect of square mesh windows (inserted square mesh net panels) position on cod absolute size selectivity conducted in directed gadoid fisheries in the North Sea and Skagerrak. Four identical 120 mm standard diamond mesh codends (henceforth referred to as a standard codend), designed as those that are commercially used, with a nominal inside mesh size of 120 mm were constructed. Windows with square meshes and a nominal mesh size of 180 mm were inserted in three codends in positions located 0ꟷ3 m, 3ꟷ6 m, and 6ꟷ9 m from the end of the codend and are all allowed to be used by the legislation. The four codends were fished pairwise in a twin trawl rig using the covered codend technique with a total of 20 paired tows (40 hauls). To identify escape proportion during towing separately from haul-back escape, an acoustically released sampler unit (named Minisampler) was attached to each codend cover. The 50 % retention length (L50) and selection range (SR) were estimated for all codends using a generalized additive modelling approach to fit monotone increasing curves. The L50 values of the codends with windows were all statistically significantly higher and around double that of the standard codend. The selectivity curves of the three codends with windows were very similarly shaped and not statistically significantly different. The Minisampler was successfully released in only some hauls, and only four or five hauls for each codend were included in the analysis, giving wide 95 % confidence limits. In the standard codend, only 21 % of the escape of cod < 35 cm occurred during towing. The escape of cod < 35 cm was higher in the codends with windows but still with a marked proportion not escaping during towing (39ꟷ64 %, <35 cm). A window was an efficient way to increase L50, and the window mesh size can be used to control the size of fish allowed to escape.