Abstract
Separator trawls utilise the behaviour of certain fish species (e.g. cod and flatfish), which stay low when a trawl passes, to separate them from others (e.g. haddock and whiting), which tend to swim upwards. Separate upper and lower codends of different mesh sizes permit differential selectivity. This paper discusses statistical aspects of the design and analysis of two trials of the gear as a means of lowering the vulnerability of immature cod to whitefish trawling in the North Sea. The first trial took place off Whitby, north-east England, in March 1994; the second in the Moray Firth, east Scotland, in November 1994. Both used the FV ‘Heather Sprig’, a three warp trawler which could tow two trawls side by side. The separator trawl fitted with the codends to be tested could therefore be towed alongside another fitted with small mesh codends which permitted selectivity to be assessed. Nominal codend mesh sizes of 100, 120 and 140 mm were tested on the port or starboard side in randomised, balanced blocks of four hauls, the two larger meshes being tested only in the lower position where they have potential for conserving cod. Selectivity and vertical separation parameters were estimated with a modified logistic regression method; the results were modelled as functions of mesh size and other factors using least squares with weighting dependent on catch size. Statistical and practical advantages of using basic statistical methods and a standard experimental design are discussed. The effectiveness of vertical separation was confirmed. Catches of cod were disappointing, particularly in the larger mesh codends, but the indications were that only about half of the cod caught in a lower codend of 140 mm mesh would be below breeding age. The results lend support to the separator trawl as a viable alternative to the standard, single codend trawl which could continue to take commercial catches but with considerably reduced proportions of immature cod.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have