Scandol et al : Experiments to reduce bycatch in an estuanne squid trawl fishery 



539 



among types of gear disappeared. Whitebait are not an 

 important concern for management in this fishery and 

 results like these were not obtained in the other experi- 

 ments; therefore no further consideration was given to 

 this matter. Results from this study indicate that the 

 composition of assemblages of bycatch was not altered 

 by combinations of experimental gear and there were no 

 detectable influences on particular species. 



These multivariate results indicated that, with the 

 exception of whitebait, none of the secondary hypotheses 

 (2A, 2B, or 2C) could be rejected. These experiments 

 did, however, have limited power to detected small ef- 

 fects because of the large variability of the catches and 

 the relatively limited replication completed. 



Discussion 



The results demonstrated that the general concepts used 

 to improve species selection in fish and shrimp trawls have 

 similar application for squid. Nevertheless, the specific 

 designs of modifications require careful consideration, 

 according to several fishery-related factors. One of the 

 starting points for reducing bycatch in all towed gears 

 involves examining simple changes to the size and shape 

 of mesh in the codend, because this is where most of the 

 selection is believed to occur (Wileman et al., 1996). 

 Broadhurst and Kennelly (1996) and Broadhurst et 



al. (2004) showed that codends made from thick twine 

 and with large posterior circumferences, like those 

 used conventionally by squid trawlers in Broken Bay, 

 have very narrow lateral mesh openings and have poor 

 selectivity. Three of the simplest methods for improving 

 selection in codends involve 1) reducing the posterior 

 circumference (e.g., Broadhurst and Kennelly [1996]), 2) 

 increasing the mesh size (e.g.. Reeves et al. [1992]), or 

 31 orienting meshes, typically orienting 60-90% of the 

 size of the existing diamond-shaped mesh on the bar so 

 that they are square shaped (e.g., Thorsteinsson [1992]; 

 Broadhurst et al. [2004]). 



These three modifications were tested during the 

 present study but, with the exception of the square- 

 mesh codend, none were demonstrated to significantly 

 improve species selection or influence the sizes of indi- 

 viduals caught. The considerable temporal variability 

 in catches and a concomitant lack of power to detect 

 small effects may partially account for the lack of dif- 

 ferences detected, but a more likely explanation is that 

 the magnitude of changes to mesh openings in relation 

 to the sizes of target and bycatch species was insuf- 

 ficient. Given that some small fish escaped from the 

 square-mesh codend, future work would probably benefit 

 from further examination of increasing codend mesh 

 openings as a means for reducing bycatch. 



In contrast to subtle modifications to codend mesh- 

 es that were tested, it was demonstrated that BRDs 



