536 



Fishery Bulletin 104(4) 



A 41/150 codend B 41/100 codend 



45/150 codend 



D 



45/100 codend 



rL Square-mesh codend 



35 N 



Approx. 

 2.4 m 



20 N 



100 T 



150T 



144 B 



1 4-mm twisted 

 30-mm mesh 

 hung on the bar 



90 B 



3-mm o braided 

 41 -mm mesh 



3-mm o braided 

 45-mm mesh 



F D, 



lamond codend 



G 90-CSMP codend 



H 75-CSMP codend 



1 75-panel codend 



T 



1.1 m 



1 



100T 



150 T 



4-mm o braided 

 90-mm mesh - 

 hung on the bar 



Figure 3 



Schematic representation of the codends and bycatch reduction devices tested during the experiments. The 41/150 codend 

 is the conventional design used in the fishery. N = normals; T = transversals, B = bars. 



mesh codend. Experiment 3 (26 August-6 September 

 2002, 10 days, three replicate hauls/day) compared the 

 41/100 and square-mesh codends (the most appropri- 

 ate non-BRD codends; see Results section), with and 

 without their respective 75-CSMP and 75-panel BRDs 

 installed. 



In each experiment, the daily position and order of 

 the codends being tested were determined randomly 

 and they were alternately used in normal commercial 

 hauls of 30-min duration between 0700 and 1500 hours. 

 Depending on the number of codends being tested, we 

 attempted two or three balanced replicates of each 

 treatment on each day. Haul location was determined 

 by the skipper and varied between the two squid trawl 

 grounds in Broken Bay (Fig. 1) according to the pres- 

 ence of jellyfish, Catostylus sp., which reduced the ef- 

 ficiency of the trawl. Data collected from each replicate 

 haul included the numbers and weights of all taxa and 

 the sizes of commercially and recreationally important 

 species. When large numbers (several hundred) of any 

 species were caught, a randomly selected subsample 

 (approximately 25-50% of the total) was measured. 

 Fish were measured to the nearest 0.5 cm fork length; 

 carapace lengths of crustaceans were measured to the 

 nearest 0.1 cm. 



Results 



Bycatch and target squid 



The primary hypothesis for this study was that the 

 devices designed to reduce bycatch would cause changes 

 in catches, by reducing bycatch but causing no important 

 reduction in the catches of squid. This hypothesis was 

 tested by using univariate statistical methods applied 

 to the results from all three experiments. In Experi- 

 ment 1, the diamond and 90-CSMP codends caught 

 less bycatch (about 74% less on average) than the other 

 types of gear (Table 1). About 50% of the bycatch were 

 discards, including teleost fish such as tailor ^Pomatonus 

 saltatrix) and mulloway iArgyrosomus japoniciis). There 

 were, however, no significant reductions in catches of 

 squid, although catches were smaller with the diamond 

 codend. As a result, later experiments were mostly done 

 with larger mesh sizes. There were no effects of altering 

 other components of the gear (circumference or mesh- 

 size of the codend). Catches of squid were significantly 

 different on the two days (Table lA), but there was no 

 interaction with the type of codend; therefore differences 

 among types of gear were consistent. Sources of varia- 

 tion that were determined to be nonsignificant at P>0.25 



