442 



Fishery Bulletin 97(3), 1999 



D 



A0-, 

 30 

 20- 

 10- 







n. II II -II -II -I 



 Compostle square 2 n = 10 

 □ Control n = 212 



10 



15 



20 



25 



40 

 30- 

 20- 

 10- 

 



H Composite square 3 n = 7 

 ^ ContfOln= 194 



10 



n n IIJI II II ll^l II II ll-n n- 



15 



20 



25 



20 1 

 15 - 

 10 



5-1 







20 

 15- 

 10 - 

 5 - 



-MJ 

 10 



E 



M 



 Composite square 2 n = 19 

 □ Control n= 121 



 jIIbJ.. 



H Composite square 3 n = 22 

 □ Control n= 167 



ijilui 



III II n nt 



15 



20 



25 



Length (cm) 



30 



35 



40 



Figure 5 (continued) 



used in the posterior section of the codend and the 

 effects that this had on distribution of catch and 

 water flow. Because the posterior diamond-mesh sec- 

 tion in this codend was 10 meshes in length, 100 

 meshes in circumference, and attached to a square- 

 mesh section 62 bars in circumference (hanging ra- 

 tio of 0.36), as the catch increased it would have 

 spread laterally, increasing its surface area that was 

 incidental to the flow and, therefore, increasing the 

 displacement of water forwards (see Broadhurst and 

 Kennelly, 1996, 1999). The composite-square-3 

 codend, however, was designed to increase size-se- 

 lectivity of prawns and was tapered to a circumfer- 

 ence of 58 bars attached to a posterior diamond-mesh 

 section that was 2 meshes in length and only 58 

 meshes in circumference (hanging ratio of 0.57). Com- 

 bined with lastridge ropes, this configuration would 

 have forced a much smaller codend diameter than in 

 the composite-square-2 codend, restricting any lat- 

 eral distribution of catch. Any consequent reduction 

 in surface area that was incidental to the flow and 

 displacement of water forwards in this codend (which 



would limit any assistance to swimming fish) may 

 account for the nonsignificant reduction of Degen's 

 leather jackets and the poorer exclusion of small in- 

 dividuals comprising discarded noncommercial 

 bycatch (Fig. 4D). It should be noted, however, that 

 these results were based on only 11 tows and that 

 further experiments may be required to provide ad- 

 ditional evidence to either support or refute the hy- 

 pothesis discussed above. 



Although the mesh configuration and water flow 

 in the composite-square-2 codend may have contrib- 

 uted to the escape of slightly more fish, in terms of 

 optimizing sizes of prawns retained, the composite- 

 square-3 codend appeared to be a better design. Al- 

 though the 95% confidence limits about the selectiv- 

 ity parameters indicated no significant differences 

 between the two designs (Fig. 6; Table 3), there was 

 some evidence to suggest that the composite-square- 

 3 codend did appear to select slightly more commer- 

 cial-size prawns. For example, compared with the 

 control, this codend caught significantly fewer 

 prawns by number (difference between means of 



