416 



Fishery Bulletin 94(3), 1996 



ent among nights in the experiment. No variables 

 displayed significant interactions among haulback 

 delay and nights (Table 2). 



One tailed, paired /-tests comparing data from the 

 control and 85-mm-long codends showed that the 85- 

 mm-long codend significantly reduced the weight of 

 bycatch and significantly increased the weight of 

 prawns during both a 10-15 s delay and no delay in 

 haulback (Fig. 3, A-B; Table 3). The 85-mm-long 

 codend significantly reduced the number and weight 

 of discarded red spot whiting compared with the con- 

 trol during the 10 to 15 s haulback delay (means re- 

 duced by 64% and 56%, respectively) (Fig. 3, D-C; 

 Table 3). There was no significant reduction in these 

 variables for this codend when there was no delay in 

 haulback, although the mean numbers and weights 

 of red spot whiting were reduced by 21% and 25%, 

 respectively (Fig. 3, D-C; Table 3). The weight of re- 

 tained cuttlefish was significantly decreased in the 

 85-mm-long codend when there was no delay in 

 haulback (Fig. 3E; Table 3). No other variables 

 showed any differences between the control and 85- 

 mm-long codends either with or without a delay in 

 haulback. Two-sample Kolmogorov-Smirnov tests 

 comparing size-frequency distributions for king 

 prawn and red spot whiting (Fig. 4) measured from 

 each sample showed no significant differences in the 

 relative size compositions between the 85-mm-long 

 and control codends during either a 10-15 s delay or 

 no delay in haulback. 



Aggregated data for the 85-mm-long codend 



Combining all the data for the 85-mm-long codend, 

 we found that this modification significantly reduced 



