Broadhurst et aL: Use and success of square-mesh codends in reducing bycatch and improving size-selectivity of prawns 



439 



^ 2000 

 f 1500 



= 1000 



c 



(0 



<u 500 



Oi 



oi 

 3 

 c 

 as 



<D 



25 

 20 



15 



10 



5 







cantly reduced the numbers of prawns 

 caught (means reduced by S.T^i and 

 12.1%, respectively), with no significant 

 reduction in weights ( although the means 

 were reduced by 1.87c and 3.7%, Fig. 4, 

 A-B; Table 1). The weights of total dis- 

 carded bycatch and discarded noncom- 

 mercial bycatch were significantly re- 

 duced by 67.7% and 69.0%, respectively, 

 in the composite-square-2 codend, whereas 

 the composite-square-3 codend reduced 

 these amounts by 58.4% and 54.2%. re- 

 spectively (Fig. 4. C-D; Table 1). Both 

 composite square-mesh codends signifi- 

 cantly reduced the numbers and weights 

 of sand trevally (by between 68.0% and 

 76.6%), red mullet, (by between 87.2% 

 and 96.9%), school whiting (by between 

 90.6% and 96.9%), and southern sand 

 flathead (by between 57.0% and 84.9%) 

 (Fig. 4, E-H; Table 1). The composite- 

 square-2 codend significantly reduced the 

 mean number and weight of Degen's 

 leatherjacket caught by 31.8%. and 23.4%, 

 respectively, and although not signifi- 

 cant, these amounts were reduced by 

 24.6% and 20.8%, respectively, in the 

 composite-square-3 codend (Fig. 41: Table 

 1 ). There was also a significant reduction 

 in the numbers of retained southern 

 calamari, by 33.9% in the composite- 

 square-3 codend (Fig. 4J; Table 1). 



Two-sample Kolmogorov-Smimov tests 

 comparing the size-frequency distribu- 

 tions for fish measured from the control 

 and composite square-mesh codends 

 showed significant differences in the rela- 

 tive size compositions of sand trevally 

 retained by both composite square-mesh 

 codends and red mullet retained by the 

 composite-square-2 codend (the control 

 codend retained proportionally more 

 larger fish. Fig. 5, A-B). No significant 

 differences were detected in the size com- 

 positions of Degen's leather jacket re- 

 tained in the composite square-mesh and 

 control codends (Fig. 5C). There were 

 insufficient data to analyze size-fre- 

 quency distributions of school whiting 

 and southern sand flathead. 



Analysis of size-selectivity of prawns 



The raw data, observed selectivity, and calculated 

 logit values for prawns captured in the codends are 



-| ^ O) 60 -| o 



kai iliJi 



25 



20 



15 



10 



5 







50 

 40 

 30 

 20 



10 

 



liLiI likLil 



LLj :.LLJ 



(UCVJ 



P 



E cr 



O « 

 O 



g 



c 



s 



dlCO 



a§ § 



ECT O 

 o « 

 O 



Figure 4 



Differences in mean catches (per 45-min tow ± SE) between the compos- 

 ite square-mesh codends and control codend: the (A) numbers and (B) 

 weights of king prawns iPenaeus latisulcatus); the weights of (C) total 

 discarded bycatch and iD) discarded noncommercial bycatch: and the 

 numbers of lE) sand trevally iPseudocaranx wrighti), (F) red mullet 

 iUpeneichthysporosiis). (Gi school whiting (SjV/ago 6asscfis;s), (H) south- 

 ern sand flathead {Platycephalus bassensis). ill Degen's leatherjacket 

 iThamnaconus degeni), and (J) southern calamari iSepioteuthis austro- 

 lis). ** = significant (P<0.01): * = significant lP<0.05). 



listed in Table 2. Figure 6 shows the corresponding 

 logistic length selection curves derived for each com- 

 posite square-mesh codend. Model deviance indicated 

 sufficient goodness of fit for both of these selection 

 curves (Table 3; Fig. 7) although some size classes 



