444 



Fishery Bulletin 97(3), 1999 



Table 3 



Computed selectivity parameters for prawns (carapace length in mm) from the two composite square-mesh codends and deviance 

 values for logistic curve goodness-of-fit. Standard errors are given in parentheses, a, b = logistic parameters (Pope et al.. 1975). P 

 = split proportion from estimated split model (Millar and Walsh, 1992). 



Composite-square-2 codend 



Composite-square-3 codend 



b 

 P 



25'7f retention (L^j) 



50% retention (Lj^) 



75% retention (L,^) 



Selection range (SR) 



Deviance 



df 



P-value 



-7.803 

 0.257 

 0.544 



26.02(1.19) 

 30.28(1.19) 

 34.56(1.76) 



8.53(0.6) 

 32.48 

 39 



0.760 



12.1%) than did the composite-square-2 

 codend (difference between means of 8.1%), 

 and there was no significant reduction in 

 weights (Fig. 4; Table 1). 



These sHght differences in size-selectivity of 

 prawns may be explained in terms of the be- 

 havior of prawns in the codend and the differ- 

 ent configurations of mesh discussed above. 

 Previous studies have shown that the response 

 of prawns to stimuli from the trawl is mini- 

 mal and after initial contact with the leading 

 edge of the footrope, the prawns are quickly 

 forced into the rear of the codend (Lochhead, 

 1961;Watson, 1976). Their escape at this point 

 is primarily determined by their probability 

 of randomly encountering openings between 

 meshes that are large enough to pass through. 

 In the composite-square-2 codend, the hang- 

 ing ratio between the posterior diamond and 

 square-mesh sections would have restricted 

 the fractional diamond-mesh openings, limit- 

 ing the opportunity for escape. Further, as the 

 catch increased and spread laterally, this would have 

 widened the diameter of the posterior section, pro- 

 viding prawns with less opportunity of encountering 

 any open meshes in the sides of the codend. In con- 

 trast, we designed the composite-square-3 codend, 

 so that it would assume a smaller diameter and cir- 

 cumference during fishing, but with a greater area 

 of square mesh, thereby providing more opportunity 

 for prawns to randomly encounter the sides of the 

 codend and the open square meshes. 



The results obtained in this study showed that 

 composite square-mesh codends in penaeid prawn 

 trawls can improve size-selectivity of the targeted 



prawns while reducing large quantities of unwanted 

 fish. It is also apparent, that such modifications may 

 improve the overall efficiency of the trawl in terms 

 of increasing catches of prawns. For example, al- 

 though the composite-square-2 and square-3 codends 

 significantly reduced the numbers of prawns caught 

 by 8.7% and 12.1%, compared with the control, the 

 differences in weights of prawns were much smaller, 

 to the point where they were nonsignificant (e.g. 1.8% 

 and 3.7%, respectively). To achieve such a result, the 

 two trawls containing composite square-mesh 

 codends caught more prawns but retained mainly 

 larger-size individuals. 



