436 



Fishery Bulletin 97(3), 1999 



duced the bycatch of juvenile prawns and some small 

 fish, although there were some concerns over suit- 

 able confignarations of square mesh. Given that in- 

 dustry had already adopted the concept of using 

 square mesh in codends, our aims in the present 

 study were to modify the square-mesh- 1 codend, af- 

 ter consultation with industry, so that it 1 ) incorpo- 

 rated a suitable BRD for unwanted fish and 2) in- 

 creased the size-selectivity of targeted prawns. 



16.5 m. Each of the two outside trawls were rigged 

 with zippers (Buraschi S-146R, pinlock sliders) to 

 facilitate codend changes. The lengths of these zip- 

 pers were calculated by assuming a fractional mesh 

 opening of 0.35 (Broadhurst and Larsen, personal 

 obs. ) X the stretched mesh length x the mesh circum- 

 ference. The middle net was not rigged in an identi- 

 cal manner to the outside trawls and therefore was 

 excluded from any analysis. 



Materials and methods 



This study was done on commercial prawn-trawl 

 grounds in Gulf St. Vincent, South Australia (Fig. 

 2), in November 1997, with a chartered commercial 

 prawn-trawler (15 m). Three locally used prawn 

 trawls (termed "Gundry trawls," Fig. 1), each with a 

 headline length of 9.14 m were rigged in a standard 

 triple gear configuration (see Andrew et al., 1991) 

 and towed at 3 knots across a combination of sandy 

 and light coral bottoms at depths ranging from 12 to 



Figure 2 



Location of Gulf St. Vincent and the area trawled. 



Codends examined 



Three codends, each attached to a zipper, were used 

 in the experiment. The first codend, termed the con- 

 trol (Fig. 3A), represented conventional codends used 

 prior to the 1996-97 fishing season and was made 

 entirely of diamond-shaped meshes (45-mm mesh 

 size, 2.35-mm diameter, 24-strand polyethylene 

 twine), measuring 81 meshes in length and 100 

 meshes in circumference. The second and third 

 codends (termed the composite-square-2 and compos- 

 ite-square-3 codends) comprised two sections of 52- 

 mni mesh (400/36 ply, UV-stabilized, high-density 

 polyethylene twine) cut on the bar. Each square-mesh 

 section consisted of an upper and lower panel, sewn 

 together so that knot directions were opposite. A 

 panel of 85-mm netting (3-mm braided polyethylene 

 twine ), measuring 6 bars x 1 1 bars, was inserted into 

 the tops of the posterior sections of each square-mesh 

 codend starting approximately 1.1m anterior to the 

 last row of meshes (Fig. 3, B and C). This panel was 

 positioned in a manner similar to that for the com- 

 posite square-mesh panel used in NSW oceanic prawn 

 trawls (see Broadhurst and Kennelly, 1996; 1997 1: that 

 is, in an area of maximum displacement of water ante- 

 rior to the end of the codend (to facilitate the escape of 

 small fish, Broadhurst et al., 1999) but sufficiently in 

 front of the anticipated build-up of catch. 



The composite-square-2 codend comprised anterior 

 and posterior square-mesh sections measuring 55 

 bars and 61 bars in length respectively. This codend 

 was similar to the industry-developed square-mesh- 

 1 codend, except that it had a circumference of 62 

 bars (1.6 m). The reduction in circumference from 

 120 bars (i.e. the industry design) to 62 bars was 

 based on the expected fishing circumference of the 

 normal diamond-mesh codend: that is, a fractional 

 mesh opening (0.35) x the mesh size (45 mm) x the 

 mesh circumference (100 T) = approx. 1.6 m. Two 

 diamond-mesh 45-mm sections (mesh size, 400/48 ply 

 UV-stabilized, high-density polyethylene twine), 10 

 meshes in length, and 100 meshes in circumference 

 were inserted between the two square-mesh sections 

 and immediately aft of the posterior square-mesh 

 section (hanging ratio of 0.36) (Fig. 3B). The ante- 





