Fishery Bulletin 97(1), 1999 



ence of meshes in the posterior section of the codend. 

 In an experiment to test this hypothesis, Broadhurst 

 and Kennelly (1996) showed that a conventional 

 codend made from an anterior section of 100 meshes 

 circumference and a posterior section 200 meshes in 

 circumference (a common commercial configuration) 

 was less selective (i.e. retained more bycatch) than a 

 conventional codend with a 100 mesh circumference 

 throughout its entire length (see also Robertson and 

 Stewart, 1988; Armstrong et al., 1990; Reeves et al., 

 1992; Galbraith et al., 1994). However, a compari- 

 son of codends with the same netting configurations 

 as above but containing composite square-mesh pan- 

 els located in their anterior sections showed that the 

 effects on selectivity due to increased circumference 

 in the posterior section were negated by a signifi- 

 cant increase in the escape of small fish (e.g. red spot 

 whiting, Sillago flindersi) through the square meshes. 



These results led to the hypothesis that increased 

 twine area and smaller mesh openings in codends 

 with posterior sections of 200 meshes circumference 

 increased the displacement of water forwards and 

 out through the meshes in the anterior section (see 

 also Watson, 1989). In turn, this water movement may 

 have 1 ) physically directed small fish out through the 

 strategically positioned composite square-mesh panel; 

 2) assisted them to maintain position in front of the 

 catch in the codend, increasing their likelihood of ran- 

 domly encountering square meshes; and (or) 3) stimu- 

 lated their lateral line receptors and thus their overall 

 escape response. The reaction of prawns to this stimuli 

 was thought to be minimal, given their inability to sus- 

 tain escape responses in trawls (see Lochhead, 1961; 

 Newland and Chapman, 1989). 



Although the results from the paper discussed 

 above led to several hypotheses about changes in 

 water flow and fish behavior due to changes in codend 

 geometry, we lacked the quantitative information on 

 flow rates necessary to support or refute them. Such 

 information is important for developing new designs 

 of codends and understanding where to position 

 square-mesh panels and other BRDs. Our goals in 

 the present study were to quantify the effects on 

 water flow at various positions in codends and un- 

 der square-mesh panels in two flume tank experi- 

 ments. We simulated commercial conditions in the 

 flume tank by using different weights of catch in 

 various codend designs. 



Materials and methods 



Two experiments were undertaken in May 1996 at 



the Australian Maritime College with the Faculty of 

 Fisheries and Marine Environment's flume tank. 



This facility consists of a recirculating flow tank of 

 fresh water, measuring 17.2 m long, 5 m wide, and 

 2.5 m deep and comprises three levels: 1) an upper 

 level where nets, etc., are placed into the tank; 2) an 

 observation level, with a continuous perspex view- 

 ing-window; and 3) a water-return channel. The two 

 lower levels feature a series of delivery bends and 

 screens that maintain constant water velocity 

 throughout the depth of the tank without any swirls 

 or vortices. Several electric motors, hydraulic pumps, 

 and impellor shafi-s provide water flow of up to 1.5 m/s. 



An electromagnetic current meter was attached to 

 the base of a stainless steel stanchion (Fig. lA) and 

 linked to a computer by means of a coaxial cable. 

 The stanchion was attached to a movable carriage 

 positioned on rails over the upper level of the flume 

 tank. This assembly enabled the current meter to be 

 repeatedly located at several predetermined positions 

 within the tank. 



A full-scale Florida flyer prawn-trawl (material: 

 18 ply twine; mesh size: 40 mm) with a headline 

 length of 5.4 m, was rigged to two fixed stanchions, 

 located on the sides of the forward section of the flume 

 tank. The trawl was rigged with a zipper (no. 10 ny- 

 lon open-ended auto-lock plastic slides) to facilitate 

 changing the codends. The codends used in the ex- 

 periments were of normal commercial size and ma- 

 terials, measuring 58 meshes long (2.3 m) and con- 

 structed from 40-mm mesh netting and 60-ply UV- 

 stabilized high-density polyethylene twine. These 

 codends comprised two sections: the anterior section, 

 which was 33 meshes long and attached to a zipper 

 and the posterior section, which was 25 meshes long 

 (for details see Fig. 1, B and C — see also Broadhurst 

 and Kennelly 1996). 



Experiment 1 



Two codend designs were compared. The codends 

 (termed the 100 and 200 commercial codends) were 

 made entirely of diamond-shaped meshes and com- 

 prised anterior sections with a circumference of 100 

 meshes, attached to posterior sections with circum- 

 ferences of 100 and 200 meshes, respectively (Fig. 1, 

 B and C). Three incisions of the same size as the 

 width of the current meter's stanchion (three meshes 

 in length), were made in the tops of each codend at 

 distances of 2200 mm, 1120 mm, and 560 mm for- 

 ward ft^om the end of the codends to facilitate place- 

 ment of the current meter inside the codends (Fig. IB). 



Experiment 2 



The two codends compared in this experiment were 

 similar to the 100 and 200 commercial codends de- 



