Broadhurst and Kennedy: Composite square-mesh panel in codends for reducing bycatch in an Australian prawn-trawl fishery 
655 
Figure 1 
Map of New South Wales showing locations of the four ports 
that were sampled (Ballina, Yamba, Southwest Rocks, and Port 
Stephens). 
netting and 48-ply twine (Fig. 2). They comprised 
two sections: the anterior section was 100 meshes in 
circumference, 33 meshes in length, and attached to 
a zipper; the posterior section was 150 meshes in cir- 
cumference and 25 meshes in length. Two designs of 
codend were compared. The control codend was made 
entirely of diamond-shaped meshes (Fig. 2 A). The 
second codend (termed the composite-panel codend) 
was similar to the control but had composite panels 
made of 60-mm and 40-mm netting cut on the bar 
and inserted into the top of the anterior section (Fig. 
2B — see also Broadhurst and Kennedy, 1996). The 
composite-panel codend was designed so that the load 
was distributed anteriorly and laterally to the panel 
of 60-mm square-mesh, allowing this 60-mm panel 
to remain open. We predicted that 1) large numbers 
of fish would escape through this panel, located at 
the point where waterflow was thought to be great- 
est and that 2) in addition to reducing load on the 
60-mm panel, the 40-mm square-mesh would also 
facilitate the escape of smaller fish. 
The two codends were compared with each other 
in independent, paired trials, with the two outside 
nets of each vessel at each location. The codends were 
used in normal commercial tows of 90-min duration 
and alternated after each shot (to eliminate biases 
between different trawls and sides of the vessels). 
Because some significant effects of a delay in haul- 
back (the period between slowing the vessel and en- 
gaging the winch to haul the trawl) were detected in 
a previous experiment (Broadhurst et ah, 1996), all 
tows were performed with no delay in haulback. The 
location of each tow was randomly selected from the 
available prawn-trawl locations that were possible 
under the fishing conditions. During a period of four 
nights at locations offshore from each of the four 
ports, we completed a total of 16 replicate tows (i.e. 
four separate paired comparisons of 16 replicate tows 
each throughout the fishery). 
After each tow, the two codends were emptied onto 
a partitioned tray. Prawns and all commercially im- 
portant species larger than the minimum legal size 
(retained commercials) were separated. The remain- 
ing bycatch (termed “discarded by-catch”) was then 
sorted. This included individuals of commercially 
important species that were smaller than the mini- 
mum legal size (“discarded commercials”). Data col- 
lected from each tow were as follows: the total weight 
of king prawns and a subsample (50 prawns from 
each codend) of their sizes (to the nearest 1-mm cara- 
pace length); the weight of the discarded bycatch; 
the weights, numbers, and sizes (to the nearest 0.5 cm) 
of retained and discarded commercial species; the 
weights and the numbers of the most commonly oc- 
curring noncommercial species; and the total num- 
bers of discarded commercial species. Several spe- 
cies (commercial and noncommercial) were caught 
in sufficient numbers to enable meaningful compari- 
sons (see Table 2). 
Data at each port for all replicates that had suffi- 
cient numbers of each variable (defined as >2 indi- 
viduals in at least 8 replicates) were analyzed with 
one-tailed, paired £-tests (i.e. four separate analyses). 
Because a previous experiment had shown that 
trawls with the composite-panel have the potential 
to retain more prawns than conventional trawls 
(Broadhurst and Kennedy, 1996), we tested the hy- 
pothesis that the composite-panel codend caught 
more prawns but less total bycatch than the control 
codend. Where analyses provided similar results for 
weights and numbers of taxa, only data about num- 
bers were included in the figures to conserve space. 
Size frequencies of prawns, as well as discarded stout 
