Heales et al : Effect of size of subsamples on estimates of catchi composition and abundance 



791 



species, their position on trawler sorting 

 trays from which the bycatch subsamples 

 were collected, had little effect on the ac- 

 curacy in representing the catches. 



Although the accuracy of subsampling 

 is a general problem for all multispecies 

 fisheries, few other studies have been pub- 

 lished on the topic. Reliable techniques for 

 subsampling are needed, however, espe- 

 cially with the demands that bycatch spe- 

 cies, as well as the target species, should 

 be ecologically sustainable. We describe 

 here research done in Australia's NPF but 

 the results and methods are applicable to 

 many sampling problems in fisheries. 



The NPF is a large tropical trawl 

 fishery that extends from Cape York in 

 Queensland to Cape Londonderry in 

 Western Australia. In addition to catch- 

 ing penaeid prawns (e.g. 8531 metric tons 

 (t) in 1997-98, Taylor and Die, 1999), its 

 bycatch component is estimated at over 

 38,000 t a year (Pender et al., 1992) or 

 more than 80% of the total catch in the 

 tiger prawn fishery (Brewer et al., 1998). 

 The neighboring Torres Straits Prawn 

 Fishery (TSPF) also has an estimated an- 

 nual bycatch of 4800 t (Williams, 1985). 



The NPF has a management require- 

 ment to assess the impact of trawling on 

 nontarget species. The bycatch of both 

 these prawn trawl fisheries (NPF and 

 TSPF) is very diverse. Ramm et al. (1990) recorded 115 

 fish taxa in their study of NPF waters, and Brewer et al. 

 ( 1998) recorded over 250 species from one area of the NPF. 

 At least 390 fish species, 234 invertebrate taxa, and 43 elas- 

 mobranch species have been recorded in a current bycatch 

 project in the NPF (Stobutzki et al., 2000). 



Despite the lack of knowledge on the ability of sub- 

 samples to accurately represent such diverse catches, 

 many studies of trawl communities in Northern Australia 

 have used subsampling techniques to estimate catch rates. 

 In two bycatch studies of the NPF, the smaller catches 

 were entirely sorted and the larger catches subsampled 

 (Poiner and Harris, 1986; Harris and Poiner, 1991). Trawl 

 catches in another bycatch study were spread evenly over 

 the sorting tray and a visually estimated fraction of the 

 catch was subsampled (Ramm et al., 1990). Other work- 

 ers simply subsampled without confirming the accuracy 

 of their subsampling techniques (Blaber et al., 1990 and 

 1994; Martin et al, 1995; Brewer et al, 1998; Wassenberg 

 etal., 1998). 



A lack of knowledge of trawl impacts on nontarget spe- 

 cies has led to the present CSIRO study that describes the 

 bycatch from the NPF and provides a framework for any 

 future bycatch monitoring program. As part of that study, 

 we made the first assessment of the accuracy of subsam- 

 pling over a range of subsample sizes as a tool for esti- 

 mating the total catch composition of a large multispecies 

 fishery. 



129" 



134 



139 



Stu(jy area 



\'L J Australia 



129" 



134 



139" 



144" 



Figure 1 



Map of the Torres Straits Fishery (area 1 — broken line) and Northern Prawn 

 Fishery (unbroken line), Australia, showing the managed area of the fisheries 

 and the regions where samples were collected. Regional codes are 1 = Torres 

 Straits, 2 = Weipa, 3 = east of Mornington Island, 4 = north of Mornington 

 Island, 5 = west of Mornington Island, 6 = east of Vanderlin Islands, 7 = south 

 of Groote Eylandt, 8 = north of Groote Eylandt, 9 = Melville Island. 



Materials and methods 



Data were collected from a series of 14 trawl samples taken 

 during two research cruises of the RV Southern Surveyor. 

 These samples were collected from one region of the Torres 

 Straits Prawn Fishery (TSPF), and from eight of the major 

 tiger prawn (Penaeus esculentus and P. semisulcatus) fish- 

 ing regions of the Northern Prawn Fishery (NPF), (namely 

 Weipa, east of Mornington Island, north of Mornington 

 Island, west of Mornington Island, north of Vanderlin 

 Islands, south of Groote Eylandt, north of Groote Eylandt, 

 and Melville Island) (Fig. 1). All trawls were undertaken 

 in either late summer 1997 (February-March, the end of 

 the wet season) or in mid-spring 1997 (September-October, 

 the dry season). We used a single 14-fathom Florida-Flyer 

 prawn trawl net so that the data would be comparable with 

 data from either of the two nets used by the twin-rigged 

 commercial NPF vessels in the tiger prawn fishery. All 

 trawls were done at night, again to be comparable with 

 the fishery Duration of trawls ranged from 1 to 3 hours 

 (Table 1), and depths ranged from 23 to 42.3 m. 



A further six trawl catches were sampled by a scientific 

 observer on board commercial NPF vessels fishing for ti- 

 ger prawns north of Mornington Island in late May 1997, 

 and north of Groote Eylandt in late September 1997. Each 

 trawl sample consisted of the entire catch from one of the 

 two 14-fathom Florida-Flyer prawn trawl nets used by 

 these vessels. All trawls were done at night. Their duration 



