798 



Fishery Bulletin 101(4) 



Rare' 95% CI 



10 20 30 40 50 60 70 80 90 100 



Percentage of catch sorted 



Figure 6 



The mean percentage of sampling error calculated when estimating the 

 relative abundance of the "rare" (<1 individual per 10-kg subsample) and 

 the "abundant" bycatch species ( 5 or more individuals per 10-kg subsample ). 

 Curves were generated from 200 random selections of the numbered order 

 in which subsamples were collected from each catch. 



The catch rates of many of these most-at-risk species are 

 extremely low. Our data suggest that extremely high sam- 

 pling errors will be incurred if monitoring catch rates of 

 species on this list is undertaken. These errors must be 

 incorporated into any models that assess the impact of 

 trawling on populations of these species. 



The range of data in the present study includes many of 

 the sources of variation likely to be encountered in setting 

 up a wide-ranging bycatch monitoring program. The results 

 we present are a valuable guide to the accuracy of subsam- 

 pling. In particular, this study emphasizes the value of col- 

 lecting large subsamples, especially when one is restricted 

 to representing the bycatch of an area by only one or a few 

 trawls. This problem is common for research cruises when 

 many regions need to be sampled in a short time (e.g. Blaber 

 et al., 1994), and is also a common problem for observers on 

 commercial fishing vessels where catch sampling is often 

 restricted by the nature of commercial practices. Because 

 there is a high level of sampling error when estimating the 

 abundances for "rare" species, reliable estimates will require 

 either taking large subsamples or sorting entire catches. 



The size of some catches in our study may be larger 

 than those of many other tropical prawn trawl fisheries in 

 Australia and overseas. However, the data in the matrix 

 on the range of cumulative species (per proportion of catch 

 sorted. Table 2) will allow managers of other trawl fisheries 

 to better understand the implications and likely accuracy 

 of bycatch sampling programs. 



Acknowledgments 



We thank S. Cook, M. Farmer, C. Liron, D. Milton, J. Salini, 

 I. Stobutzki, T. Wassenberg, G. Fry, and others for their 

 valued help in sorting the catches both on board the RV 

 Southern Surveyor and in the Cleveland laboratory. We 

 also thank D. McKay, the skipper of FV Apolloair, and 

 P. Hoschke, the skipper of FV Vcntia-a, and their respec- 

 tive crews, for their help in collecting commercial trawl 

 samples. We also thank J. Bishop, S. Blaber, B. Hill, D. 

 Milton, and V. Mawson for their valuable comments on the 

 manuscript. This work was undertaken with the support 

 of FRDC grant no 96/257 



Literature citations 



Andrew, N. L., and B. D. Mapstone. 



1987. Sampling and the description of spatial patterns in 

 marine ecology. Oceanogr. Mar. Biol. Annu. Rev. 25:39-90. 

 Blaber, S. J. M., D. T. Brewer, and A. N. Harris. 



1994. Distribution, biomass and community structure of 

 demersal fishes of the Gulf of Carpentaria, Australia. 

 Aust. J. Mar. Freshw. Res. 45:375-396. 

 Blaber, S. J. M., D. T. Brewer, J. P. Salini, and J. Kerr. 



1990. Biomasses, catch rates and abundances of demersal 

 fishes, particularly predators of prawns, in a tropical bay 

 in the Gulf of Carpentaria, Australia. Mar. Biol. 107: 

 397-408. 



