Precision on the fish stock abundance estimates based on trawl survey

The following considerations are presented based on data from 3 trawl surveys in the Bering Sea, on the Agulhas Bank and the Georges Bank: (1) any allocations except that of Neyman was not so effective compared to simple random sampling (SRS); (2) the C.V. of catch per tow (CPT) ranged from 1 for walleye pollock to about 3.5 for horse mackerel and shortfinned squid; (3) from the simulation made for the latter two species with variable CPTs, a sample size of 100-200 tows seems adequate to calculate mean CPTs in linear scale which are used as abundance indices for usually occurring species; and (4) taking it into account that the number of tows (n) are generally limited within the range of 100-200 due to man-power etc., the precision of abundance estimates (D = C.V./ square root n) mostly depends on the C.V. of CPTs for the species concerned.