THE POTENTIAL OF SAMPLING WITH PARTIAL REPLACEMENT FOR FISHERIES SURVEYS

The methodology of Nicholson et al. (ICES CM 1991 D: 11, 9pp.) for com paring sampling with fixed and random stations is extended to the case of sampling with partial replacement. The conditions under which esti mates of abundance and, more particularly, change in abundance based o n fixed samples, will be more accurate than those based on sampling wi th partial replacement and under which the latter will be more accurat e than those based on random sampling are derived. Research trawl data from NAFO Divisions 2J, 3K, and 3L for the years 1985-1991 (1985-1992 for 3L) are used to gain some idea of how the accuracy of estimates o f changes based on the three strategies (fixed, random, and partial re placement) would compare under the assumption that the stations fixed between years are selected at random. In practice, purposive selection of the fixed stations is, of course, possible.