806 
Fishery Bulletin 95(4), 1997 
were A14 (TCTGTGCTGG), A15 (TTCCGAACCC), 
A16 (AGCCAGCGAA), A17 (GACCGCTTGT), D15 
(CATCCGTGCT), E19 (ACGGCGTATG), and H17 
(CACTCTCCTC). The number of scored fragments 
varied from 1 to 6 per primer, and the size of the 
fragments from 0.6 to 2.8 kb. Fragments that could 
be scored were numbered in decreasing order of elec- 
trophoretic mobility (e.g. primer A14 fragment 1 = 
A14-1); each individual fish was scored for the pres- 
ence or absence of each fragment. Repeat tests on 
some individuals did not produce repeatable patterns 
for some weakly staining fragments, therefore pres- 
ence or absence of each fragment was not scored for 
these fragments. Omitting the DNA template from 
the PCR reaction (i.e. negative control) failed to pro- 
duce fragments. The amount of DNA in the initial 
extractions varied tenfold between samples. Excess 
DNA, 250 ng, produced different fragment patterns 
with some primers (Fig. 2), therefore all amplifica- 
tions were optimized to contain a 50-ng template of 
DNA. 
The estimated allele frequencies are given in Ap- 
pendix Table 3. Two out of 29 primer fragments (A16- 
1, E19-3) revealed a significant heterogeneity among 
populations when a heterogeneity % 2 test with a modi- 
fied probability for multiple tests was applied (Table 
4). Pairwise comparisons showed significant differ- 
ences between all pairs of populations except Ritchie 
Bank and Box at these two primer fragments (Table 
2). The heterogeneity in the total data set was con- 
firmed by the G ST tests (Table 4). The effective num- 
ber of migrants per generation was estimated to be 
7.0 from the overall G ST , minus G STnu!j due to sam- 
pling error (Table 3), as described for allozymes. 
Pairwise values varied from 6.5 (Waitaki and 
Puysegur) to 75 (Ritchie and Box). 
Discussion 
There was significant heterogeneity in the allozyme 
data set at five loci (Table 1) which indicated that 
the population samples had not been taken from a 
single panmictic stock. Pairwise comparisons showed 
that there were significant differences between all 
pairs of spawning populations except the two north- 
ern populations at Ritchie Bank and Box (Fig. 1). 
The RAPD data also showed a significant heteroge- 
neity that indicated that the population samples had 
been taken from more than one genetic unit stock. 
There were no area-specific RAPD fragments in or- 
ange roughy, as have been reported in marine prawns 
(Garcia et al., 1996) and freshwater crayfish (Mac- 
aranas et ah, 1995), but there were differences in 
