Table 2 also lists for each locus the correlation between Rogers'genetic distance and geographic distance [calculated following MOW et al. (1985) as the shortest marine route between two samples]. Mantel's nonparametric test (MANTEL 1967) was used to determine the significance of the relationship between genetic and geographic distance. Eight of the 11 independent RFLP loci exhibited significant correlations between genetic and geographic distance. For the allozymes, only the Pel locus exhibits a similar relationship. Although the mean genetic distances calculated for both sets of loci show highly significant correlations with geographic distance, Figure 4 shows that the scaling of this relationship for the allozymes differs sharply from that observed for the nuclear RFLPs. Figure 5 compares the UPGMA dendograms constructed for the two sets of polymorphisms by cluster analysis. Both dendograms show a similar population clustering and aclean break between the two samples from the western Atlanticand the four remaining samples. However, branch lengths for the RFLP loci are consistently five to six times those shown by the proteins, reflecting the detection of significant differences among populations within, as well as between, clusters at the DNA level.