The DNA markers were developed with the a priori expectation of neutrality since common alleles at all loci arose from restriction site polymorphisms located in the flanking regions of single-copy genes. In agreement witht his prediction, highly significant correlations between ge- netic and geographic distance wereo bserved att he RFLP loci (Table 2 and Figure 4), which is consistent with the outcome of random drift and limited gene flow in producing isolation-by-distance (GJ: WRIGHT 1943). Three of the RFLP loci revealed no spatial structure among populations but exhibited significant allele fre- quency heterogeneity. Two of these loci show weak dif- ferentiation that is attributable in both cases to the sig- nificant divergence of only one of the six populations (GM309 in Iceland and GM860 in Nova Scotia). This pattern is likely to reflect the action of drift. In contrast, RFLP locus GM798 detected highly significant allele frequency differences among all populations at a level exceeding that expected by chance alone. The FST val- ues calculated for the 11 independent cDNA loci are expected to be distributed as a chi-square with 10 de- grees of freedom. The mean FST valueo bserved for GM798 (0.309) exceeds the 95% confidence interval of this distribution, thus making drift an unlikely explana- tion for the heterogeneity observed. This suggests the possibility that the three polymorphisms scored by this cDNA clone may be tightly linked to a site undergoing selection. Further indirect evidence for the operation of selection at this locus based on geographic patterns of linkage disequilibrium will be addressed in a separate publication (G. H. POCSON, K. A. MESA and R. G. BOUTI- LIER, unpublished results).