In summary, our results have demonstrated significant differences between the patterns of allozyme and nuclear DNA polymorphisms among populations of a widely distributed marine fish species known to undergo large-scale migrations (e.g., GULLANDand WILLIAMSON 1962). Thesedifferences eliminate gene flow as the explanation for the homogeneous allozyme patterns observed in G. morhua. The absence of population structure atweakly polymorphic loci in this species (including mtDNA variation) may be more indicative of a recent origin of populations. The extent of variability observed at moderately polymorphic allozyme locisuch as Ldh3 or Idh-1 appears to be less than expected by chance. At other loci, such as RFLP locus GM798, the
magnitude ofallele frequencyvariation appears greater than expected by drift alone. Our interpretation of the differences between the two sets of loci as reflecting the prevalence of random driftacting at theDNA level and natural selection acting at the protein level is similar to that reached in a comparablestudy on the American oyster (KARL and AVISE1992). In suggesting that selection rather than gene flow may be responsible for the homogeneous distributions of a subset of allozyme loci, our study questions the presumed neutrality of protein polymorphism and suggests that the invariant allozyme distributions observed in other marine fishes,
or in other species, may have a similar basis.