Abstract
Sexually mature specimens of stationary and migratory cod (Gadus morhua L.) occur sympatrically at some major spawning grounds off the Norwegian coast during the spawning period, although the majority of authors suggest that the two groups do not interbreed. A commonly applied method to study population structure has been to compare sampled and Hardy-Weinberg equilibrium frequencies of haemoglobin genotypes. Significant and not significant deviation have been regarded as indicating sampling from a heterogeneous and homogeneous cod group, respectively. Statistical tests commonly applied are the Wright's fixation test (WFI) and the equivalent chi-square test. The present paper shows that the WFI test usually has unacceptably low statistical power and is therefore inadequate when testing for homogeneity in samples from two non-interbreeding populations, both in Hardy-Weinberg equilibrium. We suggest an alternative and easy to calculate binomial test, with relatively high power in a number of cases when testing for homogeneity. The binomial test can only be used when the gene frequencies of the two populations are known, as is the case with cod off northern Norway. For other cases with unknown gene frequencies, the low power performance of WFI implies that large sample sizes are required, and multi-locus methods should be considered.