292 



Fishery Bulletin 102(2) 



at http://www.univ-montp2.fr/~genetix/genetix/genetix. 

 htm) (Belkhir et al. 2 ). Significance of F ST estimates was 

 determined with 2000 randomizations. Tests of population 

 differentiation and estimations of F ST were calculated at 

 each locus individually and at all loci combined. To correct 

 for simultaneous comparisons, standard Bonferroni cor- 

 rections were applied by using a global significance level 

 of 0.05 (Rice, 1989). 



Results 



Genetic variation 



Observed numbers of alleles, allelic ranges, heterozygosi- 

 ties, and deviations from HWE are presented in Table 1. 

 All tests of genotypic linkage disequilibrium were non- 

 significant at the global and population levels. When 

 Bonferroni corrections for multiple tests were applied to 

 tests of HWE (a=0.05, P<0.0083), the pooled Georges 

 Bank sample deviated significantly at Gwol9, and the 

 Nantucket Shoals sample deviated at Gmo8 and at Gmol9. 

 Interestingly, these two loci have the greatest variation 

 based on number of alleles and heterozygosity. In each 

 case, the cause of deviation was due to an excess of homo- 

 zygotes. Population samples that generally conform to 

 expectations of random mating but show a lack of concor- 

 dance to HWE at one or more loci may be due to a number 

 of processes including null alleles, genetic drift, admix- 

 ture, selection, and insufficient sampling (e.g., Ruzzante, 

 1998). Possible explanations of homozygote excess include 

 sample admixture ( Wahlund effect) or drift; however these 

 explanations are unlikely because one would expect to see 

 similar results at all loci. More likely explanations are the 



- Belkhir K., P. Borsa, L. Chikhi, N. Raufaste, and F. Bon- 

 homme. 2002. GENETIX 4.04, logiciel sous Windows TM 

 pour la genetique des populations. Laboratoire Genome, 

 Populations, Interactions, CNRS UMR 5000, Universite de 

 Montpellier II, Montpellier i France). 



presence of null alleles or selection. Deviations of HWE 

 at Gmo8 and Gmol9 were not observed in all population 

 samples, indicating that null alleles were not present at a 

 global-level but may be present at the population-level for 

 these two loci. Subsequently, any significant population 

 structuring observed at these loci should be viewed with 

 caution (see below). 



Population structure 



Tests of population structure are shown in Table 2. Al- 

 though Gmo8 and Gmol9 showed significant deviations 

 from HWE, they did not support any significant population 

 structuring even when tests of population differentiation 

 were performed without assuming conformation to HWE 

 (i.e., permuting among genotypes rather than alleles). 

 Heterogeneity of allelic distribution was not observed 

 (P>0.05) between the 1994 and 1999 Georges Bank 

 samples at each locus individually and at all loci combined, 

 thus indicating potential genetic stability of Georges Bank 

 cod. These samples were subsequently pooled to form 

 a single Georges Bank population sample to facilitate 

 statistical analyses by allowing for better estimations of 

 allele frequencies and by reducing the number of pairwise 

 tests. Tests of population differentiation among samples 

 showed significant divergence at Gmol32 (P<0.01) and 

 at all loci combined (P<0.05). When Bonferroni correc- 

 tions were applied to pairwise measures of divergence 

 («=0.05, P<0.017), significance was observed between 

 Nantucket Shoals and Georges Bank at Gwol32 and at 

 all loci combined, and also between Nantucket Shoals and 

 Browns Bank at Gwol32. No significant differentiation 

 was observed between individual or pooled Georges Bank 

 samples and the Browns Bank sample. 



Significant among-population F ST values were esti- 

 mated at Gmol32 (0.0255, P<0.001) and at all loci com- 

 bined (0.0047, P<0.01). When Bonferroni corrections were 

 applied, significant pairwise-population F ST values were 

 estimated between Nantucket Shoals and Browns Bank 

 at Gwol32 (0.0624, P<0.001) and at all loci combined 



