372 



Fishery Bulletin 105(3) 



of B. patronus individuals from the Gulf of 

 Mexico (;;=260, Table 1). One locus (AF39661) 

 had a high overall value of F,,, but was not 

 significantly different than zero based on 

 randomization tests. However, a second locus 

 (AF39660) had lower than expected hetero- 

 zygosity at a significant level in a major- 

 ity of the populations assayed. Whether the 

 explanation for decreased levels of heterozy- 

 gosity is biological or simply sampling error 

 is unclear; however analyses excluding this 

 marker had little effect on tree topology or 

 other results. Multilocus estimates of genetic 

 divergence (F,,) were lowest within the pre- 

 defined menhaden groups; that is, the level of 

 divergence seen in the comparisons between 

 B. patronus+tyrannus and B. smithi+gunteri 

 was lower than any other pairwise comparisons among 

 species (Table 2). Similarly, the N-J tree constructed 

 from microsatellite data was deeply bifurcated (Fig. 2) 

 and the two major groupings on this tree corresponded to 

 populations of small-scaled and large-scaled menhaden, 

 exclusively. Four major clusters on this tree corresponded 

 to the populations of each of the four assayed species. 

 The greatest level of within-species divergence was 

 observed between the Atlantic and Gulf populations of 

 B. smithi. The divergence among these populations was 

 an order of magnitude higher than the mean divergence 

 from comparisons among populations of the same spe- 

 cies, and was 5.5 standard deviations removed from the 

 mean of all F^, comparisons combined (F^,*""""=0.0523, 

 F^/""'™"=0.0054, 0^ = 0.0094). 



Bayesian analysis of population structure resulted 

 in four well-defined clusters corresponding to the four 

 assayed species (Fig. 3). However, no single species 



contained exclusively one cluster; each species had in- 

 dividuals with composite genotypes including influence 

 from other clusters. In particular, admixture among the 

 two large-scaled species was indicated by individuals of 

 each species in which a majority of the genotype was 

 contributed by the complimentary species. A smaller 

 level of admixture was indicated in the small-scaled 

 species. However, one interesting result was a single 

 individual B. gunteri which had 0.723 of its genotype 

 contributed by the B. patro/iiis cluster. This is signifi- 

 cant in that, although the possibility of hybridization 

 in the western Gulf has not been ruled out (Anderson 

 and McDonald, 2007), a verified Fl B. patronus+gunteri 

 hybrid has not been documented in the Gulf. 



Estimates of genetic variability including H,, k^^, and 

 Op were higher in large-scaled species of menhaden than 

 in small-scaled species (Table 3). Multilocus expected 

 heterozygosity ranged from 0.672 in B. patronus and 



