Lynch et at: A genetic investigation of the population structure of Brevoortia tyrannus 
91 
Figure 2 
Median-joining network of 109 Atlantic menhaden ( Brevoortia tyrannus, 
n = 289) cytochrome c oxidase subunit I (COI) haplotypes. The observed haplo- 
types are sized according to frequency and coded by area. Sample sizes: New 
England (n = 50), mid-Atlantic (72 = 53), Chesapeake Bay (ti = 117), U.S. South 
Atlantic, (ti = 69). The hypothesized intermediate haplotypes are denoted by 
the smallest circles shaded dark gray. 
samples. The genotypic distributions 
of all loci in all samples conformed 
to the expectations of Hardy-Wein- 
berg equilibrium with the excep- 
tion of locus Asal6 (Table 3). Both 
the exact test in Genepop and the 
MicroChecker analysis revealed a 
significant or nearly significant de- 
ficiency of Asa 16 heterozygotes in 
all samples, indicating the presence 
of a null allele. Consequently, this 
locus was not included in any of the 
population structure analyses. 
For the seven remaining loci, the 
number of alleles (a) across all At- 
lantic menhaden samples ranged 
from 8 at Aal6 and AsaC334 to 22 
at AsaB020, and the allelic richness 
(R s ) ranged from 7.98 at Aal6 to 
21.8 at AsaB020. Allele size ranges 
were similar for Atlantic menha- 
den and Gulf menhaden across all 
seven loci. However, average a and 
R s values across all loci were lower 
in Gulf menhaden as compared to 
Atlantic menhaden (Table 3). 
AMOVAs of the COI haplotype 
data and microsatellite genotype 
data were performed to evaluate the 
temporal and spatial partitioning 
of genetic variation within Atlantic 
menhaden (Table 4). No significant 
differences were detected between 
early (May) and late (August) collec- 
tions of YOY Atlantic menhaden in 
Chesapeake Bay in 2007. Likewise, 
no differences were detected among 
year classes of Atlantic menhaden 
in the Chesapeake Bay based on 
a comparison of YOY and yearling 
menhaden collected in Chesapeake 
Bay in 2006 or a comparison of YOY 
and yearling menhaden collected in 
Chesapeake Bay in 2007. Following 
the same cohort in the same location 
across years, a comparison of YOY 
menhaden collected in 2006 with yearling menhaden 
collected in 2007 in the Chesapeake Bay, did not result 
in a significant difference based on either the mitochon- 
drial ( 0 ST ) or the microsatellite (R ST ). However, the mi- 
crosatellite (F st ) AMOVA produced a significant result, 
where 1.80% (P=0.02) of the variance was attributed to 
differences within the same cohort in Chesapeake Bay 
in successive years. 
Samples of YOY and yearling menhaden (combined) 
from four geographic regions along the U.S. Atlantic 
Coast (New England, mid-Atlantic, Chesapeake Bay, 
and U.S. South Atlantic) were compared to test for 
evidence of spatial partitioning of genetic variation. 
Only the AMOVA based on the microsatellite ^st was 
significant, attributing 0.58% (P<0.001) of variation to 
sampling location. No pairwise comparisons of <P ST , F ST 
or R st revealed statistically significant variation be- 
tween any two sampling regions of Atlantic menhaden 
after a Bonferroni correction (Table 5). 
All pairwise comparisons between Atlantic menhaden 
and Gulf menhaden collections revealed statistically 
significant variation. The mitochondrial (<P ST ) AMOVA 
between Atlantic and Gulf menhaden attributed 18.2% 
(P<0.001) of the variance to differences between puta- 
tive species. Likewise, in the microsatellite comparison, 
the F st and R ST AMOVAs attributed 11.5% (P<0.001) 
and 38.75% (P<0.001) of variance to variation between 
Gulf and Atlantic menhaden samples. 
