90 
Fishery Bulletin 108(1) 
Table 2 
Sequence statistics of Atlantic menhaden (Brevoortia tyrannus) and Gulf menhaden ( B . patronus) based on cytochorme c oxidase 
subunit I (COI) gene region by location, clade, and across all locations: number of individuals (n), number of haplotypes (N h ), 
number of polymorphic sites (S), number of transitions (Ts), number of transversions (Tv), haplotype diversity ( h ), mean nucleo- 
tide sequence diversity (n), and mean number of pairwise differences ( k ) and standard error (SE). 
Collection 
n 
N h 
S 
Ts 
Tv 
h ±SE 
n ±SE 
k ±SE 
New England 
50 
29 
47 
47 
4 
0.940 ±0.0206 
0.0258 ±0.0132 
11.8 ±5.45 
mid-Atlantic 
52 
32 
66 
67 
5 
0.956 ±0.0163 
0.0286 ±0.0145 
13.1 ±6.01 
Chesapeake Bay 
117 
50 
62 
63 
3 
0.932 ±0.0147 
0.0267 ±0.0135 
12.3 ±5.58 
U.S. South Atlantic 
69 
44 
66 
67 
2 
0.954 ±0.0166 
0.0295 ±0.0149 
13.5 ±6.16 
Atlantic menhaden 
288 
109 
97 
99 
6 
0.941 ±0.0084 
0.0274 ±0.0137 
12.6 ±5.69 
Gulf menhaden 
50 
25 
34 
33 
5 
0.879 ±0.0419 
0.0071 ±0.0041 
3.23 ±1.70 
Atlantic + Gulf menhaden 
338 
124 
99 
101 
7 
0.940 ±0.0079 
0.0258 ±0.0129 
11.8 ±5.37 
ubiquitous clade 
235 
98 
76 
76 
7 
0.924 ±0.0138 
0.0081 ±0.0045 
3.71 ±1.88 
Atlantic only clade 
100 
23 
22 
22 
1 
0.732 ±0.0456 
0.0036 ±0.0024 
1.65 ±0.98 
estimates statistical power by testing different combi- 
nations of number of samples, sample sizes, number of 
loci, number of alleles, and allele frequencies for any 
hypothetical degree of differentiation (F ST ). To match 
the number of collection locations and magnitude of F ST 
estimates in this study, five hundred replicates were 
performed on five populations by using Fischer’s method 
with the following combinations of effective population 
size (N e ) and generations of drift before sampling ( t ): 
10,000: 50, 5000: 25, 1000: 5 (F sr =0.0025); 10,000: 
201, 5000: 100.5, 1000: 20.1 (F sr =0.01); 10,000: 1025.8, 
5000: 512.9, 1000: 102.6 (F ST = 0.05). The hypothetical 
sample sizes were set to the average across all loci for 
each sampling location. An additional simulation was 
performed with t = 0, to assess a (type-I) error. The 
estimate of power was reported as the proportion of 
significant outcomes (P<0.05). 
Results 
The COI fragment was sequenced for 289 Atlantic men- 
haden and 50 Gulf menhaden. Overall, the fragment 
contained 99 polymorphic sites (97 in Atlantic menha- 
den): 5 first codon positions, 1 second codon position, and 
91 third codon positions; 101 transitions (99 in Atlantic 
menhaden), 7 transversions (6 in Atlantic menhaden); 
and produced 124 haplotypes (109 in Atlantic menhaden) 
(Table 2). All substitutions were synonymous, resulting 
in identical amino acid sequences. Haplotype diversity 
( h ) estimates for the Atlantic and Gulf menhaden sam- 
pling locations ranged from 0.879 in Gulf menhaden to 
0.956 in the U.S. mid-Atlantic, with an overall (pooled) 
estimate of 0.940 (0.941 for Atlantic menhaden). Mean 
nucleotide sequence diversity (jt) estimates for Atlantic 
and Gulf menhaden sampling locations ranged from 
0.0071 in Gulf menhaden to 0.030 in the U.S. South 
Atlantic, with an overall (pooled) estimate of 0.026 (0.027 
for Atlantic menhaden). The mean number of pairwise 
differences ( k ) ranged from 3.2 in Gulf menhaden to 13.5 
in the U.S. South Atlantic, with an overall (pooled) esti- 
mate of 12.6 for Atlantic menhaden and 11.8 for Atlantic 
and Gulf menhaden combined. 
The median-joining network for the 109 COI Atlantic 
menhaden haplotypes showed two extensive clusters 
(clades) separated by 17 substitutions and one minor 
grouping of three individuals separated by 24 substi- 
tutions (Fig. 2). In order to discriminate between the 
alternate possibilities that either the three individuals 
were one of the other North American Brevoortia spe- 
cies that had been misidentified in the field as Atlantic 
menhaden, or that there is incomplete lineage sorting 
at the COI locus, the more rapidly evolving mitochon- 
drial control region was sequenced and compared to 
those generated by Anderson (2007) for the four spe- 
cies of North American menhaden (GenBank acces- 
sion numbers EF119342-EF119454). The control region 
sequences for the three individuals unambiguously 
clustered with the “ubiquitous large-scaled” menhadens 
in an unweighted pair group method with arithmetic 
mean (UPGMA) tree with these sequences, indicating 
that the field identifications were correct and that the 
result was likely attributable to incomplete lineage 
sorting at the COI locus. Examination of the microsat- 
ellite genotypes of these three individuals supported 
this conclusion. 
The distribution of the two major COI clades dif- 
fered among Atlantic and Gulf menhaden. Clade I, 
the “ubiquitous” clade, comprised all Gulf menhaden 
samples and 64% of Atlantic menhaden samples. Clade 
II, the “Atlantic-only” clade, comprised 35% of Atlan- 
tic menhaden samples and was not detected in Gulf 
menhaden (Table 2). A contingency table of the four 
Atlantic menhaden sampling locations indicated that 
clades I and II were homogeneously distributed among 
the U.S. Atlantic Coast sampling locations (x 2 =0.478; 
X 2 0.05,3’ P>0.05). 
Eight microsatellite loci, Aal6, Asa2, Asa4, Asal6, 
AsaB020, AsaD055, AsaC334, and Sa?'BH04, were am- 
plified for the Atlantic menhaden and Gulf menhaden 
