Carson et al.: Population structure, long-term connectivity, and effective size of Lutjcinus analis 
423 
Table 2 
Distribution of individual haplotypes in 134 mutton snapper (Lutjanus analis ) sampled from four localities in the northeastern 
Caribbean Sea and one locality in the Florida Keys. Numbers below each locality indicate observed occurrence of each of 25 
haplotypes identified across all localities. GenBank® is a genetic sequence database, available at http://www.ncbi.nlm.nih.gov/ 
genbank/, accessed July 2011). 
Haplotype 
St. Croix 
St. Thomas 
Puerto Rico-east 
Puerto Rico-west 
Florida Keys 
GenBank no. 
1 
1 
0 
1 
0 
1 
JF514891 
2 
10 
10 
9 
5 
14 
JF514892 
3 
1 
0 
0 
1 
0 
JF514893 
4 
0 
0 
0 
1 
0 
JF514894 
5 
0 
0 
0 
1 
0 
JF514895 
6 
0 
1 
2 
2 
0 
JF514896 
7 
0 
0 
1 
0 
0 
JF514897 
8 
5 
6 
4 
7 
4 
JF514898 
9 
0 
1 
0 
0 
1 
JF514899 
10 
0 
0 
0 
0 
1 
JF514900 
11 
1 
0 
0 
0 
0 
JF514901 
12 
0 
0 
0 
1 
0 
JF514902 
13 
1 
0 
0 
0 
0 
JF514903 
14 
2 
1 
0 
0 
1 
JF514904 
15 
1 
0 
1 
0 
0 
JF514905 
16 
0 
0 
1 
0 
0 
JF514906 
17 
0 
1 
0 
1 
1 
JF514907 
18 
0 
2 
2 
0 
2 
JF514908 
19 
3 
1 
5 
4 
1 
JF514909 
20 
0 
1 
1 
1 
0 
JF514910 
21 
0 
0 
0 
0 
1 
JF514911 
22 
0 
0 
0 
1 
0 
JF514912 
23 
1 
0 
0 
0 
0 
JF514913 
24 
1 
1 
2 
0 
0 
JF514914 
25 
0 
1 
0 
0 
0 
JF514915 
Total 
27 
26 
29 
25 
27 
P= 0.785). Pair-wise exact tests (between samples) for 
both microsatellites and mtDNA also were nonsignifi- 
cant (data not shown but available from E. W. Carson). 
Finally, nine of 15 tests of selective neutrality were 
significant before Bonferroni correction; none remained 
significant after Bonferroni correction. 
Estimates of average long-term mutation-scaled mi- 
gration (M) between geographically adjacent pairs of 
sample localities are presented in Table 3. Estimates of 
M between adjacent localities were generated by aver- 
aging bidirectional estimates from Migrate, in part be- 
cause bidirectional estimates generally were equivalent, 
and in part because confidence intervals for estimates 
of m, generated with Migrate, are generally compro- 
mised (Abdo et al., 2004). Based on a modal mutation 
rate (//) over all microsatellites of 2.51xl0 -4 , generated 
with MS VAR, estimates of average long-term migration 
rate (m) between adjacent localities varied from 0.0033 
(PR-west vs. PR-east) to 0.0054 (SC vs. ST). Higher 
estimates of m were found between ST and SC and 
between ST and PR-east (Table 3). 
Estimates of average long-term, effective population 
size (N e ) for each locality (Table 3) were derived from 
©values generated in Migrate (N e =OZ 4//), with ji equal 
to the modal mutation rate of 2.51xl0~ 4 obtained from 
the Bayesian coalescent approach of Beaumont (1999) 
and Storz and Beaumont (2002). Initial Migrate runs 
revealed that the sample from SC had by far the low- 
est estimate of N e . Because genotypes at microsatellite 
Lem 11 did not conform to Hardy -Weinberg expectations 
in the sample from SC (Table 1), values reported in 
Table 3 reflect Migrate runs without Lanll. Average 
long-term N e among the five localities varied from a 
low of 341 (SC) to a high of 1066 (FK). Estimates from 
Migrate of 95% confidence intervals (CIs) indicate sig- 
nificant differences in average long-term N e among lo- 
calities (Table 3); the lower values for PR-west and SC 
are especially relevant because these two localities are 
close to known mutton snapper spawning aggregation 
sites in the U.S. Caribbean. 
Discussion 
Analysis of microsatellite and mtDNA variation in 
mutton snapper sampled from localities in the north- 
