422 
Fishery Bulletin 113(4) 
84° W 
> 
76“ W 
i- 
68“ W 
60° W 
52“ W 
.... i 
44“ W 36“ W 
-ml — ,L ... 
38° N- 
N 
A * 
> v. ! S . . .. Xifew®!* 
Cape Hatteras, NC, N=31 
30“ N- 
Charleston, SC; N=29 , 
O°9,10 
. 
^Cape Canaveral, FL;N=30 
22° N 
14° N 
San Juan, PR, N=75 
yj 
,X -' * 3 
Dominica, N=4 
□ Barbados; N=35 
QTrimdad and Tobago; N=4 
0 
500 
i 
Azores Islands ; N=8 
1,000 Km 
I 
Legend 
O Mark 
♦ Recapture 
Figure 1 
Tissue sample distribution and conventional tagging movements of dolphinfish ( Coryphaena hippurus ) in the western 
central and central North Atlantic. Tissue samples were taken from fish collected at different landing sites in each of the 
3 regions in the western central Atlantic, including the southeastern United States in 2012 (triangles), northeastern Ca- 
ribbean Sea during 2010-2013 (arrows), and eastern Caribbean Sea (open squares) during 1998 and 2014; samples from 
fish collected in the central North Atlantic (Azores Islands) were taken in 1998. The numbers adjacent to open circles (fish 
release locations) and closed circles (fish recapture locations) correspond to the tagging information in Table 4. 
ported to CodonCode Aligner, vers. 4. 2. 5 (CodonCode 
Corp., Centerville, MA). Sequences were aligned and 
trimmed to a common length with Mesquite, vers. 2.75 
(Maddison and Maddison, 2011). Haplotype (h) and 
nucleotide diversity (ji) were calculated with Arlequin, 
(vers. 3.5; Excoffier and Lischer, 2010). Population dif- 
ferentiation and the geographical pattern of variation 
were examined with hierarchical analyses of molecu- 
lar variance (AMOVA) (Excoffier and Lischer, 2010) 
and pairwise comparisons of <J>st through the use of 
the Tamura-Nei model (Tamura and Nei, 1993). Signifi- 
cance of (P-statistics was assessed by 10,000 permuta- 
tions of groups and haplotypes. A gamma distribution 
parameter of a=0.881 was used to run AMOVA. This 
parameter was selected by prior sequence comparisons 
in jModelTest, vers. 2.1.3, after the GTR+I+G DNA 
substitution model was selected as the best-fit model 
on the basis of the Akaike information criterion (Posa- 
da, 2008; Darriba et al., 2012). 
Fu’s F§ (Fu, 1997) was used to test for deviation 
from the neutral model of molecular evolution (Kimu- 
ra, 1968). The demographic parameters T, 0 q. and 0\ 
were estimated from pairwise sequence distribution of 
mismatches according to the demographic and spatial 
expansion models. The translation of demographic pa- 
rameters to estimates of effective population size were 
obtained by following Bowen et al. (2006). To evaluate 
a null hypothesis of a population expansion, Harpend- 
ing’s raggedness index was calculated; failure to reject 
the null hypothesis ( i.e. , nonsignificant raggedness 
values) indicates that there is no support for a stable 
(nonexpanding) population (Rogers and Harpending, 
1992). Additional demographic and spatial expansion 
parameters, including time since expansion in years 
(T), effective female population size (Ne), and immigra- 
tion rate from neighboring denies (m), were generated 
according to the method of Diaz- Jaimes et al. (2006, 
2010), using a maximum generation time of 3 years for 
dolphinfish (Mahon and Oxenford, 1999) and a muta- 
tion rate of 1.2% per million years for marine teleosts 
(Bermingham et al., 1997). 
To test for differences in genetic structure between 
seasonal runs thought to occur around Puerto Rico, 
pairwise 0 st comparisons of samples taken from the 
