16 
Fishery Bulletin 107(1 ) 
Table 1 
Allele size ranges, in total DNA base pairs, for the microsatellite markers used to characterize populations of Cynoscion are- 
narius (sand seatrout) and C. nothus (silver seatrout) from offshore Galveston, TX, in July 2007. Each marker is listed by name 
as defined in the reference paper. The size range for Sciaenops ocellatus (red drum) was obtained from Saillant et al. (2004) and 
is included for reference. 
Name 
Allele range 
C. arenarius 
Allele range 
C. nothus 
Allele range 
S. ocellatus 
Reference 
SOC050 
161-197 
173-191 
183-183 
Turner et al., 1998 
SOC243 
100-112 
106-127 
94-106 
Turner et al., 1998 
SOC410 
301-323 
299-305 
306-344 
Saillant et al., 2004 
SOC412 
117-171 
117-147 
102-168 
Saillant et al., 2004 
SOC415 
187-295 
171-207 
187-235 
Saillant et al., 2004 
SOC416 
134-206 
142-170 
141-181 
Saillant et al., 2004 
SOC419 
232-268 
224-252 
238-260 
Saillant et al., 2004 
SOC428 
163-167 
165-167 
172-242 
Saillant et al., 2004 
SOC432 
98-118 
90-128 
94-132 
Saillant et al., 2004 
to the manufacturer’s protocol, and a 400-bp internal 
size standard (Beckman Coulter, Inc., Fullerton, CA) 
was included for fragment sizing. Each microsatellite 
locus was evaluated by a prioritized set of criteria: 1) 
amplification of the PCR product resulted in adequate 
signal intensity, 2) there were no excessive stepwise 
“stutter” bands preceding actual allele peaks, and 3) 
allelic polymorphism was present in both species. From 
the initial set of 16 loci, nine satisfied these criteria and 
consistently amplified a product (Table 1). Each experi- 
mental individual, for both species, was then genotyped 
at these nine loci. 
Analysis of microsatellite data 
The program Fstat (Goudet, 1995) was used to calculate 
allele diversity (number of alleles per locus), gene diver- 
sity, and conformity with Hardy-Weinberg expectations 
at each locus within each population. The latter was 
approximated by testing the significance of the statistic 
F is (Weir and Cockerham, 1984), which can be described 
as the within-population inbreeding coefficient. Sig- 
nificant departure of F is from 0 represents significant 
deviation from Hardy-Weinberg expectations. Fstat was 
used to detect the presence of linkage disequilibrium 
between loci within populations by using a nominal level 
of a = 0.05 and Bonferroni adjustment for multiple tests. 
Finally, Fstat was used to estimate genetic divergence 
between species as 6 (Weir and Cockerham, 1984) at each 
locus and overall. The significance of 8 was assessed by 
using the exact G-test of Goudet et al. (1996) with 1000 
randomizations and an arbitrary a = 0.05. 
In order to determine which microsatellites were the 
most informative for species assignment, we used the 
critical population method of Banks et al. (2003) in as- 
sessing the discriminatory power of individual markers. 
The freeware program Whichloci (Banks et al., 2003) 
was used to generate ten random sand seatrout popula- 
tions (rc=1000) based on empirical allele frequency data. 
These populations were used in simulated assignment 
procedures with constant assignment stringency (95% 
correct assignment of group members, 5% mis-assigned 
to critical population) and two conservative log odds 
ratio (LOD) assignment scores (LOD 2 and 3). The LOD 
assignment stringency is the log of the predetermined 
acceptable ratio of correctly assigned to incorrectly as- 
signed individuals (thus LOD of 2=logl0 of the ratio 
100:1). The critical population (sand seatrout) used for 
simulations was also a conservative selection because 
it was chosen after observation of trial runs to discern 
which population routinely needed higher numbers of 
loci for correct assignment. The output from these simu- 
lations included a list of loci ranked by discriminatory 
power of assignment, the locus score based on both 
type-I and type-II errors, and the relative score of each 
locus weighted by the overall additive score of the entire 
microsatellite panel. 
To identify hybrids resulting from crosses between 
these species, we used the Bayesian framework of 
Pritchard et al. (2000). The freeware program Struc- 
ture (Pritchard et al., 2000) attempts to estimate the 
number of genetic clusters present while simultaneously 
assigning individuals to groups. This is done in part 
through progressive minimization of linkage disequi- 
librium and Hardy-Weinberg disequilibrium in iterative 
Markov chain Monte Carlo steps. Three sets of data 
were used in independent runs. In the initial run, we 
used data from the six highest-ranked microsatellite 
loci from Whichloci analyses. A second run included 
data from all nine loci. Finally, a third run included 
the six highest-ranked loci, but mtDNA haplotype data 
were used to assign individuals a priori and microsat- 
ellite data were used to improve assignment. In each 
case, model parameters and run-times were specified as 
follows. The burn-in phase was set at 25,000 iterations 
and runs lasted 175,000 iterations under the admixture 
model. These values were chosen after inspection of 
model parameter normalization in preliminary runs. 
The Dirichlet parameter (a) was inferred from the data 
and was allowed to vary between populations. Allele 
