198 



Fishery Bulletin 105(2) 



mtDNA study, significant heterogeneity was found in 

 haplotype frequencies among collection sites, indicating 

 that spotted seatrout were spatially differentiated. In 

 contrast, Gold et al. (2003) found no significant differ- 

 ences in microsatellite DNA allele frequency among 

 spotted seatrout inhabiting Texas bays. A similar study 

 of microsatellite variation among spotted seatrout of 

 the U.S. Atlantic and Florida Gulf coasts (employing 

 different loci from those of Gold et al., 2003) found 

 extensive spatial differences that coincided with known 

 zoogeographic barriers (Wiley and Chapman, 2003). 

 Surprisingly, the Indian River spotted seatrout sample 

 on Florida's Atlantic coast was genetically more similar 

 to the Choctawhatchee sample from the Florida Pan- 

 handle than to more northerly Atlantic Coast samples. 

 The overall pattern that emerges from applications 

 of molecular markers to the examination of spatial 

 genetic variability in spotted seatrout is mixed. Most 

 studies have found limited gene flow between adjacent 

 bays and moderate population subdivision, and pat- 

 terns of differentiation that may be described as "isola- 

 tion by distance." The failure of the Gold et al. (2003) 

 to find statistically significant genetic subdivision is 

 surprising, given that studies employing allozymes 

 and mtDNA have been successful in discerning genetic 

 structuring and given that microsatellites are consid- 

 ered to be among the most capable genetic markers 

 at resolving population level differentiation (Wright 

 and Bentzen, 1994). Gold et al. (2003) focused on the 

 northwestern Gulf of Mexico, and their geographically 

 limited examination may have restricted the ability of 

 the marker to discern patterns in the spatial variability 

 of spotted seatrout. It is also possible that underlying 

 genetic properties of the markers accounted for the 

 differences in observed variation, or that some mark- 

 ers (i.e., allozymes) were able to detect genetic adapta- 

 tion to environmental gradients which microsatellites, 

 assumed to be selectively neutral, were not. 



The present study is an attempt to re-examine genetic 

 variability in spotted seatrout in the northern Gulf of 

 Mexico. As did Gold et al. (2003) (who also examined 

 a portion of the present data), we used microsatellite 

 markers. Samples from Louisiana and the Gulf and 

 Atlantic coasts of Florida were added in an attempt to 

 give a greater spatial perspective that may be useful 

 in evaluating the observed genetic variability among 

 populations. In addition, samples from multiple years 

 were included for Texas and Louisiana bays, allowing 

 for evaluation of the stability and robustness of detected 

 genetic structure. 



Materials and methods 



A total of 1256 individuals were examined. Sample 

 collection sites are shown in Figure 1. All Texas sites 

 were sampled in three consecutive years, 1998-2000, 

 except Corpus Christi Bay which was sampled across 

 four years, 1998-2001. The Louisiana site was sampled 

 in 1999 and in 2000, and the Florida sites were sampled 



only in 2000. Texas and Florida samples were obtained 

 through routine resource sampling efforts of Texas Parks 

 and Wildlife Department and Florida Department of 

 Natural Resources personnel, respectively, and Loui- 

 siana samples were donated by licensed recreational 

 anglers. Soft dorsal-fin tissue was removed from the fish, 

 placed in 95% ethanol, and stored at room temperature 

 until processed. Genomic DNA was extracted by using 

 the PureGene DNA isolation kit and protocols (Centra 

 Systems, Inc., Minneapolis, MN). 



Primers designed to amplify three microsatellites 

 (Socl2, Soc50. and Soc243) originally developed for 

 red drum {Sciaenops ocellatus) by Turner et al. (1998) 

 were employed. Two additional primer pairs (C?!el33 

 and C7!el33') were designed by sequencing products of 

 the Socl33 (Turner et al., 1998) forward and reverse 

 primers and then identifying internal primer sequences 

 that amplified separate repeat regions within the origi- 

 nal iSocl33 amplicon. Primer sequences for Cnel33 and 

 C7!el33' and protocols for amplification and interpreta- 

 tion are discussed in Gold et al. (2003). 



Summary statistics were generated by using the Mi- 

 crosoft Excel add-on Microsatellite Toolkit (Park, 2001) 

 and ARLEQUIN, version 2.001 (Schneider et al., 2000), 

 which was also employed to test frequencies for devia- 

 tions from Hardy-Weinberg equilibrium by using exact 

 tests performed with Markov-chain randomization (Guo 

 and Thompson, 1992). Permutations with 1000 resam- 

 plings (Manly, 1991) were used to generate probability 

 values (P) for each test of Hardy-Weinberg equilibrium 

 for each microsatellite locus in each sample. ARLE- 

 QUIN was also used to test for linkage between micro- 

 satellite loci, and significance of P-values was estimated 

 by 1000 resamplings. Critical values for interpreting 

 significance levels for simultaneous inferential compari- 

 sons were adjusted by using the sequential Bonferroni 

 approach (Rice, 1989). 



Allelic distribution homogeneity of each microsatellite 

 was assessed with exact tests implemented with the 

 statistical package GENEPOP, version 3.4 (Raymond 

 and Rousset, 1995), and significance was estimated 

 by permutation with 1000 resamplings for each com- 

 parison. Population subdivision was estimated with 

 Weir and Cockerham's (1984) theta (B) as generated in 

 FSTAT, version 2.9.3 (Goudet, 1995), and a bootstrap 

 procedure in FSTAT was employed to calculate a 95% 

 confidence interval (CI). Although the use of theta is 

 contigent on the assumption of an infinite-alleles mu- 

 tation model (Kimura and Crow, 1964), it has been 

 shown to compare favorably with other measures of 

 genetic subdivision when employed with microsatellite 

 data (Ruzzante, 1998). The significance of population 

 differentiation across all loci was estimated as the com- 

 bined probability of P-values for Fisher's exact tests for 

 individual loci. Separate analyses were made for data 

 sets that comprised 13 samples defined by site and col- 

 lection date and 13 samples defined by site alone, with 

 date combined across all year classes. 



A hierarchical analysis of gene diversity was per- 

 formed by using the analysis of molecular variance 



