Ward et al Genetic variability in Cynoscion nebulosus, determined with DNA markers 



199 



model (AMOVA; Michalakis and Ex- 

 coffier, 1996) in ARLEQUIN (Excof- 

 fier et al., 1992). The components of 

 genetic diversity attributable to vari- 

 ance between regions (Atlantic ver- 

 sus Gulf of Mexico), variance among 

 sampling sites within regions, tem- 

 poral variance among years within 

 sampling sites, and variance among 

 individuals within samples were 

 estimated. The significance of each 

 variance component was tested with 

 nonparametric permutation proce- 

 dures (-1000 permutations; Excof- 

 fier et al., 1992). In addition, genetic 

 differentiation among all collection 

 sites for each sampling year and be- 

 tween pairs of populations within 

 sampling years was estimated by 

 using the theta statistic of Weir 

 and Cockerham (1984) accessed on 

 FSTAT (Goudet, 1995). 



Cavalli-Sforza and Edwards' 

 chord distance (Df.; Cavalli-Sforza 

 and Edwards. 1967) was used to re- 

 construct phylogenetic relationships 

 among collection sites. Estimations 

 of D(. were obtained with the statis- 

 tical package NJBPOP (Cornuet et 

 al., 1999). Takezaki and Nei (1996) 

 found D(, to be a better estimate of 

 genetic divergence with microsatel- 

 lite DNA data than with measures 

 based on the step-wise mutation model. This estimate 

 is not based on the assumption of a constant population 

 size or a constant mutation rate among loci (Takezaki 

 and Nei, 1996) and appears to accurately resolve closely 

 related populations (Paetkau et al., 1997; Angers and 

 Bernatchez, 1998). A phenogram was generated from 

 the chord-distance matrix with the neighbor-joining 

 (N-J) algorhithm. Robustness of each node was evalu- 

 ated by bootstrapping over loci for 2000 replications 

 (Hedges, 1992) with the SEQBOOT program on PHY- 

 LIP, version 3.5c (Felsenstein, 1995). The PHYLIP 

 program CONSENSE then was used to generate a 

 consensus tree which was drawn with the program 

 TREEVIEW (Page, 1996). 



Results 



The number of alleles per sample (Table 1) exceeded 

 those reported for the same loci in red drum (Sciaenops 

 ocellatus), the species of origin for the markers (Turner 

 et al., 1998). Mean observed heterozygosity (Hq) ranged 

 from 0.21 to 0.39, and there were no statistically sig- 

 nificant deviations from Hardy-Weinberg expectations 

 at any locus and sample combination after Bonferroni 

 adjustment. Without the Bonferroni correction, allele 

 frequencies at 16 of 165 comparisons would have failed to 



Figure 1 



Sampling localities for spotted seatrout ^Cynoscion nebulosus) examined from the 

 northern Gulf of Mexico and the Atlantic Coast of Florida. LL = lower Laguna 

 Madre; UL = upper Laguna Madre; CC = Corpus Christi Bay; AB = Aransas Bay; 

 SA = San Antonio Bay; MB = Matagorda Bay; EM = east Matagorda Bay; GB = 

 Galveston Bay; SL = Sabine Lake; LA = Grand Isle, Louisiana; FT = Tampa Bay, 

 Florida; FC = Charlotte Harbor. Florida; FS = St. John's River, Florida. 



meet expectations. This included all loci except Soc243, 

 which was observed to be in Hardy-Weinberg equilib- 

 rium in all samples. Soc012 most often failed to meet 

 Hardy-Weinberg expectations; six of 33 samples were out 

 of equilibrium before Bonferroni adjustment. Observed 

 heterozygosity was lower than expected in 14 of the 

 16 locus-and-sample combinations that failed to meet 

 Hardy-Weinberg expectation before adjustment. Sta- 

 tistically significant linkage disequilibrium was noted 

 for one pair of loci in one sample {Cnel33' with Socl2 

 in sample GB99) after Bonferroni adjustments. If the 

 unadjusted critical value (a=0.05) was applied, 15 of 330 

 comparisons were statistically significant. Interestingly, 

 the associated loci Cnel33 and Cnel33' did not exhibit 

 linkage disequilibrium for any sample. 



After Bonferroni adjustment, exact tests for allele 

 distribution homogeneity across all 33 samples dem- 

 onstrated statistically significant differentiation for 

 all microsatellite loci except Soc50, which approached 

 statistical significance (P=0.05). All theta estimates 

 were significantly greater than zero after Bonferroni 

 adjustments, as was the overall theta of 0.116 (95% CI, 

 0.007-0.073; P<0.001), indicating significant genetic 

 differentiation across both the spatial and temporal di- 

 mensions sampled in this study. When spatial samples 

 were collapsed to form 13 spatial samples, all loci except 

 Soc50 exhibited statistically significant deviations from 



