Ward et al Genetic variability in Cynoscion nebu/osus, determined with DNA markers 



201 



homogeneous allele distributions and the overall test 

 was significant i.P^^^^,<0.001). Estimated theta values 

 were statistically significant after Bonferroni adjust- 

 ment, except Cnel33'. The overall estimate of theta 

 (6=0.057, 95% CI, 0.005-0.062) was lower than the 

 estimate including both spatial and temporal dimen- 

 sions (0=0.116), indicating that temporal differences 

 likely contributed to overall population differentiation 

 in spotted seatrout. 



Analysis of molecular variance indicated a statisti- 

 cally significant 4.11% of the among-sample genetic 

 variance was attributable to differences between Gulf 

 and Atlantic samples (P=0.04). Statistically signifi- 

 cant variance was also detected among sampling sites 

 within the Gulf of Mexico (0.49% of the total variance, 

 P<0.001). No significant genetic variance attributable to 

 temporal differences was found within bays (P=0.42). 

 An overall Fgj of 0.046 was estimated, which was sta- 

 tistically significant (P<0.001). 



We found no significant differences among Texas 

 bays in the 1998 collections (Table 2). In 1999, spotted 

 seatrout from Texas and Louisiana were not differenti- 

 ated except for the Galveston Bay samples that differed 

 significantly from all other samples. Analyses of year 

 2000 collections, which included samples from Texas, 

 Louisiana, and Florida, revealed differences between 

 Florida samples and all samples from Louisiana and 

 Texas. The Louisiana sample differed from most Texas 

 samples, and within Texas most samples were geneti- 

 cally undifferentiated, except for samples Galveston 

 Bay which were significantly different from those of 

 all bays, except Sabine Lake and Corpus Christi. Fi- 

 nally, the upper and lower Laguna Madre samples were 

 statistically different from each other. Temporally, the 

 only statistically significant differences among years 



were seen among Louisiana and Galveston Bay sam- 

 ples. The overall theta within sampling years ranged 

 from less than 0.006 in 1998 to 0.080 in 2000, perhaps 

 reflecting the increased genetic variability introduced 

 by the Florida samples in the 2000 data set. The sta- 

 tistically insignificant differentiation among sampling 

 years within bays, with the exception of the Louisiana 

 and Galveston Bay, supports the notion that temporal 

 differentiation, at least on the limited scale reported 

 in our study, may be ignored and temporal samples 

 can be collapsed within bays. The genetic structure 

 of spotted seatrout in Galveston Bay is highly dif- 

 ferentiated, both temporally and spatially, showing 

 significant differences in 1999 and 2000 with almost 

 all other bays. 



The topology of the neighbor-joining tree based on 

 D(. for 34 site and year groups (Fig. 2) was poorly sup- 

 ported by bootstrap replications and demonstrated little 

 correspondence to geographic and temporal patterns. 

 An exception was the separation of samples FCOO 

 and FSOO (Charlotte Bay and St. John's River, FL, 

 respectively; acronyms are defined in Table 1) from 

 samples collected in the northern and western Gulf. 

 When groupings were collapsed across years (Fig. 3), 

 the distinctiveness of the Florida samples from the 

 southern Gulf Coast and the Atlantic Coast continued 

 to be supported. Spotted seatrout from Florida's Tampa 

 Bay (FTOO) were found to be more closely related to 

 spotted seatrout from Louisiana and Texas than to 

 spotted seatrout in the other Florida samples. Texas 

 and Louisiana samples formed a well-differentiated 

 grouping; however, within that grouping, there was 

 little correspondence between genetic differentiation 

 and geographic location. One exception was the south 

 coast of Texas (Corpus Christi Bay and the upper and 



