Lankford et al.; Population structure In Micropogonias undulatus 



887 



revealed significant heterogeneity between 

 Atlantic and Gulf samples (x2=24.36,p=0.020) 

 (Table 2). Parsimony analysis indicated that 

 geographically segregated haplotypes were 

 generally closely related ( 1-2 restriction site 

 changes) to the common haplotype (Fig. 1). 

 AMOVA revealed that the majority (95.4%) of 

 mtDNA variation in Atlantic croaker occurred 

 within samples (Table 3). A significant portion 

 (4.6%) was attributable to regional differences 

 (Atlantic versus Gulf, P<0.01) but variation 

 within the Atlantic region was unstructured 

 (p=0.10) and accounted for <0.1% of the total 

 genetic variance. 0gj values indicated a lack 

 of geographic structure and relatively high 

 gene flow among Atlantic localities. Gene flow 

 among Atlantic localities was estimated as 

 23.4 (DE and NO, 16.8 (NC and FL), and 12.0 

 (DE and FL) effective female migrants per 

 generation. 



O Atlantic 



DGulf 



• Atlantic & Gulf 



= 0.001 Nei's distance 



Figure 1 



Parsimony network relating 15 mtDNA haplotypes observed in 

 Atlantic croaker. Hatch marks drawn through each branch indi- 

 cate the inferred number of restriction site differences among 

 haplotypes. Branch lengths denote Nei's genetic distance between 

 haplotypes (note scale). 



Discussion 



MtDNA analysis provided no evidence that M. 

 undulatus is subdivided by Cape Hatteras into dis- 

 crete genetic stocks. Frequency- and distance-based 

 analyses both suggested a single, panmictic popula- 

 tion of croaker on the U.S. Atlantic coast. Low levels 

 of mtDNA divergence among Atlantic localities, al- 

 though not statistically significant, were more con- 

 sistent with a pattern of semi-isolation by distance 

 rather than marked subdivision by Cape Hatteras. 

 For example, AMOVA revealed that the two Atlantic 

 localities exhibiting the least genetic divergence were 

 DE and NC (0gT=OO21;p=O.89). NC and FL samples 

 were slightly more divergent but not significantly so 

 (cpg^=0.029; p= 0.180). Lack of population structure 

 was perhaps best indicated by the lack of differen- 

 tiation between DE and FL samples (<l>g^=0.040; 

 p=0.084>, two locations representing the distribu- 

 tional limits of this species' range on the U.S. Atlantic 

 coast. 



Kornfield and Bogdanowicz (1987) inferred pat- 

 terns of gene flow from distributions of unique 

 mtDNA haplotypes and their presumed precursors, 

 predicting that under restricted gene flow, unique 

 haplotypes should occur in the same population as 

 their precursors. This was the case for M. undulatus 

 at the regional (i.e. Atlantic versus Gulf) level, where 

 unique Gulf haplotypes 10, 12, and 14 each had likely 

 precursors that were confined to the Gulf region. 

 Within the Atlantic region, however, unique haplo- 

 types and their likely precursors occurred at dispar- 

 ate localities. The occurrence of rare haplotypes at 

 more than one Atlantic locality also implies that geno- 

 types arising at one locality spread rapidly to other 

 localities within the Atlantic. Combined with low lev- 

 els of nucleotide divergence, these observations sup- 

 port the hypothesis of contemporary gene flow among 

 Atlantic localities. Gene flow estimates based upon 

 Wright's island model are consistent with panmixia: 

 values ranged from N^,m =12.0-23.4, well above the 

 theoretical lower limit of Nj7i - 1 sufficient to pre- 

 clude genetic divergence of populations by random 

 drift (Slatkin, 1985). 



