672 
Fishery Bulletin 99(4) 
Figure 1 
Collection sites for southern flounder in North American waters. LLM = Lower Laguna Madre; 
MAT = Matagorda Bay; GAL = Galveston Bay; SAB = Sabine Lake; MS = Mississippi; AL = 
Alabama; STA = St. Augustine, Florida; NC = North Carolina. 
as current patterns (King et al., 1994). Support for com- 
peting models comes from examination of specific patterns 
observed in structured populations. Such patterns, if ob- 
served, may have important management implications. 
Materials and methods 
Southern flounder were collected during the summers of 
1996 and 1997 by rod and reel, flounder gigs, or Texas 
Parks and Wildlife (TPW) gill nets in four Texas bays 
(Sabine Lake, Galveston Bay, Matagorda Bay, and lower 
Laguna Madre), by pound nets in Core Sound, North Caro- 
lina, from gill nets in estuarine waters near Biloxi, Missis- 
sippi, and from commercial fish houses in Dauphin Island, 
Alabama, and St. Augustine, Florida (Fig. 1). Southern 
flounder from commercial fish houses were reported by the 
house operator to be caught locally. A majority of individu- 
als collected were adult and were not reliably assignable 
to year classes. Samples were screened by using isoelec- 
tric focusing (IEF) of sarcoplasmic proteins (methods of 
Ward et al., 1995) to insure that individuals belonging to 
other Paralichthys species were not included in our analy- 
ses (necessary because of accidental inclusion of congener- 
ics both in samples obtained from commercial sources and 
in samples of juvenile flounder obtained during routine 
resource sampling in Texas). 
Skeletal muscle, liver, heart, and kidney tissues were 
excised from fresh or frozen fish. Sample preparation and 
electrophoretic techniques and conditions followed those 
of King and Pate (1992). Gel and electrode buffers used 
were tris-borate-EDTA, pH 8.0 (Selander et al., 1971), tris- 
citrate, pH 8.0. (Selander et al., 1971), lithium hydroxide, 
pH 8.0 (Selander et al., 1971), borate buffer, pH 9.0 (modi- 
fied from Sackler, 1966), and Poulik’s discontinuous sys- 
tem, pH 8.7 (Selander et al., 1971). Histochemical meth- 
ods were primarily taken from Manchenko ( 1994). Genetic 
nomenclature followed the recommendations of Shaklee et 
al. (1990). 
BIOSYS-1 (Swofford and Selander, 1981) was used to 
generate an allele frequency table, to estimate the pro- 
portion of loci heterozygous ( H ) in the average individual, 
the proportion of polymorphic loci in individuals from 
each bay population, and genetic divergence using the 
E-statistics of Wright (1978). Exact tests calculated by 
GENEPOP (v. 3.1; Raymond and Rousset, 1995) were used 
to test for conformance of genotypic frequencies at each 
locus within a sample to Hardy- Weinberg expectations, 
genotypic linkage disequilibrium, and allelic and genotypic 
heterogeneity. Pairwise differences between samples were 
tested by using the genic differentiation randomization 
test in GENEPOP. Results were combined over loci with 
Fisher’s method (Sokal and Rohlf, 1994). Differences be- 
tween each pair of populations were summarized by us- 
ing the chord distance of Cavalli-Sforza and Edwards 
(1967). An unrooted phylogenetic tree was fitted to the 
chord distance matrix by using the neighbor-joining (NJ) 
algorithm. TreeView (Page, 1996) was used to visualize 
