Gold et a\ Genetic studies of Scomberomorus cavalla in Flonda 



495 



Figure 1 



Sampling localities for king mackerel examined in the present study. Acronyms for sample localities are 

 defined in Table 1. 



of the distribution of genotypes at each microsatellite with 

 the sex and age of individuals. Initial tests involved each 

 of the 20 samples separately. We then pooled individuals 

 sampled at Atlantic localities (nine samples), in the Flori- 

 da Keys (three samples), at Gulf localities (eight samples), 

 and over all localities (20 samples) in order to increase cell 

 sizes in individual tests. Probability (P) values for these 

 tests of independence were estimated by permutation 

 (1000 resamplings) and significance levels for simultane- 

 ous tests were adjusted with the sequential Bonferroni 

 approach. 



Tests of genetic homogeneity among samples included 

 exact tests, as implemented in genepop, the Monte Carlo 

 procedure of Roff and Bentzen (1989), as implemented in 

 the restriction enzyme analysis package of McElroy et al. 

 (1992), and the analysis of molecular variance (amova) of 

 Excoffier et al. (1992). Significance of tests of genetic ho- 

 mogeneity employed permutation with 1000 resamplings 

 per individual comparison, and significance levels for si- 

 multaneous tests were adjusted by using the sequential 

 Bonferroni approach. Tests of genetic homogeneity were 

 carried out separately for each of the seven microsatel- 

 lites. Individual tests were carried out 1) among all 20 

 samples, 2) among samples (nine) from Atlantic localities, 

 3) among samples (three) from the Florida Keys, and 4) 

 among samples (eight) from Gulf localities. Analysis of 

 molecular variance (amova) was employed to generate es- 

 timates of (genetic) variance components and <f> statistics 

 for the same comparisons. (t> statistics are a set of hierar- 



chical F-statistic analogs that consider evolutionary dis- 

 tance among alleles. Significance of <P statistics employed 

 permutation (1000 resamplings). 



Exact tests, the Monte Carlo procedure of Roff and 

 Bentzen, and amova also were used to assess genetic ho- 

 mogeneity 1) between samples from the Atlantic versus 

 those from the Gulf (excluding samples from the Florida 

 Keys), and 2) among samples from the Atlantic, the Flori- 

 da Keys, and the Gulf This design was chosen a priori, in 

 part because it was geographically logical, in part because 

 the southern Florida peninsula apparently serves (or 

 has served) as a biogeographic boundary for a number of 

 marine species (Avise, 1992; Gold and Richardson, 1998). 

 The hierarchical capability of amova also permitted a test 

 of homogeneity among samples within the three regional 

 groupings. We also carried out homogeneity testing to 

 examine the temporal stock boundaries currently used 

 in management planning for the king mackerel resource. 

 Each of the 20 samples of king mackerel was designated 

 as either Atlantic or Gulf stock according to the time of 

 year during which they were sampled. Six of the samples 

 thus were designated Atlantic stock and 14 of the samples 

 were designated Gulf stock (Table 1). Genetic homogeneity 

 was then tested between the Atlantic and Gulf "stocks," by 

 using exact tests and the Monte Carlo procedure of Roff 

 and Bentzen. 



Clustering of genetic distances, spatial autocorrelation 

 analysis, and assignment tests also were employed to 

 assess temporal and spatial variation of microsatellites. 



