498 



Fishery Bulletin 100(3) 



Table 5 



Results of tests for spatial homogeneity in allele distribution of seven microsatellites between and among pooled samples of king 

 mackerel (Scomberomorus cavalta) from the east and west coasts of Florida and the Florida Keys. Pj„,,. = probability based on 

 Fisher's exact tests, with 1000 permutations. P„j = probability based on 1000 bootstrapped replicates (after Roff and Bentzen 

 1989). <t,,^ = estimate of population subdivision based on amova; P is the probability that 0,^ differs significantly from zero (5000 

 permutations). 



' Atlantic includes JKV'^JKV-. NSB, CCN, SEB' - SEB^", FTP, and WPB (nine samples total). For explanation of these geographic abbreviation, see 



Table 1. 

 - Gulfincludes MCI. BCG, SAR'-SAR^ and PCY (seven samples total). 

 ' Keys includes KEY'-KEY' (three samples totali. 



tion. In a(i(iition. one wouW expect that eight of 162 tests 

 would be significant by chance alone at a = 0.05. Finally, 

 only one of the significant probability values occurred in a 

 pooled comparison {Sca-49 in the test of individuals from 

 the Atlantic), where larger sample sizes were expected to 

 increase robustness of tests of independence. In tests for 

 independence of allele distributions with the age (year 

 class) of individuals, eight significant probability values 

 (P<0.05) were found prior to Bonferroni correction for 

 simultaneous tests (data available from first author), two 

 of which (Sca-23 in SAR* and Sca-61 in SAE"*) were sig- 

 nificant after Bonferroni correction with the use of eight 

 simultaneous tests per microsatellite to estimate adjusted 

 a levels. In both instances, nonindependence appeared to 

 stem from an elevated incidence of specific alleles: Sea- 

 23*19 occurred at a frequency of 36% in the 1989 year 

 class, as opposed to other year classes where its frequency 

 ranged from C/f to 15%; and similarly, Sca61*12 occurred 

 at a frequency of 25-30% in the 1986 and 1989 year 

 classes, as opposed to a frequency of 0-10% in the other 

 year classes. We suspect these are anomalous instances 

 that do not reflect an age-related effect, in part because al- 

 lele distributions at Sca-23 and Sca-61 were independent 

 of year class in all other samples, and in part because al- 

 lele distributions at all microsatellites were independent 

 of year class in pooled comparisons, where larger sample 

 sizes should increase robustness of tests of independence. 

 We concluded that allelic variation at the seven microsat- 

 ellites essentially is independent of variation in both sex 

 and age (year class). 



Spatial homogeneity in allele distributions at each mi- 

 crosatellite was tested 1) over all 20 samples, 2) among 

 samples from the Atlantic, 3) among samples from the 



Florida Keys, and 4) among samples from the Gulf Only 

 three significant probability values were found prior to 

 Bonferroni correction: the exact test at Sca-23 in the com- 

 parison over all 20 samples, and both the exact test and 

 the Roff-Bentzen procedure at Sca-37 in the comparison 

 among samples from the Florida Keys (Table 4). None 

 of the probability values were significant after Bonfer- 

 roni correction. Tests for spatial homogeneity 1) between 

 pooled samples from the Atlantic versus pooled samples 

 from the Gulf, and 2) among pooled samples from the 

 Atlantic, Florida Keys, and Gulf indicated that all three 

 regional groupings differed genetically from one another 

 (Table 5). For the comparison Atlantic versus Gulf sig- 

 nificant heterogeneity prior to Bonferroni correction was 

 found at Sca-14 (all three statistical approaches), Sca-23 

 (exact test and the Roff-Bentzen procedure), and Sca-44 

 (for the probability that <Pi-.j>0). Probability values for 

 Sca-14 and Sca-23 were marginal in relation to the (ini- 

 tial) Bonferroni adjusted a of 0.007, whereas the probabil- 

 ity that 0f.j. > zero at Sca-44 was nonsignificant after Bon- 

 ferroni correction (Table 5). For the comparison of Atlantic, 

 Florida Keys, and Gulf samples, significant heterogeneity 

 was found at Sca-14 (all three statistical approaches) 

 and Sca23 (exact test only) before but not after Bonfer- 

 roni correction; heterogeneity at Sca-37 and Sca-44 in the 

 same comparison was significant both before and after 

 Bonferroni correction in at least one of the three statisti- 

 cal approaches (Table 5). Frequency differences at Sca-14, 

 Sca-23. and Sca-44 among the three regional groupings 

 are shown in Table 6 and indicate that small differences in 

 frequency of several alleles at each microsatellite appear 

 to account for observed heterogeneity among the pooled 

 sample comparisons. 



