FISHERY BULLETIN: VOL. 82, NO. 4 



TABLE 2. — Allele frequencies and heterozygosity values at seven polymorphic 

 loci in the spiny lobster, Panulirus marginatus. N = number of individuals in 

 total sample; h = per locus heterozygosity; number of genes successfully scored 

 in parentheses; localities as in Figure 1. 



1 Pooled allelic classes as follows: Est-3' includes alleles 1 05 and 1 03 while Est-3 S includes alleles 

 97, 95, and 90; Gpi m includes alleles 80, 77, 74, 72, and 68 while Gpi s includes alleles 65, 58, 54, 48, 

 and 33; Pgm* includes alleles 240, 1 86, 1 80, 1 65, 1 57, and 1 44 while Pgm s includes alleles 82, 73, 

 53, 41 , 35, and 25; Umb f includes alleles 1 33, 1 20, 1 1 5, and 1 1 1 while Umb s includes alleles 85, 80, 

 and 75. 



2 Consisting of lobsters from Kure Atoll, Maro Reef, and Necker Island. 



3 Consisting of lobsters from Oahu and Hawaii. 



4 Subject to sex-restricted allele distribution (see text). 



5 Significantly different annual samples combined (see text). 



ture of the distribution of alleles at this sex-linked 

 locus (Shaklee 1983) effectively precluded 

 Hardy- Weinberg analysis involving the two most 

 common alleles. In spite of these caveats, out of 30 

 X 2 tests (6 loci x 5 localities) only 1 significant 

 deviation from Hardy- Weinberg expectation was 

 observed; a heterozygote deficiency for Pep-1 at 

 Necker Island ( x \ = 7.63 P < 0.01). Third, the 

 data in Table 2 represent the pooled allele fre- 

 quencies observed at each locality over the 2M>-yr 

 period of the study. In three cases (Umb at Kure 

 Atoll, Maro Reef, and French Frigate Shoals), 

 there was statistically significant year-to-year 

 fluctuation in allele frequencies. The significance 

 of this finding is discussed below. Fourth, because 

 of the unusual relationship between MPI pheno- 

 type and sex in this species (Shaklee 1983), the 

 medium and slow alleles had to be pooled into one 

 class to prevent differences in sex ratio in each 

 collection from biasing the allele-frequency 

 analysis. 



The first level of analysis, involving x 2 tests of 



all pairwise comparisons, failed to reveal convinc- 

 ing evidence of stock heterogeneity in this lobster. 

 Of the 66 comparisons, only three were significant: 

 1) Est, Kure vs. Maro ( X \ = 4.76 P < 0.05); 2) 

 Pep-1, Kure vs. Oahu (x 2 x = 5.07 P < 0.025); and 

 3) Pep-1, Necker vs. Oahu (\\ = 5.73 P < 0.025). 

 Given an a = 0.05 level of significance, one would 

 expect, on the basis of chance alone, about 3 sig- 

 nificant outcomes from 60 tests. Additionally, 

 given the basic linear arrangement of islands 

 within the Hawaiian Archipelago (Fig. 1), popula- 

 tion differentiation, if it were based upon isolation 

 by distance (Wright 1943), would be expected to be 

 most pronounced between widely separated 

 localities. In contrast to these expectations, two of 

 the three observed significant outcomes involve 

 adjacent, not distant, localities. 



The second statistical test compared the allelic 

 composition of lobsters from the NWHI (Kure 

 Atoll, Maro Reef, and Necker Island samples 

 pooled) with that of lobsters from the main 

 Hawaiian Islands (Oahu and Hawaii samples 



698 



