694 



Fishery Bulletin 91(4), 1993 



from 0.74% to 1.04%, and 1.04% for the pooled samples 

 (Table 2). 



Despite the high level of within-sample variation, 

 there was little evidence for genetic differentiation 

 among the six sampling sites. The two most common 

 genotypes (1 and 2) were observed at all locations at 

 similar frequencies (30-40%, 10-25%, respectively; 

 Table 1). Of the 12 other genotypes that occurred in 

 more than a single individual, 11 were found in two or 

 more samples, and only one (genotype 14) occurred in 

 a single sample, represented by two individuals. 



Pairwise estimates of corrected nucleotide sequence 

 divergences between samples (including the Atlantic) 

 were low, ranging from 0.012% to 0.104%, with a mean 

 pairwise divergence of 0.04%. Similarly, estimates of 

 G„ were low, ranging from 0.011 to 0.025, and values 

 of N e m were correspondingly high, ranging from 19.5 

 to 44.5 females per generation. No clear pattern of 

 phylogeographic structure was revealed in cluster 

 analyses of nucleotide sequence divergences among 

 mtDNA genotypes or sampling lo- 

 cations. Chi-square analysis of het- 

 erogeneity among samples was not 

 significant, as randomizations of the 

 data into six samples were more 

 heterogeneous than the observed 

 genotypic distributions 581 of 1000 

 times (P=0.581). 



To determine an appropriate 

 number of individuals to examine 

 from each location and the number 

 of restriction enzymes to employ, we 

 compared levels of genetic variation 

 and differentiation revealed by a pi- 

 lot study of 12 individuals per loca- 

 tion with 5 enzymes with larger 

 analyses of up to 20 individuals per 

 location with 12 enzymes (Table 3). 

 Levels of variation were influenced 



more by increasing the number of enzymes surveyed 

 than the number of individuals. Increasing the num- 

 ber of enzymes increased the number of genotypes, 

 the nucleon diversities, and, to a lesser extent, nucle- 

 otide sequence diversities. Analysis of a greater num- 

 ber of individuals per location had little effect on di- 

 versity estimates with either 5 or 12 enzymes, although 

 the ranges of within-sample diversities among loca- 

 tions decreased as more individuals were analyzed. 



Increasing the number of individuals or the number 

 of enzymes had little effect on levels of genetic differ- 

 entiation. No significant differences were found in the 

 distributions of genotypes among locations in Roff and 

 Bentzen ( 1989) chi-square tests (Table 3). Furthermore, 

 increasing the number of individuals in the 12 enzyme 

 analysis did not increase the frequencies of genotypes 

 unique to a location. Instead, many unique genotypes 

 in the analysis of 12 individuals occurred in other lo- 

 cations as a greater number of individuals were used. 

 Because increasing sample sizes from 12 to 20 indi- 

 viduals did not reveal greater spatial partitioning of 

 genetic variation, we decided that 20 individuals was 

 an appropriate sample size. 



Discussion 



Population structure is typically manifested as the spa- 

 tial or temporal partitioning of genetic variation. There- 

 fore, to demonstrate if population structure exists, a 

 technique must first reveal a reasonable level of ge- 

 netic variation. RFLP analysis of mtDNA revealed con- 

 siderable genetic variation in yellowfin tuna. The over- 

 all nucleon diversity and mean nucleotide sequence 

 diversity (0.84 and 0.91%, respectively) are in the up- 



