530 



Fishery Bulletin 103(3) 



PGM-2* in the South Tanaga Island sample was due to 

 an excess of heterozygotes. 



Inbreeding coefficients (F is ) indicated deviation from 

 panmixia. The values ranged from -0.050 for IDHP-1* 

 to 0.772 for ACP*. The mean value over all loci in all 

 collections was 0.140. Statistically significant F is val- 

 ues were found at ACP* (0.521), MPI* (0.135), PGM-2* 

 (0.109), and XO* (0.524). All were the result of het- 

 erozygote deficiencies. The mean F is value for the S. 

 aleutianus type collections dropped to 0.062 and for the 

 S. sp. cf. aleutianus types, to 0.048. 



Eight of the loci showed statistically significant F sl 

 values: sAAT* (0.007), ACP* (0.572), sAH* (0.037), 

 IDDH* (0.189), MDH-2* (0.007), MPI* (0.123), PGM-2* 

 (0.206), and XO* (0.551). The mean F, value for all 

 loci in all collections was 0.215. When analyzed by 

 pure S. aleutianus-type and S. sp. cf. aleutianus- 

 type, the mean F sl values dropped to 0.013 and 0.012, 

 respectively. 



Two rougheye types 



The results of the rougheye rockfish analyses allowed 

 us to segregate rougheye rockfish individuals into two 

 types: S. aleutianus and Sebastes sp. cf. aleutianus. 

 Multidimensional scaling analysis with individual geno- 

 types (Fig. 2) yielded two distinct clusters with little 



overlap. This outcome was confirmed by the Bayesian 

 clustering model in STRUCTURE (Pritchard et al., 

 2000), which identified two types. We calculated the 

 inferred source proportions of genomes for 1027 indi- 

 viduals using 25 loci that were scored in most individu- 

 als. One hundred sixty-six individuals were missing 

 data for more than 30% of the 25 loci used and were 

 omitted from the analysis. Most of the individuals had 

 a very high proportion of their genome from one type; 

 for 851 individuals, the program assigned at least 0.80 

 of the individual's genes to one of the two ancestral 

 lines, and all had an upper 90% probability limit that 

 included 1.0. These fish were likely all purebreds. Ten 

 individuals had an inferred proportion of ancestry from 

 one lineage of less than 0.80 and two did not include 

 an upper probability interval of 1.0. These individuals 

 were possibly hybrids. If any of these 10 individuals 

 were actual hybrids of the two rougheye rockfish types, 

 none were of the first generation (i.e., heterozygotes at 

 all differentiating loci). 



Significant differences of allele frequencies (G- 

 test) were detected between the two types at 14 loci: 

 P< 0.001 for sAAT*, ACP*, ADH*, sAH*, IDDH*, MDH- 

 2*, mMEP*. MPI*, PGDH*, PGM-1*, PGM-2*, and XO* ; 

 and P<0.05 for mAAT* and GPI-B*. When the two types 

 were analyzed independently by area (Table 4), all but 

 two collections were in Hardy-Weinberg equilibrium 



