20 



Fishery Bulletin 98(1) 



(pairwise F^^j, estimate: 0.042), and the Sproat Lake 

 and Henderson Lake stocks showed the greatest dif- 

 ferentiation (pairwise Fgj. estimate; 0.091). 



Estimation of stock composition 



The three loci with the highest Fgj< estimates (Omy77, 

 Ots3, and Otsl07) also possessed the highest ratio of 



Table 5 



Average estimated stock composition C/r ) of simulated mixtures of Barkley Sound sock- 

 eye salmon based on variation at three to six microsateUite loci. True mixture percent- 

 ages were as follows: Sproat Lake 30*7^, Great Central Lake 60'7r. and Henderson Lake 

 10'7( . The three loci initially used to estimate the percentages of the mixtures were 

 Omy77, Ots3, and Otsl07. and OtslOO, Otsl08, and Otsl03 were added sequentially. 

 Each mixture was generated 100 times with replacement, and stock compositions of the 

 mixtures were estimated by resampling each baseline stock with replacement to obtain 

 a new distribution of allele frequencies, with the same sample size in the new distribu- 

 tion as in the original one. Standard deviation is given in parentheses. 



Mixture size (no. offish) 



Source and 

 loci 



100 



150 



200 



Sproat Lake 

 3 

 4 

 5 

 6 



30.7 (9.98) 

 30.7(8.21) 

 30.1(7.54) 

 31.2(7.23) 



Great Central Lake 



3 

 4 

 5 

 6 

 Henderson Lake 

 3 

 4 

 5 

 6 



60.5(11.74) 

 60.4(9.41) 

 60.1(8.26) 

 58.8(8.03) 



8.8(5.62) 



8.8(5.30) 



9.8 (4.56) 



10.0(4.33) 



31.1(8.51) 

 30.8(6.87) 

 30.6(5.87) 

 29.0(5.27) 



60.1(9.27) 

 ,59.8(7.40) 

 60.0(6.63) 

 60.9(5.96) 



8.8(4.84) 



9.3 (3..53) 



9.5 (3.73) 



10.1(3.70) 



30.8(6.81) 

 30.8(6.38) 

 30.0(4.44) 

 29.8 (4.42) 



60.2(7.40) 

 .59.7 (6.88) 

 60.1(5.11) 

 60.1(5.09) 



9.0(3.89) 



9.6(3.52) 



9.9(2.76) 



10.0(2.81) 



spatial to temporal variation (Table 4) and were there- 

 fore selected to form the core database for the analysis 

 of the simulated mixtures. The number of loci used 

 in the determination of stock composition or mixture 

 sample size had little effect upon the accuracy of the 

 estimated stock compositions (Table 5). Precision of 

 the estimated stock compositions increased as both 

 the number of loci and mixture size used in the deter- 

 mination increased. However, different options were 

 available to obtain estimates of a desired precision. 

 For example, higher levels of precision were obtained 

 with four loci (Omy77, Ots3, Otsl07, OtslOO) in con- 

 junction with a 150-fish sample size (600 units of 

 data) than with all six loci and a 100-fish sample 

 (600 units of data) (Table 5). The coefficient of varia- 

 tion for the estimated proportion of the predominant 

 Great Central Lake stock was always less than that 

 for estimated proportions of the other two stocks in the 

 mixture. For the 4 loci in 150-fish mixture analysis, 

 the coefficient of variation for the estimated propor- 

 tion of Great Central Lake fish was 12%, whereas it 

 was 22'^ for Sproat Lake fish, and 38% for Henderson 

 Lake fish. The simulations indicated that fewer than 

 six microsateUite loci could be used to provide rea- 

 sonably precise data and accurate estimates of sock- 

 eye salmon stock composition for 

 Barkley Sound fishery samples. 

 Conditional maximum likeli- 

 hood estimation can overesti- 

 mate the relative abundance of 

 rare stocks. For Barkley Sound, 

 this would likely be the Hender- 

 son Lake stock. The precision of 

 data and accuracy of stock com- 

 positions estimates were inves- 

 tigated for mixture samples of 

 100 fish composed of 2% Hen- 

 derson Lake (38% Sproat Lake, 

 60% Great Central Lake) and 

 5% Henderson Lake (35% Sproat 

 Lake, 60% Great Central), where 

 stock compositions were esti- 

 mated by using the four mic- 

 rosateUite loci (Omy77, Ots3, 

 OtslOO, and Otsl07) generally 

 used for estimation of stock 

 compositions in the 1997 fish- 

 ery samples. Estimated stock 

 compositions of the simulated 

 100 mixtures for the 2% Hender- 

 son Lake composition were 2.5% 

 (SD=3.1%) Henderson Lake, 

 34.3% (SD=9.3%) Sproat Lake, 

 and 60.4% (SD=9.3% ) Great Cen- 

 tral Lake. For the 5% Hender- 



300 



30.7 (6.48) 

 30.2(4.44) 

 29.4(4.18) 

 29.8(4.00) 



59.7(7.74) 

 60.7(5.21) 

 60.8(4.50) 

 59.9(4.31) 



9.6(3.98) 



9.1 (2.89) 



9.8(2.52) 



10.3(2.39) 



