16 



Fishery Bulletin 98(1) 



staining with 0.5 mg/mL ethidium bromide in water and 

 ultraviolet light illumination. Nelson et al. (1998) pro- 

 vide a complete description of gel electrophoretic condi- 

 tions. All gels were run for 14—18 h at 65-70 V, using 8% 

 acrylamide for analysis of OtslOO and Otsl03, and 10% 

 acrylamide for analysis of Omy77, Ots3, Otsl07 and 

 Otsl08. Twenty-nine lanes per gel were loaded. One out- 

 side lane contained a one-kb ladder (Gibco BRL), three 

 lanes contained a 20-bp ladder (Glensura Labs Inc., Del 

 Mar, CA) evenly spaced across the gel, one lane con- 

 tained a standard fish to determine precision of estima- 

 tion of allele size, and 24 lanes contained an individual 

 fish for analysis. 



Gels were scanned at a 1024 x 1024 pixel density 

 with a Kodak charge coupled device (CCD) camera 

 with low-light capability and a yellow filter. Images 

 were analyzed by using Biolmage Whole Band soft- 

 ware (Genomic Solutions Inc., 1995), where the size 

 of the amplified microsatellite alleles were reported to 

 the nearest base pair (bp) based upon the molecular 

 size grid created with the 20-bp markers. 



Because some uncertainty occuiTed in estimation of 

 allele size fi-om the 20-bp grid, we identified alleles on 

 the basis of a bimiing procedure ( Gill et al., 1990). Peaks 

 in the allele fi-equencies used to identify main alleles and 

 bin widths generally corresponding to a repeat unit were 

 set so that the main allele was located in the middle of 

 the bin. Precision of estimation of allele size was evalu- 

 ated with the standard fish analyzed for each locus. 



Data analysis 



Annual variation in allele fi-equencies within populations 

 was tested with GENEPOP version 3.1 with the Mai-kov- 



Chain approach by using;);'- probability values (Raymond 

 and Rousset, 1995). The dememorization number was set 

 at 1000, and 50 batches were run for each test with 1000 

 iterations/batch (Raymond and Ptousset, 1995 ). Each stock 

 at each locus was tested for departure from Hardy- Wein- 

 berg equilibrium by using GENEPOP. Gametic Linkage 

 disequihbrium between loci in each population was also 

 evaluated with GENEPOP. Tests of genetic differentiation 

 with three pairwise comparisons among the populations 

 wei-e also conducted with GENEPOP with the Markov- 

 Chain approach by using _;f2 probability values. Critical sig- 

 nificance levels for simultaneous tests were evaluated by 

 using sequential Bonferroni adjustment (Rice, 1989). Fgj. 

 estimates for each locus were calculated with GENEPOP, 

 and the standai'd deviation of the estimate for an individ- 

 ual locus was determined with FSTAT (Goudet, 1995) by 

 jackknifing over stocks and for all loci combined by boot- 

 strapping over loci. Estimation of variance components 

 of stock differences and annual variation within stocks 

 was determined with BIOSYS (SwoSbrd and Selander, 

 1981). Piincipal components of nine (three annual sam- 

 ples multiplied by three stocks) composite airays of allele 

 frequencies for six loci were calculated with the PRIN- 

 COMP procedure in SAS (SAS, 1989). 



Estimation of stock composition 



The effectiveness of using variation at microsatellite 

 loci for the practical assessment of stock composition in 

 mixed-stock fisheries of Barkley Sound was evaluated 

 from the stand points of precision of stock composi- 

 tion data and accuracy of stock composition estimates 

 in simulated fishery samples. Although only three 

 stocks could contribute to the fishei-y samples, we 

 wished to determine the sample 

 size required to detect accurately 

 the relatively small proportion of 

 Henderson Lake sockeye salmon 

 that were expected to be present 

 in most fishery samples. In addi- 

 tion, we wished to examine the 

 effect of the number of loci used in 

 the estimation of stock composi- 

 tion. The simulated mixtures were 

 composed of 30% Sproat Lake 

 fish, 60% Great Central Lake fish, 

 and 10% Henderson Lake fish 

 because these proportions are 

 the approximate long-term mean 

 of the Barkley Sound fishery. 



Allele fi'equencies wei'e deter- 

 mined for each locus in each 

 stock, and the model of Foumier 

 et al. ( 1984 ) was used to estimate 

 stock composition by the condi- 



