22 



Fishery Bulletin 98(1) 



tions in which annual variation has been detected, 

 the magnitude of variation has been substantially less 

 than that among populations (Nielsen et al., 1997; 

 Beacham and Wood, 1999). 



For sockeye salmon, in which the greatest geo- 

 graphic determinant of neutral genetic differentiation 

 is the nursery lake (Wood, 1995), the task of identify- 

 ing the contributions of three different lake systems to 

 a mixed-stock sample should be relatively straightfor- 

 ward. Although significant genetic variation can occur 

 among spawning sockeye salmon subpopulations iso- 

 lated by time or space (or both) within a lake system, 

 the extent of this variation is consistently much less 

 than that observed among lakes — even those lakes 

 within a single drainage system (Wood, 1995). Each 

 of the three lakes is the confluence of multiple tripu- 

 taries and may harbor genetically differentiated sub- 

 populations of sockeye salmon. The spawning ground 

 samples in our study were collected from locations 

 within each lake system at which fish ft-om more than 

 one subpopulation may have been present, and dif- 

 ferent subpopulations may have been sampled among 

 years. Thus, the departure of Omy77 (and OtslOS 

 for Henderson Lake) genotypes fi-om Hardy- Weinberg 

 equilibrium and significant annual variation observed 

 at these loci might both have reflected subpopulation dif- 

 ferentiation in allele fi-equencies. It is unlikely that the 

 heterozygote deficiency observed at Omy77 in Sproat 

 Lake and Henderson Lake sockeye salmon would be 

 a result of a null allele because genotypic fi-equencies 

 of other sockeye salmon stocks surveyed at this locus 

 have been in Hardy- Weinberg equilibrium (Beacham 

 and Wood, 1999). Nevertheless, the level of differentia- 

 tion at Omy77 was about 20 times greater among lakes 

 than was the temporal variation observed within lakes. 

 For all six microsateUite loci surveyed, differences among 

 lakes were on average 12 times greater than variation 

 within populations, confirming the relative stability of 

 the microsateUite loci in Barkley Sound sockeye salmon 

 populations over the 5—8 yr sampling period. 



The six microsateUite loci used in the current 

 study were also surveyed in nine sockeye salmon 

 stocks of the Nass River drainage in northern British 

 Columbia (Beacham and Wood, 1999). In the Nass 

 River, the three loci displaying the greatest differen- 

 tiation among stocks were OtslOO (Fgj^O.131), Ots3 

 (FsT^O.lll), and OtslOS (Fs7^0.084), whereas in the 

 Barkley Sound stocks, the three most discriminating 

 loci were Omy77 {Fg.,^0.W7), Ots3 (^^7^0.099), and 

 Otsl07 (Fgj^b.043). The fact that loci differed in their 

 relative levels of variation between the two areas 

 is not surprising given the rapid evolution of mic- 

 rosateUite loci and the likelihood that the regions 

 were founded postglacially by different sockeye salmon 

 "races" (Wood, 1995). For stock identification applica- 



tions, surveys of microsateUite variation in each geo- 

 graphic region of interest will generally be necessary 

 to determine which loci are the most effective in dif- 

 ferentiating local populations. 



Effective assessment and management of sockeye 

 salmorf production in Barkley Sound is dependent 

 upon determination of stock composition in fishery 

 catches. Previous evaluation has indicated that the 

 application of microsateUite technology to stock iden- 

 tification can provide the most reliable and cost-effec- 

 tive results (Beacham et al., 1998), but determination 

 of the feasibility of such technology for Barkley Sound 

 fisheries awaited examination of the relation between 

 the number of loci used, the sample size of the stock 

 mixture to be analyzed, and the precision of the esti- 

 mated stock contributions. For any stock identification 

 application, the optimal combination of number of loci 

 surveyed and number of fish sampled from the catch 

 is dependent on the genetic distance among stocks, 

 the desired precision for an individual stock estimate, 

 and the cost of the analysis for each locus. 



The simulated mixtures evaluated for Barkley 

 Sound sockeye salmon indicated that microsateUite 

 variation could be used to provide accurate and rea- 

 sonably precise estimates of individual stocks in the 

 catch mixtures. They ftirther indicated that although 

 genotjqaic fi-equencies at Omy77 and Otsl08 were not 

 in Hardy-Weinberg equilibrium in some stocks, but 

 assumed to be so in the stock composition estimation 

 procedure, the violation of this assumption did not 

 have a marked influence on the accuracy of the esti- 

 mated stock compositions. The precision, but not accu- 

 racy, of the estimated contributions increased with 

 both the number of loci (from 3 to 6) and the sample 

 size of the mixture (fi"om 100 to 300). For sample 

 sizes of 150 fish and larger, a greater increase in preci- 

 sion for stock contribution estimates could always be 

 achieved by increasing the number of loci surveyed to 

 six than by increasing the sample size to 300. How- 

 ever, these simulations did not include estimation of 

 the random error associated with sampling only a por- 

 tion of the catch, and this error will always be reduced 

 by increasing sample size. The level of precision of an 

 estimated stock contribution increased with the con- 

 tribution of the stock to the mixture. For estimation of 

 the more abundant Great Central and Sproat sockeye 

 salmon, the increase in precision afforded by additional 

 data was approximately equivalent whether more fish 

 (beyond 150) or more loci were analyzed (i.e. approxi- 

 mately equaUy precise stock contribution estimates were 

 achieved by analyzing four loci in 300 fish and six loci in 

 200 fish). However, estimation of the small {I07i) Hen- 

 derson Lake contribution to the mixture was more sensi- 

 tive to sample size and was more precise in the analysis 

 of four loci in 200 fish than of six loci in 150 fish. 



