252 



Fishery Bulletin 101(2) 



tions virtually did not contribute to the fishery from April 

 through June. 



Mid Fraser commercial net fishery 



Samples from the net fishery in the mainstem Fraser 

 River were obtained from areas largely upstream of the 

 confluence of the Thompson and Fraser rivers. Thus, the 

 estimated percentage of Thompson River chinook salmon 

 was <2% in this fishery (Appendix Table 2). Mid-Fraser 

 populations were estimated to comprise 63% of the samples 

 from late June and early July in 1998 in this fishery, and 

 the Chilcotin River population was the dominant popula- 

 tion. By mid July, the majority of the fish sampled origi- 

 nated from upper Fraser populations, and chinook salmon 

 from Salmon River, Bowron River, McGregor River, and the 

 mainstem Fraser River at Tete Jaune were the dominant 

 contributors to the catch (Appendix Table 2). 



Lower Fraser test fisiiery 



As was observed in the commercial gillnet fishery in the 

 lower Fraser River, chinook salmon from mid-Fraser popu- 

 lations dominated the catch in April and May, comprising 

 over 50% of the fish sampled (Appendix Table 3). The 

 Chilcotin River and Stuart/Nechako rivers populations 

 were the main populations from the mid-Fraser region. 

 However, unlike the commercial gillnet fishery, salmon 

 from the lower Thompson River comprised 5% or less of 

 the catch in April and May. North and South Thompson 

 River populations comprised <5% of the catch as well, 

 as was observed in the commercial gillnet fishery. Upper 

 Fraser River populations had largely passed through the 

 test fishery by the end of July. Chinook salmon from the 

 North and South Thompson rivers dominated the samples 

 in August, and the mainstem-spawning South Thompson 

 population was the dominant population in the fishery. By 

 September chinook salmon from the lower Fraser River 

 were the main group of fish sampled in the test fishery, 

 and they comprised 45% of the catch. By October, they 

 dominated the test fishery, comprising more than 80% of 

 the chinook salmon sampled. 



Discussion 



Evaluation of microsatellites 



The survey of microsatellite variation of Fraser River chi- 

 nook salmon was initiated to determine genetic structure of 

 chinook salmon populations within the Fraser River drain- 

 age and to provide population-specific estimates of stock 

 composition in mixed-stock fisheries in the drainage for 

 management purposes. Analysis of simulated mixtures has 

 generally indicated that the estimates of stock composition 

 are sufficiently accurate such that reliable estimates of popu- 

 lation-specific composition should be obtained when applied 

 to mixed-stock fisheries. Application to a CWT sample indi- 

 cated that the average error of the estimated percentage for 

 the seven populations present was 29( per population and for 



the five regions present was 1.4%. These levels of accuracy 

 were judged to be sufficient for management applications. 

 Indeed, there is no other technique currently available that 

 can provide current levels of accuracy in estimation of stock 

 composition for Fraser River chinook salmon. 



The 13 microsatellite loci evaluated in our survey clearly 

 differed in their ability to provide accurate estimates of 

 stock composition. Generally, loci with fewer numbers of 

 alleles (<20) were less effective for population identification 

 than were loci with greater numbers of alleles. Theoretical 

 studies of locus characteristics to guide selection for indi- 

 vidual identification suggested that a modest number of 

 independent loci was best, where each locus would have a 

 modest number of alleles and where each allele had a mod- 

 est frequency (Smouse and Chevillon, 1998). For chinook 

 salmon, loci with greater than 20 observed alleles would 

 likely be more effective for stock or individual identification 

 than loci with fewer allleles. With respect to number of loci 

 to include in stock identification applications, analysis of 

 the simulated samples indicated that bias was minimized 

 when all 13 loci surveyed were included in the analysis, but 

 the least effective loci provided only a modest increase in 

 accuracy of estimated stock compositions. Increasing the 

 number of loci included in the stock identification applica- 

 tions would be the preferred option, provided that the num- 

 ber of loci included in the analysis provided a cost-effective 

 method for fishery management applications. 



Estimation of stock composition and classification of 

 individuals to specific populations are two goals for stock 

 identification, but estimation of stock composition is the 

 more practical goal for fisheries management. In stock 

 composition analysis, the characteristics of the whole 

 sample are used to provide the most likely estimate. For 

 classification of individuals, only the characteristics of the 

 individual to be identified are used. Because more informa- 

 tion is available from a stock mixture rather than from a 

 single individual, and misallocations between individual 

 populations will cancel, estimates of stock composition will 

 generally be more accurate than classifications of individu- 

 al fish. For example, individual Chilliwack River fish were 

 correctly identified to river of origin approximately Sl'/f of 

 the time, but in the 83-fish CWT sample, the estimate of 

 the Chilliwack River component was within 4'7( of the true 

 estimate, equivalent to estimating about 25 Chilliwack 

 River fish present instead of 21 fish. Although more diffi- 

 cult, identication of individual fish to specific river of origin 

 does have some management applications for Fraser River 

 chinook salmon. Because the Birkenhead River population 

 is very distinct genetically, it is possible to identify spe- 

 cific individuals as originating from the Birkenhead River 

 with a high degree of accuracy in fisheries both within 

 and outside of the Fraser River drainage. Given current 

 conservation concerns for the Birkenhead River population 

 and with the appropriate level of sampling, it is possible 

 to identify all areas and periods in which chinook salmon 

 from this population are present. 



Given the large number of chinook salmon populations 

 spawning in the Fraser River drainage, the area of the 

 drainage, and the cost of both obtaining representative 

 samples from spawning populations and their analysis. 



