232 



Fishery Bulletin 101(2) 



son et al. (2001), and for loci in other species by Beacham 

 and Wood (1999) and Beacham et al. (2000). Precision of 

 estimation of allele size for Otsl04, OtslOJ, and Ssal97 

 was similar to that for other loci with alleles spanning a 

 similar size range. For example, estimated allele size for a 

 standard fish analyzed 515 times at Otsl07 was within a 

 4-bp interval 98'7( of the time for an estimated allele size of 

 171 bp, and 85% of the time for an allele size of 300 bp. As 

 outlined by Nelson et al. (2001), bin widths were expanded 

 at larger allele sizes to account for less precise estimation 

 of allele size as determined from the standard fish, and al- 

 leles were defined on the basis of bin width. 



With the acquisition of an automated DNA sequencer 

 (the ABI Prism 377) in our laboratory, PCR products at 

 seven additional loci — Ogo2, Ogo4 (Olsen et al, 1998), Oke4 

 (Buchholz et al., 1999), Omy325 (O'Connell et al, 1997), 

 OkilOO (K. M. Miller, unpubi. data), and Ots2, Ots9 (Banks 

 et al., 1999) — were size fractionated (96 samples on a gel) 

 on 4.5% denaturing polyacrylamide gels run at 3000 V for 

 2.25 h at a gel temperature of 51°C. PCR products at Ots2 

 and Ots9 were amplified with a single PCR reaction, as 

 were Ogo4 and Omy325, and Ogo2 and Oke4. Primer con- 

 centrations in the multiplex PCR reactions were adjusted 

 to ensure equal amplification of both loci. Allele sizes were 

 determined with Genescan 3.1 and Genotyper 2.5 software 

 (PE Biosystems, Foster City, CA). Allele frequencies for all 

 location samples surveyed in this study are available at 

 http://www-sci.pac.dfo-mpo.gc.ca/aqua/pages/bgsid.htm. 



Conversion of allele sizes between manual 

 and automated sizing systems 



With the acquisition of an automated DNA sequencer, we 

 switched the survey of variation at OtslOO, OtslOl, Otsl02, 

 Otsl04, OtslOJ, and Ssal97 from the previous manual gel 

 method to the automated sequencer, although most of the 

 alleles surveyed in this study at these loci would have been 

 analyzed on manual gels. Estimated allele sizes at these 

 loci differed between those derived from nondenaturing 

 gels stained with ethidium bromide and those derived from 

 the denaturing gels and flourescent tags on the automated 

 sequencer In order to convert allele sizes between the two 

 systems, we analyzed approximately 600 fish on both sys- 

 tems and determined the distributions of allele frequen- 

 cies. By inspection of the allele frequencies, we were able 

 to match specific allele sizes obtained from the sequencer to 

 specific allele sizes from the manual gels and then convert 

 the sizing in the automated sequencer data set to match 

 that obtained from the manual gels. Estimated allele sizes 

 from both systems were very highly correlated (r->0.987 

 for all loci). In general, sizes for the same allele from the 

 sequencer were larger than those estimated from manual 

 gels, with the differential increasing with allele size. 



Data analysis 



Each population at each locus was tested for departure 

 from Hardy-Weinberg equilibrium ( HWE) by using GENE- 

 POP (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 Rousset, 

 1995). Annual samples within populations were tested 

 separately, and 106 tests were conducted at each locus 

 (Table 1). Linkage disequilibrium between loci in each 

 population was also evaluated by using GENEPOP; 106 

 tests were conducted for each two-locus combination. With 

 13 loci, there were 78 different two-locus combinations to 

 evaluate. Critical significance levels for simultaneous tests 

 were evaluated by using sequential Bonferroni adjustment 

 (Rice, 1989). Fgj estimates for each locus were calculated 

 with FSTAT (Goudet, 1995), and the standard deviation 

 of the estimate for an individual locus was determined by 

 jackknifing over populations and for all loci combined by 

 bootstrapping over loci. All annual samples available for 

 a location were combined to estimate population allele 

 frequencies, as was recommended by Waples ( 1989). Com- 

 putation of allelic diversity (the average number of alleles 

 observed per locus), and identification of unique alleles 

 was carried out with genetic data analysis (GDA) (Lewis 

 and Zaykin, 2001). To minimize the effects of varying 

 sample sizes on estimates of allelic diversity, values were 

 reported only for population samples consisting of at least 

 50 fish. The number of unique alleles observed per region 

 was standardized by calculating the number observed 

 for each one-hundred fish sampled within each region. A 

 neighbor-joining analysis illustrating genetic relationships 

 among populations was also conducted with GDA incor- 

 porating TreeView (http://taxonomy.zoology.gla.ac.uk/rod/ 

 treeview.html) presentation. 



Estimation of variance components of regional differ- 

 ences, population differences within regions, and annual 

 variation within populations was determined with GDA. 

 Six regions were defined based upon observed population 

 structure: lower, middle, and upper Eraser River, south, 

 north, and lower Thompson River One highly distinctive 

 population, the Birkenhead River, was handled in two 

 ways. It was either included in the analysis as a separate 

 (seventh) region, or excluded from the analysis of regional 

 structure. Balanced experimental designs were required 

 for variance component analysis, and thus two or more 

 populations in each region were required (each population 

 sampled in at least two years). When the Birkenhead River 

 was treated as a region, two populations were defined from 

 the four annual samples available for the regional analysis. 

 The 1993 and 1996 samples were treated as one population, 

 and the 1997 and 1998 samples as the second population. 

 Samples included in the variance components analysis were 

 the 28 locations from which at least two annual samples 

 were available and at least 25 fish in total were sampled 

 (Table 1, Horsey and Goat excluded). Population structure 

 within tributaries of the Eraser River and Thompson River 

 was evaluated with variance components analysis for seven 

 tributaries. These were major tributaries of either the Era- 

 ser River or Thompson River for which we had surveyed 

 populations in the tributary for two or more years and 

 included the Harrison, Chilcotin, Quesnel, Nechako rivers 

 in the Eraser River drainage and the Nicola and Shuswap 

 rivers in the Thompson River drainage (Fig. 1). Variance 

 components due to sampling locations within tributaries 

 were compared with variance components due to sampling 



