612 



Fishery Bulletin 94(4), 1996 



of variation at minisatellite loci offers the potential 

 for increased population differentiation in contrast 

 with other methods of stock identification. 



Initial screening of Asian and North American 

 chum salmon populations with a single-locus 

 minisatellite probe suggested that there were three 

 regional groups of populations, namely Japanese, 

 Russian and Yukon River, and southeast Alaska and 

 British Columbia (Taylor et al., 1994). However, de- 

 tection of population structure or differentiation 

 among populations within the three major regional 

 groups of chum salmon was limited, even though 

 previous surveys of variation at protein-coding loci 

 suggested that there was at least regional structur- 

 ing within the British Columbia populations 

 (Okazaki, 1982; Beacham et al., 1987). In the cur- 

 rent study, variation at two additional minisatellite 

 DNA loci is reported, and the efficacy of using varia- 

 tion at three minisatellite loci to discriminate among 

 chum salmon populations in local areas is evaluated. 



Materials and methods 



Samples, laboratory procedures, and probes 



The stocks surveyed, their geographic locations, and 

 the methods of DNA extraction, restriction enzyme 

 digestion (Haelll), Southern blotting, and membrane 

 preparation have been outlined by Taylor et al. 

 (1994). Summarized briefly, 4 (ig of //aelll-digested 

 DNA were loaded onto agarose gels containing 30 

 wells. For the 30 lanes on each gel, 4 lanes were used 

 for A.DNA restricted with Hindlll and EcoRl as a 

 molecular weight-size marker, and 26 lanes were 

 used for genomic DNA, one of which was a control 

 fish run. DNA was size-fractionated by electrophore- 

 sis in 0.8% agarose gels in 0.5x TBE buffer (45 mM 

 Tris-borate, 1 mM EDTA, pH 8.0). After electrophore- 

 sis, the agarose gels were depurinated, alkali-dena- 

 tured, and then the DNA was transferred under 

 vacuum to nylon hybridization membranes and fixed 

 to the membranes by crosslinking under ultraviolet 

 illumination for 3 min. 



The probes used in the laboratory analysis were 

 pSsa-A33 and pSsa-A34, hypervariable minisatellite 

 single locus probes in salmonids (Taggart and 

 Ferguson, 1990; Prodohl et al., 1994). Prior to hy- 

 bridization with the target probe, the previous probe 

 used was stripped from the membrane by the high 

 temperature method outlined by Noppinger et al. 

 (1992). Following hybridization with each probe, the 

 membranes were washed twice in 2x SSC/0. 1% SDS 

 at room temperature for 15 min, once in 2x SSC/0. 1% 

 SDS at 65°C for 30 min, once in lx SSC/0. 1% SDS at 



65°C for 30 min, and were given a final wash in lx 

 SSC at room temperature for 15 min. The probes were 

 labelled with 32 P, and the membranes exposed to X- 

 ray film. Up to 42 stocks of chum salmon were sur- 

 veyed in the analysis (the term "stock" was used ac- 

 cording to the sense of Ricker (1972), where a stock 

 represents a discrete breeding unit spawning at a 

 given time or place in a river). 



Analysis of autoradiographs 



The autoradiographs were analyzed with Bioimage 

 software (Millipore Corp. Imaging Systems, Ann Ar- 

 bor, MI) and scanned with a Kodak high-resolution 

 two-dimensional charge coupled device (CCD) cam- 

 era at a 1,024 x 1,024 pixel density. The images were 

 translated into digital file images and stored at a 

 workstation for later analysis. Lambda DNA-size 

 markers (21.23, 5.15, 4.97, 4.27, 3.53, and 2.03 kilo 

 base pairs [kbp]) were run on four lanes of each gel 

 and used to create a size standard network that 

 spanned the whole image. Molecular sizes were then 

 assigned by the software to each fragment in each 

 lane on the basis of their relationship to the size stan- 

 dard network. When the pSso-A33 and pSsa-A34 

 probes were used, each lane contained either one or 

 two bands; fish with a single band were considered 

 homozygous and two comigrating bands of the same 

 size were scored for those fish. The two probes de- 

 tect alleles in chum salmon at a single locus where 

 the alleles are inherited in a Mendelian fashion (un- 

 published data), as was observed for pSsa-A34 in 

 sockeye salmon, O. nerka (Taylor et al., 1996), and 

 chinook salmon, O. tshawytscha (Stevens et al., 

 1993), and for both pSsa-A33 and pSso-A34 in brown 

 trout, Salmo trutta (Prodohl et al., 1994). 



Band assignment 



Fish from one stock could be analyzed on multiple 

 gels, and thus it was necessary to determine mea- 

 surement error of band size among gels in order to 

 combine the data. Precision of the estimation of band 

 size was determined by analyzing a standard fish 

 over all gels and by determining the relationship 

 between precision of the estimate and fragment size 

 as outlined by Taylor et al. (1994). As measurement 

 error was present in the estimation of fragment size, 

 a binning procedure (Gill et al., 1990; Galbraith et 

 al., 1991) was used to assign each observed fragment 

 to a particular size class or bin. Distribution of frag- 

 ment sizes at the Ssa-A33 locus ranged from about 4 

 to 22 kbp, whereas the distribution at the Ssa -A34 

 locus ranged from 2.5 to 8.5 kbp, with very few frag- 

 ments observed greater than 6.0 kbp (Fig. 1). Bin 



