384 
Fishery Bulletin 109(4) 
Figure 1 
Sampling locations for Pacific herring ( Clupea pallasii) collected in the 
Gulf of Alaska for genetic analysis, 2007-09. 
overwintering aggregate from Simpson Bay, in eastern 
Prince William Sound (ePWS) in December, 2007. 
All samples were collected by trawl, seine, or castnet. 
Lynn Cana and Hobart Bay samples were collected with 
a midwater trawl. Spawning herring in Berners Bay 
were collected along the shore with a castnet. Hoonah 
Sound samples were hand netted from the commercial 
roe-on-kelp fishery. In Sitka Sound, herring were col- 
lected from a purse seiner chartered by the Alaska 
Department of Fish and Game (ADFG) during the test 
fishery for the Sitka Sound sac roe fishery. Samples 
from Nichols Bay and Prince William Sound were col- 
lected with a beach seine. All samples, except those 
from Nichols Bay, were mature fish. All collections were 
shipped as whole frozen fish to the Auke Bay Laborato- 
ries in Juneau, AK. 
Laboratory procedures 
DNA was extracted from tissue samples (heart, muscle, 
or fin) by following DNeasy genomic DNA extraction 
methods (Qiagen, Valencia, CA). A master mix of 1 pL 
10x PCR buffer, 4.54 pL deionized water, 0.6 pL MgCl 2 
(25 mM), 0.8 pL dNTPs (10 mM), 0.5 pL forward primer 
(10 mM), 0.4 pL reverse primer (10 mM), 1.0 pL of fluo- 
rescent labeled primer (1 mM), and 0.16 pL of TAQ DNA 
polymerase was combined with 1 pL of DNA. Initially, 22 
microsatellite loci were amplified by polymerase chain 
reaction (PCR), by using a Gene Amp 9700 thermocycler 
(Applied Biosystems, Foster City, CA). The resulting 
PCR products were size fractionated on a DNA sequencer 
(Licor 4200 and 4300) by using known molecular size 
standards, and genotypes were scored by using SAGA 
software (Licor, inc., Lincoln, NE). Loci were double- 
scored and those with a 2-bp repeat were aligned by 
allele size and analyzed a second time to ensure data 
integrity. Individuals for which there were missing 
data for three or more of the 15 loci were dropped from 
further analyses. The numbers of individuals used in 
analyses (n) are given in Table 1. 
Genetic analysis 
Initial data from 22 loci were examined for the possibil- 
ity of scoring errors, null alleles, or large allele drop out 
with MICROCHECKER software (van Oosterhout et 
al., 2004). Deviation from Hardy-Weinberg expectation 
(HWE) was tested for each locus for each sample collec- 
tion by using Fisher’s exact test with GENEPOP soft- 
ware, vers. 4.0 (Raymond and Rousset, 1995). Linkage 
disequilibrium was tested with Slatkin’s (1994) method 
implemented in GENEPOP to confirm that loci were 
segregating independently. Markov chain parameters 
of 50,000 dememorizations, 500 batches, and 25,000 
iterations were used to calculate an accurate and reli- 
able test. The methods of Weir and Cockerham (1984) 
implemented in FSTAT, vers. 2. 9. 3. 2 (Goudet, 1995) were 
used to calculate F IS for each collection by locus, and 
each locus over all collections. A gene diversity analysis 
was conducted with GENEPOP software to examine 
observed and expected heterozygosities per locus and per 
collection. Allelic richness by collection was calculated 
in FSTAT. Effective number of alleles was calculated as 
1/(1-H e ). Chord distances (Cavalli-Sforza and Edwards, 
1967) were calculated among all pairs of collections with 
PHYLIP software, vers. 3.69 (Felsenstein, 1989), and a 
neighbor-joining tree was constructed to examine the 
relationships among collections. Data were bootstrapped 
(1000 replicates) by using allele replacement and loci 
replacement in PHYLIP, and a summary of the repli- 
cates (consensus tree) was constructed to examine the 
consistency of putative genetic partitions. 
