536 
Fishery Bulletin 99(4) 
in the case of mtDNA - N e( p) and mu- 
tation rate (p) (Kimura and Crow, 1964; 
Ohta and Kimura, 1973). However, for p 
that are relatively constant among spe- 
cies, n c orders the relative size of historic 
effective sizes of the species (Avise et al., 
1988). Estimates of n c may be influenced 
by the number of individuals analyzed, 
the scope of sampling with respect to both 
geographic range, and the number of sites 
detected in the mtDNA genome. 
The average nucleotide sequence di- 
vergence between haplotype pairs pres- 
ents a more complex picture. Accumu- 
lation of nucleotide differences between 
haplotypes requires time. If the muta- 
tion rates (p) are reasonably similar (al- 
though there is debate about the degree 
of similarity [e.g. Avise et ah, 1988; Gib- 
bons, 1997] ), the extent of haplotype di- 
vergence should order the species accord- 
ing to the time that has elapsed since 
a major species bottleneck. Such order- 
ing is a consequence of lineage sorting 
within isolates, which tends toward a set 
of related (monophyletic) haplotypes but 
fosters divergence among isolates. How- 
ever, as a result of lineage sorting, spe- 
cies composed of isolates are likely to 
exhibit higher average divergence than 
species broadly connected by gene flow 
(Neigel and Avise, 1986). The small nu- 
cleotide divergence among haplotypes in 
coho salmon suggests that extant coho 
salmon share a relatively recent common 
female ancestor. 
Two distinct clusters of haplotypes ac- 
count for most of the divergence observed 
among coho salmon. The clusters are sep- 
arated by an average of 2.72 substitu- 
tions per 1000 nucleotides, but divergence 
among haplotypes within each cluster av- 
erage about 0.87 substitutions per thou- 
sand nucleotides. 
Expanded coho mtDNA survey from 
variable sites 
By concentrating our effort on variable 
sites, we evaluated the variation present 
in a larger number of fish and from addi- 
tional populations. The two clusters of 
haplotypes observed in the preliminary 
survey persisted, each augmented by sev- 
eral new haplotypes. The average diver- 
gence between haplotypes within each 
cluster was 4.95 substitutions per 1000 
nucleotides and the clusters averaged 
10.28 substitutions per 1000 bp apart. 
By focussing on variable restriction sites, 
