436 
Fishery Bulletin 109(4) 
ing nets are indicative of the different depth habitats 
for the two species: adult S. vulpes are usually caught 
with set nets laid around 60 m depth, whereas S. zona- 
tus are usually caught with gill nets laid around 150 
m depth around the sampling locality (Sakai 1 ); Simi- 
lar habitat separation (by depth) is common for other 
closely related (sister) species of Sebastes (Narum et al., 
2004; Orr and Blackburn, 2004; Burford and Bernardi, 
2008; Hyde et al., 2008; Orr and Hawkins, 2008; Ste- 
fansson et al., 2009) — ecologically based reproductive 
isolation having often been invoked for Sebastes. For 
example, Hyde et al. (2008) showed that S. miniatus 
and “S. crocotulus” were segregated by habitat depth 
and they hypothesized a speciation model for closely 
related species pairs of Sebastes in which truncation 
of depth-related ontogenetic migration may have led to 
speciation. More detailed ecological studies may provide 
further insights into the maintenance of independent 
gene pools by S. vulpes and S. zonatus and eventu- 
ally provide clues for understanding the mechanisms 
underlying the considerable diversity within Sebastes 
(Ingram, 2011). 
Incomplete lineage sorting and introgression 
Although S. vulpes and S. zonatus are reproductively 
isolated from each other and should be treated as two 
distinct species, two specimens of S. zonatus had typical 
S. vulpes coloration. In addition, the two species shared 
two mtDNA haplotypes and did not exhibit clear sepa- 
ration in the MSN inferred from the mtDNA sequences 
(Fig. 4). This feature can be explained by 1) incomplete 
lineage sorting in mtDNA due to recent speciation, and 
2) interspecific mtDNA gene flow mediated by hybrid- 
ization and backcrossing (introgression), or both (Avise, 
2000; Funk and Omland, 2003). Incomplete lineage sort- 
ing is a source of nonmonophyletic relationship among 
rapidly radiating species in a mtDNA gene tree (Funk 
and Omland, 2003) because newly diverged species are 
initially expected to be nonmonophyletic with respect to 
any gene tree owing to allelic separations predating the 
species split, thereafter progressing to reciprocal mono- 
phyly over time as ancestral haplotypes are sorted and 
unique mutations acquired (Avise, 2000). On the other 
hand, a mtDNA gene tree is also particularly suscep- 
tible to the effects of introgression because mtDNA is 
inherited maternally and does not recombine (Funk and 
Omland, 2003). In fact, both incomplete lineage sorting 
and introgression have been frequently reported within 
Sebastes (Roques et al., 2001; Kai et al., 2002a; Narum 
et al., 2004; Buonaccorsi et ah, 2005; Hyde et al., 2008; 
Burford, 2009). 
A rigorous statistical framework accounting for the 
stochastic variance of genetic processes is generally 
required to distinguish incomplete lineage sorting from 
introgression (Peters et ah, 2007), although an ad hoc 
explanation can be given without the statistical rejec- 
tion of alternative hypotheses (Avise, 2000; Donnelly 
et al., 2004). In this study, two lines of observations 
appeared to better support incomplete lineage sorting 
as the cause of the observed nonmonophyly of mtDNA, 
although the two processes are difficult to distinguish 
unequivocally and are not necessarily mutually exclu- 
sive. 
First, the present MSN of mtDNA showed no distinct 
clades that corresponded with each species. The topol- 
ogy of the gene tree has often been used as an heuristic 
approach to determine the cause of lack of separation 
(e.g.. Baker et al., 2003; Omland et al., 2006; Zakharov 
et al., 2009). A shallow genetic divergence between spe- 
cies without distinct clades, as observed in the present 
study, is generally interpreted as indicative of recent 
speciation and incomplete lineage sorting (e.g., Baker, 
2003; Donnelley et al., 2004). The relatively small pair- 
wise sequence divergences between S. vulpes and S. 
zonatus , corresponding closely to those of intraspecific 
variations found in some other species of Sebastes (Ro- 
cha-Olivares et al., 1999; Higuchi and Kato, 2002; Kai 
et al., 2002a; Burford and Bernardi, 2008), also indicate 
recent speciation between the two species. In addition, 
haplotype and nucleotide diversities within both S. 
vulpes and S. zonatus were relatively high compared 
with those of other species of Sebastes (Rocha-Olivares 
et al., 1999; Higuchi and Kato, 2002; Kai et al., 2002a), j 
as well as other marine fishes (Grant and Bowen, 1998), 
indicating that the two species evolved from a large, ge- 
netically diverse ancestral population, thereafter main- 
taining large effective population sizes without recent 
bottlenecks (Grant and Bowen, 1998; Avise, 2000). Be- 
cause the probability of complete sorting of ancestral 
haplotypes is a function not only of stochastic processes 
and time since speciation but also effective population 
sizes (Funk and Omland, 2003), it seems plausible that 
large, stable effective population sizes of those two spe- 
cies delayed lineage sorting, resulting in their present- 
day sharing of ancestral haplotypes. 
Second, in the PCoA of AFLP, no specimens occupied 
positions intermediate between two clusters (=S. vulpes 
and S. zonatus) (Fig. 3). Because fragments detected in 
AFLP are inherited according to Mendelian expecta- 
tions (Kakehi et al., 2005), hybridized specimens (FI) 
are generally expected to have intermediate fragment 
patterns between parental species (e.g., Congiu et al., 
2001; Young et al., 2001). Therefore, the absence of 
intermediate specimens in PCoA indicated a lack of 
ongoing hybridization between the two species, although 
the possibility of historical introgression (including 
backcross) cannot be completely excluded. In fact, two 
specimens assigned to S. zonatus in the PCoA of AFLP 
had been initially identified as S. vulpes on the basis of 
body coloration (Fig. 3). One of those specimens (FAKU 
82515) was plotted near the S. vulpes cluster in the 
PCA based on measurements (Fig. 5), the count of 11 
dorsal-fin spines without basal scales for that specimen 
also being indicative of S. vulpes (usually more than 
six) rather than S. zonatus (usually one) (Table 3). Such 
equivocal morphological characters may be explained 
by traces of historical hybridization between S. vulpes 
and S. zonatus , which may have resulted in mtDNA 
introgression between them. 
