52 
Fishery Bulletin 99(1 ) 
PHYLIP 3.57c (Felsenstein 3 ), assuming that all restriction 
sites were 4 bp long (PHYLIP, Felsenstein 3 ). Nucleotide 
divergences (proportion of nucleotide substitutions) and 
their standard errors were estimated according to Nei and 
Tajima (1983), Nei (1987), and Nei and Miller (1990) by 
using REAP (McElroy et ah, 1990). 
Results 
Restriction fragment patterns from double digests were 
used to construct restriction site maps for comparisons 
of species and detection of intraspecies variation (Appen- 
dix 1). The map includes 153 restriction sites, 36 of which 
were common in all haplotypes and 28 of which were cla- 
distically uninformative because the presence or absence 
occurs only in a single haplotype. Many of the cladistically 
uninformative sites, however, were useful in species delin- 
eation. These data represent 153 restriction sites (79.3 on 
average) corresponding to 640 nucleotides (332.05 on aver- 
age) per haplotype. 
Among the 85 fish examined were 30 different compos- 
ite haplotypes (Table 3); each species had haplotypes that 
were distinct from those of other species, although S. var- 
iegatus composite haplotype 8 differed at a single site from 
S. zacentrus composite haplotype 7c (Table 4). All other 
pairs of species differed by 5 or more sites. Intraspecific 
variation was observed in nine of the seventeen species 
even when only five specimens of each species were ana- 
lyzed. The most variable species were S. zacentrus and Se- 
bastolobus alascanus, each of which had four haplotypes. 
In the study, differences between haplotypes ranged from 
a single site difference or 0.0014 nucleotide substitutions 
per site to 65 restriction site differences and 0.120 nucleo- 
tide substitutions per site (Table 4). Nucleotide divergence 
within variable species averaged 0.0024 subsitutions ( 1.56 
site changes), whereas divergences between Sebastes spe- 
cies averaged ten-fold higher, 0.0249 (15.4 site changes), 
ranging from 0.0015 (1 site change) to 0.0384 (25 site 
changes). Nucleotide divergences between Sebastes spe- 
cies and Sebastolobus alascanus averaged 0.1073 (59.2 
site changes) and divergences between Sebastes species 
and H. hilgendorfi averaged 0.0805 (43.5 site changes). 
Distribution of the variation between the two different 
mtDNA regions (ND3/ND4 and 12S/16S) reflects their 
rates of evolution. In the 12S/16S region, which is more 
conservative, 27 of 58 restriction sites were shared by all 
haplotypes. Nucleotide diversities between Sebastes spe- 
cies averaged 0.0094 nucleotide changes per nucleotide (a 
total of 3.29 sites differences in the region), divergences be- 
tween Sebastes and H. hilgendorfi averaged 0.0641 (12.67 
site differences), and divergences between Sebastes and 
Sebastolobus alascanus averaged 0.0561 (18.03 site differ- 
ences). In contrast, in the ND3/ND4 region only 9 of 95 
sites were common to all haplotypes; and nucleotide diver- 
3 Felsenstein, J. 1993. PHYLIP (Phylogeny Inference Pack- 
age) version 3.57c. Distributed by the author. Department 
of Genetics, Box 357360, Univ. Washington, Seattle, WA 
98195-7360. 
gences between Sebastes species averaged 0.0471 (12.11 
site differences) and divergences between Sebastes and H. 
hilgendorfi and between Sebastes and Sebastolobus alas- 
canus averaged 0.1373 (31.75 site differences) and 0.1929 
(40.93 site differences), respectively. The maximum like- 
lihood and majority consensus tree for the 60 maximum 
parsimony trees that imposed a cost of two for regained 
restriction sites had identical topologies (Fig. 1). The to- 
pologies of parsimony trees, which had either no addition- 
al cost or a cost of four, were somewhat different. Several 
groups of species were present in all three parsimony to- 
pologies. The S. zacentrus-S. variegatus pair, mentioned 
above, and each of four species pairs — S. melanops-S. flavi- 
dus, S. babcocki-S. heluomaculatus, S. proriger-S. brevispi- 
nis, and S. maliger-S. caurinus — clustered tightly at sub- 
terminal nodes. A more interior cluster of species included 
S. melanops, S. flavidus, S. babcocki, and S. helvomacula- 
tus. In addition, S. maliger and S. caurinus clustered sep- 
arately from all other Sebastes species and the Sebastes 
species were distinct from H. hilgendorfi and Sebastolobus 
alascanus. 
The mtDNA variation we observed among Sebastes spe- 
cies provides a tool for identifying species. From our da- 
ta, numerous schemes could be devised that distinguish 
among the Sebastes species examined. We propose a sim- 
ple scheme that minimizes the number of digests required 
and involves separation of restriction fragments from the 
ND3/ND4 PCR product on an agarose-Synergel™ gel us- 
ing only four restriction enzymes. Mho I digests produce 
11 different haplotypes (haplotypes A-K; Figure 2 A; Ta- 
ble 3); S. alutus (B), S. melanops (E), S. babcocki (G and 
H), S. ruberrimus (I), and S. caurinus (J) are species spe- 
cific. If Mbo I haplotypes A (S. heluomaculatus or S. flavi- 
dus) or C (S. maliger or S. caurinus ) are observed, digest 
the ND3/ND4 PCR product with Hind II; Hind II haplo- 
type B is specific for S. heluomaculatus and Hind II hap- 
lotype C is specific for S. maliger (Fig. 2B; Table 3). If Mbo 
I haplotypes F (S. ciliatus or S. borealis) or K (S. aleutia- 
nus, S. proriger, or S. brevispinis) are observed, digest the 
ND3/ND4 PCR product with BstN I; BstN I haplotype A 
is specific for S. ciliatus and BstN I haplotype G is specific 
for S. brevispinis (Fig. 2C; Table 3). Mbo 1 and BstN I hap- 
lotypes do not distinguish between S. aleutianus and S. 
proriger , but Cfo I haplotype B is specific for S. aleutianus 
(Fig. 2D, Table 3). The combined haplotype of Mbo I, Hind 
II, BstN I, and Cfo I can be used to identify S. borealis 
(KAFD) and S. proriger (FAFD) (Fig. 2). The single differ- 
ence between S. zacentrus and S. variegatus is the pres- 
ence of a 123-bp fragment in Rsa I digests of S. zacentrus 
(Table 2; Appendix 1). 
This simple scheme takes advantage of unique single- 
site differences for several of the species. Although a neigh- 
bor-joining tree (Saitoh and Nei, 1987) appeared stable to 
intraspecific variation for increased sample sizes of three 
species (data not shown), a single site change that produc- 
es apparent convergence between taxa in our scheme is 
conceivable. Increased certainty can be achieved by con- 
ducting digests with all four enzymes. With this strategy 
there will be at least two site differences between every 
pair of species, except S. proriger and S. brevispinis, which 
