114 
Fishery Bulletin 106(2) 
spine length, anal-fin spine 1 length, anal-fin spine 3 
length, preanal length, spinous-dorsal-fin base length, 
soft-dorsal-fin base length, and anal-fin base length. 
Counts of meristic characters were tested by using the 
Kruskal-Wallis test. 
To aid in distinguishing the two species, a standard 
principal component analysis (PCA) was conducted on 
both morphometric and meristic characters. The analy- 
ses were conducted on the covariance matrix of log- 
transformed raw morphometric data and the correlation 
matrix of raw meristic data. Differences between spe- 
cies were illustrated by plotting principal component 
(PC) 2 against PC3 of the morphometric analysis, PCI 
against PC2 of the meristic analysis, and morphometric 
PC2 against meristic PCI. 
Following the PCAs, a stepwise discriminant func- 
tion analysis (DFA) was conducted by using morpho- 
metric and meristic data to establish the relative sig- 
nificance of those characters in distinguishing between 
the species. Morphometric data were standardized by 
dividing by standard length. Only characters meeting 
assumptions of multivariate normality and that exhib- 
ited statistically significant differences were analyzed. 
The robustness of the DFA was tested by conducting 
a leave-one-out cross-validation procedure (SPSS Inc., 
Chicago, IL). 
For the data set containing all 329 specimens, only 
univariate analyses were conducted. The following char- 
acters, in addition to all those except orbit length in 
the reduced dataset, exhibited normal distributions, 
did not differ significantly in variance between species, 
and were subjected to ANCOVA: head length, lower-jaw 
length, body depth at pelvic-fin base, anal-fin spine 2 
length, caudal-peduncle depth, caudal-peduncle dor- 
sal length, caudal-peduncle ventral length, predorsal 
length, and pelvic-fin base to anal-fin origin length. 
Differences in counts of meristic characters were tested 
using the nonparametric Kruskal-Wallis test. With the 
exception of unbranched pectoral-fin rays, meristic data 
from specimens of all sizes were tested. In juveniles of 
less than 100 mm, all pectoral-fin rays were simple and 
counts for these specimens were not included in tests or 
presented in tables of lower pectoral-fin rays. Meristic 
data is presented in tables of frequency distributions 
by region. Four general regions were identified: 1) west- 
ern Pacific Ocean, for material taken in Japanese and 
Russian waters; 2) Bering Sea and Aleutian Islands, 
for material from the eastern Bering Sea and Aleutian 
Islands to Unimak Pass; 3) Gulf of Alaska, from Uni- 
mak Pass to the Alaska-British Columbia border; and 
4) the Lower West Coast, from Canada, Washington, 
Oregon, and California. 
Results 
Color pattern 
Both in life and in preservation, body color differs con- 
sistently between S. aleutianus and S. melanostictus 
(Figs. 1-4 ; see detailed description below). The spinous 
dorsal fin in S. melanostictus is almost invariably dis- 
cretely spotted. In most specimens, many small spots are 
scattered across the spinous dorsal fin, often continuing 
onto the soft dorsal fin; in some individuals, only two or 
three spots are present on the spinous dorsal fin. The 
presence or absence of these discrete spots was used in 
initial statistical analyses as the basis for identifying 
preserved specimens lacking tissue for genetic analysis. 
A few dark individuals without spots, or with spots 
apparently obscured by dark blotching, were aligned 
morphologically with other spotted individuals, and 
these dark individuals were also eventually identified as 
S. melanostictus. In contrast, all S. aleutianus examined 
were pale overall and lacked discrete spots, although 
often having diffuse mottling and blotches extending 
from the body onto the bases of the spinous and soft 
dorsal fins. 
Morphometric and meristic characters 
Specimens of S. melanostictus generally had longer 
dorsal-fin spines (Fig. 5, A and B), other spines, and 
gill rakers (Fig. 5C), whereas specimens of S. aleutianus 
had a deeper, more robust head and body and longer 
pelvic- and pectoral-fin rays. Among morphometric char- 
acters of specimens with all characters, snout length, 
suborbital depth, gill raker length, body depth at anal- 
fin origin, body depth at anal-fin insertion, dorsal-fin 
spine 1 length, dorsal-fin spine 4 length, pectoral-fin 
ray length, pelvic-fin ray length, pelvic-fin spine length, 
anal-fin spine 1 length, anal-fin spine 2 length, anal-fin 
spine 3 length, preanal length, and soft-dorsal-fin base 
length differed significantly between S. aleutianus and 
S. melanostictus (Table 1). In addition, among specimens 
of the larger data set, head length, lower jaw length, 
depth at pelvic-fin base, caudal-peduncle depth, caudal- 
peduncle dorsal length, and caudal-peduncle ventral 
length also differed significantly, and anal-fin spine 1 
and 3 lengths became nonsignificant (Table 2). 
Plots of PCA scores revealed prominent differences 
between the two species among morphometric char- 
acters, as well as slight differences among meristic 
characters. In the morphometric PCA, all characters 
were positively loaded on PCI (the size component; 
Table 3), which explained 96.8% of the total variation. 
Among the principal shape components, PC2 explained 
1.3% of variation and was heavily loaded on dorsal- 
fin spine 1 length, gill-raker length, anal-fin spine 1 
length, anal-fin spine 2 length, and dorsal-fin spine 
4 length (Table 3); PCS explained 0.3% of variation 
and was heavily loaded on anal-fin spine 1 length, 
suborbital depth, gill-raker length, prepelvic length, 
and orbit length (Table 3). In the plot of PC2 versus 
PC3, the two clusters representing S. aleutianus and 
S. melanostictus overlapped narrowly; 16 (8 S. aleutia- 
nus, 8 S. melanostictus) of 137 individuals examined 
were in this area of overlap (Fig. 6A). Of 40 genetically 
identified individuals, three S. aleutianus and three S. 
melanostictus were included in the overlap area. One 
