Orr and Blackburn: Resurrection of Sebastes variabilis and redescnption of Sebastes ciliatus 



331 



In this study, we provide morphological evidence from 

 examination of about 400 specimens collected throughout 

 the geographic and bathymetric range of the species to 

 correlate color differences with meristic and shape differ- 

 ences. In thus recognizing two species, S. ciliatus and S. 

 variabilis, previously referred to the name S. ciliatus (Tile- 

 sius, 1813), we discuss the nomenclatural consequences of 

 this decision. Both species were originally described (as 

 Epinephelus ciliatus Tilesius and Perca variabilis Pallas) 

 on the basis of early Russian collections from along the 

 Aleutian Islands (Svetovidov, 1978, 1981). The type series 

 of one species is now represented by a single extant speci- 

 men (Fig. 2A) and the other by the illustration of a single, 

 now lost, specimen (Svetovidov, 1978, 1981; Fig. 2B). 

 Although workers since the turn of the century have as- 

 sociated the name S. ciliatus with the variably light-col- 

 ored species (Jordan, 1896; Jordan and Evermann, 1898; 

 Barsukov, 1964; Orr et al., 1998, 2000; Mecklenburg et 

 al., 2002), the original description and accompanying il- 

 lustration (Fig. 2B) appear to describe the uniformly dark 

 species. We have also identified the remaining syntype 

 (Fig. 2A) ofPeiea variabilis as the light species. Therefore, 

 we refer the dark, shallow-water species (the dark rock- 

 fish) to Sebastes ciliatus (Tilesius, 1813) and resurrect the 

 name Sebastes variabilis (Pallas, 1814) for the typically- 

 light, deeper-water species (the dusky rockfish). 



Methods and materials 



Counts and measurements follow Hubbs and Lagler 

 (1958), except as noted below. Unless indicated otherwise, 

 standard length (SL) is used throughout and was always 

 measured from the tip of the snout. Depth at pelvic-fin base 

 was measured from the origin of the dorsal fin to the base 

 of the pelvic fins (at the articulation of the pelvic-fin spine); 

 depth at anal-fin origin, from the base of the last dorsal- 

 fin spine to the anal-fin origin; depth at anal-fin insertion, 

 from dorsal-fin insertion to anal-fin insertion; body thick- 

 ness, at pectoral-fin base; head thickness, at the posterior 

 orbital rim; prepelvic- and preanal-fm length, from pelvic- 

 fin base or anal-fin origin to the tip of the snout; pelvic-fin 

 to anal-fin length from pelvic-fin base to anal-fin origin; 

 caudal peduncle dorsal length from dorsal-fin insertion to 

 caudal-fin base; caudal peduncle ventral length from anal- 

 fin insertion to caudal-fin base. The small anterior notch 

 in the orbit between the frontal bone and lateral ethmoid 

 was excluded from orbit length and snout length measure- 

 ments. Accessory scales are small scales located beyond 

 the posterior field of major scales. The swimbladder mus- 

 culature was examined after dissection according to the 

 methods of Hallacher (1974). Institutional abbreviations 

 follow Leviton et al. (1985) and Leviton and Gibbs ( 1988), 

 as modified by Poss and Collette (1995). 



Individuals were identified by body and peritoneum col- 

 or (see species descriptions below) for grouping in ANOVA 

 and ANCOVA, as well as for labeling individuals in graphs 

 of principal components analysis scores. Univariate and 

 multivariate analyses were conducted by using Statgraph- 

 ics Plus 4.1 (Manugistics, Rockville, MD) and Splus 2000 



(Mathsoft, Inc., Seattle, WA). Differences were considered 

 significant at P < 0.05. 



Arcsine-transformed morphometric ratios (with SL or 

 head length as denominator) and meristic characters were 

 tested to meet the assumptions of normality required for 

 ANOVA. The following characters exhibited normal distri- 

 butions and did not differ significantly in variance between 

 species and were subjected to ANOVA: head length, orbit 

 length, snout length, interorbital width, suborbital depth, 

 gill-raker length, body thickness, pectoral-fin base width, 

 pectoral-fin ray length, caudal peduncle ventral length, 

 predorsal length, spinous dorsal-fin base, soft dorsal-fin 

 base, and counts of lateral-line pores and gill rakers. 



For morphometric characters, significant differences 

 were also identified by using an analysis of covariance 

 (ANCOVA) of log-10-transformed measurements with SL 

 or head length (HL) as covariates when assumptions of 

 normality and the homogeneity of slopes were satisfied. 

 The ANCOVA model included species as a factor, SL or HL 

 as a covariate, and a species/I SL or HL) interaction (e.g., 

 HL = C+Species+SL+(SpeeiesxSL)). A residual analysis 

 was done for each model to determine the appropriateness 

 of the model. Whenever the interaction was not significant 

 (at the 5% level), a reduced model was used by dropping 

 the interaction and forcing the slopes to be the same 

 (BD=C+Species +SL). This removed the effect of SL and 

 HL and allowed testing for significant differences between 

 species. The following morphometric characters met the 

 assumptions required for ANCOVA: head length, snout 

 length, interorbital width, gill-raker length, pectoral- 

 fin base width, pectoral-fin ray length, caudal peduncle 

 ventral length, predorsal length, spinous dorsal-fin-base 

 length, and soft dorsal-fin-base length. 



On a dataset of specimens with all characters, sheared 

 principal components analysis (SPCA) for a size-free 

 analysis (Bookstein et al., 1985) was conducted by us- 

 ing morphometric characters, and a standard principal 

 components analysis (PCA) was conducted by using all 

 meristic characters. Raw morphometric data were log- 

 transformed and the covariance matrix was subjected 

 to SPCA, as was the correlation matrix of raw meristics. 

 Differences between species were illustrated by plotting 

 scores of sheared PC2 against sheared PC3 and sheared 

 morphometric PC2 against the standard meristic PCI. 

 Separate analyses were also conducted on three group- 

 ings: 1) each species by depth, 2) each species by sex, and 

 3 ) shallow-water populations ofS. ciliatus and S. variabilis 

 primarily collected in the vicinity of the Triplet Islands 

 and Monashka Bay, on the northeast side of the Kodiak 

 Island Archipelago, and the vicinity of Lynn Canal, 

 Alaska. Shallow collections were defined as those made at 

 less than 50 m depth, and deep collections were taken at 

 depths greater than 50 m. 



These plots were also examined for groupings indicative 

 of geographic differences in body shape and meristics. 

 Geographic areas were defined as follows: British Colum- 

 bia, from the Straits of Juan de Fuca to Dixon Entrance; 

 southeast Alaska, from Dixon Entrance to Chatham 

 Strait; Gulf of Alaska, from Chatham Strait to the tip of 

 the Alaska Pennisula; Aleutian Islands and Bering Sea, 



