Pietsch and Orr: Tnglops dorothy. a new species from the southern Sea of OI<hotsk 



239 



was not named until 1837, after Reinhardt received 

 two additional specimens, also from West Greenland. 

 During the next 130 years, 18 additional species or 

 subspecies were described, but little work of a com- 

 parative nature was done. Schmidt (1930) reviewed 

 some North Pacific specimens of the group, and later 

 (Schmidt, 1950) discussed species identified in his work 

 on the ichthyofauna of the Sea of Okhotsk. Taranets 

 (1941) placed the genus, together with Sfer?uas, in its 

 own subfamily. The Arctic and North Pacific species 

 of Triglops were studied by Jensen (1944), Andriashev 

 (1949, 1954), and McAllister (1963), and a world-wide 

 revision of the genus was subsequently published by 

 Pietsch (1993). 



Within the material of Tnglops examined by Pietsch 

 (1993) are 19 specimens that were originally recognized 

 by Charles Henry Gilbert as new to science (see "Dis- 

 cussion" below). Labeled as such, but left unpublished 

 by Gilbert, Pietsch (1993) came to the same conclusion, 

 citing a lack of agreement of these specimens with the 

 diagnosis of any of the nine recognized species of the 

 genus. However, still not totally convinced that the 

 specimens represented a new species, Pietsch (1993) 

 again postponed a formal description. Labeled Triglops 

 sp. and set aside at that time, these specimens are now 

 re-examined in light of the discovery of two additional 

 specimens, and evidence is provided to formally recog- 

 nize the material as a new species. 



Methods and materials 



the ventral mid-line. Symbolic codes for institutions are 

 those provided by Leviton et al. (1985). 



Statistical significance was assessed by ANOVA; all 

 differences at P<0.05 were considered significant. To aid 

 in the discrimination of the new species and its closest 

 ally, T. pingeli, a standard principal component analy- 

 sis (PCA) was conducted on meristic characters, and a 

 sheared PCA was conducted on morphometric characters 

 for a size-free analysis. The analyses were conducted on 

 the correlation matrix of raw meristic data and the co- 

 variance matrix of log-transformed raw morphometric 

 data. Differences between species were illustrated by 

 plotting principal component (PC) 1 against PC2 of the 

 meristic analysis, sheared PC2 against sheared PCS of 

 the morphometric analysis, and sheared morphometric 

 PC2 against the standard meristic PCI. 



Following the PCA, a linear discriminant function 

 analysis (DFA) was conducted on morphometric and 

 meristic data to establish the relative significance of 

 those characters in distinguishing the species. Morpho- 

 metric data were standardized by dividing by standard 

 length. Each character was tested to verify assump- 

 tions of multivariate normality and for statistical sig- 

 nificance; those found to be distributed nonnormally 

 or that were nonsignificant were eliminated from fur- 

 ther analysis. The robustness of the DFA was tested by 

 conducting a leave-one-out cross-validation procedure. 

 Statistical analyses were performed with S-Plus (vers. 

 6.1 for Windows, 2002, Insightful, Inc., Seattle, WA) 

 and SPSS (vers. 11.5.1 for Windows, 2002, SPSS Inc., 

 Chicago, IL). 



Standard length (SL) is used throughout unless indi- 

 cated otherwise. Measurements are expressed in percent 

 of standard length or in percent of head length (HL). 

 Methods used for taking counts and measurements 

 follow those described and figured by Miller and Lea 

 (1972:7-12), with the following additions: interorbital 

 width is measured between lateral margins of the frontal 

 bones at their narrowest widths; head depth is measured 

 at mid-orbit; first infraorbital pore width is measured at 

 the greatest width, anterior to posterior; all gill-raker 

 counts are taken from the first arch; dorsolateral scale- 

 row counts include all scales (or fused clusters of scales) 

 with enlarged plate-like bases; lateral-line scale counts 

 also include scales that occasionally extend beyond the 

 posterior margin of the hypural plate. Pectoral-fin rays 

 are counted from dorsal to ventral. Counting the oblique 

 dermal folds on the ventral half of the trunk presents a 

 special problem. Because many dermal folds (the number 

 depending on the species), which originate along the 

 lateral line, terminate irregularly before reaching the 

 ventral midline, counts can be considerably higher and 

 more highly variable (than the values reported in the 

 present study) if made along a transect close to the 

 lateral line; thus, it is important to count the dermal 

 folds along a transect beginning beneath the pectoral 

 fin and continuing along, close to and strictly parallel 

 with, the base of the anal fin (see Pietsch. 1993:337, 

 Fig. 1). Dermal folds of the breast are counted along 



Systematics 



Triglops dorothy Pietsch and Orr, n. sp. 



Dorothy's sculpin 



Figures 1-4, Tables 1-3 



Triglops sp.: Pietsch, 1993:379 (probably an undescribed 

 species, description postponed). 



Holotype 



USNM (U.S. National Museum of Natural History, Wash- 

 ington, D.C.) 74578, female, 155 mm. Albatross Station 

 5013, Aniva Bay, 46 = 17'N, 143°09'E, 3-m Agassiz trawl, 

 77 m, substrata unknown, 25 September 1906. 



Paratypes 



Twenty specimens, 15 males (72.5-148.0 mm), 5 females 

 (68.5-143.0 mm). The following material was collected 

 by the U. S. Fish Commission Steamer Albatross in or 

 adjacent to Aniva Bay, Sakhalin Island, Russia, with a 

 3-m Agassiz trawl: CAS-SU 22305, 3 females (68.5-133.0 

 mm), station 5007, 46"03'N, 142°31'E, 77 m, green mud, 

 fine gray sand, 24 September 1906. USNM 74575, 2 

 males (72.5-80.0 mm), station 5006, 46°04'N, 142°29'E, 

 79 m, green mud, fine gray sand, 24 September 1906; 



