overlap values %IRI values were used as the propor- 

 tion, p. 



Morphological Comparison 



Several measurements were taken on a subsam- 

 ple of fish of each species covering all sizes en- 

 countered in this study. These included standard 

 length, head length, and gape Mean length of the 

 five gill rakers nearest the angle on the ventral limb 

 of the first gill arch, and the gap between them, were 

 also measured. 



The feeding related morphological characteristics 

 chosen for statistical comparison were head length, 

 gape, gill raker length, and gill raker spacing. Gape 

 and gill raker spacing were compared as absolute 

 measurements. Gill raker lengths were standardized 

 to a proportion of head length and head length was 

 standardized as a proportion of standard length for 

 comparison. 



Gape was compared using a one-way analysis of 

 covariance (ANCOVA). This analysis regressed the 

 natural log of the ga]je against the natural log of fish 

 standard length to allow comparison over a range 

 of fish sizes and to achieve linearity (Sokal and Rohlf 

 1969; Chen 1971). Relative head length, gill raker 

 spacing, and relative gill raker length were compared 

 among species with a Model I, one-way analysis of 

 variance (ANOVA). 



Multiple range comparisons were then made be- 

 tween individual species to detect groupings. Regres- 

 sion lines resulting from the ANCOVA were sub- 

 jected to Newman-Keuls pairwise comparisons. 

 Mean values obtained from ANOVAs were subjected 

 to Student-Newman-Keuls (SNK) multiple range 

 tests. All statistical procedures used followed those 

 presented in Sokal and Rohlf (1969) and Snedecor 

 and Cochran (1980). 



Prey Abundance 



Qualitative assessment of prey abundances was 

 made from zooplankton tows done in the kelp forest. 

 A standard 0.5 m diameter net with 0.333 ^m mesh 

 was used. Tbws were done in sets of three, 3 min 

 each: one within 0.5 m of the canopy, one in mid- 

 water, and one within 0.5 m of the bottom. These 

 were done during both day and night. Generally, only 

 presence or absence of plankton types was noted, 

 along with relative daytime vs. nightime differences 

 in abundances. Absolute abundances were not 

 estimated. Zooplankton samples were taken once in 

 mid- June 1980 and once in mid- July 1981. A total 

 of 12 samples were collected. 



FISHERY BULLETIN: VOL. 83, NO. 4 



RESULTS 



A total of 265 juveniles of the seven species was 

 examined for stomach content analysis: 27 S. 

 melanops (53 to 67 mm SL, x = 57.9); 51 S. ser- 

 ranoides (44 to 63 mm; x = 51.8); 63 5. pinniger (28 

 to 56 mm, x = 43.7); 23 S. mystiniis (46 to 72 mm, 

 X = 59.1); 38 S. atrovirens (38 to 64 mm, x = 47.1); 

 42 S. caurinus (35 to 62 mm, x = 48.7); and 21 S. 

 camatus (33 to 65 mm, x = 44.4). An often co- 

 occuring cogener of S. camatus, S. chrysomelas, was 

 not found during this study. 



Diet Analysis 



The cumulative prey type curves versus number 

 of stomachs examined leveled off asymptotically, in- 

 dicating that sample sizes for all species were suffi- 

 cient to characterize their food arrays. All species 

 were found to be zooplanktivores, consuming both 

 open water and substrate-associated prey. Open 

 water prey, which mainly float free in the water 

 column, consisted mostly of juvenile copepods, zoea, 

 and juvenile spionid polychaetes, while substrate- 

 associated prey, which were generally found direct- 

 ly associated (<0.5 m away) with a physical struc- 

 ture (rock, kelp, eta), consisted mainly of gammarid 

 amphipods, mysid and caridean shrimp, and two 

 species of isopod (Ikble 1). 



A wide variety of prey sizes were encountered 

 (Tkble 2). Prey found in open water, such as calanoid 

 and harpacticoid copepods, zoea, and larvaceans, 

 were mostly < 1 mm in length, whereas substrate- 

 associated prey, such as amphipods and decapods, 

 were 3 to 4 mm or larger. 



Mean number of prey items and prey taxa per in- 

 dividual were highly variable within and among 

 species (Tkble 3). As would be expected, species 

 which fed on small open water prey had higher 

 numbers of prey per stomach. Conversely, those 

 species which fed on large substrate-associated prey, 

 which were often large enough to fill a stomach with 

 a single prey item, had much lower numbers of prey 

 items per stomach. 



Sebastes mystimis fed mainly on the larvacean 

 Oikopleura sp., with high occurrence of copepods, 

 zoea, and juvenile polychaetes (Fig. 2). They had the 

 most cosmopolitan diet and the highest intraspecific 

 variability of prey types (a mean of almost six dif- 

 ferent prey types per individual) (Tkble 3). Because 

 its diet was composed mainly of small open water 

 prey, this species also had the highest mean number 

 of prey items per individual. 



Sebastes serranoides and S. melanops had very 



534 



