Table 3. — Mean lengths (cm) of salmon and mean size ot prey. 

 Mean lengths of sockeye and pink salmon were for those in the 

 size classes below (L,) and above (L 2 ) 55 cm, whereas those for 

 coho and chinook salmon were those below (L,) and above (L 2 ) 

 60 cm. The prey ratio v 7 V 2 / V, is assumed to be indicative of ratios 

 in prey lengths between the two groups of predators. The two most 

 frequent fish and invertebrate prey species listed are euphausiids 

 (EU), Parathemisto (PA), sand lance (SL), and herring (HR). 



ferences in prey abundances by depth. The numbers 

 of salmon caught with non-empty stomachs were 

 tabulated by depth zone of capture (Table 4). Coho 

 salmon were most abundant in water depths of <18 

 m, whereas sockeye and pink salmon were most 

 abundant between depths of 18 and 36 m, and 

 chinook salmon most abundant in depths >18 m. 

 Coho and chinook salmon have similar diets, but are 

 found at significantly different depths (x 2 = 714.7, 

 P < 0.01). Thus partitioning of the diets among 

 salmon species is not related simply to water depth. 

 Morphological characters of the salmon species 

 were compared with their food preferences. Chinook 

 and coho salmon have fewer, shorter, and more wide- 

 ly spaced gillrakers than those of sockeye and pink 

 salmon (Table 5). As gillrakers are used to strain food 

 organisms from water passing over the gills (Lagler 

 et al. 1962), I expected salmon species feeding on 

 planktivorous prey to have more gillrakers that are 

 longer and more closely set than those in primarily 

 piscivorous salmon species. Similar arguments could 

 be made for tooth size (Table 5). Partitioning of the 

 diet among the species of salmon investigated is 

 clearly a reflection of morphological differences 

 among the species. 



DISCUSSION 



The calculation of unit volumes for individual prey 

 classes is an important component of the analysis. 

 Prey types were assumed to be in a similar state of 



FISHERY BULLETIN: VOL. 84, NO. 1 



digestion for the different size classes of each species 

 of salmon so that calculated unit volumes would be 

 comparable Violation of this assumption may ac- 

 count for the inverse predator-prey size relationship 

 found for sockeye and pink salmon with euphausiids 

 and Parathemisto. The analysis of relative sizes of 

 the species eaten assumes that different prey types 

 were not more or less digested than others. This is 

 unlikely to be strictly true, but it was assumed that 

 differential digestability of the prey species did not 

 significantly alter their relative sizes. 



Previous work on diet description of Oncorhynchus 

 species has indicated that there can be considerable 

 variability in dietary components of a particular 

 species. However, some general conclusions can be 

 drawn. Sockeye salmon are the least piscivorous of ' 

 the northeast Pacific Oncorhynchus species (Allen 

 and Aron 1958; LeBrasseur 1966; Foerster 1968). 

 Euphausiids have been reported consistently as a 

 major contributor to the diet of pink salmon (Maeda 

 1954; Ito 1964; Takagi et al. 1981). The fish compo- 

 nent reported has been variable, ranging from <1% 

 to over 90% of stomach volume (Takagi et al. 1981). 

 Chinook and coho salmon tend to be the most 

 piscivorous (Allen and Aron 1958; Prakash 1962; 

 Reimers 1964; LeBrasseur 1966; Machidori 1972). 

 For chinook salmon, fish were reported to provide 



Table 4. — Number of salmon caught with non-empty stomachs and 

 depth of water (m) in Strait of Juan de Fuca, British Columbia. 

 Salmon were caught by troll gear. Numbers in parentheses are per- 

 cent of each species caught in each depth zone. 



Table 5. — Comparisons of morphometric and 

 meristic characters of Pacific salmon whose dietary 

 components were investigated in this study. 



'From Hikita (1962). 



2 From Morrow (1980). 



3 Gillraker length as percent of postorbital-hypural length. 

 Gillraker length is from Hikita (1962), postorbital-hypural 

 length from Beacham and Murray (1983). 



"From Vladykov (1962), Hikita (1962). 



86 



