402 
Fishery Bulletin 108(4) 
selected for larger prey. However, prey selection by coho 
and pink salmon showed little ontogenetic variation, 
although we had far fewer data for these taxa to make 
a definitive statement. Patterns of prey selectivity were 
fairly similar between stations for each juvenile salmon 
species, although there were occasionally some differ- 
ences, e.g., for 100-149 mm Chinook salmon, which 
selected euphausiids at the deep station but avoided 
them at the shallow station. 
Juvenile salmon diet overlap 
Juvenile salmon diets rarely overlapped significantly. 
Diet overlap for the most frequently co-occurring (n=ll 
hauls) species, chum and Chinook salmon, ranged from 
10.5% to 66.8%, and averaged 38.3%. PSI was significant 
only on June 1985 and August 1987, when both species 
consumed euphausiids, hyperiid amphipods, and C. 
pacificus. We observed the greatest diet overlap between 
coho and Chinook salmon, which both ate decapod prey 
in June 1986 and 1987, for an average PSI of 77.9%. Diet 
similarities varied for the other juvenile salmon species 
and were based on only one or two co-occurrences. For 
juvenile chum and coho salmon, the PSI was only 10.8%; 
for juvenile pink and chum salmon, the PSI was 58.9% 
in April 1986. 
Salmon diets also varied spatially; however, the PSI 
was not significant for any of the 13 intraspecific com- 
parisons between stations, and averaged only 26.5%. 
Chinook salmon PSI was approximately 30% (/?= 7), and 
that of chum (n = 5) and pink (n= 1) salmon was <20%. 
Discussion 
Juvenile salmon diets, zooplankton abundance and com- 
position, and salmon feeding selectivity in Dabob Bay 
exhibited considerable spatial, seasonal, and interan- 
nual variability. Some notable patterns and trends were 
nevertheless evident, and provide important implications 
for our understanding of feeding ecology and potential 
resource competition among juvenile salmon as they 
migrate from nearshore environments to more open 
waters. 
Food resources and salmon gut fullness 
Dramatic differences in the zooplankton communities 
between the deep water and nearshore stations occurred 
in two of three years. The most striking difference was 
the greater biomass of zooplankton, often an order of 
magnitude higher, at the deep station compared to the 
shallow station (Fig. 4; Bollens et ah, 1992b), primar- 
ily because of the greater abundance of large, verti- 
cally migrating zooplankton at the deeper station (e.g., 
euphausiids and large calanoid copepods). Because zoo- 
plankton biomass represents a measure of overall food 
abundance (although see below for an alternative inter- 
pretation), we expected to see differences in gut fullness 
between the two stations if juvenile salmon were food 
limited and did not often migrate between the sites (a 
distance of 9 km). However, despite this large difference 
in prey resources between sites, we found no evidence 
for fuller salmon guts at the deep station. In addition, 
different zooplankton communities were consistently 
observed between stations (Fig. 5), which could explain 
the low diet overlap of each salmon species between sta- 
tions (mean PSI=27%). 
Gut fullness can be a useful indicator of feeding suc- 
cess. However, gut fullness does not account for poten- 
tial differences in prey nutritional quality (Brodeur, 
1992; Armstrong et al., 2008). Given equal gut fullness, 
feeding success may be greater for fish that consumed a 
larger proportion of high quality prey than were eaten 
by congener species. We observed very high gut fullness 
(9% of body weight) in juvenile chum salmon during 
early summer 1986, but much of the diet was composed 
of low-quality food items such as larvaceans. Data on 
prey quality would enhance future comparisons of ju- 
venile salmon diets and gut fullness. 
Electivity patterns and ontogeny 
The nearshore feeding ecology of the earliest marine 
stages of juvenile salmon <100 mm has been studied well 
in the past (e.g., Kaczynski et al., 1973; Sturdevant et 
al., 1996), and juvenile salmon feeding ecology, species- 
specific feeding preferences, and prey selection for fish 
>100 mm in length captured in highly variable near- 
shore coastal environments have been examined in more 
recent studies (Moulton, 1997; Landingham et al., 1998; 
De Robertis et al., 2005, Armstrong et al., 2008). Ours 
is one of few studies where larger juvenile fish from 
transitional inland marine habitats have been examined 
(Willette, 2001; Sturdevant et al., 2004; Armstrong et 
al., 2008). 
A variety of zooplanktivorous fish select prey dis- 
proportionate to their abundance in the environment 
(e.g., Lazzaro, 1987; Gerking, 1994). For salmon and 
other fishes, these patterns have been attributed to 
multiple factors, including: prey size (Brodeur, 1991); 
prey pigmentation or other visual indicators (Peterson 
et al., 1982; Schabetsberger et al., 2003); and verti- 
cal migration behavior of predator and prey (Bollens 
and Frost, 1989; Viitasalo et al., 2001). Our three-year 
study of juvenile salmon feeding in Dabob Bay provides 
additional evidence that juvenile chum, Chinook, coho, 
and pink salmon exhibit ontogenetic shifts in prey size 
selection and that they select for larger and more visu- 
ally conspicuous prey. Previous studies showed that diel 
vertical migration is an important mediator of plank- 
tivore trophic interactions in Dabob Bay (e.g., Bollens 
and Frost, 1989; Frost and Bollens, 1992; Bollens et 
al., 1993). 
Juvenile salmon in Dabob Bay used a diverse prey 
field and demonstrated species-specific prey preferences. 
Chum and Chinook salmon both highly preferred in- 
sects, cephalopods, decapod larvae, hyperiid amhipods, 
