Fishery Bulletin 120(1) 
index values for dietary overlap observed within small 
(<66 mm SL: SI=0.42) and large (£66 mm SL: SI=0.34) size 
classes was similar to that observed across the full size 
range of sampled fish (SI=0.39). Results of comparisons 
of overlap across size classes within species indicate more 
dietary overlap between size classes of northern rock sole 
(SI=0.70) than those of yellowfin sole (SI=0.40). 
Foraging characterization 
Results of PCA reveal similarities and differences in for- 
aging patterns between yellowfin sole and northern rock 
sole. The first principal component (PC1) explains 41.8% 
of the variance and reflects differences in general foraging 
habitat between yellowfin sole and northern rock sole 
(flatfish species was not itself used as a response variable) 
22.6%) 
co) 
® 
{ 
& 
ror 
x< 
a) 
® 
cS) 
Cc 
© 
— 
© 
EZ 
N 
oO 
oO 
(Fig. 5). Yellowfin sole consumed prey in areas with higher 
temperature, shallower depths, and finer sediments than 
the areas where northern rock sole consumed prey (PC1: 
|coefficient of correlation [r]|>0.72). The PC1 is also 
strongly correlated with stomach fullness and time of 
sampling, reflecting the diel pattern of feeding and overall 
higher levels of stomach fullness for northern rock sole. 
The second principal component (PC2) explains 22.6% of 
the variance and is associated with the gradient between 
fine sand and larger sediments (medium sand, coarse 
sand, granules, and shells). 
The PC1 scores for euphausiids and mysids consumed 
by both species are notably higher than those for other 
prey items, reflecting that their consumption occurred in 
deeper, cooler habitats. Euphausiids had the greatest mean 
depth and lowest temperature of occurrence in the diets of 
Prey habitat Species 
a Pel © YFS 
© Hyp e NRS 
4 Epi 
© End 
Granule 
4 
PC1 (variance explained=41.8%) 
Figure 5 
Principal component analysis of foraging patterns of yellowfin sole (Limanda aspera) 
(YFS) and northern rock sole (Lepidopsetta polyxystra) (NRS) captured in coastal areas 
of the southeastern Bering Sea in August 2012. Analysis is based on the mean value of 
biotic and abiotic conditions averaged across all occurrences of each prey type in stom- 
achs of each predator species. Prey are grouped by taxonomy and habitat associations 
relative to the benthos: endobenthic (End), epibenthic (Epi), hyperbenthic (Hyp), and 
pelagic (Pel). Traits of predators are standard length (SL), stomach (gut) fullness (GF), 
and Fulton’s condition factor (K). Traits of sampling stations are depth; temperature 
(Temp); sediment fractions of silt or clay (Silt), fine sand (Fsand), medium sand (Msand), 
coarse sand (Csand), granules, and shells; time of sampling (Hour); and catch per unit of 
effort (CPUE) for the predator species. Prey taxa include Polychaeta (Poly), Amphipoda 
(Amph), Cnidaria (Cnid), Copepoda (Cop), Bivalvia (Biv), Cumacea (Cum), Decapoda 
(Deca), Mysida (Mys), Euphausiacea (Euph), and Pisces (Pisc). 
