290 



Fishery Bulletin 100(2) 



J) 40 



20 



All (n = 545) 



75 50 25 

 % Weight 



Southwest (n = 272) 



50 25 



% Weight 



East(n= 116) 



100 75 50 



% Weight ' 



Southeast (n= 115) 



50 25 



% Weight 



Figure 4 



Three measures of diet importance for 13 prey categories (defined in Table 2) in three 

 areas of the eastern Pacific Ocean (Fig. 1 1. and for all samples pooled. MsF = mesopelagic 

 fishes, F = flyingfishes. EC = epipelagic cephalopods, G = Gempylidae, A = Auxis spp. 



Diet measures by area 



We present three measures of diet importance, '^rW, '^iN. 

 and %0. in Figure 4 for all samples pooled, and by area. 

 We included the data for trace quantities of hard parts 

 (cephalopod mandibles and fish otoliths) in the analysis of 

 %N and ^^O, but not o{''}W. 



The graphical representation mdicates that, when all 

 data are pooled, most components of the diet appear to 

 be quite rare (close to the origin of the graph. Fig. 4, 

 all I. However, when the data are examined by area, the 

 diet proves to be more varied. Flyingfishes and epipelagic 

 cephalopods were clearly the dominant prey. Overall, fly- 

 ingfishes were eaten by more of the dolphinfish (29'~f oc- 

 currence) than any other category, followed by epipelagic 

 cephalopods C2V-i occurrence). Because 274 stomach sam- 

 ples were totally empty, these are equivalent to 58'7f and 

 42% occurrence, respectively, in the stomachs that con- 

 tained food or hard parts. Prey counts contributed more 

 than prey biomass to the apparent importance of mesope- 

 lagic fishes and epipelagic cephalopods in the diet. This is 

 partly due to the accumulation of digestion-resistant hard 

 parts of these two taxa in the stomachs. 



Area was an important source of variation in the '7 W of 

 three prey groups. The regi-ession tree for epipelagic cepha- 

 lopods indicated that 25'/ of the apparent variation in the 

 %W of that prey was explained by area (southeast vs. oth- 



ers). Area was also an miportant predictor of fiyingfish pre- 

 dation; 15'f of the apparent variation in the 'iW was ex- 

 plained by area ( north, west, and southeast vs. other areas ). 

 The regression tree for Tetraodontiformes indicated that 

 4Kf of the apparent variation in predation on that taxon 

 was explained by area (west vs. others). The gravimetric im- 

 portance of the 10 other prey taxa could not be modeled by 

 regression trees (i.e. the trees pruned back to the overall 

 mean %W for those prey) owing to their infrequency in the 

 diet or low sample size ( or to both ). Nevertheless, we pres- 

 ent our results by area to illustrate the substantial spatial 

 variability of the diet of common dolphinfish in the EPO. 



In the east area, the stomachs of 116 common dolphin- 

 fish were sampled from 1 dolphin set and 22 floating-ob- 

 ject sets. Sixty-two of these stomachs were empty and 2 

 contained only trace hard parts (SS.C^ of the females, 

 52.5'^i of the males). The flyingfishes were the dominant 

 prey in the east area in terms of all three indices (Fig. 4, 

 east). Flyingfishes were eaten by 3591 of the dolphinfish 

 sampled and comprised 49*^/ of the total weight of the 

 stomach contents. Epipelagic cephalopods were also eaten 

 by substantial numbers of dolphinfish (16% occurrence), 

 and accounted for 24% of the prey counts. The epipelagic 

 cephalopods were numerically important, but less so by 

 weight, because their mandibles resist digestion and may 

 accumulate in the stomachs over time. The other prey taxa 

 were fairlv rare in the east area. 



