Foods of Yellowfin and Blackfin Tuna 47 



23.9% and 76.1%, respectively, of the yellowfin he examined from the 

 Pacific. Similar occurrences were reported for yellowfin from the Atlan- 

 tic (Dragovich 1970), and for skipjack tuna (Alverson 1963; Nakamura 

 1965; Batts 1972); bluefin tuna, T. thynnus (Pinkus et al., 1971); and 

 albacore, T. alalunga (Pinkas et al., 1971) from the Pacific. 

 Comparative Diets 



Since temporal and spatial variations in the diets were so great 

 (data collected over a period of three years, and from several widely 

 different geographical locations), we believed that only by analyzing 

 small, discrete samples could we detect important differences in them. 

 To achieve this, we used only stomach contents of the two species col- 

 lected together off Oregon Inlet on 10 different days from May through 

 September 1981 (Table 3). 



Index of Relative Importance. — Indexes of Relative Importance 

 (IRI), which present the combined contributions of volume, frequency 

 of occurrence, and numbers of each food item to the diet (Table 3), 

 showed that, surprisingly, invertebrates were very important foods for 

 both species. The first five categories (ranks) for yellowfin were Teuthid- 

 ida (squids), unidentifiable fish, Raninidae, Scombridae, and unidentifi- 

 able crustaceans. For blackfin they were unidentifiable fish, Teuthidida, 

 Raninidae, Stomatopoda, and unidentifiable crustaceans. Obvious dif- 

 ferences were more clupeids and unidentifiable diogenid crabs in black- 

 fin, and more scombrids and squids in yellowfin. Other items were also 

 different, but their respective IRI values were relatively small (i.e., exo- 

 coetids for yellowfin = 9.7, for blackfin = 0.0). 



Correlation Coefficients. — Data from Table 3, ranked by IRI values, 

 were used to obtain quantitative comparisons of local food habits of the 

 two species. Three different measures were used: Spearman Rank Corre- 

 lation Coefficient (Fritz 1974); Kendall Rank Correlation Coefficient 

 (Bray and Ebeling 1975); and Pearson Product-moment Correlation 

 Coefficient (Goodall 1973). The first two require no assumption of 

 normality with regard to the distribution of the two predator species, 

 whereas the latter does. Cailliet and Barry (1978), who compared the 

 three methods of analyzing diets that have different distributions of prey 

 items, found that Spearman and Kendall correlation coefficients are 

 somewhat unpredictable when there are 1) a large number of ties, 2) a 

 considerable nonoverlap of prey items, and 3) high prey richness and 

 evenness (i.e., diversity). They felt that the Pearson method was best. 

 Although our data have a fairly low richness and evenness, there are 

 relatively few ties (2 for yellowfin, 3 for blackfin) and there is a fairly 

 good overlap in the diets. For these reasons all three methods of meas- 

 uring diet similarity are probably appropriate. Qualitatively, both spe- 

 cies feed extensively on epipelagic and mesopelagic fishes and inverte- 



