Chase: Differencesin diet of Thunnus thynnus at seasonal feeding grounds off New England 



173 



Comparison of study areas The top 12 prey items, 

 overall, were lanked i'or each area by frequency of occur- 

 rence, and area differences were tested with Spearman 

 rank correhition. Of the ten painvise comparisons, only 

 Stelhvagen Bank and Great South Cliannel showed 

 a significant association in the ranking of prey items 

 (;=0.98. P<0.02). attributable to a high rank of sand 

 lance and a similar ranking of squid. Atlantic herring. 

 and Atlantic mackerel for both areas. 



Stomach-contents biomass 



Comparison of study areas Large differences in stom- 

 ach-contents biomass were found: Jeffreys Ledge aver- 

 aged nearly 2 kg, followed by approximately 1 kg for 

 Stellwagen Bank and Great South Channel, and less 

 than 0.5 kg for the remaining areas. Stomach-con- 

 tents biomass data from the five areas were positively 

 skewed and heteroscedastic. The natural logarithm- 

 transformed data for Stellwagen Bank, Cape Cod Bay, 

 and Great South Channel were normal (Wilk-Shapiro 

 test, P>0.05). Transformed biomass data for the other 

 two areas still differed significantly from normality. 

 Transformation of biomass data reduced the inequal- 

 ity of variances, but significant differences (Levenes 

 test, P<0.05) remained, which precluded use of analy- 

 sis of variance. The Brown-Forsythe test for unequal 

 variances showed a significant effect of area on stom- 

 ach-contents biomass (P<0.0001). Painvise compari- 

 sons of the equality of prey weight means were made 

 with the Welch test (Bonferroni corrected significance 

 level of P=0.005). All area paii-wise comparisons of 

 stomach-contents biomass were significantly different 

 except that between Cape Cod Bay and Great South 

 Channel(P=0.816). 



The similarity in the amount of food found at Cape 

 Cod Bay and Great South Channel was also indicated 

 by the geometric mean of stomach-contents biomass 

 (Table 4). The arithmetic mean of stomach-contents 

 biomass was higher at the Great South Channel than 

 Cape Cod Bay, but this value was biased by a few 

 samples with large amounts of prey. The stomach- 

 contents biomass range for Cape Cod Bay did not exceed 

 3.0 kg, in contrast to the wider range for the Great South 

 Channel up to 16.0 kg, including 13 samples over 3.0 

 kg. The use geometric means reduced the bias of skew- 

 ness and indicated that samples from Jeffreys Ledge con- 

 tained the most prey and that the amounts declined mov- 

 ing southward. 



Effect of tuna size Increased stomach-contents biomass 

 with increasing body size (due to increasing gape and 

 stomach size) was not clearly demonstrated from these 

 data. Correlation of stomach-contents biomass to bluefin 

 weight was not significant for Gulf of Maine samples. A 

 scatterplot of data for the Gulf of Maine samples revealed 

 that a large majority of the samples contained small 

 amounts of food, regardless of body size (Fig. 3). A size- 

 related trend was observed in that only very large bluefin 

 tuna (>250 kg) contained more that 6 kg of food. Size 



Osteichttiyes 

 93% 



Elasmobranchii 

 3% 



Cephalopoda 

 2% 

 Ponfera 

 2% 



B 



other invertebrates 

 sand lance 2% 



23% ^^ I f—.. other fish 



squid 



2% 



Atl mackerel 



3% 



Atlantic herring 

 53% 



Figure 2 



Percent prey weight composition in stomach contents of bluefin 

 tuna caught off New England during 1988-92 ln=568). The taxo- 

 nomic composition (A) includes 0.05'^ Crustacea and O.Ol^r Uro- 

 chordata. The comparison by major prey type (B) comprises the 

 five most common prey and all remaining fish and invertebrates. 



effects were also compared by using the ratio of stomach- 

 contents biomass and tuna weight C* kg/kg, wet weight) to 

 tuna length. The percentage of food to body mass declined 

 with increasing bluefin tuna length (Fig. 4). Food obsei-ved 

 in 120-149 cm bluefin tuna averaged over 1** of their body 

 weight, and declined to approximately 0.5% for bluefin 

 tuna over 230 cm. The high ratios primarily resulted from 

 large meals of Atlantic herring or sand lance. Cape Cod 

 Bay ratios were consistently the lowest among the four 

 areas, ranging from 0. 1 to 0.2%. 



Characteristics of prey species 



Prey size Prey size was evaluated for 190 stomach sam- 

 ples that contained measurable prey from the four Gulf of 

 Maine areas. A total of 1866 prey items were measured, of 

 which 95% were either sand lance, Atlantic herring, squid, 

 or Atlantic mackerel (Table 5). A significant positive cor- 



