Chase: Differencesin diet of Thtmmis thvnnus at seasonal feeding grounds off New England 



177 



80 



50 

 40 



sample and intersample variation, prey 

 weiglit skewiu'ss, and variance dependency 

 are widespread findings in food habit stud- 

 ies that reflect natural feeding interactions 

 (Smagula and Adelnian, 1982; Amundsen 

 and Klemetson, 1986; and Hodgson ct al., 

 1989). Large sample variation was evident 

 in the present study and hmited the appli- 

 cation of parametric statistics. 



Differential digestion rates can result 

 in an underestimate of the relative im- 

 portance of a major prey item (Hyslop, 

 1980). Soft-bodied squid likely undergo di- 

 gestion quicker than bony fish, and, there- 

 fore, prey weight measurements may have 

 been negatively biased. Postcapture de- 

 composition of prey has not been mea- 

 sured for bluefin tuna but may negatively 

 bias stomach-contents weight for samples 

 held a long time after capture (Holliday. 

 1978 ). The effect of bias on the prey weight 

 data is suspected to be minor because of 

 strong market incentives to chill harvested bluefin tuna 

 and because samples were collected soon after capture. 



The separation of chum bait from natural prey and the 

 exclusion of chum from diet analyses were potential min- 

 imal sources of bias. Because bluefin tuna swallow prey 

 whole and chum is cut into small pieces, cut chum was 

 readily separated from natural prey in stomach samples. 

 Samples of small bluefin tuna that fed on the discards of 

 trawlers in the area south of Martha's Vineyard did pres- 

 ent a problem, and many of these samples were removed 

 from analyses because natural prey and discards could not 

 be distinguished. Other study areas did not indicate an as- 



n=1866, /•=0.357, P=<0.001 







Relationship 

 Maine sampl 

 ach samples 



50 100 150 200 250 300 350 



Bluefin tuna lengtti (CFL, cm) 



Figure 5 



between bluefin tuna length and prey length for all Gulf of 

 es during 1988-92 that contained measurable prey (190 stom- 



sociation between tuna fisheries and trawl fishing; conse- 

 quently, no evidence of discard feeding was obsei-ved. It is 

 possible that the availability of chum could limit feeding 

 on natural prey, although no evidence of this was found. 

 Most chum was obsei-ved in the stomach contents of blue- 

 fin tuna from Stellwagen Bank and Jeffreys Ledge, but 

 these areas also had the highest mean stomach-contents 

 biomass and lowest frequency of empty stomachs. 



Most of these biases were negative, possibly decreasing 

 prey weight measurements in stomach contents, and could 

 not easily be separated from natural conditions that lead 

 to suboptimal feeding habits. Butler and Mason (1978) re- 



