176 



Fishery Bulletin 100(2) 



tuna were sampled prior to 1970 (Crane, 1936; Kiumholz, 

 1959; Dragovich, 1970; Matthews et al.') were limited 

 because of small sample sizes. In pre- 1970 studies, sand 

 lance or benthic organisms were not found to be prey 

 items. Mason (1976) was the first to record sand lance in 

 bluefin stomach contents. In three later studies, sand lance 

 were found to be the most frequently occurring prey, fol- 

 lowed by squid (Holliday, 1978; Eggleston and Bochenek, 

 1990; and the present study). 



Two limitations common to previous diet studies were 

 the small number of samples collected and the broad geo- 



120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 

 Tuna length (cm) 



Figure 4 



Relationship between the ratio of stomach-contents biomass to tuna 

 weight O wet weight [kg/kg]) and tuna length (cui-ved fork length). All 

 Gulf of Maine samples for bluefin lengths of 120-299 cm (/i=519) were 

 combined and arranged in lO-cni bins. 



graphical range over which they were collected. In con- 

 trast, large numbers of samples were collected in the pres- 

 ent study, at discrete feeding grounds on the New England 

 continental shelf; moreover this study is the first in which 

 the diet of bluefin tuna in the Gulf of Maine was quantified 

 during their seasonal feeding migration. The major con- 

 tribution of Atlantic herring and the common occurrence 

 of demersal organisms as prey have not been previously 

 recorded for bluefin tuna in the western North Atlantic 

 Ocean. 



Sampling biases 



Regiu'gitation, diel and seasonal effects, differ- 

 ential digestion rates, postcapture digestion, 

 and predator size effects can be problematic 

 to food habit studies and can bias data col- 

 lection. (Bowen, 1986). This section discusses 

 the potential effect that sampling biases may 

 have on the findings of significant differences 

 in the stomach-contents biomass and prey 

 associations. A majority of the empty stom- 

 achs had distended stomach rugae, indicating 

 that regin-gitation had occurred. Similar obser- 

 vations were made during previous studies 

 (Mason, 1976; and Dragovich, 1970), which 

 would cause a negative bias for stomach-con- 

 tents weight measurements. Diel and seasonal 

 effects on feeding are difficult to assess and 

 are large sources of variation. Large inters- 

 ample variance can be expected given the 

 complex interactions between predator popu- 

 lations and the temporal and spatial distri- 

 bution of prey (Smagula and Adelman, 1982; 

 Hodgson et al., 1989). Increasing sample size 

 often may reduce variability, but high intra- 



Table 5 



Summary statistics for length data from major prey items m stomach contents and bluefin tuna sampled from Gulf of Maine 

 (GOM), July-October, 1988-92. Prey lengths are total length, except for squid, which are mantle 'engths. Bluefin tuna lengths are 

 curved fork length (CFL). Area 1 = Jeffreys Ledge; area 2 = Stellwagen Bank; and area 4 = Great South Channel. ?i = number of 

 samples taken. 



Species 



Area 



Prev 



All prey 



Atlantic mackerel 

 Squid 



Atlantic herring 

 Atlantic herring 

 Atlantic herring 

 Sand lance 

 Sand lance 

 Sand lance 



GOM 

 GOM 

 GOM 

 GOM 



1 



4 

 GOM 



2 



4 



