58 
Fishery Bulletin 108(1) 
size sufficiency was reached if the difference between 
the slopes was not significantly different (P>0.05). 
The effects of within-winter period, year, and size 
class on bluefin tuna diets were determined by using 
row x column ( RxC ) tests of independence and using 
counts of stomachs with a particular prey species (Sokal 
and Rohlf, 1981). Prey species were grouped by family 
for RxC tests and degrees of freedom were calculated 
as (rows- 1) (columns- 1). 
Gastric evacuation rate and daily ration 
Individual stomach fullness (kg prey/kg predator) values 
were highest in early afternoon and lowest during the 
early morning for both large medium and giant bluefin 
tuna. These high and low periods were used as begin- 
ning and ending points, respectively, to estimate gas- 
tric evacuation rates. Stomach fullness values were 
assigned to one-hour time periods (with the exception of 
the first and final hours when stomachs were at values 
before maximum and after minimum stomach fullness, 
respectively) to examine diel feeding patterns; data from 
2004-05 ( /? = 64) and 2005-06 (n = 114) were pooled for 
this analysis to increase sample size. Gastric evacuation 
rates ( GER ) of bluefin tuna were estimated using an 
exponential decay model (Elliott and Persson, 1978), 
S t = S 0 • e- GER -\ (1) 
where S t = the individual stomach fullness at time t\ 
S 0 = the stomach fullness at time t- 0; 
GER = the instantaneous rate of gastric evacuation 
(rate per hr); and 
t = the time in hours after peak gut fullness. 
The number of fish used in this analysis was lower 
than the total collected because time of capture was 
not always provided by the fishing crew. Although other 
evacuation rate models have been used with tuna (e.g., 
linear, Olson and Boggs, 1986), we chose the exponen- 
tial model because we could not test between multiple 
evacuation models given the gap in stomach contents 
data at night. Additionally, an exponential decline may 
better describe the evacuation rates of fish which require 
rapid digestion rates because of high metabolic demands 
(Bromley, 1994). The difference (GER di ^ erence ) between 
GER iarge medium and GER giant was estimated by using the 
nonlinear (NLIN) procedure of SAS (SAS, 1996). If the 
confidence interval of GER di ^ erence contained the value 
of zero then it was assumed that GER large medium and 
GER„ iant were not significantly different. 
Daily ration (kg prey/kg predator/day) estimates of 
large medium, giant, and pooled bluefin tuna size class- 
es were calculated by using the Eggers (1977) approach, 
DR = 24 • S • GER, (2) 
where DR = the daily ration estimate; and 
S = the mean stomach fullness of the hourly 
means. 
For time points at which no stomachs were collected 
(1900-0300 Eastern Standard Time [EST]), stomach 
fullness values were estimated from the gastric evacu- 
ation model (described above). 
Population-level consumption 
Annual population-level consumption of Atlantic men- 
haden by bluefin tuna during their residency in North 
Carolina (C Pop ) was estimated as 
G Pop = R BFT ' R NC ’ ER * ' T NC > (3) 
where B bft = bluefin tuna biomass (kg); 
R nc = proportion of the bluefin tuna population 
in North Carolina during the winter; 
DR = the estimate of bluefin tuna daily ration 
(kg prey/kg predator/day) in this study; 
W Mh = the proportion by weight of Atlantic men- 
haden in bluefin tuna stomachs; and 
r nc = the time (days) that bluefin tuna and 
Atlantic menhaden are both present in 
the coastal waters of North Carolina. 
In order to determine the precision of C p , a distribu- 
tion of C Pop estimates was obtained by using simulation 
software (@RISK, vers. 5.0, Palisade Corp.). The Monte 
Carlo simulation approach is described in Overholtz 
(2006); briefly, a distribution was created for each vari- 
able in the C Pop equation above. Then, a random draw 
was made from each distribution and a new estimate of 
C Po p was made. This process was repeated 5000 times. 
The range of C p estimates was then compared to the 
range of consumption estimates previously published on 
other known Atlantic menhaden predators (e.g., bluefish, 
striped bass, weakfish, and the Atlantic Coast com- 
mercial harvest). The consumption estimates presented 
are annual and cover the entire U.S. East Coast. Given 
past bluefin tuna diet studies conducted on summer 
and fall feeding grounds (e.g., Chase 2002), we assume 
that Atlantic menhaden are not a major prey of bluefin 
tuna in other areas at other times of the year. Thus, 
our estimates off North Carolina during winter are 
likely indicative of the annual coastwide consumption 
of Atlantic menhaden by bluefin tuna. 
Bluefin tuna captured in North Carolina were esti- 
mated to be predominantly age 6+ fish based on size-at- 
age regressions (Murray-Brown et al. 3 ). The most recent 
(2005) population estimate was 94,836 age-6+ bluefin 
tuna (ICCAT, 2007). These values were coupled with 
individual mean weight-at-age 3 estimates to calculate 
the total biomass (B bft ) of age 6+ bluefin tuna from 
the western Atlantic. A pert distribution of the mean, 
minimum, and maximum B bft values was constructed 
by using one standard deviation below and above the 
3 Murray-Brown, M., S. McLaughlin, and C. Lopez. 2007. His- 
tory of United States Atlantic bluefin tuna size class clas- 
sification and changes. NOAA Fisheries Final Report, 14 p. 
U.S. Dep. Commerce, Silver Spring, MD. 
