430 
Fishery Bulletin 99(3) 
dictions. When ABT schools are captured by purse seine, 
it is assumed that nearly the entire school is captured. 2 If 
the net intersects an edge of the school while it is being de- 
ployed, the entire school will change its direction of travel 
in unison resulting in either the entire school being encir- 
cled in the net and captured or in the entire school escap- 
ing into open water. As a result, the estimate of total NFS 
from this data point is likely to be an accurate count of 
the number of individuals in the school. It is encouraging 
that our model so closely estimated the NFS of large gi- 
ants considering that it was constructed from data for age 
2+ fish that are a fraction of the size of large giant bluefin 
tuna. The accuracy of our prediction indicates the poten- 
tial for generality of ABT school structure across both tu- 
na size and NFS. However, substantial verification of our 
models is necessary before they may be applied to abun- 
dance estimation. 
Predicting NFS from school dimensions 
An alternative to using N 1 and N 2 to predict N s is to use 
maximum school length and width. Identifying the longest 
and widest axis of the surface of a school and counting the 
number of individuals along these axes may yield more 
accurate estimates of the total NFS. The maximum dimen- 
sions of bluefin schools (length, width) had greater power 
as predictors of N s than N 1 and N. 2 ( model 3 versus models 
4, 5, 6, Table 2). Moreover, a combination of length and 
width to predict N s produced a more confident estimate of 
school size than models using either variable individually. 
Irregularity in length and width at small school sizes likely 
introduced variation that reduced the individual predic- 
tive power of these variables. Inclusion of both length and 
width in the model to predict N s could allow accurate pre- 
diction of small schools that are elongate or wide. 
The regression diagnostics for the model using school 
length and width to predict school size (model 6, Table 
2) suggest that it is the more reliable model for estimat- 
ing NFS with aerial photographs. For the single open- 
ocean estimate available in the literature (Lutcavage and 
Kraus, 1995), the length and width of the school in num- 
ber of individuals could not be determined. Model 6 may 
have the potential to yield a more accurate estimate of 
school size with a wider range of photograph qualities (as 
affected by sea state, water clarity, sunlight, etc. ) because 
of its ability to predict NFS from partial surface counts. 
However, the utility of school length and width to predict 
N s will remain uncertain until field data are available for 
thorough evaluation. 
Enclosure effects 
The effects of capture and captivity on school structure 
and behavior were points of concern in our study. Evaluat- 
ing the effects that the enclosure had on school structure 
is problematic because the same factors that led to the use 
2 Genovese, M. (captain). 1995. Personal commun. FV White 
Dove Too , 600 Shunpike Rd., Cape May, NJ 08210. 
of an enclosure to make the observations preclude a direct 
in situ comparison. Because tuna are highly mobile and 
unpredictable in their movements, it would be difficult 
to obtain a number of school observations comparable to 
that collected from the captive fish in our study. Further- 
more, the ability to approach schools in nature without 
significantly disturbing them is questionable, and effec- 
tive observation at a distance that would not cause dis- 
turbance would be unlikely because of turbidity and the 
normal movement of schools. In this respect, a group of 
tuna that has become comfortable with the presence of 
human observers may be a better source of accurate school 
structure observations than a school of noncaptive fish 
that may perceive a human or mechanical presence as a 
predatory threat and react accordingly. Evidence of the 
acclimation of the study specimens to the enclosure was 
seen in their active and aggressive feeding behavior (Han- 
rahan and Juanes 3 ) that was similar to the available anec- 
dotal accounts of their open-water feeding behavior. The 
relation between fish length and enclosure dimensions 
may cause the impression that school formation was con- 
strained heavily by captivity and that multiple schools 
could not achieve meaningful separation from one another 
(Fig. 1 illustrates a school of tuna at a scale of 1:1 to the 
enclosure). However, the relatively low density of fish in 
the enclosure (0.05 kg/m 3 ) allowed individuals to move 
in an uninhibited manner within the enclosure. Although 
the extent of enclosure-induced behavioral modification 
cannot be quantified in our study, the ability of our simple 
linear model to predict accurately the NFS for a noncap- 
tive school is very encouraging. 
Acknowledgments 
We thank S. Belle, P. Sylvia, and the volunteers of the 
New England Aquarium bluefin tuna research project 
for invaluable collaborative support. This manuscript was 
greatly improved by comments provided by J. Boreman, E. 
Brainerd, R. Rountree, K. Friedland, and four anonymous 
reviewers. This work is the result of research sponsored 
by NOAA National Sea Grant Office, Department of Com- 
merce, under grant NA 46RG0470, Woods Hole Oceano- 
graphic Institution Sea Grant project no. 22800058 to F. J. 
Additional support was provided by NOAA’s Cooperative 
Marine Education and Research Program, the Department 
of Natural Resources Conservation (UMass, Amherst), 
and the Manasquan River Marlin and Tuna Club. 
Literature cited 
Block, B. A., J. E. Keen, B. Castillo, H. Dewar, E. V. Freund, 
D. J. Marcinek, R. W. Brill, and C. Farwell. 
1997. Environmental preferences of yellowfin tuna ( Thun - 
3 Hanrahan, B., and F. Juanes. In prep. Atlantic bluefin 
tuna (Thunnus thynnus thynnus) prey size-selectivity and 
feeding behavior. 
