Hanrahan and Juanes: Estimating the school size of Thunnus thynnus thynnus 
427 
interaction of predation risk, food availability, migratory 
status, and the species’ life history (Hager and Helfman, 
1991; Pitcher and Parrish, 1993). Elective group size for 
a species differs continuously under natural conditions 
with varying abiotic and biotic factors. At a discrete point 
in time when two schools were present in the enclosure, 
the size of one group determined the maximum number 
of individuals that could be in the other. However, there 
were no barriers to formation of groups numbering any- 
where from 2 to 50 individuals. Therefore, the modes in 
group number that we observed should have resulted from 
underlying behavioral tendencies rather than enclosure- 
induced limitation of group size. If elective group sizes 
form in response to environmental conditions, then the 
range of environmental conditions during the study period 
was probably insufficient to detect environmental influ- 
ences on NFS, and therefore elective group sizes. 
The range of school sizes that we observed in the enclo- 
sure was limited by the number of individuals available to 
join schools. Photographs of ABT in the northwest Atlan- 
tic Ocean over a 50-day period in 1993 revealed that sur- 
face school counts ranged from 5 to 1294 individuals, and 
that the median school size was 84 individuals (Lutcav- 
age and Kraus, 1995). The median value is well in excess 
of the maximum number of fish observed in the surface 
layer of our schools, emphasizing the importance of verify- 
ing the accuracy of our model predictions for larger schools 
with field data. When very large schools occur in relatively 
shallow water, the vertical depth of the school would be 
confined by the maximum water depth. A similar effect 
could be imposed by physical and chemical barriers such 
as vertical stratification in temperature and dissolved ox- 
ygen. Tagging studies may reveal more of the individual 
and group behaviors of this species in relation to the en- 
vironment and assist in further understanding the way 
in which school structure may be affected by the physical 
and chemical environment. 
Our results show that the vertical distribution of fish (in 
intervals) varies little across the range of observed school 
sizes. The ANCOVA of N t on N s illustrated that the slope of 
each regression was significantly different from all others 
(all P<0.001), but the biological importance of this differ- 
ence is questionable because the slopes varied by less than 
10%. Although the rate at which individuals are added to 
each interval varies as school size increases, there is no 
consistent trend in slopes among intervals. Furthermore, 
the statistical significance of the difference in slopes may 
be driven by the small standard error for each regression 
and extremely high power (>0.99). 
The shape of large schools of bluefin tuna was less vari- 
able than that of small schools. Schools of less than 15 in- 
dividuals are less vertically expansive, and generally one 
to three fish deep; larger (> 15 individuals) schools are more 
than three individuals deep (Fig. 6). The pattern of in- 
creasing vertical depth continues to the largest school sizes 
observed, which are nearly always more than five individ- 
uals deep. The weakly cone-shaped profile of the model 
school depicted in Figure 3 is representative of the shape 
of most schools with more than 3 intervals. Smaller schools 
tend to be distributed in a vertically shallow, loosely oval 
profile. Our findings related to school structure are consis- 
tent with the observations of other investigators who have 
