Hanrahan and Juanes: Estimating the school size of Thunnus thynnus thynnus 
429 
measurements (Table 1), the time when school structure 
data were collected, or low statistical power from small 
sample sizes. However, on a very limited number of occa- 
sions, the thermocline became situated within the enclo- 
sure at approximately 10 m. Only a few individual fish 
traversed the thermal boundary, and such excursions were 
very brief. Entire schools were not observed to cross to the 
cold side of the thermal boundary. The thermal profile is 
an important factor in determining the vertical distribu- 
tion of Pacific yellowfin (Thunnus alhacares) and bigeye 
(T. obesus ) tuna (Holland et ah, 1990). It is reasonable to 
assume that the thermal profile is important in the verti- 
cal distribution of Atlantic bluefin tuna as well, and that 
this effect was not detected in our study because of the 
periodicity of observations, the limited variation of envi- 
ronmental conditions within the enclosure, or the spatial 
constraints imposed by the enclosure. 
Predicting NFS from surface counts 
Our least-squares regression model predicts the number 
of fish in ABT schools from the number of individuals at 
the surface of the school without attempting to describe 
the fine-scale structure within schools in terms of interindi- 
vidual spacing and orientation. The number of fish in the 
surface interval alone accounts for 74% of the variation in 
school size of three-dimensional schools similar to a densely 
packed dome. Because the only school type that we observed 
was similar to the densely packed dome, the applicability of 
our model to three-dimensional schools of other configura- 
tions is questionable and may be determined only by study- 
ing ocean schools of other configurations. Furthermore, the 
application of our model to two-dimensional schools such as 
parabolas, echelons, and surface sheets is not appropriate 
or necessary given the apparent lack of a three-dimensional 
component in their structure. 
The only data point available to verify our model pre- 
dictions is from a school described as “dome-shaped” that 
was photographed and subsequently captured by purse- 
seine (Lutcavage and Kraus, 1995). Thirty-two fish were 
counted at the surface of this school from an aerial photo- 
graph, and 125 “large giant” ABT were subsequently cap- 
tured by the purse-seine vessel. We applied our model that 
predicts NFS from the number at the surface of the school 
to this datum to produce a NFS prediction. The prediction 
with our model of 134 individuals differed by 7% from the 
purse-seine capture of 125 individual fish, and was well 
within the 95% confidence intervals (Fig. 7) of model pre- 
