426 
Fishery Bulletin 99(3) 
Table 1 
Mean, minimum, maximum, and standard deviation of environmental data (pH; °C=degrees Celsius; mg/L=milligrams/liter; 
ntu=turbidity units; ppt=parts per thousand) recorded during schooling observation periods. The r 2 and P-values for linear regres- 
sions of each school measure on each environmental variable are provided. Measurements of dissolved oxygen were discontinued 
due to instrument failure. 
Date 
Observation 
start time 
pH 
Temp. 
(°C) 
Dissolved oxygen 
(mg/L) 
Total suspended solids 
(ntu) 
Conductivity 
(ppt) 
27 Jun 1996 
10:00 AM 
— 
— 
— 
— 
— 
21 Jul 1996 
3:30 PM 
8.51 
22.90 
9.37 
— 
27.45 
6 Aug 1996 
10:11AM 
8.42 
22.18 
8.62 
0.12 
28.65 
7 Aug 1996 
8:45 AM 
8.43 
24.18 
8.39 
0.23 
28.98 
15 Aug 1996 
3:45 PM 
8.44 
23.84 
— 
0.60 
28.06 
17 Aug 1996 
10:48 AM 
8.46 
23.86 
— 
0.77 
28.42 
22 Aug 1996 
11:55 AM 
8.47 
23.40 
— 
0.05 
27.63 
23 Aug 1996 
11:15AM 
8.50 
23.50 
— 
0.45 
29.63 
Mean 
8.46 
23.41 
8.79 
0.37 
28.40 
Minimum 
8.42 
22.18 
8.39 
0.05 
27.45 
Maximum 
8.51 
24.18 
9.37 
0.77 
29.63 
Standard deviation 
0.04 
0.68 
0.52 
0.28 
0.76 
N s 
0.1* 
0.15* 
0.84* 
0.09* 
0.04* 
Coefficient of determination 
School length 
0.01* 
0.19* 
0.91* 
0.02* 
0.01* 
School width 
0.09* 
0.22* 
0.46* 
0.1* 
0.03* 
: /'>(). 10 . 
Table 2 
Regression equations employed to predict school size. The results of partial F-tests indicated that model 6 
school size UV S ). N v N 2 , school length, and school width were measured in individual fish. 
is the best predictor of 
Model 
Equation 
n 
r 2 
partial F 
1. N j vs. N s 
N s 
= (0.0337867 + 3.3173368 x NJ+ N l 
74 
0.67* 
2. N j vs. N s (weight= 1/variance) 
N s 
= (0.2042865 + 3.1849359 x NJ+ N 1 
74 
0.74* 
3. N v N 2 vs. N s (weight= 1/variance) 
= (-1.412557 + 1.8516536 xN x + 0.4190831 x N2)+ AT,+ N 2 
74 
0.70* 
4. Length vs. N s 
N s 
= -6.769737 + 5.5334126 x length 
74 
0.74* 
5. Width vs. N s 
N s 
= -6.70494 + 5.7894971 x width 
74 
0.79* 
6. Length, width vs ,N S 
N s 
= -9.788046 + 3.6463236 x width + 2.7083604 x length 
74 
0.86* 
(6 vs. 4) * 
(6 vs. 5) * 
* P<0.0001. 
ing an r 2 of 0.79 (P<0.0001) (Eq. 5, Table 2). The final, 
multiple regression model used school length and width 
to predict N $ (Eq. 6, Table 2, Fig. 5) with an r 2 of 0.86 
(P<0.0001). Partial F - tests revealed that the final multiple 
regression model explained significantly more of the vari- 
ation in school size than either length or width separately. 
No significant relationships were seen when school dimen- 
sions were considered in terms of environmental factors 
(Table 1). 
Discussion 
School structure 
The non-normal frequency distribution of NFS and the 
appearance of modes in our data (Fig. 4) suggest the 
formation of elective groups of particular number. Previ- 
ous investigations in school size show that fishes actively 
assemble into elective group sizes dependent upon the 
