126 
Fishery Bulletin 113(2) 
■ 0.1 lx □ 10,000 lx 
jyiM 
Total length (mm) 
Figure 5 
Burst [swimming] speed (BS; measured in millimeters per sec- 
ond) of juvenile white rockfish ( Sebastes cheni) in each size class 
(total length in millimeters). Error bars indicate standard de- 
viations. The effect of fish length on BS was significant only at 
the light intensity of 10,000 lx (Kruskal-Wallis test, P<0.05). As- 
terisks indicate significant difference in BS between the treat- 
ments with different light intensities (Wilcoxon test: * P<0.05; 
** PcO.Ol). 
Schooling behavior 
The NND and SA have been referred to as indices of 
the density of individual fish and the orderliness of a 
fish school (Masuda, 2006; Miyazaki et al., 2000). De- 
velopment of schooling behavior has been observed to 
decrease predation risk of larval and juvenile fishes 
(Masuda, 2006, 2011). The sizes of experimental tanks 
were expected to affect the NND with smaller values in 
smaller tank sizes. The NND of juvenile white rockfish 
at 30.3 mm TL (tank size: 30 L) was greater than val- 
ues observed for fish at larger sizes (49.4 and 58.6 mm 
TL) that were put in larger tanks (100 L) in our study. 
These results indicate that the decrease in NND for 
fish at larger sizes was not attributed to the difference 
in tank size in our experiments. 
Compared with data from tank experiments avail- 
able for other fish species at similar body lengths and 
light conditions, NNDs of juvenile white rockfish from 
our study were greater. The NND (absolute value in 
millimeters) of white trevally juveniles in a previous 
experiment was 15.5 mm (SD 1.4) at 20 mm TL (Mi- 
yazaki et al., 2000), a much shorter NND than that 
of juvenile white rockfish (62.5 mm [SD 27.9] at 23.4 
mm TL) in our study under the same light intensity 
(0.1 lx). The standardized NND (divided by fish total 
length) was also smaller in white trevally (0.75 mm 
TL-1 at 20 mm TL) than in white rockfish (2.69 mm 
TL-1 at 23.4 mm TL). The decrease in NND with in- 
crease in body size (from 23.4 to 41.1 mm TL) indi- 
cates the development of school formation. The ability 
of white rockfish juveniles to avoid predation through 
the use of schooling behavior would be mini- 
mal during the postsettlement period (about 
20-40 mm TL) compared with the ability of 
white trevally because that behavior is not 
well developed in white rockfish. 
It has been suggested that development of 
mechanosensory organs, such as lateral lines, 
contribute to decreases in NND in school for- 
mations of juvenile fish (Masuda and Tsuka- 
moto, 1998). Formation of scales and change 
in body color both have been reported to occur 
in cultured rockfishes at about 30 mm TL, the 
size at which physiological and morphological 
metamorphosis is complete (Suzuki and Aida, 
1999; Nagasawa, 2001). The decrease in NND 
observed for fish up to 41.1 mm TL at all light 
levels in our study appears to coincide with 
the size at which white rockfish complete their 
metamorphosis from larvae to juveniles. For 
fish in the size classes of 23.4 and 30.3 mm 
TL, the NND was shortest under intermediate 
light intensity (100 lx). Juvenile white rockfish 
of these sizes are considered daylight feed- 
ers (Shoji et al., 2011) and swim faster under 
high light intensities than under intermedi- 
ate intensities, when compared with the same 
size class. It seems that small juveniles can- 
not form schools with small NNDs under low 
light conditions because the complete sensory functions 
of the eyes and lateral lines are undeveloped at the 
2 smallest size classes (Suzuki and Aida, 1999; Naga- 
sawa, 2001). 
Separation angle of fish in early life stages is a proxy 
of orderliness of a fish school (Masuda et al, 2003). 
Japanese anchovy ( Engraulis japonicus) juveniles form 
schools with SAs <30° (Masuda, 2011). Juveniles of a 
piscivorous fish, Japanese Spanish mackerel ( Scomb - j 
eromorus niphonius), form schools in which individuals 
are more dispersed with an SA of 33-44° (Masuda et 
al., 2003). In our study, SAs observed for white rock- 
fish in all length classes and at light intensities tested 
(>40°) were larger than previously reported SAs of oth- 
er species. These observations from our study indicate 
that juvenile white rockfish form loosely aligned aggre- 
gations that maintain their body angle in inconsistent 
directions during the juvenile period. 
Antipredator behavior 
Predation rate of juvenile fishes is controlled by the 
probability of encounter with, and capture by, their 
predators (Bailey and Houde, 1989). Capture success 
of predators is affected by the BS of prey fish (Bai- 
ley and Houde, 1989). In our study, BS of the juvenile 
white rockfish under the low light intensity (0.1 lx) was 
significantly lower than BS of fish under high light in- 
tensity (10,000 lx) in all size classes tested. In addi- 
tion, the BS of fish under the highest light intensity 
increased as white rockfish grew, but BS of fish under 
lowest light intensity was stable regardless of juvenile 
