Nakano et al.: Development of swimming speed and schooling behavior in juvenile Sebastes cheni 
123 
meter (400-TST-902, Sanwa Supply Inc. 1 , Okayama, 
Japan) at the surface of each tank. A video camera 
(Handycam DCR-DVD505, Sony Corp.) was mounted 
vertically above the center of the experimental tanks 
to record fish behavior. Red light was used for ob- 
servations at 0.1 and 1 lx. Fish were not fed for 1 
day before the experiments. No fish died during these 
experiments. 
Experiment 1 
Two circular polycarbonate tanks of different sizes 
were used for the experiments, and fish were put in 
one of the tanks depending on fish size. A 30-L tank, 
with a water depth of 28 cm, was used for the experi- 
ment with fish at 23.4, 30.3, and 41.1 mm TL, and a 
100-L tank, with a water depth of 40 cm, was used 
for the experiment with fish at 49.4 and 58.6 mm TL. 
Fish behavior was observed under 6 light levels (0.1, 1, 
10, 100, 1000, and 10,000 lx). Twenty fish in each size 
group were introduced into each tank and were accli- 
mated for 1 h at 100 lx, a level that approximates the 
light intensity in their natural habitat (seagrass and 
macro-algal beds) during day. Fish were acclimated at 
each light intensity for 30 min before each observation. 
Fish behavior was recorded for 5 min by the video 
camera. Fish behavior observed in the 10 s of video 
collected immediately after the first 1 min of record- 
ing was used for analysis of cruising speed (CS). Two 
parameters of schooling behavior, nearest neighbor dis- 
tance (NND) and separation angle (SA), were defined 
according to the method described by Masuda et al. 
(2003). NND and SA are descriptors of parallel orien- 
tation and aggregation, respectively, calculated from a 
still frame of video at 1 min from the start of record- 
ing. NND was divided by TL to facilitate the compari- 
son among different size groups. SA, ranging from 0° 
to 180°, was expected to be 90° when fish were located 
in a random direction and was expected to decrease 
as they developed a parallel orientation. Aeration was 
stopped during this experiment. Individual fish were 
used in only one treatment and were not reused in this 
study. 
Experiment 2 
Burst [swimming] speed (BS) was measured by using 
sound stimuli according to the methods in Masuda et 
al. (2002). A 10-L, circular, polycarbonate tank, with a 
water depth of 15 cm, was used for observations of fish 
in all size classes. Ten fish were introduced into the 
tank and were kept for 1 h at 100 lx for acclimation. 
The fish were acclimated for 30 min under experimen- 
tal light intensity before the start of each experiment. 
A sound stimulus was produced with a steel nut (10 g) 
hung by a string (40 cm); the nut was released from a 
Mention of trade names or commercial companies is for iden- 
tification purposes only and does not imply endorsement by 
the National Marine Fisheries Service, NOAA. 
distance of 20 cm from the side of the tank to hit the 
tank wall. The experiment was conducted at the low- 
est and highest light intensities (0.1 and 10,000 lx). 
The BS was calculated from the distance of fish move- 
ment for 0.25 s after the sound stimulus. Aeration was 
stopped during the experiment. Individual fish were 
used in only one treatment and were not reused in this 
study. 
Statistic 
The CS, NND, and SA of juvenile white rockfish were 
compared at the different light intensities (6 levels) 
within the same size class and among results for fish 
in different size classes within the same light intensity 
by means of Kruskal-Wallis tests with Steel-Dwass’s 
tests for multiple comparisons. BS was compared for 
fish at the different light intensities (2 levels) within 
the same size class by the Wilcoxon test and among 
results for fish in the different size classes within the 
same light intensity by the Kruskal-Wallis test with 
Steel-Dwass’s test for multiple comparisons. All these 
statistical analyses were performed with the software 
programs R, vers. 3.1.1 (R Core Team, 2014) and JMP, 
vers. 8, (SAS Institute Inc., Cary, NC). 
Results 
Mean CS was highest under higher light intensi- 
ties (1000 or 10,000 lx) within all size classes of fish 
(Fig. 2). There was a significant effect of light intensity 
on CS in each fish size class (Kruskal-Wallis test with 
Steel-Dwass’s test for multiple comparisons: PcO.OOl). 
For fish within the same light intensity, size had a sig- 
nificant effect on CS. Across all light intensities, the CS 
of fish in the 2 largest size classes (49.4 and 58.6 mm 
TL) was significantly larger than the CS of fish in the 
smaller size classes (PcO.OOl). 
Mean NND decreased as fish grew. At 23.4 and 30.3 
mm TL, the NND under 0.1 lx was significantly greater 
than that under 100 lx (Kruskal-Wallis test with Steel- 
Dwass’s test for multiple comparisons: PcO.OOl; Fig. 3). 
Light treatment had no significant effect on NND in 
the three largest fish size classes (41.1, 49.4, and 58.6 
mm TL: P>0.05). Within the same light intensity, the 
NNDs of the one (23.4 mm TL under 1 and 10 lx) or 
two smallest size classes (23.4 and 30.3 mm TL under 
0.1 and 10,000 lx) were significantly greater than those 
of other size classes (PcO.OOl). 
The variability in SA within size and light-intensity 
groups was large enough that no significant effect of 
light intensity and fish size was detected (Kruskal-Wal- 
lis test: P>0.05). Mean values of SA ranged between 
29.8° (SD 17.7) for fish under 100 lx and in the size 
class of 49.4 mm TL and 102.0° (SD 51.6) for fish under 
0.1 lx and in the size class of 30.3 mm TL (Fig. 4). 
The effect of fish size on BS was significant under the 
highest light intensity of 10,000 lx (Kruskal-Wallis test: 
