274 
Fishery Bulletin 95(2), I 997 
However, with the data from spring releases, mean 
recapture frequency of the smallest fish released (45 
to 60 mm) was statistically similar to frequencies of 
some of the larger fish released (70 to 85 mm and 
those >110 mm) (Fig. 7; P= 0.33). Fish from groups 2 
and 4 (60 to 70 mm and 85 to 110 mm when released) 
had marginally greater recapture frequencies than 
those for small fish (P <0.03; spring releases). Fish 
from the two largest size intervals (fish >85 mm) 
released in summer exhibited mean recapture fre- 
quencies about twice as high as those for any size 
fish from spring releases (P<0.02). 
Interaction between size-at-release effects and re- 
lease season effects was statistically significant 
(P=0.01, season x size interaction term, Table 5). A 
significant interaction term indicates dependence of 
one factor upon the other; in this case, size-at-release 
affected recapture rate (P<0.001), but the degree of 
that effect depended upon release season. 
Size structures of released cultured Mugil cephalus 
and wild recruits A comparison of the sizes of fish 
in cast-net scamples revealed that similar-size indi- 
viduals schooled together. One month after spring 
releases, most of the smaller tagged striped mullet 
collected were schooling with relatively large num- 
Size-at-release interval 
Figure 6 
Mean recapture frequencies (± SE, «= 6 lots) for the five 
sizes of fish released into Kaneohe Bay during summer 
releases (see Fig. 4 for description of fish size-at-release). 
Data are mean recapture frequencies per release lot ([num- 
ber recaptured/number released] x 100%) summed over 
collections made between 16 and 36 weeks after release. 
See Figure 4 for description of fish size-at-release codes. 
Letters above bars indicate results of multiple compari- 
sons of means; size-at-release intervals that share the same 
letter were not significantly different. 
bers of wild M. cephalus similar in size to the tagged 
individuals. However, the larger cultured fish re- 
leased found relatively few counterparts in size 
among wild individuals at that time (Fig. 8). The size 
structure of cultured fish released in spring was 
clearly out of phase with the wild recruitment pulse 
at that time. Whereas we had timed spring releases 
Size-at-release interval 
Figure 7 
Recapture frequencies (± SE, n=6 lots) for the five sizes of 
fish released into Kaneohe Bay during spring releases (see 
Fig. 4 for description of fish size-at-release). Data are mean 
recapture frequencies per release lot ([number recaptured/ 
number released] x 100%) summed over collections made 
between 16 and 36 weeks after release. Letters above bars 
indicate results of multiple comparisons of means; size-at- 
release intervals that share the same letter were not sig- 
nificantly different. 
Table 5 
ANOVA table (randomized-block design, lots=blocking vari- 
able) for evaluation of release season and size-at-release 
effects on recapture frequencies after 4 months in the wild. 
Data (means per release lot) were combined here over the 
20-week period following 4 months in the wild (weeks 16 
to 36). Recapture frequencies are percent of the total num- 
ber of fish released that were recovered during this pe- 
riod; these proportions were arc-sin square-root trans- 
formed prior to analysis. 
Source of 
variation 
Sum of 
squares 
df 
Mean 
square 
F-ratio 
P 
Release lot 
0.009 
2 
0.004 
4.309 
0.030 
Release season 
0.000 
1 
0.000 
0.021 
0.886 
Size-at-release 
0.030 
4 
0.007 
7.173 
0.001 
Season x size 
0.019 
4 
0.005 
4.577 
0.010 
Error 
0.019 
18 
0.001 
