276 
Fishery Bulletin 95(2), 1997 
mate in the wild, then cultured individuals should 
have occupied several of the other tributaries 
sampled. However, few tagged fish were collected 
farther than 1 km away from either release site, and 
the difference in proportions of cultured fish retrieved 
outside of release sites, compared with proportions 
collected at release sites, did not increase over time. 
These data (Table 4) provide circumstantial evidence 
that, following 2 weeks of acclimation in the wild, 
cultured striped mullet then tended to stay at or near 
the stream they occupied for the duration of the study. 
Strong site fidelity (during the juvenile stage) has 
also been documented in marine nursery habitats 
following hatchery releases of lobster (Bannister and 
Howard, 1991; Latrouite and Lorec, 1991) and cod 
(e.g. Nordeide et al., 1994). 
Impact of release season 
Fish size-at-release is clearly an important media- 
tor of the effect of hatchery releases on stock abun- 
dance (Hager and Noble, 1976; Bilton et al., 1982; 
Tsukamoto et al., 1989; Liu, 1990; Svasand and 
Kristiansen, 1990; Ray et al., 1994; Leber, 1995; Wahl 
et al., 1995; Willis et al., 1995). At all of the release 
sites tested in Hawaii, size-at-release has been an 
important factor affecting recapture probability of 
cultured striped mullet (Leber, 1995; Leber et al. 1 ). 
In previous studies with striped mullet, where re- 
leases were conducted in summer and fall, recapture 
rate was directly related to size of fish at the time of 
release. 
As expected (Leber, 1995), in this study recapture 
rates after summer releases of small fish (individu- 
als <60 mm long) approached zero and were an or- 
der of magnitude less than recapture rates of the 
larger fish released. Thus, when releases are made 
in summer in Kaneohe Bay, small (<60 mm) cultured 
striped mullet do not significantly affect juvenile re- 
cruitment in Kaneohe Bay. It is important to note 
that the fish in the different size intervals released 
were produced from multiple rearings and that the 
smallest fish released in summer were not merely 
the slowest growing individuals; rather, size-at-re- 
lease was related primarily to age. 
A new finding revealed by this study was that the 
seasonal timing of striped mullet releases can sub- 
stantially alter size-at-release effect on recapture 
rate. Compared with recapture rates after summer 
releases, recovery of the smallest individuals released 
was significantly greater when releases were timed 
to coincide with peak recruitment of small wild indi- 
viduals (in the spring). This was the first evidence 
that releases of relatively small (45 to 60 mm TL) 
individuals could make any lasting contribution to 
striped mullet abundances in nursery habitats on 
Oahu. Subsequently, Leber and Arce (1996) showed 
that some of the small fish released in spring did 
survive to adult size and contribute to the commer- 
cial fishery catch in Kaneohe Bay. The latter study 
also revealed that the smallest individuals in sum- 
mer releases from this study apparently suffered to- 
tal mortality. Because of the obvious economic im- 
portance of our findings, we replicated part of this 
study in a follow up study, with spring releases of 
the same size groups studied here; the results were 
identical — small fish (<60 mm) did contribute to ju- 
venile recruitment when releases were made in 
spring (Leber et al., 1996). 
It is not clear how one is to interpret the lack of a 
strong correlation between size-at-release and recap- 
ture rates following the spring releases. On the ba- 
sis of cast-net samples alone, we cannot rule out the 
possibility that a direct relation existed between size- 
at-release and survival after spring releases. Cast 
nets are biased in favor of collecting small individu- 
als (Leber et al. 1 ). Thus, a weak size-at-release ef- 
fect following spring releases could be masked by 
sampling bias. Indeed, for fish from the spring re- 
leases, data from subsequent samples of adult cul- 
tured fish caught in the Kaneohe Bay mullet fishery 
revealed a (nonsignificant) trend towards a direct 
size-at-release effect (Leber and Arce, 1996). As in 
this study of juveniles, the data for adults revealed a 
highly significant effect of size-at-release on recov- 
ery rates when releases were made in summer. 
On the basis of this study and on subsequent data 
on adult recruitment to the commercial fishery (Leber 
and Arce, 1996), striped mullet < 60 mm should not 
be released during summer in Kaneohe Bay. How- 
ever, early (spring) releases of 45-60 mm striped 
mullet can make a contribution both to juvenile re- 
cruitment (this study) and to adult recruitment 
(Leber and Arce, 1996). Maximum recovery from 
summer releases will occur when individuals are >85 
mm at the time of release. To determine optimal size- 
at-release, an economic analysis is needed to evalu- 
ate benefits and costs of releasing larger individuals. 
Bilton et al. ( 1982) showed an interaction between 
release timing and size of juvenile coho salmon, 
Oncorhynchus kisutch, released in British Columbia. 
In that study, returns would be maximized from early 
release of large juveniles. The effect of the seasonal 
timing of releases on size-dependent recapture rates 
may not be universal (e.g. Willis et al., 1995); never- 
theless, release season could be a key factor in suc- 
cessful enhancement of many marine species. 
What processes could account for the seasonal 
change in size-at-release dependent recapture rates? 
Size structures of cultured and wild fish suggest that 
