Leber et al.: Influence of release season on size-dependent survival of Mugi! cephalus 
277 
schooling behavior of striped mullet may partly con- 
trol the release-season effect. Schools of juvenile 
striped mullet are usually aggregated according to 
size (Leber, 1995). Because of the difference in size 
structures between wild and cultured fish in the 
spring, schools of larger striped mullet, after spring 
releases, contained mostly cultured fish and few wild 
fish. We hypothesize that at the time of spring re- 
leases, the large individuals were more susceptible 
than smaller ones to mortality from predation. We 
reason that, because the smallest fish released in 
spring had merged with relatively large numbers of 
small wild striped mullet, the smallest fish should 
have been afforded greater refuge from predators 
than that provided the large fish in our spring re- 
leases, because there were more small wild fish than 
large ones (i.e. refuge effect from schooling behav- 
ior; e.g. Parrish, 1989, 1992; deVries, 1990; Ranta et 
al., 1994). 
This pattern was reversed following summer re- 
leases, when size structures of the larger cultured 
and wild individuals were equivalent. By summer, 
most wild juveniles had grown larger than the size 
range of the smallest cultured individuals released. 
Thus, few small wild juveniles were available to form 
schools with small cultured fish and thus the advan- 
tage of refuge that such schooling behavior would 
provide to small cultured fish was reduced. 
The results of this study are consistent with the 
hypothesis that size-selective predation is a primary 
mechanism controlling recapture rates following 
hatchery releases in Kaneohe Bay (Leber, 1995). Al- 
though, after summer releases, large wild fish were 
not as abundant as small wild fish in the spring (thus 
reducing the advantage gained by cultured fish from 
schooling with large wild fish), larger cultured fish 
would have the added advantage of size in escape 
from predators. Whatever the cause(s) of size-at-re- 
lease impact on recovery rates, it was clear from this 
study that release season can influence the underly- 
ing mechanism. 
Conclusions 
The importance of conducting test releases to evalu- 
ate release strategies prior to conducting full-scale 
hatchery releases cannot be overemphasized. This 
study documented that release season can have a 
significant effect upon recovery of cultured striped 
mullet in the wild by affecting size-at-release depen- 
dent recapture rates. To optimize the impact of full- 
scale releases, marine stock-enhancement programs 
should perform test releases to evaluate interaction 
of release season with size-at-release effects. 
We hypothesize that survival of cultured fish will 
be greater when releases are timed so that size-at- 
release coincides with modes in population size struc- 
tures of wild stocks. A corollary to this is that the 
fewer cultured fish there are in a particular size in- 
terval at the time of release, the lower survival will 
be for wild fish in that interval. 
These results need to be related to the hatchery 
costs of rearing fingerlings to various sizes and also 
to the increased production allowed by releasing 
small fingerlings in the spring, because spring re- 
leases would make nursery tanks or ponds available 
to grow more fish for summer releases. 
Although the mechanism underlying the direct 
relationship between survival and size-at-release is 
not well understood, it is clear that in Hawaii, fish 
size-at-release can determine release success follow- 
ing summer releases of striped mullet. Based on this 
study, critical release size (CSAR, the size-at-release 
below which probability of survival approaches zero; 
Leber, 1995) for enhancing striped mullet in Kaneohe 
Bay appears to be lower when releases are made in 
spring (CSAR <45 mm) than when releases are made 
in summer (CSAR <60 mm). 
Acknowledgments 
We thank Johann Bell, Laurie Peterson, and the 
anonymous reviewers for their comments on and 
improvements to the manuscript. Thanks to Dave 
Sterritt for help with graphics, data management, 
and field work. We thank Ryan Takushi and the Oce- 
anic Institute finfish program for their dedicated 
assistance with fish production for this study; Anton 
Morano and Dan Thompson for help with tagging, 
field collections, and tag decoding; and Joyce Gay for 
her assistance in preparing the tables. This paper 
was funded by a grant from the National Oceanic 
and Atmospheric Administration (NOAA). 
Literature cited 
Bannister, R. C. A., and A. E. Howard. 
1991. A large-scale experiment to enhance a stock of lob- 
ster ( Homarus gammarus L.) on the English east 
coast. ICES Mar. Sci. Symp. 192:99-107 
Bilton, H. T., D. F. Alderdice, and J. T. Schmute. 
1982. Influence of time and size at release of juvenile coho 
salmon ( Oncorhynchus kisutch) on returns at maturity. 
Can. J. Fish. Aquat. Sci. 39:426-447. 
Blaber, S. J. M. 
1987. F actors affecting recruitment and survival of mugilids 
in estuaries and coastal waters of Southeastern Africa. 
Am. Fish. Soc. Symp. 1:507-518. 
