466 



Fishery Bulletin 94(3). 1996 



was also considerable in streams 3 km away from 

 the release site. Decrease in proportions of cultured 

 fish in samples after 8 months coincided with the 

 seasonal period (March) when yearlings begin to 

 move out of nursery habitats into deeper water and 

 when new recruits begin to arrive (Major, 1978; 

 Leber, 1995; Oceanic Institute 4 ). 



The striped mullet collected in this study that were 

 released in 1991 (Leber et al. 3 ) represented a large 

 portion (28%) of the total striped mullet caught in 

 net samples at Kaneohe Stream. These data docu- 

 ment that a portion of the cultured striped mullet 

 released in Kaneohe Bay can be found in their nurs- 

 ery habitats up to two years after release. 



It is evident from this study and from follow-up 

 creel interviews in which fishery landings were sur- 

 veyed that there is significant potential to increase 

 the striped mullet population in Kaneohe Bay with 

 relatively small-scale hatchery releases. Cultured 

 mullet released in spring 1992 were first detected in 

 the commercial fishery in Kaneohe Bay in October 

 1993, when a 367-mm-TL fish was recovered during 

 contact interviews with commercial fishermen (Leber 

 and Arce, in press). By the seasonal closure of the 

 fishery in December 1994, 119 of the mullet released 

 in 1992 had been recovered through contact inter- 

 views with commercial fishermen. According to tag 

 data from interviewing fishermen in Kaneohe Bay, 

 fish released in this study accounted for 9% of the 

 commercial catch during fall 1994, at which time the 

 proportion of cultured fish to wild fish was increas- 

 ing logarithmically. Ninety-four of the cultured fish 

 caught were checked for maturity; 44 males were ripe 

 with milt and 2 females were gravid with mature 

 eggs (Leber and Arce, in press). 



Survival and hatchery impact on abundance The 



best gauge of the immediate effect of the 1992 hatch- 

 ery releases on mullet population size is actual abun- 

 dance of released fish in the wild. But actual num- 

 ber of survivors after 11 months is unknown. The 

 >50% hatchery contribution to abundance at the re- 

 lease site is impressive, but from an economic view- 

 point, data on actual increases in yields and popula- 

 tion size are needed to compare the benefits of enhance- 

 ment with costs. An estimate of survival would be a 

 better gauge of stock -enhancement impact than would 

 the proportion of hatchery fish in the population. 



However, it is logistically difficult to quantify ac- 

 tual survival of released cultured fishes in open 

 coastal environments. There is substantial literature 

 on evaluating animal survival, much of which is 

 based on change in relative abundance (catch per unit 

 of effort [CPUE]) over time in mark-release-recap- 

 ture experiments. The analysis theory for release- 



recapture experiments dates back to Flicker's ( 1945, 

 1948) relative recovery-rate method. A more general 

 theory based on maximum likelihood was developed 

 independently by Seber (1970) and Youngs and 

 Robson (1975). Brownie et al. (1985) and Burnham 

 et al. (1987) have provided a detailed discussion of 

 maximum-likelihood methods. These methods for 

 estimating survival are based on change in abun- 

 dance in samples over time and are confounded in 

 open environments by dispersal into and out of the 

 population (Connor et al., 1983; MacCall, 1990; 

 Nichols and Pollock, 1990; Frank, 1992) and by gear 

 bias, as capture probabilities change with changes 

 in fish size and habitat selection (e.g. Kjelson and 

 Colby, 1976; MacLennan, 1992; Thompson, 1994; 

 Acosta and Appeldoorn, 1995; Leber et al., in press). 

 Without reliable estimates of immigration and emi- 

 gration, one can use change in capture rate over time 

 to measure loss from a population (i.e. the sum of mor- 

 tality + emigration ) but not to estimate mortality alone. 



In this study, decrease in abundance over time in 

 cast-net samples is a poor indicator of actual sur- 

 vival. Striped mullet move out of their nursery habi- 

 tats as they approach maturity (Blaber, 1987), and 

 juveniles move from shallow water to deeper water 

 during their first year (Major, 1978; Leber et al. 3 ). 

 Overall decline in mullet abundance in samples over 

 the 11-month study period was due to a combination 

 of mortality, dispersal, and gear bias as mullet grew 

 and moved into deeper water where cast nets are 

 not effective sampling devices. 



Postrelease mortality prior to initial sampling is 

 also a source of error in estimating survival follow- 

 ing hatchery releases. A key question is whether 

 mortality of cultured fish is intense during the first 

 day or so after a release (e.g. what percent of released 

 individuals are consumed by predators?). Initial 

 mortality could be an important factor in enhance- 

 ment dynamics and should be accounted for in sur- 

 vival estimates, especially in open environments 

 where mortality can be confused with emigration. 

 But initial postrelease mortality has received little 

 attention in "release-recapture" literature. There is 

 evidence, however, that until cultured fish have been 

 conditioned by exposure to predators in the wild, in- 

 adequate predator avoidance behavior can result in 

 increased mortality (Parker, 1968; Healey, 1982; Olla 

 and Davis, 1988; Olla and Davis, 1989; Tsukamoto, 

 1993; Olla et al., 1994). Initial mortality following hatch- 

 ery releases of Pacific salmon can be severe and is most 

 intense during the 48-h period following releases. 6 Ini- 



6 Blankenship, H. L. 1995. Washington Department of Fish 

 and Wildlife, 600 Capitol Way North, Mail Stop 43149, Olym- 

 pia, WA. Personal commun. 



