stages combined (Zq = 1.54) between 1979 and 

 1987 was not significantly different (Table 9) from 

 the total density-dependent mortality rate (Zq^j.^ = 

 1.18) estimated during the prejuvenile period from 

 1967 to 1982 (Table 3), suggesting that nearly all 

 of the density-dependent mortality for prejuvenile 

 American shad occurs before the midlarval stage. 



DISCUSSION 



Although American shad eggs and early larvae 

 experience high mortality (15-40%/day) in the Con- 

 necticut River, the average density-dependent mor- 

 tality rate (Z^p^g = 1.18) during those stages com- 

 prised a relatively small percentage (18%) of the 

 total prerecruitment mortality. This suggests that 

 most of the annual variability in American shad 

 recruitment is explained by density-independent fac- 

 tors, which is consistent with the significant positive 

 correlation between mean June flow and egg and 

 early larval mortality rates, and with the significant 

 inverse correlation (r = - 0.74, P < 0.001) between 

 mean June river flow and adult recruitment from 

 the 1967 through 1982 year classes (Crecco and 

 Savoy 1984, 1987b). Whereas these data illustrate 

 that density-dependent mortality plays a minor role 

 in governing annual variability in American shad 

 recruitment, this does not mean that compensatory 

 processes are trivial. As pointed out by Ricker (1954) 

 and others (Gushing 1974; Garrod and Horwood 

 1984), only a small amount of density-dependent 

 mortality is required to stabilize the growth poten- 

 tial of fish populations because density-dependent 

 effects become progressively more effective at 

 higher egg and larval densities (Shepherd and 

 Gushing 1980; Murray 1982). 



Although our results on American shad support 

 the Gulland (1965) hypothesis that density-depend- 

 ent mortality persists throughout the prerecruit- 

 ment period, most (82%) of the density-dependent 

 mortality occurs during the egg and early larval 

 stages. Our average estimate of density-dependent 

 mortality {Zq = 1.54) during the early larval 

 periods does not differ significantly from the mean 

 Zd value (1.34) estimated among the 5-d larval 

 cohorts in 1983 and 1984 (Grecco and Savoy 1987a), 

 or from the mean density-dependent mortality rate 

 Zj) (1.21) estimated by stock-recruitment methods 

 for the entire prerecruitment period (Lorda and 

 Grecco 1987). That year-class strength is established 

 early in the ontogeny of American shad is supported 

 further by the significant positive correlation (r = 

 0.84, P < 0.009) between the relative abundance of 

 early larvae and all subsequent stages from 1979 



FISHERY BULLETIN: VOL. 86, NO. 3 



through 1987 and adult recruitment from those year 

 classes (Table 8). 



The main causes of density-dependent mortality 

 are thought to be predation, competition, and canni- 

 balism (Ricker 1954). Since adult American shad are 

 not thought to feed in freshwater (Walburg and 

 Nichols 1967), we can probably eliminate cannibal- 

 ism as a mechanism for significant density-depend- 

 ent mortality during the egg and early larval stages. 

 Therefore, density-dependent mortality among early 

 American shad larvae is most likely caused by intra- 

 specific competition for food or space and predation. 

 The exact underlying density-dependent mortality 

 mechanisms are difficult to quantify because the out- 

 come of competition and predation may depend on 

 June flow effects shown here (Table 3) and else- 

 where (Grecco and Savoy 1987a, b) to be the prin- 

 cipal density-independent factor. High June river 

 flows have been shown to reduce June river tem- 

 peratures (Grecco and Savoy 1984) and the growth 

 rates of shad eggs (Watson 1968) and early larvae 

 (Grecco and Savoy 1985b). Because slower growing 

 larvae may be susceptible to predation for a longer 

 period of time (Hunter 1976), periods of high flow 

 may indirectly enhance egg and larval predation. 

 Additionally, since high flows reduce the spatial 

 patchiness (Grecco and Savoy 1987a), abundance 

 and availability of river zooplankton (Whitton 1980; 

 Threlkeld 1986), high flows may result in increased 

 levels of competition among American shad larvae 

 for available prey, especially if shad larvae are 

 capable of depleting local aggregations of edible 

 zooplankton. Periods of high runoff that coincide 

 with peak larval production, such as in June 1982 

 and 1984, may advect larvae and their zooplankton 

 prey from eddies and backwaters where they are 

 normally found (Gave 1978) to areas of high pred- 

 ator abundance. In light of the many ways in which 

 June flows potentially mediate larval mortality, it 

 is unlikely that a single compensatory mechanism 

 is responsible for the relatively high density-depend- 

 ent mortality rate (Z^pre) among early larvae. 



Since egg and prolarval shad have endogenous 

 food reserves, density-dependent mortality of these 

 stages is likely due to predation and competition for 

 sites among spawning adults. As spawning stocks 

 reach high densities, such as in 1982-84, crowding 

 of adult fish on the spawning grounds may result 

 in a reduction in the number of eggs released ((jood- 

 year 1980). In addition, since the amount of spawn- 

 ing habitat in the Connecticut River can be con- 

 sidered fixed from year to year, larger spawning 

 stocks are more apt to deposit an increasing per- 

 centage of eggs in unfavorable areas. Layzer (1974) 



478 



