SAVOY AND CRECCO: MORTALITY OF AMERICAN SHAD 



mean June river flow (Crecco and Savoy 1984, 

 1985b). Mean June flows (m^/s) were measured 

 within the major spawning areas (Leggett 1977) by 

 the United States Geological Survey (U.S. Geo- 

 logical Survey 1967-84). We estimated total egg 

 production (Eggs,) in Equations (3a) and (3b) as the 

 product of the mean fecundity of a female American 

 shad times that year's parent stock (PARf) of 

 female shad (Table 1). The average fecundity was 

 reduced from 269,000 ova (Leggett 1969) to 200,000 

 to reflect the average rates of egg retention and in- 

 complete fertilization (Watson 1970; Reed and Russo 

 1976^). Since Leggett (1969) showed that the aver- 

 age fecundity of American shad varied by less than 

 10% from 1966 through 1973, we were justified in 

 using an average fecundity for all years. 



The mean density-independent mortality rate 

 (Z/pre) present during the prejuvenile stage was the 

 y-axis intercept (a) in Equation (3b) plus the slope 

 (c) times the overall geometric mean June flow 



'Reed, R. J., and A. Russo. 1976. American shad research 

 Connecticut River, Massachusetts, 1976. 1. Fecundity, egg reten- 

 tion, sex ratio, and age class composition. Unpubl. manuscr., 16 

 p. Massachusetts Cooperative Fisheries Research Unit, Univer- 

 sity of Massachusetts, Amherst, MA 01003. 



(GJFLOW) from 1967 through 1987: 

 Zipre = a + c(GJFLOW), 



(4) 



The mean density-dependent mortality rate (Zq^^^) 

 was the slope (6) of Equations (3a) and (3b) times 

 the geometric mean egg production (GEgg) from 

 1967 through 1987. A positive and statistically sig- 

 nificant slope (6) of prejuvenile mortality on egg 

 abundance would support the Gushing hypothesis 

 that density-dependent mortality takes place before 

 the juvenile stage. If the b estimates of Equations 

 (1), (3a), and (3b) were all positive and significant, 

 our results would support Gulland's (1965) hypoth- 

 esis that density-dependent mortality occurs over 

 the entire prerecruitment phase. 



Results 



Although total juvenile abundance varied 14-fold 

 from 1967 through 1982 (Table 2), the total post- 

 juvenile mortality rates {ZA() exhibited relatively 

 low variability (95% G.I.: 4.70-5.00) about the esti- 

 mated mean {EZA = 4.85) (Fig. 2). The slope of the 

 linear regression between total postjuvenile mortal- 



6.6 



5.4 



5.2 



5.0 



< 



N 4.8 



4.6 



4.4 



4.2 



4.0 



50 



1 1 1 



100 160 200 



Juvenie Abundance (thousands) 



260 



300 



Figure 2.— Relationship between American shad postjuvenile mortality (ZA,) and relative abundance of juveniles 



(J() from 1967 through 1982. 



471 



