FISHERY BULLETIN: VOL. 86, NO, 3 



Table 2.— Mean juvenile indices, scaled juvenile abundance from 

 Equation (2) and total egg production for American shad in the Con- 

 necticut River from 1967 through 1987. SE = standard errors 

 about the estimates. 



can shad has a similar 4-6 yr postjuvenile phase as 

 sockeye salmon, we tested Peterman's hypothesis 

 for American shad with key factor analysis (Bellows 

 1981; Rosenberg and Doyle 1986). We fixed the 

 postjuvenile period for American shad in the Con- 

 necticut River between 101 days and 5 years, corre- 

 sponding to the average age (101 days) at which 

 juvenile shad leave the river (Crecco and Savoy 

 1985b) and the average age (5 years) when they 

 return to the river as mature adults. The total post- 

 juvenile (age 101 days to 5 years) mortality rates 

 (ZA() for the 1967-82 year classes were related to 

 the scaled juvenile indices (J<) for those year 

 classes (Table 2) in a linear model: 



ZA, = 



a + 



b{Jt), 



(1) 



where ZAf = -logg(i2(/J<). If significant density- 

 dependent mortality is present during the post- 

 juvenile stage, the slope (6) of Equation (1) would 

 be positive and differ significantly (P < 0.05) from 

 zero. 



The juvenile indices were scaled to thousands of 

 fish (Jf) so that the y-axis intercept in Equation (1) 

 directly estimates the mean density-independent 

 mortality (Zj^g^i), and the slope (b) times the geo- 

 metric mean juvenile abundance (GM) from 1967 

 through 1987 is the mean density-dependent mor- 

 tality rate (Z^post). Total juvenile abundance (J() 



was estimated by multiplying the juvenile indices 

 (IND,) by a scalar (SC): 



SC = [GR exp{-EZA)  GM] = 3,518.3, (2) 



where GR is the geometric mean total adult 

 recruitment for 1967 through 1982; GM is the geo- 

 metric mean juvenile index from 1967 through 1987; 

 EZA (4.85) is the mean total instantaneous mortality 

 among postjuveniles from 1967 through 1982. We 

 estimated EZA as the sum of mortality during the 

 late juvenile period (age 101-365 days), and the 

 subadult stage (age 1-5 years). The mean total in- 

 stantaneous mortality rate during the late juvenile 

 stage was estimated as 2.65 (0.01 • 265 days), using 

 a mean daily mortality rate of 0.01 (SE = 0.002) 

 extrapolated from the 1979-84 larval and juvenile 

 survivorship curves (Crecco and Savoy 1985b). The 

 mean total instantaneous mortality rate of subadult 

 shad was 2.2, based on an annual instantaneous 

 natural mortality rate of 0.45 from the method of 

 Pauly (1980) plus 0.10 to reflect oceanic fishing mor- 

 tality (2.2 = 4(0.45 -I- 0.10)). We estimated the an- 

 nual natural mortality rate (0.45) by substituting the 

 ;^(0.25, SE = 0.03) and L (55 cm FL, SE = 3 cm) 

 parameters of the von Bertalanffy equation for male 

 and female shad combined and preferred ocean tem- 

 perature (14 °C) of American shad (Leggett and 

 Whitney 1972) into Pauly' s multiple regression 

 model. The oceanic fishing mortality estimate (0.10) 

 was based on tagging studies in Delaware Bay 

 (White et al. 1969; Zarbock 1969) and off the New 

 York-New Jersey coast (Nichols 1958). 



To determine if density-dependent mortality takes 

 place during the egg and larval stages as Cushing 

 (1980) hypothesized, we related total prejuvenile 

 mortality rates {ZEJt) to annual egg production 

 (EggSj ), and to both egg production (Table 2) and 

 mean June river flow (JFLOW) from 1967 to 1987 

 (Table 1) in linear regression models: 



ZEJt = a -H 6 (Eggs,) 



(3a) 



and 



ZEJt = a + ^(Eggs,) -H c(JFLOW,), (3b) 



where ZEJt = -log^(J( /Eggs,). Mean June river 

 flows (m^/s) were included in Equation (3b) because 

 previous studies (Crecco and Savoy 1985b, 1987a) 

 have shown that high June flows reduced prejuvenile 

 survival rates, leading to a significant inverse cor- 

 relation (r = - 0.74, P < 0.01) between the juvenile 

 indices of abundance from 1967 through 1980 and 



470 



