328 



Fishery Bulletin 93(2), 1995 



temperature, supporting the conclusion that late- 

 hatched cohorts, which grew faster than average, 

 were the principal contributors to striped bass re- 

 cruitments in 1987-89. Although cohort-specific 

 mortality rates were not significantly related to co- 

 hort growth rates or to any measured variable, the 

 annual median GIZ ratios (i.e. for combined cohorts) 

 were positively correlated with the juvenile recruit- 

 ment index (Rutherford et al. 4 ). This result indicated 

 that, while the relationships between growth and 

 survival of individual cohorts are complex and diffi- 

 cult to demonstrate, the effect of reduced stage du- 

 ration on larval production and on potential recruit- 

 ment did occur and could be discerned when cohorts 

 were aggregated. Chesney's (1993) simulation model 

 of Potomac River striped bass larval dynamics in 



1987 predicted good growth and survival of cohorts 

 hatched late in the season when temperatures and 

 prey densities were high. Research on other species 

 of temperate estuarine and freshwater fishes also has 

 demonstrated that survival and recruitment are 

 highest for fast-growing cohorts (Rice et al., 1987a; 

 Crecco and Savoy, 1985; Jennings et al., 1991), sup- 

 porting Cushing's (1973) "single-process" concept, in 

 which fast larval-stage growth enhances recruitment 

 success through shortened stage durations. 



Differences in mean temperatures between 1987, 

 1988, and 1989 in the Potomac River may have led 

 to significant differences in production of 8.0-mm- 

 SL larvae, owing solely to effects on stage duration. 

 Egg productions in the Potomac River were approxi- 

 mately 10 billion in 1987 and 1989, and 6.7 billion in 



1988 (Houde and Rutherford, 1992). If larval mor- 

 tality rate had been equal in all years (e.g. Z=0.25), 

 then the mean effect on larval growth and stage du- 

 ration of the observed 4.0°C higher mean tempera- 

 ture in 1987, compared with 1988 or 1989, could have 

 accounted for a 3.7-fold greater production of 8.0-mm- 

 SL larvae in 1987 than in 1988, and a 2.6-fold greater 

 production in 1987 than in 1989. Our estimated pro- 

 duction of 8.0-mm-SL larvae in 1987 was 7.8 and 1.7 

 times higher, while the juvenile recruitment index 

 (Schaefer et al. 1 ) was 16.0 and 2.9 times higher in 

 1987 than in 1988 or 1989, respectively. Although 

 our results point to temperature as a critical factor 

 and Chesney's (1993) simulation supports this view, 

 other simulation models of major factors thought to 

 influence year-class strengths of Potomac River 

 striped bass suggest that temperature may be less 

 important than maternal size or zooplankton prey 

 abundance (Cowan et al., 1993), or, under special cir- 

 cumstances, than contaminant levels (Rose et al. , 1993 ). 



Previous studies of striped bass larval growth have 

 suggested that growth rates are correlated with tem- 

 peratures in the spawning areas (Dey, 1981; Uphoff, 



1989; Low 3 ). However, most growth estimates were 

 based upon modal analysis of larval lengths and, 

 consequently, may have been inaccurate. Without 

 benefit of otolith-increment analysis, among-cohort 

 variability in growth rates was unevaluated. In the 

 Choptank River of the Chesapeake Bay, Uphoff 



( 1989) estimated annual mean growth rates of striped 

 bass larvae to be 0.37-0.56 mmd" 1 from 1981 to 1986, 

 on the basis of length-frequency distributions. These 

 rates were higher than our mean estimates in the 

 Potomac and Upper Bay and were higher than 

 growth rates of all except 4 of the 46 cohorts that we 

 analyzed, although temperature ranges and prey 

 densities are similar in these Chesapeake Bay tribu- 

 taries. Larval cohort growth rates that were 

 backcalculated from otolith-aged, juvenile striped 

 bass from South Carolina also were higher (0.35 to 

 0.68 mmd -1 ; Secor, 1990) than most of our estimates. 

 However, larvae from South Carolina experience 

 mean temperatures during spawning and larval de- 

 velopment that are 3-5°C higher than temperatures 

 encountered by Chesapeake Bay striped bass (Secor, 

 1990). Growth rates of Hudson River larvae in 1973- 

 76, estimated from the weekly seasonal increases in 

 larval mean lengths from a designated, arbitrary 

 hatch date until 15 July, were 0.10 to 0.20 mmd -1 

 (Dey, 1981). Those rates were lower than our mean 

 rates and generally lower than our individual cohort 

 growth rates, even though mean temperatures en- 

 countered by larvae in the Hudson and Chesapeake 

 estuaries are similar. Mean annual growth rates of 

 larvae in the Sacramento-San Joaquin River system, 

 estimated for the 1968-86 period from modes in 

 length-frequency distributions, ranged from 0.29 to 

 0.46 mm-d -1 (Low 3 ). These rates are higher, on aver- 

 age, than our mean rates, although many cohorts of 

 Chesapeake Bay larvae grew at rates in this range. 



Individual growth rate variability 



Our analysis of individual larval growth histories to 

 detect evidence of growth compensation and size-se- 

 lective mortality may have been compromised some- 

 what by the back-calculation method. Campana's 



( 1990) biological intercept method will provide accu- 

 rate estimates of mean back-calculated growth rate 

 even in the presence of a "growth effect" but will tend 

 to linearize individual growth rates and mask growth 

 inflections (Campana, 1990; Secor and Dean, 1992). 

 Campana (1990) demonstrated through simulation 

 analysis that time-varying changes in the body 

 length-otolith radius relationship caused by increas- 

 ing somatic growth rate could result in underesti- 

 mated lengths at earlier ages and overestimated 

 lengths of older larvae, giving the appearance of com- 



