612 



Fishery Bulletin 100(3) 



data for SL and estimated age at time of capture. To stabi- 

 lize the variance of length over the observed age interval, 

 length data were log-transformed and model parameters 

 were estimated from the log-transformed version of the 

 growth equation (Warlen, 1992). Because the conventional 

 3-parameter fit produced two cases where the estimates 

 of the model intercept (L^^^) were biologically unreason- 

 able, all data sets were rerun by using a 2-parameter (i.e. 

 A|Q, and a) fitting procedure and by setting the intercept 

 (third parameter) at 3.5 mm. This size at hatching was an 

 intermediate value between the estimates of Powell and 

 Phonlor { 1986) for Atlantic menhaden hatching at 16° and 

 24°C. The overall average growth rate from hatching to 

 a given age was the quotient of the predicted size at that 

 age minus the size at hatching (3.5 mm) divided by that 

 age. Differences in population growth curves between 

 sampling locations in each year were tested by using a 

 2-parameter Hotelling's T'^ test of the model parameters 

 (Bernard |1981| as modified by Hoenig and Hanumara-^). 



The birthdate (=spawning date) of each larva was back- 

 calculated by subtracting its estimated age from the date 

 of capture. Birthweek cohorts were defined as larvae 

 spawned in a given calendar week. Density values and 

 percentage age composition of each week's catch were 

 used to estimate the estuarine recruitment of birthweek 

 cohorts for each year and each location by using the meth- 

 ods of Warlen (1994). Birthdate distributions were used 

 to estimate the percentage of larval recruits to the New 

 Jersey estuary that could have originated in the SAB. The 

 menhaden spawning season in North Carolina and New 

 Jersey was estimated from the birthdate distributions of 

 larvae that survived to enter the estuaries. The percent- 

 age distribution of spawning by week was based on the 

 relative abundance of larvae collected throughout each 

 recruitment year 



We assumed that larval Atlantic menhaden caught each 

 week were newly recruited to the estuary and that they 

 were in transit past the sampling sites to upper portions 

 of the estuaries. This assumption is supported by the 

 arguments in Warlen (1994), Churchill et al. (1999), and 

 Forward et al. (1999) for the North Carolina sampling 

 site and the patterns observed for other shelf-spawned 

 estuarine dependent species such as summer flounder 

 iParalichthys dentatus) (Keefe and Able, 1993; Able and 

 Kaiser, 1994) and other species for New Jersey (Witting 

 etal., 1999). 



ResuKs 



Timing and abundance of larval recruitment 



The magnitude of larval Atlantic menhaden recruitment 

 appeared to differ substantially between the two loca- 

 tions (Fig. 2). The sum of the weekly mean larval densities 



' Hoenig, N. A, and R. G. Hanumara. 1983. Statistical consid- 

 erations in fitting seasonal growth models for fishes, 25 p. 

 ICES council meeting 1983/D. 



Sep 'Nov 'Jan' 'Mar' 'Hay' 

 Aug Oct Dec Feb Ape Jun 



Date 



Figure 2 



Weekly mean density (larvae/100 m^) (line graph) 

 and birthdate frequency distributions (bar graph) 

 as calculated from relative abundances of larval 

 Atlantic menhaden iBrevoortia tyrannus) in collec- 

 tions at Pivers Island, North Carolina, and Little 

 Sheepshead Creek, New Jersey, during 1989-90, 

 1990-91, and 1992-93. 



