previous years, daytime samples after April were 

 excluded, except for estimating larval entrainment 

 with station EN data. During this period, daytime 

 samples were not collected in the river because 

 these samples underestimated abundance due to 

 the fact that older larvae apparently remained near 

 the bottom during the day and were not suscep- 

 tible to the bongo sampler (NUSCO 1987). 



Typically, the temporal distribution of larval 

 abundance was skewed, with a rapid increase to 

 a maximum followed by a slower decline. This 

 skewed density distribution results in a sigmoid- 

 shaped cumulative distribution where the time of 

 peak abundance is the time at which the inflection 

 point of the sigmoid occurs. The Gompertz 

 growth function (Draper and Smith 1981) was 

 chosen to describe the cumulative distribution of 

 the abundance data because the inflection point 

 of this function is not constrained to the central 

 point of the sigmoid curve. The form of the 

 Gompertz function used was: 



(abundance curve). This density function has the 

 form: 



C, = a(exp[ - Pe "']) 



(4) 



where Q = cumulative density at time t 



a = total or asymptotic cumulative den- 

 sity 



P = location parameter 



K = shape parameter 



t = time in days since February 15 



The origin of the time scale for our data was 

 set to the 15th of February, which is when winter 

 flounder larvae generally fu-st appear in the Niantic 

 River. Least-squares estimates and asymptotic 

 95% confidence intervals for these parameters 

 were obtained by fitting the above equation to 

 the cumulative abundance data (based on the 

 weekly geometric means) using nonlinear regres- 

 sion methods (SAS Institute 1985). 



The derivative of the Gompertz function with 

 respect to time yields a "density" function which 

 directly describes the larval abundance over time 



4 = aPK(exp(-K?{-|3e "'}]) 



where d, = density at time t 



(5) 



and all the other parameters are the same as in 

 Equation 4, except for a, which was rescaled by 

 a factor of 7 because the cumulative densities 

 were based on weekly geometric means and thus 

 accounted for a 7-day period. Time of peak 

 abundance was estimated as the date tj corre- 

 sponding to the inflection point of the function 

 defmed by its parameters P and k as: 



(log«P) 



(6) 



The a parameter was used as an index to compare 

 annual abundances. The k parameter was used to 

 compare the steepness of the abundance curve 

 (Equation 5), where k increases as the peak of 

 the curve increases. 



Winter flounder larvae were reared in the lab- 

 oratory during 1986 to determine growth rates at 

 various temperature regimes. Eggs were stripped 

 from a female and fertilized with milt from two 

 males. larvae that hatched within 24 hours of 

 each other were placed in 39- L aquaria held in a 

 water bath. The water temperature in each regime 

 was gradually increased during the holding period 

 to mimic the seasonal temperature increase during 

 larval winter flounder development. Photoperiod 

 was similar to natural conditions. Larvae were 

 fed rotifers {Brachionus plicatilus) and brine shrimp 

 nauplii {Artemia salina) ad libitum. Known-age 

 larvae were routinely sacrificed and measured to 

 the nearest 0.1 -mm standard length to obtain in- 

 formation on growth rate. For comparisons with 

 other laboratory growth studies on larval winter 

 flounder, length was converted to weight (|ig) by 

 the length- weight relationship of Laurence (1979): 



weight = 0.045( length) 



(7) 



156 



