Ovaries in development stage 4 from 1973 and 

 1974 cruises were exsected, weighed (±0.01 g), and 

 preserved in a modified Gilson's fluid (Simpson 

 1951) for subsequent fecundity estimation. 



Egg counts for fecundity estimation were made 

 by a method similar to Bagenal's (1957), but mod- 

 ified as follows. After 4-6 wk preservation the 

 ovaries were washed over a 1.0 mm mesh screen 

 which permitted all the eggs to pass through it but 

 retained the ovarian tissue. After repeated wash- 

 ings and decanting the eggs were diluted in water 

 to a known volume, randomly stirred, and three 10 

 ml aliquots, each containing from 200 to 400 eggs, 

 were extracted and the eggs counted using a mi- 

 croscope. To test the accuracy of the sampling 

 method two aliquots were extracted from an egg 

 batch, counted, and replaced until 20 aliquots 

 were counted. The mean and standard deviation 

 (SD) of all counts was 255 and 38.3, and the 

 coefficient of variation (CV) was 15.0%. For the 

 means of three samples the SD was 20.4 and the 

 CV was 8.7%. The mean egg number of the three 

 aliquots times the total dilution volume divided by 

 the aliquot volume was used to estimate fecundity. 



Estimates of length at maturity (length at 

 which 50% are mature = L^^) were calculated 

 using probit analysis (Finney 1971). Samples used 

 in this analysis were collected from September to 

 November (Table 1; cruises 1-4, 6) which 

 minimizes the affects of seasonal growth on length 

 at maturity. The proportion of mature fish for each 

 centimeter group was calculated for each year by 



sex. Table 2 shows the proportion mature by cen- 

 timeter group for the interval between and 100% 

 mature. The proportions were transformed to 

 probits (Fisher and Yates 1964) and the iterative 

 method was used to calculate the weighted linear 

 regression equation, Y = a + bX,hy least squares 

 for the logarithm (base 10) of length iX) and the 

 probit (7). Chi-square tests indicated no sig- 

 nificant (x^o.oi ) heterogeneous deviations from the 

 regression lines; therefore, the regression 

 coefficients were used to determine L^,.. The var- 

 iance of L,o was estimated as: 



50' 



V(L5o) = — 



1_ 

 62 



l,nw 



( Xnwx V 



„ / l,nwx\ ' 

 Znwix-— J 



\ Lnw / 



where n is the number of observations at each 

 centimeter group, w is the weighting coefficient 

 (Z2/PQ) (Finney 1971, table II) for probit ( 7), andx 

 is the logarithm (base 10) of the length. 



In order to test for differences in Lg^ between 

 sexes and between years, 2-values (Natrella 1966) 

 were calculated using the equation: 



'501 ^502 



Vv(L50i) + V(L50i) 



Table 2. — Percentage mature used in probit analysis by total length group for male (M) and 

 female (F) Atlantic croaker collected in 1973-76. TV = number of specimens, L^^ = length at 507f 



191 



