SHEPHERD AND GRIMES: GROWTH VARIATIONS IN WEAKFISH 



statistical comparisons of growth curves; therefore, 

 we compared von Bertalanffy curves following Ber- 

 nard (1981). This method incorporates a Hotelling 

 T 2 test to compare matrices of parameter estimates, 

 variances, and correlation coefficients. 



RESULTS 



We made age determinations and annulus 

 measurements on scales from 1,240 weakfish, 647 

 males and 593 females, ranging in size from 50 to 910 

 mm TL. The validity of growth rings on weakfish 

 scales has been previously established by Taylor 

 (1916) and Perlmutter et al. (1956). Our data show 

 that annulus formation occurred from April to June, 

 the point at which mean marginal increments 

 approached minimum values (Fig. 2), which cor- 

 roborates the earlier findings of annuli validity. 



The relationship between fish TL (cm) and scale 

 size (SS) per area (see Methods) was best described 

 by a logarithmic equation. The six equations were 



Area 1 In TL = -1.620 + 1.054 In SS 



r 2 ^ 0.874 n = 23 

 Area 2 In TL = -1.830 + 1.117 In SS 



r = 0.956 n - 132 

 Area 3 In TL = 0.622 4- 1.073 In SS 



r 2 = 0.958 n = 171 



Area 4 In TL = -2.145 + 1.178 In SS 



r 2 = 0.879 n = 524 

 Area 5 In TL = -1.174 + 0.971 In SS 



r 2 = 0.857 n= 190 

 Area 6 In TL = -0.327 + 0.785 In SS 



r 2 = 0.639 n = 200. 



The logarithmic transformation was justified follow- 

 ing examination of the residual values (Draper and 

 Smith 1966). 



Body length-scale size regressions were compared 

 between the six geographical areas to determine 

 which equations should be used in back- calculating 

 lengths at age. The equivalence of TL-SS equations 

 between areas, sexes, and interactions was tested 

 using analysis of covariance. Significant differences 

 {P < 0.001) were found in TL-SS equations be- 

 tween areas but not between sexes or interaction 

 effects. Adjacent areas with similar slopes were com- 

 bined into a single region. Area 1 was excluded from 

 the analysis because of a small sample size, and the 

 range of lengths in other areas was truncated to 

 include equal sizes. The differences between areas 

 were evident when the slopes of the six equations 

 were examined (Fig. 3). Areas 2, 3, and 4 were statis- 

 tically similar (P > 0.329) and were therefore pooled 

 into one region (Region I). Area 1 was included in the 

 final pooled data for Region I, as the inclusion did not 





mar apr may jun jul aug sep oct 



Month 



nov 



dec may 



FIGURE 2. — Mean marginal scale increments ±95% confidence intervals of weakfish, Cynoscion 

 regalis, in the Middle Atlantic Bight for all ages combined. Sample size given for each month. 



805 



