FISHERY BfLLETIN: VOL. 83. NO. 3 



LARVAL HERRING 



1976- 1977 



15-1 



LU 

 < 



> 



DC 10 



< 



cc 



LU 

 CO 



INCREMENT EVERY 2 DAYS 



o DAILY DEPOSITION 



\ I \ \ r 



MAY 1 JUN 1 JUL 1 AUG 1 SEP 



OCT 1 



I \ \ \ f- 



MAY 1 JUN I JUL 1 AUG 1 SEP 



DATE OF HATCHING 



NOV 



OCT 1 



NOV 



Figure 2. -Frequency of Atlantic herring hatching during the 1976-77 study, calculated under two assumptions of 



otolith increment count and age relationship. 



< 0.01). The average length achieved per increment 

 was 0.33 mm and 0.10 mm for early- and late-hatch- 

 ed larvae, respectively. Bartlett's test confirmed that 

 variances were homogeneous. Analysis of residuals 

 showed that the last three residuals, corresponding 

 to the three largest larvae, were below the average. 

 The exclusion of these points did not alter the results 

 of the analysis. 



Analysis of the subset of larvae with fewer than 60 

 increments (Table 2) showed that data were better 

 fitted with two different regression lines {P < 0.01) 

 and that the slopes were significantly different {P < 

 0.01). Regression plots are shown in Figure 4. 

 Change in length of early-hatched larvae was 0.33 

 mm/increment and 0.15 mm/increment for late- 

 hatched larvae. Bartlett's test showed variances to 

 be homogeneous and residuals showed no trends, ex- 

 cept for the two youngest late-hatched larvae which 

 fell below the regression line. Late-hatched larvae 

 were slightly larger than early-hatched larvae for the 

 lowest increment counts. 



Differences in length-at-increment count were 

 apparent for data whose calculated hatch dates in- 

 cluded deposition rates of from 0.5 to 1.0 incre- 

 ments/d (Table 3). The change in length of early- 



hatched larvae was 0.33 mm/increment compared 

 with 0.17 mm/increment for late-hatched larvae. 

 Data were again better fit with two regression lines 

 (P < 0.01) whose slopes were significantly different 

 (P < 0.01). 



The von Bertalanffy growth equation fit the late- 

 hatch larval data well (Table 4). L^ was estimated at 

 29.81 mm, with a 95% confidence interval of 26.41 

 to 33.22 mm. Fit to the early-hatched larval data was 

 poor. L^ was estimated at 35.59 mm, with a con- 

 fidence interval of 17.76 to 53.41 mm. These data 

 were adequately fit with a straight line, and there is 

 little justification for fitting with a curvilinear func- 

 tion other than it has been traditionally used for 

 adult fish. Beverton and Holt (1954), however, 

 stated that the von Bertalanffy equation should not 

 necessarily be used during the early life stages. 

 Nevertheless, when the parameter estimates from 

 the two curves were compared, they were signifi- 



FlGURE 4. -Regression plot of length-at-otolith increment count for 

 Atlantic herring. Only lengths for larvae with 60 or fewer otolith 

 increments have been included for analysis. Data from Lough et 

 al. (1982). 



292 



