394 



Fishery Bulletin 94(3), 1996 



1-e-' 



Figure 3 



Comparison of the regression lines for the seven field sea- 

 sons analyzed (6 = rate of exponential decay; t = age in days). 



temperature regimes and that the difference was 

 likewise reflected in daily otolith increment widths. 



Several studies have observed a poor correlation 

 between otolith length and fish length (Secor et al., 

 1989; Moksness and Wespestad, 1989), and this was 

 partly the case in the present work. Since increments 

 may continue to be deposited during periods of slow 

 or negative somatic growth, caution must be exer- 

 cised in estimating age based solely on length rela- 

 tionships. After 120 days of growth, the fish in the 

 1988-89 and 1993-94 seasons showed smaller mean 

 otolith diameters than did fish in 1989-90, 1990-91, 

 1991-92, and 1992-93 seasons, despite their having 

 larger mean standard lengths. There is, however, a 

 strong relationship between mean otolith increment 

 width and the somatic rate of growth as expressed 

 in mm/d of SL. From 60 to 120 days of age, larval 

 herring grew very slowly in 1976-77 and exhibited 

 thin increments; whereas in 1988-89, growth was 

 more rapid and the increments were wider. Secor et 

 al. ( 1989) found evidence of a lag effect between food 

 ration and its reflection in the otoliths and, until this 

 Finding can be demonstrated in herring, it is perhaps 

 best to cite trends rather than attempt to relate a 

 given increment to a specific event. 



The predictive model for Atlantic herring growth 

 presented above has to be viewed as general in na- 

 ture, and the widening of the confidence intervals 

 with increasing length (Tables 4 and 5; Figs. 4 and 

 5) must be kept in mind. Natural variability of length 

 at age and difficulty in the reading of otoliths increase 

 as fish become older and make precise age determi- 

 nations extremely difficult. For example, the ability 

 to predict correctly the age of an individual herring 



larva at the 70% confidence level decreases from ±2 

 days at 8 mm SL to ±2 weeks at 30 mm. In spite of 

 this problem, otolith ageing of field-caught larvae pro- 

 vides a degree of precision not possible with indirect 

 methods based on size-frequency analysis (Ebert, 1973 ). 

 The inflection noted in the rate of otolith growth 

 (Fig. 7) occurs at an age when the otic bullae would 



