Ahrenholz et al.: Otolith ageing of Brevoortia tyrannus 



213 



increment count regressions (transverse sections) 

 revealed that these parameters were not significantly 

 different between treatments. 



We tested for the homogeneity of slopes for the in- 

 crement count-age regressions between sectioning 

 orientations separately for the known-age experi- 

 ment and both of the marking experiments. None of 

 the slopes were significantly different (P for known 

 age=0.0583, for OTC=0.188, and for ALC=0.667). 

 Tests for differences in intercepts for the same ex- 

 perimental sets revealed none (P for known 

 age=0.345, for OTC=0.082, and for ALO0.526). 

 Therefore we pooled the increment count results 

 within each experiment. 



We used regressions to compare the results for the 

 known-age and chemically marked otoliths for about 

 the same time duration (i.e. 136 days for known-age 

 fish, 147 days for the OTC group, and 131 days for 

 the ALC group; Table 1, Fig. 3). The intercepts of the 

 three increment-count regressions were not signifi- 

 cantly different from zero, and none of the three slope 

 estimates were significantly different from 1.0 (Table 

 1). The results for the ALC and the OTC groups had 

 sufficient power (>0. 80) to detect a difference in slope 

 of 0.1 from a value of 1.0. The standard error of the 

 slope was relatively greater for the known-age group, 

 and thus the power estimate was less than that for 

 the ALC and OTC groups (Table 1). 



We examined the results from day 161 to 237 for 

 the ALC experiment in parallel fashion. A test for 

 homogeneity of slopes for increment count on days 

 postmark revealed a (marginally) significant differ- 

 ence between the sectioning orientations (P=0.044). 

 Estimates of the slopes from separate regressions for 

 the oblique-transverse and transverse sections were 

 significantly different from 1 (P=0.016 andP<0.001). 

 While these observations begin to define limits for 

 applying daily ageing techniques to juvenile Atlan- 



tic menhaden, they do not reduce the usefulness of 

 the technique over a relatively broad time period. 

 With the minor exception of the period when incre- 

 ment counts were less than daily in the ALC trial, 

 the results for the ALC-marked and OTC-marked test 

 groups were equivalent for either section orientation 

 with slopes near one and with good statistical power 

 to detect a small deviation from one (Fig. 3, Table 1 ). 



Discussion 



Atlantic menhaden, on the average, form one growth 

 increment per day through at least an estimable age 

 of 200 days (131 days postmark + approximately 70 

 days in age at marking) and a size of nearly 150 mm 

 fork length. We could reliably age menhaden up to 

 200 days old to within about 7 days when juvenile 

 growth rates were high (e.g. above 0.6 mm-d -1 over 

 summer months). At moderate juvenile growth rates 

 (approximately 0.5 mm-d -1 ), we still detected incre- 

 ments at approximately one per day through a 250 

 day time period for the OTC fish ( 147 days postmark 

 + an average of 103 days of age at marking), but the 

 variability of an individual age estimate increased 

 in comparison with fish with higher growth rates. 

 For the OTC and known-age test groups respectively, 

 95% confidence intervals increased to approximately 

 ±16 and 21 d for similar-aged menhaden with slower 

 growth rates (Table 1, Fig. 3). As growth rates de- 

 clined further (below 0.3 mm-d -1 ), our increment 

 counts declined to less than one per day, and vari- 

 ability in estimated age increased; this may be due 

 to decreases in increment width or to reduced peri- 

 odicity as has been found for starved larval Atlantic 

 menhaden (Maillet and Checkley, 1990). After day 

 131 postmark (ALC), declining growth rates and an 

 increment periodicity of less than one per day (Fig. 



Table 1 



Least squares linear regression analysis for increment counts from known-age, oxytetracycline (OTC) marked and alizarine 

 (ALC) marked Atlantic menhaden, Brevoortia tyrannus. otolith sections. The null hypotheses tested are that the intercept=0 and 

 the slope=l. (SE=standard error.) 



' Estimate of percent statistical power to detect a deviation of 0.1 from a slope of one at the P = 0.05 level. 



2 95% confidence interval (±days) for an age estimate based on individual ring counts. The 95% confidence intervals for indi- 

 vidual age estimates were constant over the range of values used to generate the regression. 



3 Through day 131 postmark. 



