FISHERY BULLETIN: VOL. 85, NO. 2 



Table 2. — Parameters for weighted regressions of increment counts on days from hatch of striped bass larvae 

 reared under four feeding regimes. SE indicates standard error of the estimate, C.I. indicates confidence 

 interval, N.S. indicates slope not significantly different than 1.0, * indicates P = 0.05. 



The slope of the regression Une for starved lar- 

 vae (condition 2, n = 43), 0.469 increments/day, 

 differed significantly from 1.0 increment/day. Incre- 

 ments appeared regularly spaced. Otoliths of 

 starved larvae did not appear aberrant under the 

 light microscope. 



The regression of increment counts versus true 

 daily age for larvae, which v^ere starved then fed 

 (condition 3, n = 12), had a slope of 0.930 incre- 

 ments/day with confidence intervals which included 

 1.000 increments/day (Table 2). However, the re- 

 gression intercept was - 10.430, an overestimate of 

 age at first increment deposition. This leads to a 6-d 

 underestimate of true age because depositional rates 

 were underestimated during the first 2 weeks of life. 



The slope of the regression line for intermittent- 

 ly starved larvae (condition 4, n = 12) was 0.873 in- 

 crements/day. The slope of 1.0 increment/day fell 

 at the very edge of the confidence interval. If a 

 slightly smaller alpha level had been chosen, deposi- 

 tion would not have been assumed daily. 



Initial increment formation began at 4 days after 

 hatching with a 95% confidence interval that ranged 



Table 3. — Counting bias for larvae starved 

 for the first 15 days after hatch (calculated 

 as estimated age - true age^ Underesti- 

 mate of age is indicated by - ; overestimate 

 indicated by + . SEI = secondary elec- 

 tron image; BEI = backscattered electron 

 image. 



from 3 to 5 days. Yolk-sac absorption occurs at 7 

 days after hatching at 18°C and first feeding begins 

 at approximately the same time. However, initial 

 increment deposition does not appear to be con- 

 nected to these events. Two or three weakly defined 

 increments were observed within the core in many 

 SEM preparations. They were not counted in light 

 or SEM readings. 



Scanning Electron Microscopy 



Results from the SEM study are qualitative rather 

 than quantitative due to the small sample sizes, n 

 = 13, used for the SEM. With SEM, otohth incre- 

 ment counts for condition 3, larvae which were 

 starved then fed (Table 3), and for condition 4, lar- 

 vae which were intermittently starved (Table 4), 

 yielded more accurate counts than those obtained 

 on the same specimens with light microscopy. With 

 light microscopy counts from larvae which were 

 starved for 15 days resulted in an underestimate of 

 true age by 10 days (Table 3). The variability was 

 also high (SE = 7.9 days). SEM counts underesti- 

 mated true age by 2 days. Variability was small; the 

 standard error was 3.4 and 4.0 days for SEI and 



Table 4. — Counting bias for larvae intermit- 

 tently starved (calculated as estimated age 

 - true age'). Underestimate of age is in- 

 dicated by - ; overestimate indicated by 

 + . SEI = secondary electron image; BEI 

 = backscattered image. 



^Estimated age = number of Increments + 

 mean age at first increment formation. 



'Estimated age = number of increments + 

 mean age at first increment formation. 



174 



