212 



Fishery Bulletin 93(2). 1995 



methods. Mean counts from the known-age fish were 

 increased by five to estimate time from spawning 

 rather than from first feeding (Warlen, 1992). ALC- 

 marked material was viewed under a microscope at 

 400-lOOOx, counted with a hand-held counter, and 

 the median of five serial counts was taken. 



Statistical analysis 



Regression and analysis of covariance (ANCOVA) 

 computations were conducted with SAS statistical 

 programs (SAS Institute, Inc., 1985). Analysis of co- 

 variance was used to test for common regression 



B 



Figure 2 



Photomicrographs of juvenile Atlantic menhaden, 

 Brevoortia tyrannus, sagittal otolith sections showing fluo- 

 rescent marks (arrows). (A) Transverse section of otolith 

 from a post-larval fish 12 days after immersion in alizarin 

 complexone solution (ALC). The maximum dimension of 

 this section (dorsal/ventral) is 462 urn. (B) Oblique-trans- 

 verse section of otolith from a juvenile 42 days after im- 

 mersion in oxytetracycline solution (OTC). The maximum 

 dimension of this section ( posterior-dorsal/anterior- ventral ) 

 is 938 urn. 



parameters (ring count versus known days) for the 

 low and high food treatments and between transverse 

 and oblique-transverse sectioned material for each 

 marking-validation method (Ott, 1977). Mean growth 

 rates were estimated as the slopes of simple linear 

 regressions of length on age. 



We tested the null hypothesis that growth incre- 

 ments in otoliths of larval and juvenile Atlantic men- 

 haden are formed daily The null hypothesis is ac- 

 cepted if the regression of estimated increment count 

 on known age in days since marking is significant, 

 the slope is not significantly different from one, and 

 the intercept is not significantly different from zero. 

 We also calculated the appropriate statistical power 

 to detect a relatively small difference from a slope of 

 one (Rice, 1987). Student's-^ test was used to test for 

 significance of the slope and intercept. Statistical 

 power to determine a deviation of 0.1 (a=0.05) from 

 a slope of one was estimated for each linear regres- 

 sion (Rice, 1987; Neter et al., 1989). 



Results 



Otolith preparations were generally readable for the 

 ranges in sizes and ages for the elapsed times. 

 Known-age fish were sampled at ages 52, 66, 81, 122, 

 131, 132, and 136 days; they ranged from 10 to 64 

 mm in fork length. Mean growth for the known age 

 group was 0.55 mm-d -1 , with a standard error (SE) 

 of 0.048 over the interval from March to June. OTC- 

 treated fish were sampled 13, 18, 42, 110, and 147 

 days after treatment; they ranged from 27 to 98 mm 

 in fork length and ranged in estimated age from 62 

 to 130 days with a mean age of 103 days at marking, 

 resulting in mean growth of 0.49 mm-d -1 (SE=0.011) 

 for the interval April to August. ALC-marked fish 

 were sampled 12, 42, 71, 100, 131, 161, 190, 205, and 

 237 days after treatment; they ranged from 26 to 175 

 mm FL and were approximately 70 days old at mark- 

 ing. Mean growth rates of the ALC-marked fish were 

 0.67 and 0.95 mm-d- 1 (SE=0.020 and 0.015) through 

 day 131 postmarking in low and high food tanks re- 

 spectively Growth visibly declined for the interval 

 from day 161 to day 237 postmark and was 0.29 

 mm-d -1 for both the low and high food treatments 

 (SE=0.037 and 0.195). 



It was readily apparent from scatter plots that the 

 ALC growth-increment count beyond day 131 (day 

 161 to day 237) postmark was less than one per day 

 and the variance about an individual sampling date 

 substantially greater. We pooled data up through day 

 131 postmark from the ALC high and low food treat- 

 ments because tests for homogeneity of the result- 

 ing slopes (P=0.667) and intercepts (P=0.831 ) for age- 



