94 



Fishery Bulletin 94( 



1996 



02 04 06 08101214161820 

 Increment width (urn) 



Figure 5 



Frequency distributions of the widths of the first 

 five daily increments (W, - W s ) measured with 

 optical microscopy and SEM. 



Good agreement among age estimates was also 

 reflected in a high level of certainty in the data. Con- 

 fidence codes for 679r of all age readings indicated 

 high confidence or better ( codes >4 ), 97% were at least 

 moderate confidence (codes >3), and only one sample 

 was discarded as unreadable (code=l). Moreover, 

 among the three age readers, the distributions of 

 codes were alike, indicating similar perceptions with 

 respect to the overall clarity of otolith microstruc- 

 ture. There were some differences, however, in the 

 confidence codes of very recently spawned fish and 

 those of older larvae. Fish 1-3 d old were deemed 

 the most difficult to age, whereas ages assigned to 

 older fish (>7 d) were viewed most confidently. 



Discussion 



We have shown that a check mark forms in the 

 otoliths of S. jordani larvae at parturition. This ex- 

 trusion check can be identified by its optical charac- 

 teristics and by its location in the otolith ( = 17 /jm 

 radius). Penney and Evans (1985) first described an 

 extrusion check in Sebastes on the basis of observa- 

 tions of larval redfish in the northwest Atlantic 

 Ocean. These authors described it as "a heavy ring 

 composed of a wide, translucent band followed by a 

 prominent, high-contrast dark band." The check was 

 absent in gestating larvae but was consistently ob- 

 served in planktonic larvae. Our results are in agree- 

 ment with theirs and confirm the findings of Laidig 



tional to the mean. A linear (Y=a + (bx)) least- 

 squares regression of logtNL) on age yielded 

 estimates of b = 0.0272 (s /) =0.000250) and a = 

 1.640 (s Q =0.00231) with an r 2 value of 0.849 and 

 a 2 = 0.00598. A back transformation of pre- 

 dicted length at age, with bias correction (Miller, 

 1984), is presented in Figure 6. Also shown in 

 the figure are the original untransformed data 

 and the predicted length-at-age relationship 

 reported by Laidig et al. (1991). 



When different readers examined the same 

 otoliths there was good agreement among age 

 estimates ( Fig. 7 ). For example, among the com- 

 bined total of 87 cross-validations, the three age 

 readers agreed to the day 41'^ of the time. More- 

 over, they were within ± 1 d 849r of the time and 

 ±2 d 93 f # of the time. In addition, all paired t- 

 tests for systematic age differences among the 

 readers were not significant (Fig. 7). The sta- 

 tistical power of these tests was such that mean 

 differences in age in the range 0.38-0.88 d were 

 detectable at the a - 0.05 level of significance. 



14 



130 



120 



11 



100 



■in 



8.0 



70 



6 i ) 



5.0 



4 C 



Observed mean (*1 standard deviation) 

 Predicted (this study) 

 Predicted (Laidig et al, 1991) 



y 



tTVt J- 



T /$> T 



T Jr*' I * ' 



I,- 



10 



1 I ' 



20 



25 



30 



35 



Age (d) 



Figure 6 



Observed and predicted growth of ethanol-preserved larval 

 shortbelly rockfish, Sebastes jordani (means are bracketed by ±1 

 standard deviation). 



