Hobbs et a\ Modification of the biological intercept model to account for ontogenetic effects in Hypomesus transpacificus 



35 



B 



r = 068 



-1 ' 1 ' — 1 ' 1 ' 1 ' 1 ' 1 ' 1 



150 IKi -50 50 100 150 200 



-200 -100 100 200 300 



Residuals of otoiitfi radius-on-age 



Figure 3 



Correlation analysis of residuals evaluating age-independent variability and growth-rate 

 effects in the relationship of otolith size to fish size for young delta smelt fHypomesus trans- 

 pacificus): (A) fish standard length versus otolith radius residuals, (B) age versus standard 

 versus otolith radius residuals. Pearson (r) correlation values are shown on graph. 



Comparison of the TVG, Bl, and MF models 



To account for ontogenetic change in the OS-FS rela- 

 tionship, we applied the BI model separately for pre- 

 flexon and postflexon larvae. Mean back-calculated 

 size-at-age for the TVG model varied from -10.3% 

 to -hO.93% of the observed mean-length-at-age and 

 consistently underestimated size from 20 dah to 100 

 dah. The stage-specific BI model varied from -3.5% 

 to +6.4%, centering closely around zero, and the MF 

 model varied from +2.4% to +13.1% and consistently 

 overestimated size from 15 dah to 100 dah (Table 4; 

 Fig. 5). These results indicate that the ontogenetic 

 stage-specific BI model more accurately describes the 

 observed mean size-at-age trajectories for delta smelt. 

 However, the observed size-at-age was variable from 

 50 dah to 100 dah, resulting in statistically reason- 

 able fits for all the models, with r'~ values ranging 

 from 0.994 to 0.935 (Table 4). 



Discussion 



Validation of daily otolith increment formation 



Accurate size-at-age estimates are important for mea- 

 suring subtle differences in growth during the early 

 life history of fishes because they are used to examine 

 factors affecting recruitment success for many species 

 (Rice et al., 1987; Sirois and Dodson, 2000). However, 

 variability in age estimates can occur. Much of this 

 variability in age estimates depends largely on when 



the first increment is estimated to have formed (Cam- 

 pana and Neilson, 1985; Campana, 2001). In this study, 

 increment formation in laboratory-reared delta smelt 

 began six dah, about the time of first feeding. Numerous 

 studies of other smelt species have also revealed that 



