36 



Fishery Bulletin 105(1) 



daily increments form at first feeding (Sepulveda, 1994; 

 Hirose and Kawaguchi, 2001). However, for some species 

 of smelt, increment formation does occur at hatching 

 (Ohama, 1990). Otterlei et al. (2002) demonstrated that 

 temperature could further increase age variability by 

 altering the timing of the first increment depending upon 

 when, in the season, a fish was born. Delta smelt spawn 

 from late February through June, potentially resulting 

 in cohorts experiencing temperature differences of more 

 than 10°C at birth (Moyle, 2002). This temperature 

 difference could result in up to a two-fold change in the 

 embryonic development and lead to a 2-3 day difference 



Table 4 



Comparison of growth curves generated from otolith size- 

 at-age data. r-= coefficient of determination for estimates 

 of size-at-age and known size-at-age for each age group. 

 "% range of mean size-at-capture" represents the vari- 

 ability between estimated size and known mean size at 

 time of sampling. 



Model 



% range of 

 mean size at capture 



Biological intercept 0.994 -3. .5 to +6.4 



Modified Fry 0.935 +2.4 to +13.1 



Time-varying growth 0.038 -10.3 to +0.90 



Estimated adjusted age (days) 



Figure 5 



Growth curves back-calculated from delta smelt ^Hypomesiis transpaci- 

 ficicus) otoliths. Size-at-age data for laboratory-reared fish are shown 

 as points with ±1 standard deviation. Three models of back-calculation 

 were used to generate growth curves from otoliths: the modified Fry 

 (MF), the stage-specific biological intercept (BI), and the time-varying 

 growth (TVG) model. 



in the timing of first feeding for delta smelt (first author, 

 unpubl. data). 



Otoiith-size— fish-size relationship 



The results of this study showed that minimal age-inde- 

 pendent variability occurred in the OS-FS relationship 

 and that growth rate effects were negligible (Fig. 3, A 

 and B). Removing the effect of age resulted in a strong 

 correlation between otolith growth and fish growth, 

 indicating that at the preflexion stage, otolith growth 

 was proportional to somatic growth. However, the OS-FS 

 relationship showed a significant interruption in linear 

 growth due to an ontogenetic shift at the postflexion 

 stage, thus demonstrating that the assumption of con- 

 stant proportionality was violated (Fig. 4). Moreover, 

 this change in proportionality corresponded with the 

 transition from preflexion larvae to postflexion-stage 

 juveniles. Several studies, across broad taxonomic orders 

 have demonstrated similar deviations from proportional- 

 ity in the OS-FS relationship (e.g., flatfish (Rhombosolea 

 tapirina and Atnmotretis rostratus) [Jenkins, 1987], 

 rainbow smelt iOsmerus mordax) [Sirois et al., 1998], 

 and Atlantic cod {Gadus morhua) [Otterlei et al., 2002]), 

 which were also associated with life-stage transitions. 

 Measured growth-rate effects were weak for delta 

 smelt (Fig. 3B) and, thus, are not likely to further ex- 

 plain the age-independent variability in the OS-FS rela- 

 tionship. In contrast, Sirois et al. (1998), found signifi- 

 cant growth-rate effects in the OS-FS relationship for 

 rainbow smelt. Furthermore, Hare and 

 Cowen (1995) found both growth-rate ef- 

 fects and ontogenetic shifts in the OS-FS 

 relationship for bluefish larvae iPomato- 

 mus saltatrix). These studies highlight 

 the different mechanisms responsible for 

 the variability in the otolith and somatic 

 growth relationship, and, thus, potential 

 violations in the constant assumed pro- 

 portionality for linear back-calculation 

 models. 



Other factors that can influence the 

 OS-FS relationship include temperature 

 and salinity. In the field, delta smelt can 

 experience a broad range of tempera- 

 tures during their larval stage. The ef- 

 fects of differing temperatures at this 

 life stage on the OS-FS relationship and 

 the influence on otolith back-calculations 

 from field-caught fish is unknown. It is 

 assumed that otolith growth is allome- 

 tric to fish growth during seasonally 

 variable temperatures. For our study, 

 delta smelt were reared in temperature 

 and salinity controlled conditions, and 

 therefore temperature and salinity were 

 considered to have negligible effects and 

 did not influence the variability in the 

 OS-FS relationship. However, tempera- 

 ture variability can result in different 



