298 



Fishery Bulletin 97(2), 1999 



r=0.53. p<0 01 , n=49 



•: V 



r=0.69, p<:0,001. n=59 



• -s 



r=0.68. p<0.001, n=53 



700 



R.sagitta 



. ..f- 



r=0.86, p < 0.001 , n=56 



r=0.60, p<0.001, n=50 ' 









r=0.66, p<0.0C1, n=54 



600 



L.sagitta 



.; .•/• 



• .; 



900 



200 



R.lapillus 



J L 



100 



800 



Llapillus 



( 200 I 



Figure 2 



Scatter plots of individual measures of otolith area (|jm-) for newly hatched cod larvae 

 collected on the Scotian Shelf Correlation coefficients (/) for each relationship are given 

 in each panel. 



lapilli (Fig. 6). However, the residuals were not sig- 

 nificantly related to either measure for sagittae. 



Standard length (SL) at hatching was negatively 

 related to sea temperature at collection: SL = 5.26 

 (±0.07) - 0.939(±0.007) X temp, n = 44, r- = 0.79, P = 

 0.001. To explore the potential relationship between 

 otolith size and temperature, we regressed our most 

 predictive estimate of otolith size, lapillar area, 

 against the residuals from the SL-temperature rela- 

 tionship. There was no clear relation between tem- 

 perature residual and lapillar area (Fig. 7). 



Discussion 



Otoliths can be used to infer the standard length of 

 cod larvae at hatching. However, our data suggest 

 that the precision with which such inferences can be 

 drawn may not be sufficient to permit accurate com- 

 parisons of the size distribution at hatching between 



survivors and the population at large. On the Scotian 

 Shelf newly hatched cod larvae varied between 2.4 

 and 6.1 mm SL (Miller et al., 1995). The best regres- 

 sion relationship we developed in our study related 

 SL to area of the lapillus, and explained 359f of the 

 variation in the data. Accordingly, 65% of the varia- 

 tion remained unexplained. The regressions we de- 

 veloped suggest that the 95*^ confidence intervals of 

 individual back-calculated size at hatching for a fish 

 are wide. Predicted SL for larvae with average-size 

 lapilli ranged from 3.66 to 5.41 mm. This covers fully 

 48.7'7c of the total range in initial sizes at hatching 

 observed in cod larvae from the Scotian Shelf More 

 strikingly, based on estimates of mean and variation 

 of SL given by Miller et al. ( 1995), this range includes 

 917( of all newly hatched cod larvae on the Scotian 

 Shelf Back calculations from lapillar otoliths larger 

 or smaller than average size will be even more im- 

 precise. Thus, we conclude that estimates of initial 

 size in cod larvae are unlikely to be sufficiently pre- 



