266 



Fishery Bulletin 90(2). 1992 



variates, the growth increment 

 model was used to investigate 

 sex differences in growth. This 

 was done by fitting a model with 

 only go, gi, and indicator vari- 

 ables for sex. The P-value for the 

 intercept coefficient for sex was 

 0.059, but the P-value for the 

 slope term was 0.998. By the 

 criteria established earlier, 

 neither coefficient would be con- 

 sidered statistically significant, 

 although the P-value for the in- 

 tercept coefficient is close to the 

 critical value. The estimate of 

 the intercept coefficient for sex 

 (0.058) indicates that the females 

 grow approximately 6% (e^^ss) 

 more than males on an annual 

 basis, regardless of age. A differ- 

 ence of this magnitude is suffi- 

 cient to account for the greater 

 asymptotic size of the females. 

 Figure 4 shows the fitted sex- 

 specific curves for the annual 

 growth increment. Despite the 

 lack of statistical significance of 

 the intercept term for sex, it was 

 retained in the model while eval- 

 uating the significance of envi- 

 ronmental covariates on growth. 

 The larger size attained by the 

 female Pacific whiting is com- 

 pelling evidence that there are 

 sex-specific differences in Pacific 

 whiting growth. Including this 

 term in the baseline model is 

 important because it accounts 

 for this sex-specific variability in 

 growth. 



Table 5 shows the analysis 

 of variance using the annual 

 growth-increment regression 

 model. The model was built in a 

 forward stepwise fashion, adding the environmental 

 term to the baseline model that resulted in the largest 

 reduction in the residual sum of squares. Temperature 

 and population biomass were significant covariates in 

 the model. Temperature had significant intercept (P< 

 0.001) and slope terms (P 0.026). For biomass, only the 

 slope term was significant (P 0.002). 



The parameter estimates in Table 5 indicate that a 

 0.5°C increase in mean summer sea-surface tempera- 

 ture will bring about a 24% reduction in the annual 

 growth increment at age 1. At age 4, the same increase 



in temperature would be expected to produce a 12% 

 reduction in annual growth, and by age 7, the percent 

 reduction would be close to zero. The model predicts 

 an increase in growth due to increasing temperature 

 above age 7, but as annual growth is very slight by this 

 age, the consequences of this prediction are not impor- 

 tant. One concern about the reliability of these results 

 is that they may be overdependent on growth during 

 the 1983 El Nino, when sea-surface temperature was 

 the highest during the study. To investigate this pos- 

 sibility, a model was fit to the data excluding the 



