BOEHLERT ET AL.: TIME SERIES OF GROWTH IN SEBASTES 



The second principal component of S. plnniger 

 growth variability extracted the coherent trends 

 in the time series evident prior to 1965 (Fig. 3). 

 The loadings (Fig. 5A) were negative for age 

 groups 2-4 and positive for age groups 5 and 6; 

 the loading value for age group 1 was zero, indi- 

 cating that its variability is not related to this 

 mode. The principal component time series of 

 this second mode of variability (Fig. 6) generally 

 showed a gradually increasing trend from 1945 

 to 1965, followed by a decreasing trend there- 

 after. This time variability described well the 

 trends evident in Figure 3 for age gi'oups 5 and 

 6, for which the loading values (Fig. 5A) were 

 positive. Since the loading values of the second 

 mode were negative for age gi'oups 2-4 (Fig. 

 5A), this mode of variability (the product of the 

 loading value and the principal component time 

 series, see Equation (1)) corresponded to a 

 gradually decreasing trend in age groups 2-4 

 gi'owth rates prior to 1965. followed by an in- 

 creasing trend thereafter. We have no explana- 

 tion for the difference in the early trends in 

 growth between age groups 5 and 6 and age 

 groups 2-4. Further, growth increments in 

 1964-68 (Fig. 3) reinforced this difference and 



produced a peak in the principal component time 

 series in those years (Fig. 6). This second mode 

 of variability describing the trends in gi'owth 

 rates accounts for 24% of the variance in the 

 data. 



The loading values of the third principal com- 

 ponent of S. pinniger gi'owth variability (Fig. 

 5A) showed no obvious coherent relationship 

 among age gi'oups. The values oscillated from 

 positive to negative among the age gi'oups, and 

 the corresponding principal component time 

 series (Fig. 6) showed no remarkable character- 

 istics. Thus, this mode of variability has no ob- 

 vious physical interpretation. 



The first four principal components of S. dip- 

 loproa variability explained 817c of the total 

 variance summed over the six growth anomaly 

 time series; the loadings for the first three 

 modes are shown (P^ig. 5B), with corresponding 

 principal component time series (Fig. 7). The 

 loading values of the first principal component, 

 accounting for 307c of the variance, were approx- 

 imately the same for age groups 2-6 and zero for 

 age gi"oup 1. This highlighted the fundamental 

 difference between gi-owth rate variability in 

 age group 1 and that in the other age gi'oups. 





C\2 

 U 



Oh 



CO 



o 



10 1920 1930 1940 1950 1960 1970 1980 



Figure 7. — Principal component time series for the three modes explaining 

 64.6% of the total variance in growth of Sebastes diploproa from the six time 

 series in Figure 4. The curve represents the double .5-yr running average. 



799 



