FISHERY BULLETIN: VOL. 87, NO. 4, 1989 



The principal component time series of this first 

 mode of variabiHty (Fig. 7) showed gi'owth rates 

 as nearly normal in 1915-45, moderately de- 

 creasing in 1945-70, and abruptly increasing 

 after about 1972. This was consistent with the 

 earlier interpretation of variability in the indi- 

 vidual time series (Fig. 4). The increased growth 

 rates after 1972 coincided with the same signal 

 described by the first principal component of S. 

 pinniger growth rates (Fig. 6). 



The second principal component of S. dip- 

 loproa growth rates extracted the independent 

 nature of variability in age group 1 and ac- 

 counted for 19% of the total variance over the six 

 growth anomaly time series. The loading values 

 of this second mode of variability (Fig. 5B) were 

 near zero for all age groups except age group 1. 

 The corresponding principal component time 

 series (Fig. 7) was almost identical to the time 

 series of S. diploproa age group 1 (Fig. 4); peri- 

 ods of positive, normal, and negative growth 

 rates in both time series closely corresponded. 

 This further supported the interpretation of this 

 mode of variability as describing age gi'oup 1 

 gi'owth rates. As will be expanded upon in the 

 following section, the lack of hnkage between 

 gi'owth rates in age gi'oup 1 and the other age 

 gi'oups is Hkely related to the fact that S. dip- 

 loproa inhabits an environment during the first 

 year of life that is quite different from that dur- 

 ing later states of life (Boehlert 1977) and is thus 

 subject to diffei'ent environmental factors in- 

 fluencing gi'owth rates. 



As with S. pinniger, the third principal com- 

 ponent of S. diploproa growth rates has no ob- 

 vious physical interpretation. The loading values 

 (Fig. 5B) oscillate from positive to negative 

 across the age gi'oups, and the principal com- 

 ponent time series (Fig. 7) was nearly normal 

 over the entire record length. 



DISCUSSION 



The study of growth in marine fishes has 

 typically been concerned with relatively short- 

 term trends in gi'owth of cohorts or populations, 

 most often with fished stocks. Differences in 

 growth may have existed between stocks (Tem- 

 pleman and Squires 1956; Borisov 1979), geo- 

 graphical regions (Boehlert and Kappenman 

 1980), stock densities, and years (Margetts and 

 Holt 1948; Jones 1983). Such growth differences 

 may be due to genetic factors (Borisov 1979), 

 density dependence (Margetts and Holt 1948; 

 Peterman and Bradford 1987), or environmental 



factors, most importantly temperature (Brett 

 1979; Kreuz et al. 1982). The dramatic reduction 

 in stock sizes of many Sebastes spp. (Archibald 

 et al. 1983; Ito et al. 1987) has led several 

 authors to suggest density-dependent increases 

 in gi'owth (Gunderson 1977; Boehlert and Kap- 

 penman 1980). Recent changes in ageing meth- 

 odology for Sebastes (Boehlert and Yoklavich 

 1984), however, have rendered long-term com- 

 parisons of growth difficult without a historical 

 collection of otoliths. 



Analogous to historical time series derived 

 from tree rings (Fritts 1976), the technique 

 described in our paper uses uniform method- 

 ology to develop time series of growth for long- 

 lived fishes. Reading of otoliths, however, has 

 more inherent variability than that for tree 

 rings, and the resultant ageing biases described 

 earlier must be considered. The impacts of age- 

 ing errors will be most apparent in high fre- 

 quency signals in the time series and are prob- 

 ably not well resolved by our data; this is the 

 main reason we used the running averages to 

 low-pass filter the data, making the low fre- 

 quency signals more apparent to the human eye 

 (Figs. 3, 4, 6, 7). 



An interesting biological feature of our re- 

 sults is the differing pattern of gi'owth of age 

 group 1 S. diploproa (Fig. 4), the only age 

 group not contributing significantly to the vari- 

 ance described by the first principal component 

 for either species. In contrast, first year growth 

 in S. pinniger was similar to that in age gi'oups 

 2-6. Sebastes pinniger have a relatively narrow 

 seasonal spawning peak (Westrheim 1975; 

 Gunderson et al. 1980), and pelagic young ap- 

 parently recruited to their juvenile benthic 

 habitats within about 6 months (Richardson and 

 Laroche 1979). Sebastes diploproa, however, 

 seemed to spawn during most months of the 

 year (Snytko 1975), and pelagic prejuveniles 

 were present year-round, at least in the South- 

 ern California Bight (Boehlert 1977). Assuming 

 that the first annulus is laid down on a seasonal 

 basis, first year growth was probably quite 

 variable. Further, S. diploproa are deepwater 

 (200-500 m) members of this genus, but their 

 first year is spent in surface waters, probably in 

 the upper meter of the water column (Boehlert 

 1977, 1981); thus, the factors influencing growth 

 in the first year may differ from those affecting 

 growth in subsequent years. Temperature can 

 have an important impact on juvenile rockfish 

 gi'owth (see summary in Boehlert and Yoklavich 

 (1983)), but it may not show coherent cycles be- 



800 



