Abstract.- The relationship be- 

 tween length and age of larval and 

 juvenile shortbelly rockfish Sebastes 

 jordani, determined from otolith 

 microstructure, is complex. Models 

 that assume size increases smoothly 

 with age may not accurately describe 

 growth in young-of-the-year rock- 

 fish. A segmented (piece-wise linear) 

 regression model relating somatic 

 and otolith size was used to back-cal- 

 culate body length-at-age. The seg- 

 ments of this model coincide with 

 different growth stanzas, which are 

 separated by distinct life-history tran- 

 sitions. The composite function of 

 this model, and a Gompertz curve re- 

 lating otolith size and age, yielded a 

 good fit to the back-calculated stan- 

 dard length-at-age data. Comparison 

 of back-calculated with actual growth 

 showed no evidence of size-selective 

 mortality. The change in body length, 

 as the number of otolith increments 

 increased, was equal to the observed 

 increase in length per day of a se- 

 quentially sampled cohort, validating 

 the daily periodicity of the incre- 

 ments. 



Dynamics of Growth in the 

 Early Life History of Shortbelly 

 Rockfish Sebastes jordani 



Thomas E. Laidig 

 Stephen Ralston 

 James R. Bence 



Tiburon Laboratory, Southwest Fisheries Science Center 



National Marine Fisheries Service. NOAA 



3 1 50 Paradise Drive, Tiburon. California 94920 



Manuscript accepted 24 June 1991. 

 Fishery Bulletin, U.S. 89:611-621 (1991). 



The abundance of young-of-the-year 

 fish, and ultimately the number that 

 reach maturity, varies greatly among 

 years. Much of this variation is un- 

 related to the size of the spawning 

 stock, and understanding its causes 

 remains a central focus of fisheries 

 research (e.g., Hjort 1914, Sissen- 

 wine 1984, Rothschild 1986, Hol- 

 lowed et al. 1987, Sinclair 1988). Mor- 

 tality agents, such as starvation 

 (Hjort 1914, Lasker 1975 and 1978, 

 Houde 1977 and 1978, Theilacker 

 1978 and 1981, Grover and Olla 1986) 

 and predation (Hunter 1981, Hunter 

 and Kimbrell 1980, Sissenwine 1984, 

 Bailey and Houde 1989), can operate 

 strongly during early stages, and it 

 is clear that mortality rates of young 

 fish are higher than those experi- 

 enced by older fish (Miller et al. 

 1988). Interannual variation in year- 

 class strength can depend upon the 

 rate of growth early in life (Houde 

 1987 and 1989, Underwood and Fair- 

 weather 1989). Under most condi- 

 tions, the more rapidly fish grow 

 through early, high-risk stages, the 

 fewer die; small changes in growth 

 rate can thus lead to a major change 

 in recruitment (Houde 1987). Knowl- 

 edge of the processes affecting 

 growth during the early life history 

 may help predict the occurrence of 

 strong year-classes. 



Numerous mathematical models 

 have been developed to describe the 

 growth process, including the Gom- 

 pertz, von Bertalanffy, logistic, and 



exponential functions (Ricker 1979). 

 Although these models usually per- 

 form well on adult stages, they com- 

 monly falter in predicting growth 

 during the first few weeks of life. For 

 example, Phillips (1964) used a von 

 Bertalanffy function to model growth 

 of shortbelly rockfish Sebastes jor- 

 dani. The model predicted growth of 

 adults well, but its accuracy deterior- 

 ated for young-of-the-year fish. Be- 

 cause somatic growth during early 

 stages can be affected by abrupt 

 physiological changes (e.g., flexion, 

 juvenile transformation, and settle- 

 ment), models that predict growth 

 during this period should reflect (or 

 describe) this complexity (Ricker 

 1979). 



Rockfish (Sebastes spp.) are an im- 

 portant component of the west coast 

 groundfish fishery (PFMC 1989). 

 Growth rates of young rockfish have 

 been estimated from linear models 

 relating length to age (Boehlert 1981, 

 Boehlert and Yoklavich 1983, Penney 

 and Evans 1985, Haldorson and 

 Richards 1987, Woodbury and Ral- 

 ston 1991). Because most of these 

 studies examined a narrow range of 

 ages, linear models fit the data ade- 

 quately. Although Penney and Evans 

 (1985) showed that a linear model ex- 

 plained much of the variation in 

 length during larval and juvenile 

 stages of redfish, back-calculated 

 growth rates varied systematically 

 with age; growth rate was relatively 

 slow for young larvae, more rapid 



