672 



Fishery Bulletin 103(4) 



converted through interpolation to provide depth at 

 specific temperatures (Hourston 1 ). 



Aging and growth determinations 



Ages were determined by using the otolith burnt-section 

 technique (MacLellan, 1997) with a minor modification. 

 A survey directed at studying juvenile rockfish in 1991 

 captured two 17-cm silvergray rockfish. An examination 

 of these otoliths indicated that the previous application 

 of the method had incorrectly assigned the first annulus 

 to the age count in specimens. Therefore, some previ- 

 ously aged specimens were probably under-estimated by 

 one year (MacLellan 5 ). A faint first annulus is consis- 

 tent with the late spring to mid-summer parturition of 

 silvergray rockfish that appears to preclude significant 

 summer growth in its first year. The method was modi- 

 fied in August of 1992, and we added one year to all 

 previously aged specimens in the data set. 



Most (85%) of the otoliths were aged by one reader. 

 The remaining 15% were aged by two readers to moni- 

 tor consistency. If there was a disagreement, the two 

 readers agreed on a "resolved" age. 



Age and length data were fitted to a generalized 

 growth model (Schnute, 1981) (Appendix 1). Growth 

 dimorphism was calculated as the ratio of the mid- 

 points of fork length (maximum observed length minus 

 minimum observed length) between males and females 

 (Lenarz and Wylie Echeverria, 1991). 



by tracking the proportions in each maturity stage by 

 month. Lacking histological confirmation for character- 

 izing maturity, we followed the suggestion of Wylie Ech- 

 everria (1987) and used only those specimens collected 

 from the reproductive or gestation period of March to 

 August. Within this subset, we grouped female stages 1 

 and 2 as immature, and stages 3-7 as mature. Because 

 most mature females exhibited fertilized eggs by March, 

 we assumed that females with small, nondeveloped 

 ovaries in March through August would not complete 

 parturition in the same calendar year. 



We assumed that stage 1, during which testes are 

 translucent and string-like, was the only male imma- 

 ture stage. Subsequent stages 2 and 4-7 were grouped 

 as mature (stage 3 was not used in the field). The pro- 

 portion of stage-2 males (in relation to males in other 

 mature stages) decreased rapidly during the mating 

 season (September- January) indicating that many of 

 the specimens classified as stage 2 would become ma- 

 ture within the same calendar year. We emphasize, 

 however, that without histological support for these 

 classifications, the assumptions of maturity-at-age or 

 maturity-at-length remain tentative. 



The estimated proportions of maturity at age were 

 computed by fitting a generalized additive model (GAM) 

 to the binomial maturity classes (0=immature, ^ma- 

 ture) (Hastie and Tibshirani, 1990). A logistic link with 

 a binomial error structure was applied, as well as a 

 second-degree nonparametric LOESS smoother. 



Reproductive maturity 



Maturity stage was classified macroscopically in the field 

 (Table 1). We examined the annual reproductive cycle 



4 Hourston, R. 2003. Personal commun. Institute of Ocean 

 Sciences. Fisheries and Oceans Canada. 9860 West Saanich 

 Road, P.O. Box 6000. Sidney, British Columbia. V8L 4B2, Canada. 



5 MacLellan S. 2000. Personal commun. Pacific Biological 

 Station, Fisheries and Oceans Canada. Nanaimo, British 

 Columbia. V9T 6N7, Canada. 



Fecundity 



Fecundity was estimated from a single sample (??=132) 

 of females captured by commercial bottom trawl in Sea 

 Otter Trough in April 1989 (Fig. 1). The catch was stored 

 in refrigerated seawater for four days prior to sampling. 

 Sampling was stratified by length to obtain a range of 

 ages, and from each fish we obtained measurements of 

 fork length, gonad weight, and somatic weight. We also 

 collected otoliths and counted the number of cysts con- 



