Norton and MacFarlane: Nutritional dynamics of reproduction in Sebastes flavidus 



301 



data to compensate for size variation. Tissue and 

 component masses were converted to natural loga- 

 rithms [\n(X + 0.01)] and subjected to the general 

 linear model analysis of covariance (ANCOVA) by 

 using the natural log of standard length (In SL) to 

 adjust for individual size differences among monthly 

 samples. The adjusted means and standard errors 

 of tissue and component masses were backtrans- 

 formed by taking their antilogs. ANCOVA also de- 

 termined the statistical significance of tissue and 

 component variation across the reproductive cycle 

 within each sex. We determined differences among 

 specific monthly means by Duncan's multiple con- 

 trast test with a = 0.05. Differences in monthly means 

 between sexes were assessed by Mests. All analyses 

 were performed within the Number Cruncher Sta- 

 tistical Software package (NCSS, Provo, UT). 



We estimated the contribution of somatic nutri- 

 tional components to female reproduction by deter- 

 mining changes in tissue component masses between 

 the onset of rapid ovarian growth and parturition. 

 The decrease in somatic component quantities from 

 maximal values before December to minimal levels 

 in March was calculated for each tissue for each sex. 

 Likewise, ovarian accumulation of nutritional com- 

 ponents was calculated as the gain in mass from 

 November to maximal values in December or Janu- 

 ary. Since male yellowtail rockfish are sexually qui- 

 escent during this interval (Eldridge et al., 1991), 

 declines in their somatic nutritional components re- 

 flect utilization for adult metabolic 

 maintenance. Differences in net 

 losses of somatic components be- 

 tween males and females were con- 

 sidered estimates of allocations spe- 

 cific to female reproduction. 



Temporal dynamics 



Ovarian development was largely synchronous 

 within the yellowtail rockfish population from the 

 onset of recrudescence in May through late vitello- 

 genesis in November, as documented histologically 

 by Bowers (1992) and MacFarlane et al. ( 1993) (Fig. 

 1). During December, females were in late vitello- 

 genesis and migratory nucleus stages. Since fertili- 

 zation occurred in January, both late oocyte and early 

 embryonic stages were represented. In February, all 

 females were in mid to late gestation. Parturition, 

 or larval release, was completed in March when ova- 

 ries returned to a small, spent condition. 



In males, testicular development was evident in 

 August and September; in all other months testes 

 were regressed and quiescent (Fig. 1). Copulation 

 appeared to have occurred in September and Octo- 

 ber; sperm were stored in ovaries prior to fertilization. 



During the annual reproductive cycle, ovarian 

 mass increased 11 -fold from September through De- 

 cember (Fig. 1). The most significant increase oc- 

 curred during late vitellogenesis between November 

 and December. Ovaries reached maximum size by 

 January and retained mass through February. In 

 comparison, testes reached maximum size in Sep- 

 tember and were less than 5% of ovarian mass be- 

 tween December and February. 



The masses of liver, muscle, and mesenteric fat var- 

 ied significantly (P<0.01) across the annual repro- 



Results 



Size variation 



Monthly mean standard lengths 

 (±SE) for females and males were 

 36.5 ±0.7 cm and 35.3 ±0.4 cm, re- 

 spectively (Table 1). Differences in 

 length between sexes and among 

 males over the entire study interval 

 were not significant (P>0.05). Be- 

 cause there were differences among 

 monthly mean lengths of females 

 (P<0.01), analysis of covariance 

 eliminated tissue or component mass 

 variation due to size. 



