208 
Fishery Bulletin 107(2) 
Table 1 
Sample sizes for determinations of body composition and fatty acid profiles of quillback rockfish ( Sebastes maliger) embryos (early 
and middle stages) and hatched, preparturition larvae (late stages). The sample unit was one maternal female, from which sub- 
samples of embryos or larvae were obtained for use in biochemical analyses. Sample sizes varied due to inadequate subsample 
masses being available for some analytical procedures. 
Variable 
Sample size 
Early stages (1-3) 
Middle stages (4-9) 
Late stages (10) 
Developmental stage 
5 
6 
4 
Oil globule volume 
3 
4 
4 
Wet tissue mass 
3 
4 
4 
Moisture, protein 
3 
4 
4 
Ash 
2 
4 
3 
Lipid 
3 
4 
4 
Fatty acids 
4 
4 
4 
in Watanabe, 1982). Fish are capable of selectively ca- 
tabolizing particular fatty acids (FAs) while retaining 
others (reviewed in Tocher, 2003). Differences in rates 
of individual FA use during embryogenesis would be 
reflected by changes in overall FA profiles as embryos 
develop into hatched larvae. Assessing net differences 
in the amounts of individual FAs present may reveal 
which FAs potentially contribute to variability in larval 
survival (e.g., due to deficiencies in particular EFAs 
resulting from inadequate maternal provisioning). 
Our study was driven by three objectives. First, we 
sought to describe the amount and sources of energy 
consumed during quillback rockfish embryogenesis, 
by measuring changes in lipid and protein levels from 
early to late stages of development. Second, to assess 
the usefulness of OGV as an indicator of the energetic 
status of embryos and preparturition larvae, we inves- 
tigated how well changes in OGV were correlated with 
changes in stage and biochemical composition. Last, we 
reduced lipids to their FA components and compared the 
overall FA profiles of embryos to preparturition larvae, 
to determine whether all FAs were used at the same 
rate as the total lipid during embryonic development, or 
whether some were used disproportionately fast while 
others were conserved. 
Methods 
Sampling 
Quillback rockfish were caught 15-28 April 2006 by 
hook and line in southeastern Alaska on the northwest 
side of Chichagof Island (58°10'N, 136°21'W). Fish were 
caught within approximately 1 km of shore at depths of 
30 to 75 m. Fifteen gravid females ranging in size from 
360 to 480 mm (fork length) were transported live to 
Auke Bay Laboratory in Juneau, where they were kept 
in flow-through seawater tanks at 3.5-4°C. During a 
2-week holding period, the females did not feed and 
did not release larvae naturally. Females were then 
sacrificed and a sample of embryos or larvae was manu- 
ally expressed from each fish. Sample sizes available 
for biochemical analysis varied occasionally because 
each analytical procedure was destructive and required 
separate subsamples of embryos or larvae, and sample 
masses were below the minimum needed to ensure 
accurate analysis in some cases (Table 1). One sample 
of stage seven embryos was omitted from analysis due 
to the apparent degradation and possible resorption of 
embryos by the parent. 
Changes in lipid and protein levels during development 
Developmental stages We ranked embryos or larvae 
from each female in order of development (stages 1-10, 
from immediately after fertilization through posthatch- 
ing; Fig. 1) following the descriptions of kurosoi rockfish 
(S. schlegelii ) by Yamada and Kusakari (1991), and 
incorporating our own observations for quillback rockfish 
(Table 2). In quillback rockfish, we found that the retina 
went through many stages of pigmentation and that 
body pigment appeared relatively early in development 
and became more pronounced through time. Yamada 
and Kusakari (1991) include only one stage for retinal 
pigmentation and one for peritoneal pigment (stages 25 
and 29, respectively), so we further divided the embryo 
stages based on these characteristics. 
Developmental stages were then used for tracking 
changes in body composition during embryonic devel- 
opment. Because the durations (in days) of stages vary 
widely (Eldridge et al., 2002), they are not strictly ap- 
propriate for statistical analyses with linear models. 
In any model using developmental stage categories, 
the assumption that the stages represent equal inter- 
vals can distort the true patterns of change over time. 
Quantitative statements about rates of change in body 
composition ideally would be based on time since fer- 
