210 
Fishery Bulletin 107(2) 
The remaining embryos or larvae from a female were 
placed in a 20 ml glass vial capped with nitrogen and 
stored in a freezer at -80°C to prevent oxidation and 
tissue degradation prior to further processing. 
Moisture, protein, and ash content A subsample of 
approximately 2-4 g (wet mass) of embryos or larvae, 
representing a composite of thousands of individuals, 
was used from each sample for analysis of moisture, 
protein, and ash (inorganic components such as phos- 
phorous, calcium, and other minerals). To determine 
percent moisture, samples were placed in crucibles in a 
Leco Thermogravimetric Analyzer 601 (TGA 601) (Leco 
Corporation, St. Joseph, MI), heated to 135°C to boil off 
moisture, and wet and dry sample masses were com- 
pared. Percent ash was determined gravimetrically by 
further heating samples to 600°C to combust all organic 
components and weighing the remaining mass. 
The dry mass percent protein was calculated as the 
observed nitrogen content multiplied by a factor of 6.25, 
based on the assumption that nitrogen accounts for 
16% of the protein mass (Craig et al., 1978). Nitrogen 
content was determined following the Dumas method 
(Horwitz, 2002), using a Leco FP 528 nitrogen analyzer 
(Leco Corporation, St. Joseph, MI), with approximately 
0.1 g dry sample mass burned at 850°C and the re- 
leased nitrogen measured by thermal conductivity. 
A National Institute of Standards and Technology 
Standard Reference Material (SRM) 1546 (pork and 
chicken homogenate) was used to calibrate the Leco 
TGA 601, and Leco calibration sample ethylenediamine- 
tetraacetic acid (EDTA, 9.57 ± 0.04% nitrogen) was used 
to calibrate the Leco FP 528. Two quality assurance 
samples, Chinook salmon ( Oncorhynchus tshawytscha) 
homogenate and walleye pollock (Theragra chalcogram- 
ma) homogenate, were subjected to proximate analysis 
along with the larval samples to verify the accuracy 
of protein, moisture, and ash measurements. Repli- 
cate measurements of nitrogen content were taken as 
a check for precision, with a target error limit of less 
than 15% coefficient of variation. 
Carbohydrate content was not analyzed in this study 
because fish eggs typically have very low levels of car- 
bohydrates, averaging 2.6% of dry mass (Kamler, 1992). 
Adult fish also do not typically store carbohydrates in 
any appreciable quantities (Brett, 1995). 
Lipid content A subsample of 0.2 to 0.3 g wet mass 
containing hundreds of embryos or larvae was used from 
each of the samples for lipid analysis. Samples were 
processed by a modified Folch’s method as described 
by Christie (2003). A 2:1 solution of chloroform and 
methanol, with 0.1 g/L butylated hydroxytoluene (BHT) 
to minimize oxidation, was used to extract lipids under 
high temperature (120°C) and pressure (1200 psi) on a 
Dionex ASE 200 Accelerated Solvent Extractor (Dionex 
Corporation, Sunnyvale, CA). Extracts were washed 
with 0.88% KC1 followed by a 1:1 (by volume) metha- 
nol/deionized water solution, both added at 25% of the 
extract volume, to remove co-extractables (e.g., glycerol) 
from the solution containing the extracted lipids. The 
resulting extract volume was reduced to less than 1 mL 
by evaporating excess solvent with a Yamato RE 540 
rotary evaporator system (Yamato Scientific America, 
Inc., Santa Clara, CA), then drawn up by electronic 
pipette with sufficient chloroform to bring the volume to 
1000 qL. For gravimetric analysis of total percent lipid, 
a 500-pL aliquot of the extract was placed in aluminum 
weighing pans in a fume hood overnight, allowing the 
solvent to evaporate and leave behind the extracted 
lipids. The remaining half of the extract was capped 
with nitrogen and stored at -80°C to minimize oxidation 
until further processing for FA analysis. 
In quality assurance tests, the extraction method con- 
sistently yielded wet tissue lipid concentration values 
not exceeding 15% error compared with the certified 
value for Standard Reference Material 1946 (lake trout 
[ Salvelinus namaycush]). Three quality control samples 
were also processed concurrently with the larval sam- 
ples. A method blank containing no sample was used 
to verify that any contaminants or residues that could 
bias the observations of lipid mass were less than 0.01% 
of the average sample mass. As a check for accuracy, 
extraction of Pacific herring ( Clupea pallasii ) reference 
tissue yielded lipid concentrations that varied by less 
than 8% from the average value established in prior 
analyses. As a check for precision, one larval sample 
was split into two portions that yielded percent lipid 
values with less than 1% coefficient of variation. 
Energy estimates 
Total energy content, energy density, and the relative 
energetic contributions of protein and lipid were esti- 
mated from protein and lipid masses. Protein mass was 
expressed as its energy equivalent by calculating the 
product of protein mass and an energy density of 20.1 
J/mg, and a similar calculation was made for lipids using 
an energy density of 36.4 J/mg — figures which are con- 
versions of the average energy density values reported by 
Brett (1995). For samples having both protein and lipid 
analyses completed, these were combined to estimate the 
total energy content per individual embryo or larva, and 
expressed in relation to sample wet and dry masses to 
obtain energy density values. 
Oil globule volume 
Subsamples of 16 to 37 embryos or larvae (mean=24) 
from each female were placed in Petri dishes and photo- 
graphed digitally under a dissecting microscope. Using 
the Clever Ruler 3.0 software (shareware published by 
zcstar.com), we measured two perpendicular oil globule 
diameters for each larva from the photos. An average 
oil globule volume (OGV) for each was then converted 
to millimeters using a stage micrometer at the same 
magnification. The change in OGV was determined as 
the difference in average OGV between early embryonic 
stage samples and hatched larval samples. To describe 
trends and variability in OGV across all stages of devel- 
