38 
Fishery Bulletin 1 10(1 ) 
To identify structures within embryos and preparturi- 
tion larvae, the samples were placed in a petri dish 
with a small amount of fresh water and raised above 
the microscope stage less than 2.5 cm. An object was 
placed under the dish to block some light coming from 
the stage below to provide proper shading for identi- 
fication of internal structures. Often this procedure 
worked best when the light was covered under half of 
the embryo. Oil globule diameter was calculated by 
measuring two perpendicular bisecting diameters and 
averaging them. 
We followed Yamada and Kusakari’s (1991) criteria 
for developmental staging of Sebastes schlegeli embryos 
and preparturition larvae and added additional charac- 
teristics, such as eye and body pigmentation, to further 
divide late stages (Table 1, Fig. 2). One developmental 
Table 1 
Descriptions of developmental stages observed in quill- 
back rockfish (Sebastes maliger ) embryos and pre-partu- 
rition larvae and the corresponding stages from Yamada 
and Kusakari (1991) (Y&K). 
Stage 
Y&K 
Description 
1 
14 
Late gastrula 
2 
16 
Head fold 
3 
17 
Optic vesicle 
4 
20 
Optic cups 
5 
21 
Auditory placodes 
6 
22 
Lens 
7 
23-25 
Otoliths, heartbeat, black pigment in 
retina and iris is strongest in periphery. 
8 
25 
Entire retina and iris translucent 
black, small, black spots of pigment on 
ventral side of tail. 
9 
25 
Black retina with scattered silver pig- 
ment in iris, spots on ventral side of 
tail darkened and multiplying. 
10 
29 
Iris silver but black still visible 
throughout, dark spots of pigmenta- 
tion on gut and peritoneal wall, yellow 
pigment may be present on tail. 
11 
29 
Iris is completely silver, yellow pig- 
ment on tail, dark ventral pigment 
may have spread to form a line, lower 
mandible appears as a nub, but is not 
detached; when mechanically stimu- 
lated, will respond with a twitch. 
12 
28 
Lower mandible detached and mouth 
open, yellow pigment appears on top of 
the head, embryos hatch easily when 
disturbed and are able to swim, the top 
of the cranium is a defined bulb. 
13 
28 
Lower jaw becomes angular and 
defined, lower mandible opens and 
closes in a gulping motion. Embryos 
may be hatched. 
code was assigned to each female because for nearly ev- 
ery female all embryos were at the same developmental 
stage. In the few cases where more than one stage was 
present, owing to one group of embryos being unhealthy 
and arrested in development, only healthy embryos were 
photographed and analyzed. 
Fecundity measurements were taken from fish sam- 
pled near Little Port Walter in 2008 to examine the 
relationships between weight-specific fecundity and 
maternal age, length, and weight. Fecundity estimates 
were determined by the gravimetric method where sub- 
samples of the ovary are related to ovarian weight (e.g., 
Jennings et al. 2001). 
Analysis 
Because gravid females were sampled at different times 
and embryos were at varying stages of development, 
these factors had to be considered when examining the 
relationship between OGD and maternal age, length, 
or weight (maternal factor). OGD was related to the 
developmental stage (i.e., OGD decreases as the embryo 
develops). This trend was similar among sampling 
periods. Our data also showed that younger, smaller 
fish have earlier stage embryos (within a sampling 
period) than older, larger females. This finding indi- 
cated that older, larger females develop larvae earlier 
than younger, smaller ones. Because developmental 
stage is confounded with both the maternal factors and 
OGD, comparisons across fish at different developmental 
stages can mask any relationship between the maternal 
factors and OGD. Therefore, it is necessary to remove 
the stage effect from both the maternal factors and the 
OGD to reveal the effects of the maternal factors on 
OGD. Our approach was to develop adjusted measures 
of OGD and maternal factors that removed the stage 
effect and to use the adjusted measures to visualize and 
statistically test the relationships between OGD and a 
given maternal factor. Alternatively, a general linear 
model (GLM) with OGD as the dependant variable, 
and with stage, maternal factor (either age, length, 
or weight), sampling period, and interaction terms as 
independent variables could be used to account for these 
confounding relationships. However, significant interac- 
tion terms in a GLM require that separate models be 
run for each factor (Lehman et ah, 2005). With multiple 
significant interactions, a multitude of models would 
be required. 
The OGD observations were adjusted by subtracting 
the expected OGD based on polynomial expression for 
stage, 
OGD = intercept + /3 ; S ( + jS 2 S ( 2 , (1) 
where S ( = the stage of all embryos and pre-parturition 
larvae within a female. 
The intercept and /3 parameters were estimated regres- 
sion coefficients. 
Each maternal factor was also adjusted to eliminate 
