Conrath: Reproductive potential of Sebastes variabilis and S. polyspinis 
145 
Diagnostic plots of residual trends verses fitted values 
and standardized residuals versus theoretical quantiles 
(Q-Q plots) were examined visually for trends for each 
model. Generalized linear models with a binomial distri¬ 
bution were used to examine the effect of maternal size 
on reproductive failure. To test for overdispersion, each 
binomial model was repeated by using quasi-binomial dis¬ 
tribution, and the dispersion parameter was examined. 
A backward stepwise approach was used to populate the 
models with explanatory parameters that were statisti¬ 
cally significant (a=0.05). The GSI values were compared 
by site in each season. Effects of season, fish length, and 
site were considered in models fitted to relative fecun¬ 
dity data. Effects of fish length and site were considered 
in models fitted to values of fertilization failure and 
incomplete fertilization for fish captured in the month 
of May and to values of total reproductive failure for fish 
captured in May and December. Reproductive parame¬ 
ter values from the 2 sites sampled were combined if the 
site was not significantly related to the parameter being 
examined (a=0.05). The Akaike information criterion 
was used to determine which model best fit the data. The 
generalized linear models were completed by using the 
packages stats4 (vers. 3.2.0) and multcomp (vers. 3.2.5; 
Hothorn et al., 2008) in statistical software R, vers. 3.2.0 
(R Core Team, 2015). 
Results 
Catches of northern and light dusky rockfish within both 
study sites were dominated by adult individuals with very 
few immature individuals and no individuals exhibiting 
abortive maturation. No adult northern rockfish were cap¬ 
tured at the site on Snakehead Bank; therefore, compar¬ 
isons between sites were limited to data for light dusky 
rockfish captured in both seasons. During May, all mature 
light dusky rockfish collected throughout both of the sam¬ 
pled sites that lacked evidence of reproductive failure had 
ovaries that were dominated by eyed embryos. This finding 
was also true for nearly all northern rockfish; one female 
contained embryos at the embryo body stage. During 
December, all mature individuals of both species had 
ovaries that were dominated by oocytes that were either in 
the late migratory nucleus or early ovulation stage unless 
there was evidence of reproductive failure. Both species 
were found to have highly synchronous oocyte and embryo 
development, and the timing of this development was very 
similar between the 2 species. 
The GSI values for northern rockfish were generally 
higher than those for light dusky rockfish, and these val¬ 
ues for both species were higher in May prior to parturi¬ 
tion than in December (Table 1). The mean GSI for light 
dusky rockfish in May was not significantly related to site 
(P=0.71). The mean GSI for light dusky rockfish captured 
in the month of December was considerably higher at 
49 Fathom Pinnacle than at Snakehead Bank. There were 
no obvious trends in residuals in the model diagnostic 
plots for the GSL 
Estimates of fecundity were generally lower for north¬ 
ern rockfish than light dusky rockfish during both sea¬ 
sons, with higher values found during the winter (Table 1). 
Relative fecundity was calculated to compensate for the 
relationship between fecundity and size and was also 
greater for both species in December. Light dusky and 
northern rockfish had similar values of mean relative 
fecundity during each season. The mean relative fecundity 
of northern rockfish was 109.6 embryos/g for specimens 
captured in May and 165.1 oocytes/g for specimens cap¬ 
tured in December. The mean relative fecundity of dusky 
rockfish was 108.1 embryos/g for fish captured in May and 
152.1 oocytes/g for fish collected in December. Parame¬ 
ters tested in the model to explain variation in relative 
fecundity included site (for light dusky rockfish only), 
season, and FL (Table 4). Relative fecundity estimates 
were significantly higher (P<0.001) in December than in 
May for both species and increased with size for both spe¬ 
cies (Fig. 4). 
Total and partial reproductive failure occurred in 
both light dusky and northern rockfish and were found 
in both seasons. Reproductive failure was more prevalent 
and easier to discern in the specimens caught in May 
because of the advanced state of embryo development; 
therefore, only the data from the cruise conducted in 
May were used for this analysis. The proportion of adult 
individuals in which total reproductive failure had occurred 
was very similar for both species (16.3% for northern rock¬ 
fish and 15.6% for light dusky rockfish), but the proportion 
of fish undergoing partial reproductive failure was higher 
in northern rockfish (14.9%) than in light dusky rockfish 
(3.7%). The most common type of total reproductive failure 
in northern rockfish, with 68.4% of cases, was the fertiliza¬ 
tion failure type of skipped spawning, and 31.6% of cases 
of total reproductive failure were due to the resting type 
of skipped spawning. For light dusky rockfish, more than 
half of total reproductive failure was due to the resting 
type of skipped spawning (55.2%), a smaller proportion 
was due to the fertilization failure type of skipped spawn¬ 
ing (41.4%), and just one fish (3.4%) was experiencing the 
resorbing type of skipped spawning. Rates of reproductive 
failure of light dusky rockfish were not significantly differ¬ 
ent between sites; therefore, failure rates for light dusky 
rockfish from both sites were combined in this analysis 
(total reproductive failure: P-0.63; fertilization failure 
and incomplete fertilization: P=0.38). 
Larger fish of both species (>42 cm FL) were signifi¬ 
cantly less likely to experience total reproductive failure, 
particularly failure due to resting skipped spawning 
(Fig. 5). The relationship of total reproductive failure 
caused by resting skipped spawning and FL was sig¬ 
nificant for both species (light dusky rockfish: P=0.011; 
northern rockfish: P=0.044), but the relationship between 
total reproductive failure caused by fertilization failure 
was significant only for northern rockfish (light dusky 
rockfish: P=0.238; northern rockfish: P=0.002). Partial or 
total reproduction failure due to incomplete fertilization 
or fertilization failure, however, was not significantly 
related to FL for either species (light dusky rockfish: 
