Johnson et al.: Reproductive success of Clupea pallasi after the Exxon Valdez oil spill 
751 
the 1989 year class were present; therefore the num- 
ber of larvae analyzed per female was doubled. Sitka 
and St. Mathews Bay were excluded because of an 
insufficient number of females from the 1989 year 
class. Lateral views of larvae were displayed digi- 
tally, and specimens were rotated to align eyes in 
order to minimize variance. Yolk shapes were gener- 
ally elliptical; major and minor axes were measured 
perpendicular to the body axis. Yolk volume was es- 
timated from these linear measures according to the 
method of Hourston et al. ( 1984). Yolk-sac edema was 
indicated if the anterior margin of the yolk membrane 
was bounded by an area of clear fluid. Pericardial 
edema was scored if the pericardium was unusually 
large or convex ventrally. 
Data processing and statistics 
To assess the general health of parent fish, condi- 
tion factor (K) was calculated for each female accord- 
ing to the method of Bagenal and Tesch (1978): 
„ 100(W) 
A = r 9 
FL h 
where W = somatic wet weight in g; 
FL = fork length in cm; and 
b = the value determined by site from 
length-weight regressions. 
Gonad weight was substracted from body weight to 
avoid variation in spawning condition. 
Times of hatching among sites, which were tem- 
perature dependent, were compared by using peak 
hatching times as the estimator. Peak hatching day 
was defined as the day the most larvae hatched from 
eggs of a given female; if two hatch peaks of equal 
magnitude occurred, the first peak was reported. 
Mean incubation temperature for eggs from each fe- 
male was calculated by weighting mean water-bath 
temperatures by the number of eggs hatched daily. 
This method avoided possible under or over estimates 
of mean incubation temperature caused by early or late 
hatching of eggs as seasonal temperature increased. 
Most observations were expressed as percentages. 
The denominator used to calculate percentages var- 
ied by response parameter (Table 1). Percentages of 
eggs fertile and initially dead were based on the to- 
tal number of eggs counted near the beginning of the 
experiment. Percentages of eggs that hatched were 
based on the total number of hatched larvae plus the 
number of dead eggs determined at the endpoint. The 
number of hatched larvae was subdivided into num- 
ber live, moribund, and dead. Hearts of moribund 
larvae were beating, but these larvae were incapable 
Table 1 
Description of key response parameters used to evaluate 
reproductive impairment in Pacific herring collected from 
Prince William Sound and southeast Alaska. Herring were 
collected in 1995, artificially spawned, and reared in a labo- 
ratory until hatching. Moribund larvae were alive (heart 
beating) but incapable of swimming. 
Parameter (%) 
Description 
Hatch 
100 • (total number of eggs that 
hatched) /(total number of eggs 
that hatched + total number of 
dead eggs) 
Live (viable) 
100 ■ (total number of live larvae 
excluding moribund larvae)/ (total 
number of eggs that hatched) 
Effective swimmers 
100 ■ (total number of effective 
swimmers) / (total number of live 
larvae excluding moribund larvae) 
Spinal abnormalities 
100 • (number of live + moribund 
larvae with spinal defects) / (total 
number of live + moribund larvae) 
of movement. Accordingly, percent live was the num- 
ber of living larvae (excluding moribund larvae) di- 
vided by the total number hatched. Swimming of live 
larvae was categorized as effective, ineffective, or 
incapable. Effective swimmers were active, fre- 
quented the water column, and avoided capture. In- 
effective swimmers were generally more lethargic 
than effective swimmers and were more likely to be 
found on jar bottoms. Incapable swimmers were un- 
able to swim in a straight line and were often ca- 
pable only of spasmodic twitching. Swimming of 
moribund and dead larvae was, by definition, non- 
existent; thus the number of live larvae was used as 
the denominator for swimming categories. Because 
larvae quickly became distorted after death, spinal 
aberrations were assessed only in live and moribund 
larvae. Percentage of spinal abnormalities, therefore, 
was the number of larvae with spinal aberrations 
divided by the sum of live and moribund larvae. 
One-way analysis of variance (ANOVA) was used 
to examine differences among sites, among age 
classes, and between regions. Each reproductive pa- 
rameter was tested separately by individual age class 
and for all age classes combined; percentage data 
were arc-sine transformed and corrected for small n 
as necessary (Snedecor and Cochran, 1980). To ac- 
count for variance among sites, the E-test compari- 
son between regions was: 
jp IMS between regions 
IMS 'among sites 
