752 
Fishery Bulletin 95(4), 1997 
All year classes 
Hatching Effective swimmers 
Figure 2 
Mean (±SE) percent hatching, live, effective swimmers, and spinal abnor- 
malities of larval Pacific herring by site and region in Alaska, 1995. S = Sitka, 
M = St. Mathews Bay, K = Ketchikan, F = Fish Bay, C = Port Chalmers, R = 
Rocky Bay, Y = Seymour Canal; sites in chronological order of spawning date. 
Sample size is shown in hatching graph. No significant differences existed 
between regions for any reproductive parameter (P>0.50). 
where MS = mean square. 
Somatic weight, FL, and K were ana- 
lyzed similarly. Age-class responses 
within site were compared because 
in PWS different age classes were po- 
tentially exposed to varying levels of 
oil (Table 2). When the overall 
ANOVA was significant (P<0.05), a 
priori multiple comparisons were 
used to identify which ages differed: 
pi _ between age classes 
MS error 
Maternal age was used as the stan- 
dard in all age comparisons because 
ages frequently differed in the male 
and female crosses. Age-3 and age-4 
herring were not exposed to oil at any 
life stage in PWS; therefore they were 
combined as site-specific controls. 
Few older age fish were captured; 
thus, ages >age 9 were combined and 
reported as 9+. 
Because cold storage of adult fish 
(mean time of fish capture to mean 
spawning time) varied among sites 
(0.7-12.9 h), we also examined the 
possible effect of storage time on all 
reproductive parameters with storage 
time as a covariate in the ANOVA. 
Storage times up to 7 h did not signifi- 
cantly affect any of the key reproduc- 
tive parameters (P>0.376, except 
P=0.084 for % live larvae). We repeated 
the ANOVA for regional differences 
with storage times in the model as a covariate and re- 
stricted the analysis to include only those fish sampled 
within the same time period (<7 h). 
Scored yolk-sac edema was analyzed with the 
Kruskal-Wallis nonparametric test (SAS Institute 
Inc., 1989). Yolk-sac edema was also re-expressed as 
a percentage by female, arcsin-transformed, and 
analyzed by ANOVA. Yolk volume was analyzed by 
ANOVA. 
Results 
Regional comparison 
Herring sampled from all sites appeared healthy and 
showed no obvious external signs of disease. For fish 
of the same age, there were no significant differences 
in FL (P>0.09), weight (P>0.09), or condition factor 
(P>0.41) between regions. For herring in PWS, mean 
FL ranged from 196 to 260 mm and mean weight 
from 60.7 to 151.7 g, whereas in SE, mean FL ranged 
from 198 to 253 mm and mean weight from 65.1 to 
140.4 g (Table 3). 
For all age classes combined, reproductive success 
of herring did not differ significantly between regions 
(P>0.50); the best and worst responses generally oc- 
curred in SE, whereas PWS sites were intermediate 
(Fig. 2). Statistical power of these tests was high 
(>0.99) and remained high for most analyses. Re- 
stricting the analysis to fish stored for <7 h did not 
alter the overall results; no significant (P>0.39) re- 
gional differences existed for any reproductive pa- 
rameter. In SE, mean responses ranged from 63 to 
91% for hatching success, 95 to 98% for live larvae, 
90 to 96% for effective swimmers, and 1 to 7% for 
