The Exxon Valdez Oil Spill Trustee Council concluded that oil caused mortal
ity of pink salmon (Oncorhynchus gorbuscha) eggs in Prince William Sound st
reams. Their conclusion was based primarily on Alaska Department of Fish an
d Game (ADF&G) studies which reported that mean mortality of embryos in egg
s was higher in oiled than non-oiled streams when sampled shortly after spa
wning completion. However, developing embryos are vulnerable to shock morta
lity for a period of 20 days after fertilization, and the embryos in eggs f
rom the latest spawners were still in the sensitive period at the time samp
ling took place. We argue that the original ADF&G analysis should have incl
uded sample timing in statistical comparisons of mortality between streams.
Analysis of a subset of the ADF&G data showed that sampling shock was a ma
jor source of embryo mortality in these samples, and that source of mortali
ty in the original survey would likely have been mistakenly interpreted as
an oiling effect. Compensating for sample timing removed all statistical ev
idence for an oiling effect in the data subset. We conclude that the ADF&G
study design confounded the ability to assess for the effect of oil exposur
e on pink salmon eggs.