Page 879 - The Toxicology of Fishes

Reproductive Impairment of Great Lakes Lake Trout by Dioxin-Like Chemicals 859

The temporal association of either the lack of recruitment or the symptoms of dioxin-like toxicity
with HAH contamination in Great Lakes salmonines has been consistent and supports the hypothesis.
The best example of temporal consistency of the association has come from the retrospective risk analysis
of HAH exposure in lake trout from Lake Ontario (Cook et al., 2003). The declines in lake trout
populations during the 1930s in Lake Ontario were temporally consistent with elevated HAH exposure
(Figure 21.9). Additionally, declines in sac fry mortality during the late 1970s and early 1980s were
temporally associated with a decline in HAH exposure of lake trout (Figure 21.9). Further evidence of
a consistent association between HAH exposure and dioxin-like toxicity observations in Great Lakes
lake trout over time comes from the observations of fry and sac fry mortalities in lake trout from Lake
Michigan (Mac and Edsall, 1991). These authors reported consistent improvements in hatchability and
sac fry mortality in lake trout from Lake Michigan over the period of 1975 to 1984. This was at a time
when the concentrations of PCBs were declining in the aquatic ecosystem of Lake Michigan in general
and lake trout from Lake Michigan in particular (Figure 21.3).
The geographic distribution of symptoms of dioxin-like toxicity in lake trout was consistent with the
relative degree of contamination in the Great Lakes. Dioxin-like effects were observed in early life stages
of lake trout in all three of the lower Great Lakes with historic populations, while problems of reduced
hatchability, or elevated sac fry mortality were not observed in progeny of Lake Superior lake trout (Mac
and Edsall, 1991). Thus, consistency of association is supported based on the geographic distribution of
elevated contaminant burdens, particularly PCBs and dioxins (DeVault et al., 1989; Hickey et al., 2006)
and the observations of dioxin-like endpoints of toxicity in developing fry from those lakes (Cook et
al., 2003; Mac and Edsall, 1991; Mac et al., 1985, 1993).
Consistency of association is also supported by the fact that the same signs of toxicity were observed
in different species at the same locations. Skea and coauthors (1985) observed sac fry mortality consistent
with dioxin-like toxicity in rainbow trout, Chinook salmon, and lake trout from Lake Ontario. Researchers
from New York observed similar effects in lake trout and rainbow trout collected in this same time period
(Cook et al., 2003). Similarly, the Michigan Department of Natural Resources reported that Chinook
salmon fry from Lake Michigan were observed to have similar stage-specific mortality from 1979 to
1981 (Mac and Edsall, 1991), consistent with the observations in lake trout at the same time (Mac et
al., 1985, 1993). The fact that multiple species from the same locations expressed elevated fry mortality
supports the consistency of association criterion.

Predictive Performance
The criterion of predictive performance requires that a deductive hypothesis drawn from the association
can predict an unknown fact or consequence (Fox, 1991). In two cases, the association of HAH contam-
ination in Great Lakes lake trout and the lack of recruitment observed during the last half of the 20th
century led to predictions that were confirmative of this relationship. One of the predictive models that
strongly supported this association was the retrospective analysis of Lake Ontario outlined above (Cook
et al., 2003). In that case, sediment core HAH profiles were predictive of the population levels of lake
trout, the signs of sac fry mortality, and reductions in fry survival. This is a strong affirmation of the
association, and provided a predictive model for the outcomes observed in lake trout from Lake Ontario.
This criterion was also satisfied through the fulfillment of Koch’s postulates. Heinrich Herman Robert
Koch, the noted German bacteriologist and epidemiologist, published a series of four postulates that,
when satisfied, provide evidence for causality between an agent and a disease. These postulates have
been adopted by environmental toxicologists to help identify causal linkages between chemical contam-
inants and adverse effects in exposed organisms (Giesy et al., 1994). Adapted for toxicology, Koch’s
postulates state that: (1) an agent (chemical) must be present in the affected organisms, (2) the agent
must be correlated with the observed symptoms of the disease, (3) the agent must be isolated and identified
(measured) from the affected organism, and (4) the agent must be introduced into healthy individuals
and cause the untoward symptoms of the disease. These postulates have been satisfied through the isolation
and extraction of chemicals from lake trout from Lake Michigan and the reintroduction of these extracts
into healthy, freshly fertilized trout eggs (Tillitt and Wright, 1997; Wright and Tillitt, 1999). The extract
of lake trout from Lake Michigan produced graded, dose-dependent symptoms of blue sac syndrome in
lake trout embryos from the hatchery (Tillitt and Wright, 1997) and rainbow trout embryos (Wright and