Page 784 - The Toxicology of Fishes

764 The Toxicology of Fishes

99 99
Cumulative Probability (%) 98 Cumulative Probability (%) 98
95
95
90
90
80
80
70
70
50
50
30
30
20
20
Acute AWQC
10
5
Species LC /2
2 Chronic AWQC 10 5 2 Dissolved Cu
50
1 1
10 100 10 100 1000 10000
Copper Concentration (µg/L) Copper Concentration (µg/L)
A B
FIGURE 18.4 Rather than choosing single numbers to represent exposure or effect and then calculating hazard quotients,
probability distributions can be used to express risk. In panel A, dissolved copper concentrations measured over time in a
copper-contaminated river are compared to single-value effect benchmarks (freshwater acute and chronic ambient water
quality criteria [AWQC]). This shows the percent of time that concentrations in the river exceed these values (i.e., chronic
AWQCs are exceeded about 10% of the time). In a more in-depth analysis, effect benchmarks can also be shown as
distributions. Panel B shows the same dissolved copper dataset as it compares to the distribution of acute toxicity thresholds
(estimated as the LC 50 divided by 2) for 62 species of freshwater aquatic organisms. Arrows show that 5% of the time,
dissolved copper concentrations exceed the acute toxicity thresholds for about 15% of tested species.
estimates may take a variety of forms depending on the approaches used for the exposure and effects
analyses; for example, the use of point estimates for exposure and effects analysis yields deterministic
(single-point) risk, whereas risks associated with a range of exposures can be calculated if a full exposure-
response curve is available. When single-point exposure and effect estimates are used in risk assessments,
relative risk is often expressed as a hazard quotient (HQ), which is simply the ratio of the projected
exposure concentration divided by the expected effect concentration. HQ values greater than 1 (exposure
greater than effect benchmark) suggest that effects might be expected, while values less than 1 suggest
the absence of effects. Although appealingly simple, this approach does not lend itself easily to the
incorporation of uncertainty (e.g., is HQ = 0.99 clearly a nonproblem and is HQ = 1.01 clearly a
problem?). As our ability to quantify expected ecological effects increases, the sophistication of risk
assessment procedures is also increasing. Much of the recent effort in ecological risk assessment has
focused on developing probabilistic approaches to risk expression, such as “there is a 90% chance that
growth of rainbow trout will be reduced by 20% or more” (see Figure 18.4).
Discussion of uncertainties associated with each component of the risk assessment is also an important
element of the risk characterization. The description of uncertainties may be qualitative or quantitative
and serves to convey the degree of confidence in the risk estimate (for more on uncertainty, see
Uncertainty in Risk Assessment section below). The completed risk characterization, including the risk
estimate, assumptions and uncertainties associated with the analysis, and the ecological relevance of the
findings, is used as the basis for making risk management decisions as well as for communicating risks
to interested parties and the public (see Figure 18.5).

Risk Management
Output from the four components of risk assessment discussed above is an articulation of the ecological
risks that occur, or are predicted to occur, under the conditions assessed; however, in most situations
this expression of risk is not the goal. Instead, the goal is to use this information to help decide how the
risks should be managed; for example, are the risks posed by an area of contaminated sediment sufficient
to warrant removing the sediments by dredging? The process of evaluating risk information in the context