Page 791 - The Toxicology of Fishes

Ecological Risk Assessment 771

the value for each input variable is selected randomly from the applicable distribution, then risk under
that set of assumptions is calculated. This process is repeated thousands of times to generate a probability
distribution of expected risk. In general, risk predicted at the 95% confidence limit by this approach will
be lower than that predicted by simply making worst-case assumptions at each step along the way; this
is because the Monte Carlo simulation recognizes that even if the true value of one of the inputs is near
the worst case, it is unlikely that all inputs will be. For this reason, many risk assessors believe that
simulations such as Monte Carlo analysis give a much more realistic representation of risk. Such
simulations can also be used to evaluate the sensitivity of the risk assessment to various input parameters;
for example, one could decrease the uncertainty (variability) of one of the input parameters by some
margin, rerun the analysis, and determine what effect that has on the predicted risk distribution. This
can help a risk assessor determine whether collecting more data to better define a distribution is likely
to improve uncertainty in the overall risk assessment enough to warrant the additional work.
Simulation techniques are quite valuable in exploring the uncertainty in risk predictions and assessing
the relative sensitivity of risk predictions to different components of the risk calculation. Nonetheless,
one must guard against a false sense of accuracy that could be inferred from the elaborate projections
that come from these analyses. Many uncertainties in risk assessment are difficult to describe in terms
of sampling distributions (matrix effects on bioavailability of chemicals, uncertainty in lab-to-field
extrapolation) or may not even be a part of the simulation model at all (e.g., knowledge gaps).

Challenges for Toxicology to Advance Ecological Risk Assessment
The nomenclature, techniques, and applications of ecological risk assessment have advanced rapidly in
the past two decades, but significant hurdles must yet be overcome. The following text explores several
such areas that offer challenges to ecotoxicologists as we move forward.

Describing Ranges of Effect Rather Than Single Thresholds

Many risk assessments use specific benchmarks to distinguish between acceptable and unacceptable. As
described previously, risk is then expressed as a hazard quotient (ratio of exposure concentration to
benchmark concentration) or, if the exposure distribution is known, as the probability or percentage of
time that the benchmark will be exceeded (see panel A in Figure 18.4). In either case, one can make
statements about whether risk exists, but not about the specific biological effects that can be expected.
In other words, what are the expected biological differences between two systems, one with a benchmark
exceeded 5% of the time and the other with exceedances 25% of the time? Further, the magnitude of
exceedance is not considered, yet clearly this magnitude will alter the type of biological effects that can
be expected from the exposure. Where the costs of mitigative measures (e.g., wastewater treatment,
contaminated sediment removal) are not excessive, it may be possible to ignore these shortcomings and
simply manage exposures such that thresholds for effects are essentially never exceeded; however, when
expenditures are large (or even prohibitive), a more refined understanding of expected effects is necessary
to properly evaluate cost/benefit for different management alternatives.

Interpreting Sublethal Effects
As discussed above, ecological risk assessments typically focus on measures of effect that can be
explicitly linked to changes in populations or communities. Many additional measures (e.g., histological,
physiological, behavioral) may be associated with changes in populations when present at sufficient
intensity but may not be at lower intensity, and these relationships are not typically understood quanti-
tatively. Many toxicants, for example, induce cellular changes in tissues at exposure concentrations well
below those that cause measurable changes in toxicity tests measuring survival, growth, or reproduction,
making the ecological significance unclear. Moreover, at low intensities, these measures may actually
be part of mitigating against higher level effects. Enhancing our ability to quantitatively use these
additional endpoints in ecological risk assessment is a key need.