Tier IV chronic bioassays

subtle sublethal end points related to the maintenance of a viable population. The

interpretation of chronic bioassays should clearly define the relationship between

chronic sublethal end points, such as growth, to population end points, such as

survival and reproduction (Moore and Dillon 1993).

Population or demographic modeling is one method that can be used to

interpret the results of chronic bioassays (U.S. Army Engineer Waterways

Experiment Station (USAEWES) 1993). These models use life history

information (e.g., survival, growth, reproductive rates) to make projections about

potential population-level impacts (e.g., abundance over time). Using these

models in conjunction with chronic bioassay results, one can compare estimated

rates of population increase for dredged material and reference sediment.

Population modeling requires information on survivorship and fecundity of

the target population, information that can be obtained from well-designed

laboratory studies. In the environment, however, species are subject to both biotic

and abiotic forces contributing to their ability to reproduce, grow, and survive.

As with previous tiers, Type I and Type II error is involved in the comparison

between reference and dredged sediments. In addition, the analytical form of the

population model introduces model uncertainty, and estimates of survivorship

and fecundity introduce both variability and parameter uncertainty. Variability

can typically be addressed in the population model by using a stochastic

framework. One existing approach (Ferson 1991) models environmental

stochasticity through (a) random fluctuations in age- or stage-specific fecundities

and survivorships, (b) random fluctuations in carrying capacities (the maximum

sustainable population), (c) random fluctuations in dispersal rates, and (d) two

types of local or regional catastrophes.

The variability of each vital rate and carrying capacity is modeled with a

standard deviation. Each population can have a separate set of standard

deviations. The random fluctuations can be normally or lognormally distributed,

and can be correlated among populations. Within a population, survivals,

fecundities, and carrying capacities can be uncorrelated, perfectly correlated, or

negatively correlated. Descriptive statistics such as standard deviations on the

vital rates are obtained from the bioassay data; however, these values incorporate

both uncertainty (measurement error) and variability (population heterogeneity).

First-order uncertainty analysis within a matrix framework can be used to

evaluate confidence in the predictions. Other options available for evaluating

uncertainty include the following:

a. Examine "within year" or "within season" measurements to estimate

measurement error.

b. Examine "among year" or "among season" measurements to estimate

variability.

c. Conduct a sensitivity analysis on each of the parameters in the

population model to evaluate the range of possible results predicted by

the model.

Chapter 4 Uncertainty in Tiered Evaluation of Dredged Material