the operational definition of acute or chronic toxicity may not account for certain adverse effects.
The species now used in EPA-approved bioassays were originally chosen based on the ease of culturing them and the availability of data on their sensitivities to various compounds. The same taxa were used to develop federal and state water-quality criteria. Even though these organisms were generally the most sensitive of the species evaluated, no data support the assumption that they are more sensitive than all other species in receiving ecosystems. In fact, most studies based on laboratory cultures use "weed" species—species that are easy to work with because they are relatively insensitive to changes in physical and chemical conditions. Consequently, it is likely that species evaluated as candidates for EPA-approved protocols had below-average sensitivity to changes in environmental conditions. An added complication exists when test protocols specify particular clones. Using clones increases the reproducibility of results but sacrifices information on the genetic variation within populations. Because clones can vary in sensitivity to toxins, it is unlikely that the clones used in bioassays are the most sensitive. With time, new, more-sensitive test species have been approved, but the selection process is generally driven by a combination of convenience considerations rather than a desire to identify the most sensitive species.
To compensate for the possibility that sensitive species are not protected, effluent bioassay protocols use "application factors" to calculate acceptable effluent concentrations for compounds. Application factors are essentially safety factors that reduce permitted effluent limits below those that show toxicity in bioassays (e.g., Peltier and Weber, 1985, p. 79). For example, if the LC50 (the concentration that kills 50 percent of test organisms) of an effluent is 1 part per million, a permit may require that the effluent concentration in the receiving ecosystem remain below 0.01 part per million. However, there is little biological or theoretical basis for choosing application factors. They may be either excessively or insufficiently protective (Forbes and Forbes, 1993).
Even if bioassays used the most-sensitive species as test organisms, several features of effluent bioassays would nevertheless complicate the use of test results as predictors of community- and ecosystem-level consequences of effluent releases. Most effluent bioassays measure changes in the survival, growth rate, reproduction, or behavior of individuals. However, these measures are insufficient to predict population dynamics in a taxon such as Ceriodaphnia because they do not measure density-dependent feedbacks. For example, feedback mechanisms may modify reproductive behavior at high population densities and low food concentrations, or high infant mortality may be offset by modifications in numbers of eggs produced or in the sizes of and nutrients present in individual eggs. Such effects are not incorporated into existing bioassay protocols.
Although it is difficult to make quantitative predictions of changes in population dynamics on the basis of toxicity to individuals, it is even more difficult to