Predictions of sediment toxicity using consensus-based freshwater sediment quality guidelines

The objectives of this study were to compare approaches for evaluating the combined effects of chemical mixtures on the toxicity in field-collected sediments and to evaluate the ability of consensus-based probable effect concentrations (PECs) to predict toxicity in a freshwater database on both a national and regional geographic basis. A database was developed from 92 published reports, which included a total of 1,657 samples with high-quality matching sediment toxicity and chemistry data from across North America. The database was comprised primarily of 10- to 14-day or 28- to 42-day toxicity tests with the amphipod Hyalella azteca (designated as the HA10 or HA28 tests) and 10- to 14-day toxicity tests with the midges Chironomus tentans or C. riparius (designated as the CS10 test). Mean PEC quotients were calculated to provide an overall measure of chemical contamination and to support an evaluation of the combined effects of multiple contaminants in sediments. There was an overall increase in the incidence of toxicity with an increase in the mean quotients in all three tests. A consistent increase in the toxicity in all three tests occurred at a mean quotient > 0.5, however, the overall incidence of toxicity was greater in the HA28 test compared to the short-term tests. The longer-term tests, in which survival and growth are measured, tend to be more sensitive than the shorter-term tests, with acute to chronic ratios on the order of six indicated for H. azteca. Different patterns were observed among the various procedures used to calculate mean quotients. For example, in the HA28 test, a relatively abrupt increase in toxicity was associated with elevated polychlorinated biphenyls (PCBs) alone or with elevated polycyclic aromatic hydrocarbons (PAHs) alone, compared to the pattern of a gradual increase in toxicity observed with quotients calculated using a combination of metals, PAHs, and PCBs. These analyses indicate that the different patterns in toxicity may be the result of unique chemical signals associated with individual contaminants in samples. Though mean quotients can be used to classify samples as toxic or nontoxic, individual quotients might be useful in helping identify substances that may be causing or substantially contributing to the observed toxicity. An increase in the incidence of toxicity was observed with increasing mean quotients within most of the regions, basins, and areas in North America for all three toxicity tests. The results o