Evaluating methods for assessing sediment quality in a Great Lakes embayment

A probability-based, sediment quality assessment was conducted during 1995 in the lower St. Louis River Area of Concern, located in western Lake Superior. A regional application of the intensified sampling grid developed for the United States Environmental Protection Agency's Environmental Monitoring and Assessment Program was used to randomly select 90 sites for measuring the following sediment quality indicators: sediment chemistry, physical parameters, sediment toxicity, and benthic macroinvertebrate community structure. Screening methods were used to assess sediment chemistry and sediment toxicity at all sites, whereas more conventional metrics were used at a subset of sites. In addition, sediment quality data were collected from 20 a priori training sites, 10 in low impact areas and 10 in high impact areas. Mean probable effect concentration quotients were calculated for sediment chemistry variables at each site. As the range of mean probable effect concentration quotients values increased, the incidence of sediment toxicity increased. Benthic data from the training sites were used to establish standard criteria for developing two benthic integrity indices based on multimetric analysis and discriminant function analysis. Based on the training site results, the discriminant function analysis categorized the macroinvertebrate community at all random sites as 45 percent low impact and 55 percent high impact. A multimetric approach categorized 55 percent of the random sites as low impact and 36 percent as high impact. Due to the overlap of 95 percent confidence intervals, the multimetric approach also placed 9 percent of the random sites into an indeterminate category. The incidence of high impact sites appears to be primarily due to physical habitat characteristics. This finding was supported by the sediment quality triad assessment of 52 random sites that indicated alteration of the benthic community at 71 percent of sites was probably not due to chemical contamination.