A cautionary note on the use of species presence and absence data in deriving sediment criteria.

In recent years, a variety of approaches to deriving sediment quality guidelines have been developed. One approach relies on establishing an empirical relationship between the concentration of a contaminant in sediment and the condition of some biological indicator, for example, combining measured sediment concentrations of contaminants combined with data on colocated benthic species to measure in situ community effects of contamination. Biological threshold concentrations derived in this manner are being considered or have already been adopted by some regulatory agencies as a means for deriving sediment guidelines (e.g., Canada's Provincial Sediment Quality Guidelines). In order to test the validity of this method, we constructed several Monte Carlo simulations to illustrate that the methodology used to develop these guidelines is flawed by the effects of sampling and statistical artifacts that emerge from undersampling a lognormal density function. As a case study, this paper will present the screening level concentration method used by the Ontario Ministry of the Environment (Toronto, ON, Canada) and provide the results of several probabilistic exercises highlighting these issues. We present a word of caution on the applicability of methods that rely exclusively on statistical and mathematical relationships between invertebrate data and sediment concentrations to derive sediment quality guidelines.