A case study comparing static and spatially explicit ecological exposure analysis methods.

Abstract

Exposure to chemical contaminants must be estimated when performing ecological risk assessments. A previous article proposed a habitat area and quality conditioned population exposure estimator, E[HQ]P, and described an individual-based, random walk, Monte Carlo model (SE3M) to facilitate calculation of E[HQ]P. In this article, E[HQ]P was compared with exposure estimates from a baseline risk assessment that evaluated mink and great blue heron exposure to fluoride at a federal Superfund site. Calculation of E[HQ]P took into consideration a receptor's forage area, movement behavior, population size, and the areal extent and quality of suitable habitat. The baseline assessment used four methods that did (total and unit Tier 2) and did not (total and unit Tier 1) consider habitat area or quality; where "total" included all exposure units on site and "unit" only a given exposure unit. Total Tier 1 estimates were consistently higher than E[HQ]P (e.g., 169.1 mg/kg x d versus 21.6 mg/kg x d). Risk managers using total Tier 1 results for decision making would be unlikely to underestimate exposure; however, implementability of correspondingly lower remedial objectives could be challenging. Unit Tier 1 estimates were higher (e.g., 96.5 mg/kg x d versus 61.6 mg/kg x d) or lower (e.g., 3.5 mg/kg x d versus 51.1 mg/kg x d) than E[HQ]P depending on variations in landscape features. Total Tier 2 and E[HQ]P estimates were similar (e.g., 20.7 mg/kg x d versus 21.6 mg/kg x d) when an ecologically questionable average exposure was assumed. Unit Tier 2 estimates were consistently well below E[HQ]P (e.g., 17.8 mg/kg x d versus 61.6 mg/kg x d) when an average exposure was not assumed. Risk managers using unit Tier 1 or 2 results could be basing their decisions on potentially large underestimates of exposure. By forgoing average exposure assumptions, and explicitly addressing landscape heterogeneity, SE3M appears capable of yielding exposure estimates that are not as potentially misleading to risk managers as those produced with traditional averaging methods.