Prediction of ecotoxicity of hydrocarbon-contaminated soils using physicochemical parameters

The physicochemical properties of eight hydrocarbon-contaminated soils were used to predict toxicity to earthworms (Eisenia fetida) and plants. The to xicity of these preremediated soils was assessed using earthworm avoidance, survival, and reproduction and seed germination and root growth in four pl ant species. No-observed-effect and 25% inhibitory concentrations were dete rmined from the earthworm and plant assays. Physical property measurements and metals analyses of the soils were conducted. Hydrocarbon contamination was characterized by total petroleum hydrocarbons, oil and grease, and GC b oiling-point distribution. Univariate and multivariate statistical methods were used to examine relationships between physical and chemical properties and biological endpoints. Soil groupings based on physicochemical properti es and toxicity from cluster and principal component analyses were generall y similar. Correlation analysis identified a number of significant relation ships between soil parameters and toxicity that were used in univariate mod el development. Total petroleum hydrocarbons by gas chromatography and pola rs were identified as predictors of earthworm avoidance and survival and se ed germination, explaining 65 to 75% of the variation in the data. Asphalte nes also explained 83% of the variation in seed germination. Gravimetric to tal petroleum hydrocarbons explained 40% of the variation in earthworm repr oduction, whereas 43% of the variation in plant root growth was explained b y asphaltenes. Multivariate one-component partial least squares models, whi ch identified predictors similar to those identified by the univariate mode ls, were also developed for worm avoidance and survival and seed germinatio n and had predictive powers of 42 and 29% respectively.