You are here:
Development of Interspecies Correlation Models for Petroleum Hydrocarbons
Citation:
Bejarano, A. AND M. Barron. Development of Interspecies Correlation Models for Petroleum Hydrocarbons. SETAC North America 35th Annual Meeting, Vancouver, BC, CANADA, November 09 - 13, 2014.
Impact/Purpose:
In this study, ICE models are developed and validated for oil and dispersant products.
Description:
Estimating the consequences of petroleum products to water column organisms has commonly been hampered by limited acute toxicity data, which exists only for a relatively small number of test species. In this study, we developed petroleum-specific Interspecies Correlation Estimation (ICE) models, which are mathematical relationships between surrogate and predicted test species, with the goal of facilitating prediction of toxicity to a broader number of aquatic species. These analyses showed that compared to ICE models developed for non-specific groups of chemicals, these petroleum ICE models have a greater adjusted coefficient of determinations (adj-R2), lower Mean Square Errors (MSE) and a greater cross-validation success rate, indicative of greater a predictive power. For instance, most model cross-validation values (>90%) were within 2-fold of the measured values, which fall within the fold difference commonly found during inter-laboratory comparisons with the same species. One of the added benefits of developing petroleum ICE models is that these facilitate the construction of ICE-based Species Sensitivity Distributions (SSDs), which are probabilistic distributions of toxicity data across species. SSDs are useful in that protective levels or hazard concentrations (HC) values, assumed to be protective of a large number of species, can be derive. Comparison of several empirically based SSDs with ICE-based SSDs showed that HC values between these two approaches were within the same order of magnitude of each other. The development of both petroleum ICE models and ICE-based SSDs has an immediate applicability to assessments of the environmental consequences of petroleum products and oil spill to water column organisms. Specifically, these predictive models can provide information when the costs of toxicity testing are prohibitive, or when species-specific toxicity testing is restricted or not feasible.