Nonparametric identification of petrogenic and pyrogenic hydrocarbons in aquatic ecosystems.

Novel nonparametric models developed herein discriminated between oiled and nonoiled or pyrogenic and oiled sources better than traditionally used diagnostic ratios and can outperform previously published oil identification models. These methods were compared using experimental and environmental hydrocarbon data (sediment, mussels, water, and fish) associated with the Exxon Valdez oil spill. Several nonparametric models were investigated, one designed to detect petroleum in general, one specific to Alaska North Slope crude oil (ANS), and one designed to detect pyrogenic PAH. These ideas are intended as guidance; nonparametric models can easily be adapted to fit the specific needs of a variety of petrogenic and pyrogenic sources. Oil identification was clearly difficult where composition was modified by physical or biological processes; model results differed most in these cases, suggesting that a multiple model approach to source discrimination may be useful where data interpretation is contentious. However, a combined nonparametric model best described a broad range of hydrocarbon sources, thus providing a useful new analytical assessment tool.