Evaluation of degradation processes, particularly, biodegradation, is a key element of chemical regulation around the globe. Technical complexity and costs associated with biodegradation testing, particularly for UVCB substances, necessitates advancement of non-testing methods (e.g., quantitative structure-property relationships (QSPRs)). A critical limitation of current models is the inability to incorporate test system and environmental conditions, creating uncertainty in model relevance and reliability. This work highlights a novel model (HC-BioSIM) for predicting primary degradation rates (DT50s) of hydrocarbons integrating chemical structure as well as system-dependent parameters, using an expanded database of high-quality petroleum hydrocarbon (HC) DT50 data in water, soil, and sediment systems (N=728, 1033, & 838, respectively).
The HC-BioSIM model (RMSE=0.34-0.57; R2 = 0.52-0.68) significantly outperformed BioHCWin (RMSE= 0.75-1.2; R2=0.13-0.38) in water, sediment, & soil. Average predicted DT50 errors were reduced by 2.5x-3.9x. No significant bias in HC class, carbon number, or test system parameters were observed. K-fold cross-validations demonstrated low variability in model performance (R2, RMSE) as well as parameter importance. This strongly supports a high degree of generalizability of the model for application to external data. Additionally, the HC-BioSIM models provide improved accuracy of persistence categorization, with correct classification rates of 85-97% for water, sediment, & soil compartments. This represents a significant improvement over the existing BioHCWin model (using intermedia extrapolation factors).
For the first time, system-specific and environmental effects on biodegradation of petroleum HCs can be quantitatively evaluated in water, sediment, & soil systems. The model reduces uncertainty and provides a better basis for assessment of environmental degradation for hydrocarbon substances. The model can be applied broadly across regulatory frameworks for risk management. The curation of high quality sediment & soil databases improves understanding of the relative persistence properties of hydrocarbons between environmental media, including assessment of the intermedia extrapolation factors used in persistence assessments and exposure models.