Bioconcentration factors for hydrocarbons and petrochemicals: understanding processes, uncertainty and predictive model performance
Camenzuli, L.; Davis, C.W.; Parkerton, T.F.; Letinski, D.J.; Butler, J.D.; Davi, R.A.; Febbo, E.J.; Paumen, M.L.; Lampi, M.A. (2019). Bioconcentration factors for hydrocarbons and petrochemicals: understanding processes, uncertainty and predictive model performance. Chemosphere 226: 472-482. https://hdl.handle.net/10.1016/j.chemosphere.2019.03.147
In: Chemosphere. Elsevier: Oxford. ISSN 0045-6535; e-ISSN 1879-1298, more
|  |
| Author keywords |
Bioconcentration; BCF; Uncertainty; PBT assessment; Hydrocarbons; Petrochemicals |
| Authors | | Top |
- Camenzuli, L., more
- Davis, C.W.
- Parkerton, T.F.
|
- Letinski, D.J.
- Butler, J.D.
- Davi, R.A.
|
- Febbo, E.J.
- Paumen, M.L.
- Lampi, M.A.
|
| Abstract |
Fish bioconcentration factors (BCFs) are often used to assess substance-specific bioaccumulation. However, reliable BCF data are limited given the practical challenges of conducting such tests. The objectives of this paper are to describe nine rainbow trout studies performed in our lab using tailored dosing and test designs for obtaining empirical BCFs for 21 test substances; gain insights into the structural features and processes determining the magnitude and uncertainty in observed BCFs; and assess performance of six quantitative structure property relationships (QSPRs) for correctly categorizing bioaccumulation given current regulatory triggers. Resulting mean steady-state BCFs, adjusted to a 5% lipid content, ranged from 12 Lkg-1 for isodecanol to 15,448 Lkg-1 for hexachlorobenzene which served as a positive control. BCFs for hydrocarbons depended on aromatic and saturated ring configurations and position. Uptake clearances appeared to be modulated by gill metabolism and substance bioavailability, while elimination rates were likely influenced by somatic biotransformation. Current approaches for quantifying uncertainty in experimental BCFs, which take into account only variability in measured fish concentrations, were found to underestimate the true uncertainty in this endpoint with important implications for decision-making. The Vega (KNN/Read-Across) QSPR and Arnot-Gobas model yielded the best model performance when compared to measured BCFs generated in this study. |
|
