Multivariate statistical methods for evaluating biodegradation of mineral oil

Two methods were developed for evaluating natural attenuation and bioremediation of mineral oil after environmental spills and during in vitro experiments. Gas chromatography-mass spectrometry (GC-MS) in selected ion monitoring (SIM) mode was used to obtain compound-specific data. The chromatographic data were then preprocessed either by calculating the first derivative, retention time alignment and normalization or by peak identification, quantification and calculation of diagnostic ratios within homologue series of polycyclic aromatic compounds (PACs). Finally, principal component analysis (PCA) was applied to the preprocessed chromatograms or diagnostic ratios to study the fate of the oil. The methods were applied to data from an in vitro biodegradation experiment with a North Sea crude oil exposed to three mixtures of bacterial strains: R (alkane degraders and surfactant producers), U (PAC degraders) and M (mixture of R- and U-strains) over a 1-year-period with five sampling times. Assessment of variation in degradability within isomer groups of methylfluorenes (m/z 180), methylphenanthrenes (m/z 192) and methyldibenzothiophenes (m/z 198) was used to evaluate the effects of microbial degradation on the composition of the oil. The two evaluation methods gave comparable results. In the objective pattern matching approach, principal component 1 (PC1) described the general changes in the isomer abundances, whereas M samples were separated from U and R samples along PC2. Furthermore, in the diagnostic ratio approach, a third component (PC3) could be extracted; although minor, it separated R samples from U and M samples. These results demonstrated that the two methods were able to differentiate between the effects due to the different bacterial activities, and that bacterial strain mixtures affected the PAC isomer patterns in different ways in accordance with their different metabolic capabilities.