Figure 5. Scores plot of PC-11 vs. PC-2 for Ml , Se , andSa as neat species.
After understanding the behavior of the Ml , Sa , andSe bacteria as neat species, the spectra for the mixtures were evaluated. Models shown in Figure 6 were generated to provide a clear view of a complex environment where a combination of these species might appear on the surface. Figure 6A shows the PCA model for theSa/Ml mixture with a PC-1 score of 53% and a PC-2 score of 25%. An SNV pre-processing algorithm was applied to achieve class separation. The vibrational signatures of the mixture show a class that lies between the neat substances of the mixture. A difference is shown for the Se class, which appears distanced from the mixture and components. The same analysis was carried out for the Se/Sa mixture, Figure 6B, where the mixture classification showed a tendency for the Se class. The mixture is separated from the Ml component. Finally, a third PCA model was generated for the Se/Ml , shown in Figure 6C, by applying SNV and SG2 pre-processing algorithms. A second pre-processing step was expected since Se showed a higher turbidity than Ml . The effect may be attributed to a lower bacterial Ml concentration. The model also showed a tendency for the mixture towards the Se neat component. A further evaluation was performed by analyzing the models’ contributing signals. This analysis required comparing the vibrational signature of the neat species with the loadings spectral values used for the PCA model.
(A)