Molecular formula characteristics and results of the principal component analysis of the FTICR-MS data. Formulae were grouped a) in a frequency plot to show the number of formulae that were detected in each of the various samples. The van Krevelen diagrams in panels b) and c) show all 3751 unique formulae in black points compared to b) the 844 formulae unique to any single sample (pink) and to c) the 560 formulae common to all 8 samples (blue). The PCA biplot shows the d) samples’ scores and e) variables’ loadings, along with f) the van Krevelen diagram color-coded according to the boxes drawn in the loadings biplot. Specific characteristics of the formulae responsible for the PCA variance are given in Table 4.