In this PCA calculation, the loading plots are directly related to
the wavenumbers. They indicate the spectral features that have an
important contribution to the principal components, in other
words, the ones which influence the score plots. Fig. 2 shows the
loading plots of PC1 and PC2 and the Raman CH stretching region
of five samples representative of the different chemical families.
A high contribution of the 2845–2910 cm−1 region (A on Fig. 2) to
PC1 underlines its important role in the exploited variance. This
might be related to the high intensity of the oils spectra in this
region (high PC1 contribution on the score plot on Fig. 4) and the
weak intensity of the glues spectra (very low PC1 contribution on
the score plot on Fig. 4) for instance. Around 3080 cm−1 (B on
Fig. 2), one feature specific to the diterpenic resins spectra has a
positive important contribution to PC2 and a low and negative
contribution to PC1. This can explain the distribution of the
diterpenic resins on the score plot. Another point that can be
highlighted is the important contribution to both PC1 and PC2 of
the 2740–2800 cm−1 region and the 3100–3184 cm−1 region (C on