Metabolomic Profiling of Hormone Treatments

To assess general statistical importance of hormone treatments in the metabolite profiles, a Principal Component Analysis (PCA) was performed on the normalized data (Fig. 2). The hormone type appeared to be responsible for the variation observed in PC1 (which explained 17.7%) while variation between the specific phytohormones and the treatment concentrations appeared to be responsible for PC2 (9.18% explained variance). Overall, a clear separation between growth hormones (pink-purple) and stress-response hormones (blue-green) was observed. A PCA of the full feature table was also conducted and a similar separation was observed (Supplementary Information).
Fig. 2

PCA of the normalized feature table. Circles were added manually with growth treatments (coloured pink-purple) circled in pink and stress-response treatments (coloured blue-green) circled in blue. As the broken stick test revealed that the PC1 and PC3 axes explained more variance than would be expected by randomly dividing the variance into parts, PC1 and PC3 were chosen for the plot. Results of the broken stick test are available in the Supplement and Zenodo

Analysis of the chemodiversity showed unique patterns with respect to the features detected in each treatment (Fig. 3). The number of features detected was consistent between all the treatments except for NAA10 which had significantly fewer features (Fig. 3a). The Pielou’s evenness (J) for all treatments was consistent, with the exception of SA1 and NAA10 which had significantly higher and lower J values, respectively (Fig. 3b). No significant differences were observed in the Shannon diversity (H’) (Fig. 3c). Determining the Functional Hill diversity, which also does take the dissimilarity of features into account (Petrén et al., 2022 ), showed that the significantly lower diversity in NAA10 was explained by structurally similar features with less abundance than the other treatments and a slightly higher dissimilarity of features in BAP10, G, and S (Fig. 3d).
Fig. 3

Diversity indices of the features detected in the MS1 data. A: number of features, B: Pielou’s evenness (J), C: Shannon diversity Index (H’). D: Functional Hill diversity. S: stress-response control, G: growth control