Statistics and data analysis
Data were statistically analysed with SPSS v. 21 (IBM Corp., Armonk, NY, USA). The microbial-based biostimulant effects were analysed by an unpaired Student’s t -test. The compound intensity table exported from MSDIAL v. 4.0 (Riken, Tokyo, Japan) (Tsugawa et al., 2015) was uploaded into MS-FLO (Riken, Tokyo, Japan) (DeFelice et al., 2017) to reduce false positives and duplicates. The output was imported into R v. 3.6.0 for centring, scaling, PCA, and calculation of fold changes, Benjamini-Hochberg false discovery rates (FDR), and corrected p-values. Venn diagrams were plotted to identify metabolites common to two sampling points but not exclusive to a particular growth stage. Compounds with P < 0.05 were imported into ChemRICH (Barupal & Fiehn, 2017) for enrichment analysis based on their chemical similarity and MetaMapp (Barupal et al., 2012) for chemical network analysis. Cytoscape (Saito et al., 2012) displayed exported MetaMapp data and plotted the final images.