Fig. 2: Co-expression network of the transcriptome, metabolome, and microbiome.

The k-means clustering algorithm and Pearson’s correlation analysis (two-sided; r ≥ 0.7, P values < 0.01) divided poplar gene expression profiles (red), flavonoid metabolome expression profiles (blue), and microbiome (ASVs; orange) into six clusters. The X-axis depicts 27 samples from nine poplar species, and the Y-axis depicts the Z-scores standardized for each gene, flavonoid, and ASV. The bold line is the mean. The numbers shown in each box (for example, 3672 genes, 11 flavonoids, and 865 ASVs for Cluster I) come from the number of genes, flavonoids, and ASVs for all 27 samples in each cluster. The numbers on the X-axis represent the samples: Leuce (1–3, Pto-M; 4–6, 84K; 7–9, Pal-Y; 10–12, LM50); Aigeiros (13–15, H3-1; 16–18, 107); Tacamahaca (19–21, Pot-M; 22–24, Psz-Z); Turanga (25-27, Peu-H). The genes of Cluster I were enriched in Turanga (higher expression than at least one section; |log2FC| > 1, two-sided, FDR adjusted P values < 0.05) and the genes of Cluster IV were enriched in Leuce (higher expression than at least one section; |log2FC| > 1, two-sided, FDR adjusted P values < 0.05). Source data are provided as a Source Data file.