Fig. 2

Constrained analysis of seed bacterial and fungal communities’ principal coordinates. a–c Variation of bacterial communities constrained by a domestication status, b genome group, and c breeding line. d–f Variation of fungal communities constrained by d domestication status, e genome group, and f breeding line. Cumulative sum scaling (CSS)/log transformed reads were used to calculate Bray–Curtis distances. Full datasets were used to investigate the variation by domestication. Wild and domesticated rice categorization explained 6.9% (bacteria) and 13.9% (fungi) of the overall variance (PERMANOVA, p < 0.001). To analyze the effects of genome groups and breeding lines, the subset consisting of 17 wild Oryza accessions and two varieties of O. sativa (Tongil and Nipponbare) and the subset of 25 varieties were used, respectively. Rice genome group explained 24.2% (bacteria) and 20.3% (fungi) of the total variance in the subset (PERMANOVA, p < 0.001). The breeding line explained 24.7% and 17.6%, bacteria and fungi each, of the variability of seed microbiome in 25 cultivars (PERMANOVA, p < 0.001). All CAP analyses were significant (Permutest, p < 1e-5)