Fig. 2

Co-occurrence networks and linkage with CH₄ emissions. a The whole co-occurrence network structure of methanogenic community across all 39 paddy fields. The local co-occurrence network was also constructed in each field separately (Fig. S3). b The predictions of climatic variables (MAT, MAP), main soil geochemical variables (TOC, TN, TP, and pH), and methanogenic archaeal communities network and diversity to CH₄ emissions base on random forest regression analysis. R² means decision coefficient, and “% var explained” means the goodness of fit of the model. The red column indicates the factor that has a significant effect (P < 0.05; ANOVA, Duncan test), and the dark gray column indicates the factor that has no significant effect. The network index used in the model is the first component (56.9%) from the principal component analysis of eight main topological attributes of co-occurrence networks (node number, edge number, modularity, positive correlations, negative correlations, average clustering coefficient, network diameter, characteristic path length). Microbial diversity is Shannon index. c Spearman’s correlation analysis of network topological attributes and MAT. Blue shaded area is a 95% confidence interval. Correlations with other environmental factors are shown in Fig. S4