Figure legends
Fig. 1. Bipartite incidence networks of Frankia OTUs withinAlnus host individuals (a) and soils (b). Samples and detectedFrankia OTUs from each sample are represented x- and y-axes, respectively. (b) gray area represents OTUs from riparian soils and white area represents OTUs from host rhizosphere soils. Colors indicates each river area (blue: BT; red: DR; green: SE; yellow: UT).
Fig. 2. (a) Non-metric multidimensional scaling of Frankiacommunity among nodules and rhizosphere of the A. hirsuta , and riparian soils. Circles indicated Frankia communities in hosts’ root–nodules. Triangles indicated the communities in rhizosphere soils. Squares indicated the communities in riparian soils. (b) FrankiaOTU richness in host’s root nodules and null model by random sampling. The asterisk represents P-value of difference of OTU richness between in nodules and null model by GLM (***: P < 0.001).
Fig. 3. (a) Correlation between genetic distance of A. hirsutaand symbiosis filtering difference. (b) Proportions of unique and shared deviance explained on symbiosis filtering difference. GD: Nei’s genetic distance of hosts (black); OTHER: measured predictors including spatial distance among hosts, soil pH, amount of inorganic nitrogen in soils, and community dissimilarity of Frankia in rhizosphere soils based on Jaccard index (gray); BOTH: both predictors (white).