Host genetic variation in symbiosis filtering
Symbiosis filtering differences were significantly correlated with host genetic distance (Fig. 3a). This correlation remained significant even when spatial distance among alder individuals was corrected using a partial Mantel test (r = 0.444, P < 0.001). Furthermore, GDM analysis quantified the explanatory proportions of the genetic distances of the hosts and other predictors, including spatial distance among hosts, soil pH, amount of inorganic nitrogen in soils, and community dissimilarity of Frankia in rhizosphere soils, for symbiosis filtering differences (Fig. 3b). The genetic distance of alders and other factors absolutely explained the 36.64% deviance in the symbiosis filtering differences.
Where the deviance explained by the model with genetic distance and other factors was 100 %, the genetic distance of alders uniquely explained 47.94% of the deviance while the other factors uniquely explained 1.31% (Fig. 3b). This result demonstrates that the host genetic distance was the best of the measured predictors for symbiosis filtering differences.
Taken together, our findings suggest that differences in symbiosis filtering force among host individuals could generate slight differences in symbiont composition among host individuals in order to alter members of mutualistic bacteria obtained from rhizosphere soils, while symbiosis filtering might work to preserve high homogeneity of symbiont composition in host individuals. Moreover, the differences in symbiosis filtering force were likely to covary with the genetic variation of host plants rather than the soil environments surrounding the hosts.