Abstract Predicting the composition of rhizosphere microbiomes has become increasingly important for sustainable agriculture. A key open question is whether a plant’s rhizosphere community is shaped more by the specific traits or host phylogeny, under different soil conditions. We conducted a greenhouse experiment on 15 legume species, including three pairs of crop-wild relative pairs, under different phosphorus conditions. We then sequenced the bacterial and fungal rhizosphere communities. Using Bayesian models, we found rhizosphere composition was shaped by individual species identity, independent of host phylogeny (intraclass correlation=0.40-0.79). This suggests that closely related plants do not necessarily share similar rhizosphere microbiomes. These patterns remained consistent across host intraspecific variation and nutrient treatments. Using a custom-built root imaging platform, we quantified root architectural traits and applied machine learning to correlate with rhizosphere community composition (R2=0.46–0.80). Root diameter and carbon content were the strongest drivers. Notably, these key root traits were largely uncorrelated with phylogeny, yet strongly explained variation in rhizosphere community composition. Our results indicate that even closely related legume species may host divergent rhizosphere communities.
Stewart et al. (Tue,) studied this question.