Abstract
Plant-soil feedbacks (PSFs) drive plant community diversity via interactions between plants and soil microbes. However, we know little about how frequently PSFs affect plants at the seed stage, and the compositional shifts in fungi that accompany PSFs on germination.We conducted a pairwise PSF experiment to test whether seed germination was differentially impacted by conspecific versus heterospecific soils for seven grassland species. We used metagenomics to characterize shifts in fungal community composition in soils conditioned by each plant species. To investigate whether changes in the abundance of certain fungal taxa were associated with multiple PSFs, we assigned taxonomy to soil fungi and identified putative pathogens that were significantly more abundant in soils conditioned by plant species that experienced negative or positive PSFs.We observed negative, positive, and neutral PSFs on seed germination. Although conspecific and heterospecific soils for pairs with significant PSFs contained host-specialized soil fungal communities, soils with specialized microbial communities did not always lead to PSFs. The identity of host-specialized pathogens, that is, taxa uniquely present or significantly more abundant in soils conditioned by plant species experiencing negative PSFs, overlapped among plant species, while putative pathogens within a single host plant species differed depending on the identity of the heterospecific plant partner. Finally, the magnitude of feedback on germination was not related to the degree of fungal community differentiation between species pairs involved in negative PSFs. Synthesis. Our findings reveal the potential importance of PSFs at the seed stage. Although plant species developed specialized fungal communities in rhizosphere soil, pathogens were not strictly host-specific and varied not just between plant species, but according to the identity of plant partner. These results illustrate the complexity of microbe-mediated interactions between plants at different life stages that next-generation sequencing can begin to unravel.