Because arbuscular mycorrhizal (AM) fungi proliferate through extraradical hyphal growth and produce spores asexually, it is often assumed that little intraspecific variation exists within these fungi.  While distinct ecotypes have been found in different geographic locations, the possibility of more local variation has been little explored.  We conducted an experiment to determine if functional variation exists among collections of Gigaspora gigantea originating on different host plant species within an East Coast serpentine grassland.  We collected G. gigantea spores from under three grasses in the field—Andropogon gerardii, Schizachyrium scoparium, and Sorghastrum nutans—and tested their ability to promote growth of S. scoparium and So. nutans. The same sterilized field soil was used for all plants; other soil microbes were added back to the soil.  Only two G. gigantea spores were used to inoculate a single plant, but those were taken at random from spores collected under six widely separated individuals of a particular plant species.  After three months, both plant species showed significant differences in shoot biomass as a function of the host plant on which G. gigantea originated.  Interestingly, while G. gigantea collected under S. scoparium in the field promoted the least growth in S. scoparium, it promoted the most growth in So. nutans.  Thus, not only are there functional differences in the isolates collected on different host plant species, there appears to be negative feedback between the plants and the fungi.  These results revealed a previously unrecognized level of fungal variation which could help maintain the species diversity of plant communities.