We proposed a theoretical framework predicting mutualistic outcomes for the arbuscular mycorrhizal (AM) symbiosis based on host provenance (crop versus wild). To test the framework, we grew two isolates of Rhizoglomus irregulare (commercial versus an isolate locally isolated), with five crop plants and five wild plants endemic to the region that co-occur with the locally sourced fungus. While inoculation with either isolate had no effect on plant biomass, it decreased leaf P content, particularly for wild plants. All plants associating with the commercial fungus had lower leaf P. Overall, our data shows that wild plants may be more sensitive to differences in mutualistic quality among fungal isolates.