Root responses of grassland species to spatial heterogeneity of plant-soil feedback

Plant roots selectively forage for soil nutrients when these are heterogeneously distributed. In turn, effects of plant roots on biotic and abiotic conditions in the soil, which result in so-called plant-soil feedback can be heterogeneously distributed as well, but it is unknown how this heterogeneity affects root distribution, nutrient uptake and plant biomass production. Here, we investigate plant root distribution patterns as influenced by spatial heterogeneity of plant-soil feedback in soil and quantify consequences for plant nitrogen uptake and biomass production. We conditioned soils by four grassland plant species to obtain own' and foreign' soils that differed in biotic conditions similar as is done by the first phase of plant-soil feedback experiments. We used these conditioned soils to create heterogeneous (one patch of own and three patches of foreign soils) or homogeneous substrates where own and foreign soils were mixed. We also included sterilized soil to study the effect of excluding soil biota, such as pathogens, symbionts and decomposers. We supplied N-15 as tracer to measure nutrient uptake. In nonsterile conditions, most plant species produced more biomass in heterogeneous than in homogeneous soil. Root biomass and N-15 uptake rates were higher in foreign than own soil patches. These differences between heterogeneous and homogeneous soil disappeared when soil was sterilized, suggesting that the effects in nonsterilized soils were due to species-specific soil biota that had responded to soil conditioning. We conclude that plants produce more biomass when own and foreign soils are patchily distributed than when mixed. We show that this enhanced productivity is due to nutrient uptake being overall most efficient when own and foreign soils are spatially separated. We propose that spatial heterogeneity of negative plant-soil feedback in species diverse plant communities may provide a better explanation of overyielding than assuming that plant-soil feedback effects are diluted.