1. Many studies that provided evidence for a positive relationship between plant diversity and productivity have proposed that this effect may be explained by complementarity among species in resources utilization, or selection of particularly productive species in high-diversity plant communities. Recent studies have related the higher productivity in diverse plant communities to suppression of pathogenic soil biota. If soil biota plays a role in diversity–productivity relationships, the question remains about how they may influence complementarity and selection effects. 2. Here we examine how complementarity and selection effects may depend on soil biota using a plant–soil feedback approach. We used monocultures and mixtures of early successional plant species, which are known to have mostly negative plant–soil feedback effects, and mid-successional plant species, which generally have neutral plant–soil feedback. 3. We found that plant–soil feedback effects differed between monocultures and mixed plant communities, as well as between early and mid-successional plants. This resulted in a significant interaction effect between diversity and successional stage. In monocultures, plant–soil feedback tended to be negative for early and positive for mid-successional plant species. Interestingly, the community feedback responses of the mixed communities were opposite, being positive for early and negative for mid-successional community. 4. Plant–soil feedback differentially affected complementarity and selection effects of early and mid-successional plant communities: it enhanced complementarity effects of early and decreased selection effects of mid-successional species. 5. Synthesis. Soil biota that drive plant–soil feedback effects can influence the diversity–productivity relationship not only through decreased biomass production in monocultures compared to mixtures, but also through influencing complementarity and selection effects among species in mixed plant communities. Our results reveal that biodiversity–productivity relationships depend on plant–soil feedback interactions, which depend on the successional position of the plant. We propose that including successional position and trait-based analyses of plant–soil feedback in diversity-functioning studies will enhance understanding consequences of biodiversity loss for productivity and other ecosystem processes.

Biomass data of early and mid successioinal plant species in monoculture and mixture as influenced by plant-soil feedbackData were collected in the greenhouse two-stage (conditioining & feedback) biodiversity experiment in Wageningen, the Netherlands. Abbreviated headlines are as follows: "stage"=succession stage;"species"=plant species (including ten early succesional plant species and ten mid successional plant species);"soil"including conditioned soil and unconditioned soil;"monoculture"=biomass production in monoculture(g pot-1);"mixture1""=biomass production in plant community assemblage1(g pot-1);"mixture2""=biomass production in plant community assemblage2(g pot-1);"mixture3""=biomass production in plant community assemblage3(g pot-1);"mixture4""=biomass production in plant community assemblage4(g pot-1);"mixture5""=biomass production in plant community assemblage5(g pot-1).Jing et al JOE data.xlsx