Plant-plant and plant-soil interactions can help maintain plant diversity and ecosystem functions. Changes in these interactions may underlie experimentally-observed increases in biodiversity effects over time via the selection of genotypes adapted to low or high plant diversity. However, little is known about such community-history effects and particularly the role of plant-soil interactions in this process. Soil-legacy effects may occur if co-evolved interactions with soil communities either positively or negatively modify plant biodiversity effects. We tested how plant selection and soil legacy influence biodiversity effects on productivity, and whether such effects increase the resistance of the communities to invasion by weeds. We used two plant selection treatments: parental plants growing in monoculture or in mixture over 8 years in a grassland biodiversity experiment in the field, which we term monoculture types and mixture types. The two soil-legacy treatments used in this study were neutral soil inoculated with live or sterilized soil inocula collected from the same plots in the biodiversity experiment. For each of the four factorial combinations, seedlings of eight species were grown in monocultures or 4-species mixtures in pots in an experimental garden over fifteen weeks. Soil legacy (live inoculum) strongly increased biodiversity complementarity effects for communities of mixture types, and to a significantly weaker extent for communities of monoculture types. This may be attributed to negative plant-soil feedbacks suffered by mixture types in monocultures whereas monoculture types had positive plant-soil feedbacks in both monocultures and mixtures. Monocultures of mixture types were most strongly invaded by weeds, presumably due to increased pathogen susceptibility of mixture types and thereby reduced biomass and altered plant-soil interactions. These results show that biodiversity effects in experimental grassland communities can be modified by the evolution of positive vs. negative plant-soil feedbacks of plant monoculture vs. mixture types.