Soil bacterial taxonomic diversity is critical to maintaining the plant productivity.

Soil microbial communities play a central role in driving multiple ecosystem functions and ecological processes that are key to maintaining the plant productivity. However, we lack sound evidence for the linkage between soil microbial diversity and plant productivity, which hinders our ability to predict the consequences of microbial diversity loss for food security under the context of global environmental change. Here, we used the dilution-to-extinction approach to test the consequences of soil microbial diversity loss for the aboveground plant biomass in a glasshouse experiment. Compared with original soils, the bacterial alpha-diversity (Observed operational taxonomic units and Shannon index) significantly decreased in treatments with serially diluted inoculum. Principal coordinates analysis showed that the overall bacterial community compositions (beta-diversity) in original soils were clearly separated from the treatments with serially diluted inoculum. The aboveground biomass of lettuce harvested from the original soils was significantly higher than that from the sterilized soils regardless of the inoculation. The ordinary least squares regression model showed a significant linear relationship between the plant biomass and bacterial alpha-diversity, indicating that reduction in soil microbial diversity could result in a significant decline in the biomass of lettuce. No significant correlation was observed between plant biomass and soil processes including soil basal respiration and denitrification rates. Structural equation models suggested that the effects of soil microbial diversity on the plant biomass were maintained even when simultaneously accounting for other drivers (soil properties and biological processes). Our study provides experimental evidence that soil microbial diversity is important to the maintenance of the plant productivity and suggests that the functional redundancy in soil microbial communities may be overestimated especially in the agroecological system.