Background and aims N enrichment is a major driver of plant species loss worldwide. Nevertheless, its impacts on local species assemblages (i.e., beta-diversity), another essential but often overlooked index of biodiversity, remain uncertain. Moreover, whether or how C input changes the effects of N enrichment on plant biodiversity are poorly understood. Methods We addressed the questions using location test of continuous input of a gradient of both N (n = 5) and labile C (n = 3) in a semi-arid grassland, conducted a five-year survey of plants biodiversity in each plot annually, and analyzed their variations and correlations. Results N enrichment made plant alpha-diversity lose and plant production increase, which enhanced over time. The responses of different species varied, and plant communities shifted from stochastic to deterministic assembly during the five-year survey. The effects of N enrichment on plant biodiversity could be significantly relieved by a lower rate of labile C input. In this case, a lower rate of labile C enhanced coexistence of plant species (based on data of number and height), especially for certain taller species; but a higher rate of labile C input was opposite. Conclusion A lower labile C treatments under N enrichment alleviated the negative responses of plant biodiversity over time. Plant species that could adjust their height strategically may better adapt to varied N and C addition and play important roles in plant communities assembly. All provided novel prospects in prediction and prevention of biodiversity reduction in semi-arid grasslands.