Grasslands, covering 40% of Earth's surface, provide crucial ecosystem services. However, land-use intensification has led to widespread degradation, reducing biodiversity and ecosystem functioning. Soil organisms are vital for nutrient cycling and plant production. Studies show that inoculating degraded land with soil from healthy ecosystems can alter microbial communities and promote plant growth. However, it's unclear if inocula from similar but distinct ecosystems will have different effects, and the optimal inoculum amount is unknown. This study addresses these gaps by investigating the impact of two grassland soil inocula (meadow steppe and upland meadow) at varying amounts on a degraded grassland's soil and plant communities over three years. The hypotheses are: 1) different donor soils steer communities towards their own composition; 2) upland meadow soil leads to better restoration; and 3) higher inoculum amounts accelerate these effects.

Previous research demonstrated that soil inoculation from late-successional fields can alter soil microbial communities and promote the establishment of late-successional plant species through plant-soil interactions. Studies showed the composition of soil and plant communities shifting towards a target community when using soil inoculum from a semi-restored grassland. Evidence exists that inoculation with soil from distinctly different ecosystems induces changes in both soil and plant communities. However, research on the impact of inoculum amounts and the potential for different effects from similar ecosystems is limited. Existing studies employed varying inoculum amounts, raising questions about the relationship between inoculum amount and restoration success. While pot experiments show a positive correlation between inoculum amount and plant biomass, the field effects over longer time scales remain uncertain. The need to optimize inoculum amounts in restoration projects to minimize donor site damage further motivates this research.

This three-year field experiment was conducted in a degraded grassland near Erguna Forest-Steppe Ecotone Research Station, China. Two donor grasslands (meadow steppe and upland meadow) were selected, with three replicate sites for each. The top 10 cm of soil was excavated, homogenized, and transported to the degraded site. The degraded grassland's top 5 cm was removed. Three inoculation amounts (1 cm, 3 cm, 5 cm) of each donor soil were used, with a control group. Each plot was 2x2 m². Soil samples were collected in August of each year for various analyses. Soil physicochemical properties (moisture, pH, total carbon, nitrogen, and phosphorus) were measured. DNA extraction and amplicon sequencing were performed for bacteria (16S rRNA gene V4 region) and fungi (ITS2 region). Nematode communities were extracted and identified to genus level. Plant species richness, cover, and biomass (above- and belowground) were recorded annually. Data were analyzed using linear mixed models, PERMANOVA, and network analysis. Specifically, Response Ratio (RR) was calculated to compare treatment effects against control and donor sites. Co-occurrence networks were constructed using Spearman rank correlation to understand the interactions among bacterial, fungal, and nematode genera. Keystone taxa were identified based on node degree in the network.

Both donor soil identity and inoculation amount significantly affected soil microbiomes and nematode communities. After three years, bacterial and fungal communities diverged towards the donor soil composition, with this divergence amplified by higher inoculum amounts. Nematode community effects were delayed, becoming significant in years two and three. Soil inoculation also significantly impacted plant communities, steering their composition towards the donor sites. The cover of perennial plants increased over time and was higher in upland meadow soil treatments. Inoculation suppressed degenerate indicator species and increased target species (*Leymus chinensis*) cover. Belowground biomass was significantly higher in inoculated plots, increasing with inoculum amount. Soil nutrient concentrations and soil moisture also increased with inoculum amount. Similarity analysis showed increased dissimilarity from the control with higher inoculum amounts for bacteria and plants. Inoculation with meadow steppe soil increased similarity to the donor meadow steppe, and upland meadow soil increased similarity to the upland meadow donor for fungi and plants. Network analysis revealed that upland meadow soil introduced more central genera in the soil network, resulting in higher network complexity compared to meadow steppe soil.

This study confirms that soil inoculation can steer the development of soil and plant communities in a degraded grassland. The results highlight the importance of donor site selection in restoration projects, as even similar ecosystems can lead to different outcomes. The stronger effects of upland meadow inoculation compared to meadow steppe inoculation suggest distinct differences in colonization success between the two soils. The linear relationship between inoculum amount and restoration success indicates that higher amounts lead to faster and more pronounced effects, likely due to increased density of introduced organisms and improved establishment conditions. The identification of key genera, particularly in the upland meadow inoculum, points to specific microorganisms that might drive the observed changes in soil communities and plant performance. Future research should focus on identifying the mechanisms by which these key genera influence other community members and plant growth.

Soil inoculation effectively promotes grassland restoration, but the direction and rate depend heavily on both donor soil identity and inoculum amount. Upland meadow soil showed superior restoration capacity compared to meadow steppe soil. Higher inoculum amounts generally accelerated the restoration process. Identifying key microbial genera will allow for the development of targeted inoculation strategies. Long-term studies are needed to fully understand the long-term interactions between inoculated soil and plant communities and their influence on ecosystem functioning.

The study is limited to a specific geographic location and soil types. The three-year timeframe might not capture the full long-term dynamics of ecosystem recovery. The focus on specific indicator species might not fully represent the complex interactions within the plant community. Further research using different geographical regions and longer time frames would greatly enhance the generalizability of this research.