Climatic shifts critically regulate soil organic carbon (SOC) dynamics, biodiversity, and productivity in grasslands. However, the mechanisms linking abiotic/biotic factors to plant diversity–SOC mineralization synergies remain unclear. To study this mechanistic relationship, we established 12 sampling sites along an east–west precipitation gradient across the China-Mongolia steppe. We measured plant diversity and biomass for both species and functional groups in 12 grassland sites. In the laboratory, we analyzed soil nutrients and microbial communities. We also measured the SOC mineralization potential using 28-day incubations. The results indicated that: (1) Aboveground biomass (AGB) increased through two opposing strategies, enhancing or reducing plant diversity, with thresholds at Shannon–Wiener indices of 1.14 (arid west) and 2.19 (humid east). AGB shifts altered resource competition and microenvironments, directly impacting diversity. (2) Plant diversity was primarily regulated by soil pH, SOC, and mean annual temperature (MAT). (3) Perennial grasses dominated productivity, while perennial forbs drove diversity via niche complementarity. (4) Microbial biomass carbon (MBC) was the direct driver of SOC mineralization, modulated by mean annual precipitation (MAP) through SOC mediation. (5) SOC mediated contrasting ecosystem effects by suppressing plant diversity through pH-driven nutrient limitations while simultaneously enhancing mineralization rates via stimulation of microbial decomposer activity. SOC properties and precipitation govern divergent changes in grassland diversity and carbon cycling. Strategic management of SOC pools, coupled with precipitation adaptation and biodiversity conservation, can enhance ecosystem resilience under climate change. This mechanistic framework advances understanding of grassland responses to global change.

Study area

The research was carried out in the steppe of Inner Mongolia, located in northern China, and serves as a prime example of the Eurasian steppe concerning its climate, soil, and vegetation (Bai et al., 2008). The geographical coordinates of the study area range from 39°01′ to 49°32′ N and 101°37′ to 120°02′ E, with elevations between 653 m and 1,478 m. Annual temperature averages fluctuate from 2°C to 8°C, with January recording the lowest monthly averages (-9°C to -26°C) and July the highest (19°C to 24°C). The region experiences mean annual precipitation varying from 100 to 380 mm, with approximately 80% of it falling during the growing season (May to August), aligning with peak temperatures (Bai et al., 2008). The soil types found here include chestnut, calcareous brown, and desert soils. A total of twelve sites across the east-west gradient were selected to represent four types of plant communities: meadow grassland, typical grassland, desert grassland, and desert, with a trend of decreasing annual precipitation. The eastern meadow steppe is characterized by dominant species such as Leymus chinensis, Stipa baicalensis, and Filifolium sibiricum. In contrast, the central grasslands feature Leymus chinensis, Stipa grandis, and Artemisia frigida as predominant species. The desert grasslands are primarily composed of Stipa klemenzii and Allium polyrhizum, while the western desert areas are characterized by Salsola passerina, Hololachna songarica, and Nitraria tangutorum (Mi et al., 2015).

Study site setup and plant-soil sampling methodology

This study involved fieldwork and sampling in August 2012. Twelve sites were selected along an east-west belt transect (Mi et al., 2015). Each site covered an area of 100 m × 100 m, in which ten 1 m × 1 m quadrats were systematically positioned. Various measurements were conducted within these plots, focusing on species composition, both aboveground and belowground biomass, soil microbial communities, and the physical and chemical characteristics of the soil.

The quantity of plant species and their respective abundances were documented in each sampling plot. The plant species were classified into four categories according to their life forms: annual and biennial herbs (AB), perennial grasses (PG), perennial forbs (PF), and shrubs or semi-scrub plants (SHS). Counts and biomass measurements were recorded according to plant functional groups, and relative abundance and biomass were subsequently calculated.

Plant specimens within the sample plots were clipped at ground level. After collecting the above-ground biomass, three soil cores were randomly collected using a 7 cm diameter root auger and placed into a root bag in each sample plot. Samples were obtained from three different soil depths: 0-5 cm, 5-10 cm, and 10-20 cm. Soil in the root bags was washed with running water. Roots were separated using a 1 mm sieve, eliminating dead roots and other debris. Aboveground biomass and root samples were dried in an oven at 65°C for 48 hours until they reached a constant weight, after which they were weighed.

Using a soil auger with a 5 cm diameter, soil samples were taken. The cores from three augers were merged to create a composite sample, which was subsequently sieved through a 2 mm mesh, air-dried, and cleaned of gravel and other impurities prior to analyzing the physical and chemical properties. The pH of the soil was determined with a potentiometric method, using a 2.5:1 liquid-to-soil ratio and the FE28 pH meter (Mettler Toledo, Switzerland). Total carbon (TC) and total nitrogen (TN) content (g/kg of dry soil) were measured by a CHON elemental analyzer (Elementar VARIO EL III, Hanau, Germany). Inorganic carbon (IC) was determined using a calcium carbonate analyzer (Eijkelkamp, Giesbeek, Netherlands). Soil organic carbon (SOC) was then calculated by subtracting IC from TC. The carbon-to-nitrogen ratio (C/N) was obtained by dividing SOC by TN.

Microbial biomass carbon (MBC) was assessed through the chloroform fumigation-K₂SO₄ extraction technique, while the analysis of the soil microbial community was conducted using the phospholipid fatty acid (PLFA) method.

Meteorological data and soil organic carbon mineralisation

Meteorological data were sourced from nearby meteorological stations at the sampling locations within the study area. Precipitation and temperature readings for each sampling site were derived and processed using kriging spatial interpolation within the Geographic Information System (ArcGIS). Data on soil organic carbon mineralisation and carbon dioxide accumulation from incubated soils under conditions of 25°C and 60% water-filled pore space (WFPS) across the 12 sampling sites were retrieved from our prior research for additional analysis (Mi et al., 2015).

Statistic analysis

Data organization was carried out using Excel 2021. A one-way regression analysis was performed to correlate the Simpson diversity index, Shannon-Wiener diversity index, and Margalef richness index of all sample plots with the total and aboveground biomass using SPSS version 23.0. The regression equation was fitted by selecting the function that yielded the highest coefficient of determination (R²). Mantel test and Pearson’s test were used to investigate the correlation between diversity indices as well as soil organic carbon mineralisation accumulation and environmental variables through using R version 4.1.2 (Lucent Technologies, Jasmine Mountain, New Jersey, USA). Pathway analysis of the effects of climatic factors, vegetation traits, soil physico-chemical traits and soil microbial properties on biodiversity and soil organic carbon mineralisation accumulation was analysed using structural equation modelling (SEM). The SEM analysis incorporated MAP, MAT, aboveground and belowground biomass, soil organic carbon (SOC), the carbon-to-nitrogen ratio (C:N), microbial biomass carbon (MBC), the fungal-to-bacterial ratio (F:B), and pH levels. The SEM analyses were performed using Amos version 17.0.2 (Amos Development Corporation, Chicago, IL, USA), applying chi-square (χ²) tests, comparative fit indices (CFI), root mean square error of approximation (RMSEA), and goodness-of-fit index (GFI) to evaluate model performance.

Species diversity is frequently assessed using the Simpson index, Shannon-Wiener diversity index, and richness index, with the formulas for each being:

Simpson index: D=1-∑Pi²

Shannon-Wiener diversity index: H′=-∑Pi lnPi

Margalef richness index: R=(S-1)/lnN

where S denotes the total number of species within the community; N represents the overall count of individuals across all species in the community; N_i indicates the number of individuals for each specific species, and P_i = N_i/N. The variable i signifies the relative density of plant species, calculated as the number of individuals of a particular species divided by the total number of individuals of all species.