Soil net nitrogen (N) mineralization is a key biogeochemical process influencing plant available N and net primary productivity (NPP) in terrestrial ecosystems. However, the spatial variations and controlling factors of soil net N mineralization (RPNM) in arid and semi-arid grasslands are less studied and unclear. In this study, we investigated the soil RPNM by performing a laboratory incubation experiment. Soil samples were collected from 30 sites in three east-west transects on the Inner Mongolia Plateau (MP), Loess Plateau (LP), and Tibetan Plateau (TP) along a 3,200 km arid and semi-arid grassland gradient, with each transect containing three different grassland types (meadow steppe, typical steppe, and desert steppe, respectively). Results showed that the average RPNM values ranged from -0.37 to 1.29 mg N kg–1 d–1, with a significantly lower RPNM found in the desert steppe (0.08 ± 0.01 mg N kg–1 d–1) compared with those in the meadow steppe (0.30 ± 0.03 mg N kg–1 d–1) and in the typical steppe (0.33 ± 0.03 mg N kg–1 d–1) in the MP and LP transects (p < 0.05). This difference could be explained by variations in climatic and soil factors, such as hydrothermal index (HT), the soil pH, soil organic matter (SOM) and precipitation. However, no significant differences in RPNM were found among different grassland types in the TP transect, possibly due to the similarly low microbial activity, as indicated by the MBC values. Across all three grassland transects, HT, SOM, and microbial variables were the major factors controlling RPNM, which together explained 20.7% of the variation in RPNM. Further SEM analysis indicated HT was an integral predictor of RPNM, directly or indirectly via SOM, under different conditions of precipitation and temperature. Our findings provide field evidence and parameters for biogeochemical cycling to better predict future N transformation processes under changing precipitation and temperature regimes across a wide range of arid and semi-arid grassland ecosystems.

Soil sampling was conducted between July 18 and August 23, 2018. We first randomly selected eight plots (1 m × 1 m), with 10 m apart from each plot at each site. Then, topsoil (0-10 cm) from five soil cores (10 cm in diameter) were collected after removing surface litter from each plot. Subsequently, the 5 soil samples were combined to form a composite sample. In total, 240 soil samples (3 transects × 10 sites × 8 replicates) were collected from 30 sites. Visible roots and litter residues were manually removed from each soil sample, and fresh soil samples were sieved using a 2 mm mesh sieve. The sieved soil samples were split into two subsamples; one subset was used for the laboratory incubation to estimate the RPNM and was stored at 4°C. The remaining soil was air-dried to analyze soil basic properties including soil organic matter (SOM), total carbon (TC), total nitrogen (TN), pH value, soil microbial biomass carbon (MBC), and microbial biomass nitrogen (MBN).

Soil organic matter was measured by wet oxidation with KCr₂O₇ + H₂SO₄ and titrated with FeSO₄. The TC and TN content were analyzed using an elemental analyzer (EA3000, Euro Vector, Pavia, Italy). Soil pH (1:2.5 soil to water ratio) was measured using a pH meter (FE20K, Mettler-Toledo, Greinfensee, Switzerland). MBC and MBN were determined based on chloroform fumigation methods (Vance et al., 1987). Briefly, fumigated and unfumigated soil samples were first extracted using 0.5 M K₂SO₄ and then the extracted solution was analyzed using a C/N analyzer (multi-N/C 3100 Analytic Jena AG, Jena, Germany). The MBC and MBN were calculated by subtracting the difference of dissolved organic C and N content between fumigated and unfumigated soil samples and dividing it by the conversion coefficient. The conversion coefficients used were 0.45 for MBC and 0.54 for MBN.

Soil samples (30 g, dry weight) were placed in 150-ml plastic bottles (eight replicates for each site) and were adjusted to approximately 60% water holding capacity (WHC) with deionized water. Then, the incubation bottles were sealed with parafilm and pre-incubated for 24 h at 25°C for short-term equilibration. Then, the pre-incubated soil samples were divided into two subsamples. One soil subsample was used for measuring the initial NH₄⁺-N and NO₃^--N concentrations (c(NH₄⁺-N) t0 and c(NO₃^--N) t₀) and the other subsample was incubated for 7-days in the chambers. The optimum temperature for soil microbial activity in the three transects might be different because of the long-term microbial adaption to different climatic conditions (especially temperature and precipitation). We incubated the collected soil samples at 15℃ for samples from the Tibetan Plateau, at 20℃ for samples from the Inner Mongolia Plateau, and at 25℃ for samples from the Loess Plateau. The incubation temperature for soils from each plateau was chosen based on the mean highest daily temperature from mid-July to mid-August, corresponding to the hottest time period in the year for each plateau. This also covered the period during which the field work was conducted. This gives the maximum potential N mineralization rate that can realistically occur for the three steppes in each plateau. During the 7-day incubation period, the soil samples were maintained at a constant WHC with deionized water every 2–3 days. After the incubation, the NH₄⁺-N and NO₃^--N concentrations (c(HN₄⁺-N) t₁ and c(NO₃^--N) t₁) were measured.

The RPNM on a dry mass basis was calculated as the difference in NH₄⁺-N and NO₃^--N concentrations before and after incubating the samples (Wang et al., 2006):

R_PNM = (∆c (NH₄⁺-N) + ∆c(NO₃⁻ -N)) / ∆t               (1)

where ∆t is the incubation time.

The hydrothermal index (HT) was used to determine the equilibrium state of hydrothermal resource allocation (Zhang et al., 2020). HT is the ratio of the normalized mean annual precipitation (MAP_norm) to the normalized mean annual temperature (MAT_norm), as shown in Equations 2–4 according to Zhang et al. (2020):

MAP_norm = (MAP − MAP_min) / (MAP_max − MAP_min)           (2)

MAT_norm = (MAT − MAT_min) / (MAT_max − MAT_min)         (3)

HT = MAP_norm / MAT_norm                       (4)

where MAP_??? is the minimum of the average annual precipitation across the 30 grassland sites included in this study, MAP_??? is the regional maximum, MAT_??? is the regional minimum of the average annual temperature across the 30 grassland sites in this study, and MAT_??? is the regional maximum temperature.