The experiment was conducted in a phytotron at Inner Mongolia Agricultural University, Hohhot, China, with the following controlled climatic conditions: 12 h light/12 h dark photoperiod, 40% day/40% night relative humidity and 25°C average indoor temperature. The seeds of S. breviflora were collected from the desert steppe located in Siziwang Banner, Wulanchabu city, Inner Mongolia autonomous region, China (41°46′43.6″N, 111°53′41.7″E). In order to eliminate errors caused by seed size, uniform seeds were selected by weighing based on the average size of each batch (0.31 g per 100 seeds). The studied soil was the topsoil (0–20 cm) collected from the same site. After air-drying, soil properties were determined. The soil type was chestnut soil, with a pH of 8.29 and a field capacity of 23%, Other soil properties were shown in Supplementary Table S1. The soil was passed through a 2 mm sieve and used to fill pots of 15 cm in diameter and 12 cm in height that can hold 1.5 kg of soil. The seeds were sterilized in 10% H₂O₂ for 30 min and germinated in Petri dishes (120 mm in diameter) on wet filter papers.

Five N addition treatments (0, 7.69, 15.38, 23.08, and 38.46 mg N kg⁻¹ soil) with four replicates each were used in this study. The rates were calculated to equal atmospheric N deposition rates of 0, 10, 20, 30, and 50 kg N ha⁻¹ yr.⁻¹, respectively (Supplementary Figure S1). Based on the ratios of organic N:inorganic N of deposition (30%:70%) and NH₄⁺-N:NO₃⁻-N (1.63:1.00) in desert steppe (data from our N deposition monitor net), we designed the N fertilizer mixture to be CO(NH₂)₂:NH₄HCO₃:Ca(NO₃)₂ = 7.41:27.73:17.96 in order to accurately mimic the N forms in deposited N.

All of the N fertilizers were dissolved in deionized and were then added to the 1.5 kg of soil in the pot. After air-drying, the N and soil were mixed thoroughly. Germinated seeds were sown in the pots at a rate of 11 seeds per pot. Soil moisture in each pot was maintained at 70% of field capacity by weighing every 2 days. When the plants were 5–6 cm high, they were thinned to three plants per pot.

The number of seeds emerging is recorded at 5 days after sowing (DAS) to calculate the emergence rate. The height of S. breviflora was measured with a standard ruler at 30, 60, 90, and 120 DAS. The number of S. breviflora leaves was also counted by forceps at the same time. At harvest, shoots were cut at soil surface. Roots were lifted out of soil and washed by deionized water carefully. All the plant materials were dried at 70°C for 3 days and weighed.

The fresh soil in the pots was collected and separated into two parts, one for soil total N analysis and the other for soil inorganic N measurement. The NH₄⁺-N and NO₃⁻-N in the fresh soil were extracted using KCl solution and the concentrations were determined using a continuous flow analyzer (SEAL, Germany). The soil inorganic N concentration (SIN)was the sum of the ammonium and nitrate concentrations. Plant materials were ground with a ball mill (Retsch MM 400, Germany) and air-dried soil samples were digested in a mixture of concentrated H₂SO₄ and H₂O₂, and the N concentration was then determined using the micro-Kjeldahl procedure (Gallaher et al., 1976). The N content was calculated by multiplying plant N concentration and biomass (Li et al., 2010). Soil pH was measured with a pH meter (Leici, China) at a soil:water ratio of 1:2.5. The equations for calculating plant N uptake rate and soil residual N rate were listed as follows: Plant N uptake rate = Plant N content Root biomass × Growth days Soil residual N rate = SIN in N added treatments − SIN in the control N addition rate

SPSS software version 24.0 (IBM, US) was used for one-way analysis of variance (ANOVA) to determine the significance of plant height, leaf number, biomass, soil inorganic N concentration, soil pH, plant and soil total N, and N content of the plants (LSD, α = 0.05). Average values are reported as the arithmetic mean. Origin 2021 software (OrginLab, United States) was used to create the figures.

Compared with the control without N addition, N addition did not increase the emergence rate and plant height of S. breviflora plants during 120 days of growth irrespective of the amount of N added (Figure 1; Supplementary Figure S1). The average plant height was 18.1 cm at 30 DAS. Plant height increased by 45.4% at 60 DAS compared with the height at 30 DAS. After 60 DAS, the plants cannot grow higher any more, and maintained at 26.5 cm on average.

Added N significantly suppressed leaf emergence when N addition rates were 30 and 50 kg ha⁻¹ yr.⁻¹ at 30 DAS compared with the control (Supplementary Figure S2). There were six leaves on the S. breviflora plants on average in these two treatments, and this was 27.1% lower than the number in the other treatments. This discrepancy was amplified at 60 DAS when the average leaf number was nine. However, leaf numbers in all the treatments were similar at 90 and 120 DAS. From 90 to 120 DAS, leaf number increased by 42.4%, and reached to 99 leaves (including growing leaves). The leaf emergence rate was much higher from 90 to 120 DAS compared to the other growth periods.

Similar to the situation with plant height, N addition also did not change plant biomass including shoot and root biomass (Figure 2). The shoot and root biomass of each pot averaged 2.04 g pot⁻¹ and 2.40 g pot⁻¹ at 120 DAS, respectively. The ratio of root:shoot biomass of the plants was ~1.

Added N did not significantly improve shoot and root N concentrations relative to the control at harvest. Shoot N concentration was 2.33%, and was 2.01 times higher than the average root N concentration of 1.16% (Supplementary Figure S3). The shoot N contents were similar among all treatments, which were 47.3 mg pot⁻¹ on average. The same trend was found in root N content and plant total N content, which averaged 30.6 and 77.8 mg pot⁻¹, respectively. Shoot N content was 54.6% higher than root N content (Figure 3).

At harvest, no significant changes were found for soil NH₄⁺ concentration in any of the N addition treatments relative to the control (Figure 4). However, N addition increased soil NO₃⁻ concentration by 380.2% compared with the control when the addition rate was 50 kg ha⁻¹ yr.⁻¹ (Figure 4). In contrast, soil NO₃⁻ concentration in the other treatments with relatively low N addition rates were not significantly different than the control. No significant changes in soil total N were measured for the different N addition rates (Figure 4). Compared to the control, the soil pH decreased significantly by 0.15 pH units only when the N addition rate reach to 50 kg ha⁻¹ yr.⁻¹, and there were no significant differences in the other treatments (Supplementary Figure S4).

Added N did not change plant height, leaf number, and biomass, although there were changes in leaf number before 60 DAS at N addition rates of 30 and 50 kg ha⁻¹ yr.⁻¹. This indicates that the growth of S. breviflora plants did not respond to N addition, as we hypothesized, and is consistent with the result of Wu et al.,who show that continuous N addition did not change the growth of S. breviflora in natural system, even at 100 kg ha⁻¹ yr.⁻¹ in an experiment spanning 16 years (Wu et al., 2020). In many studies, N addition rate can be as high as 200 kg N ha⁻¹ yr.⁻¹,which results in a change in the growth of some plants (Kazanski et al., 2019). It is possible that the same things could occur in S. breviflora if the plants were exposed to a much higher N addition rate than we used. However, it is difficult to achieve such high N deposition rates in the desert steppe.

In a previous study conducted in the desert steppe, N addition also fails to increase the plant height and biomass in Alhagi sparsifolia and Lycium ruthenicum (Li et al., 2022). In contrast, Wan et al. (2019) and Ren et al. (2020) found that N addition promotes plant height and biomass accumulation in Solidago canadensis, Artemesia argyi, and Pterocypsela laciniata. Furthermore, the sensitivities of plants to N addition in the same grassland system can be different. The two dominant species in the desert steppe have obvious differences in sensitivity to added N; a positive effect is observed in Artemisia capillaris but not in Stipa tianschanica when the N addition rate is 100 kg N ha⁻¹ yr.⁻¹. The results of our study leave no doubt that S. breviflora is insensitive to N addition. In terms of individual plants, N addition is beneficial to nitrophilic species (Bobbink et al., 1998), but not to plants with a conservative N use strategy (Suding et al., 2005; Bobbink et al., 2010; Lü et al., 2020).

In plants that grows in the desert steppe, leaf N concentration stay between 19.0 and 30.0 mg g⁻¹, while root N concentration is lower, and ranges from 7.0 to19.0 mg g⁻¹ (Li, 2014); this includes S. breviflora, Cleistogenes songorica, Convolvulus argenteum, Artemisia frigida, and Kochia prostrata. The ranges are stable even though the geographical locations are different (Ma et al., 2019; Li et al., 2022; Liu et al., 2022). Leaf and root N concentrations in S. breviflora in the desert steppe are determined to be 19.6 and 7.2 mg g⁻¹, respectively, in a previous study conducted in natural conditions (Li, 2018). In this study, the leaf and root N concentration of S. breviflora plants were 23.3 and 11.6 mg g⁻¹ respectively, which were slightly higher than the result of the previous study. This discrepancy could be due to the better growth condition in our study. Compared with the other plant species that grow in the desert steppe, S. breviflora has the highest biomass with the lowest N concentration (Li, 2018). Therefore, the efficient N utilization of S. breviflora is one of the reasons that it became a constructive species.

In two previous studies, N addition increases the N concentration of plants (Yuan and Chen, 2015; Huang et al., 2018), however, the N concentration of S. breviflora plants was not affected by added N in our study. Similar results are found in the desert steppes for Lycium ruthenicum Murr and Alhagi sparsifolia Shap as constructive species. Plants that adopt a conservative strategy for N utilization can provide a guarantee for maintaining ecosystem stability and effectively reduce the disturbance caused by N addition to grassland ecosystems.

At N addition rates equaled to or less than 30 kg N ha⁻¹ yr.⁻¹, SIN did not show a significant change. The budget between input and output determines the change in soil inorganic N concentration. Added N was the only source of input N in this study because the effects of N deposition can be ignored. Although 30% of the N added to the soil is urea, it is unlikely to maintain this form at harvest because urea is converted into NO₃⁻-N in calcareous soil within 1 week after addition (Tong et al., 1992). N output occurs through plant uptake, NO₃⁻ leaching, and NH₃ emission. In the N addition treatments, the uptake of S. breviflora plants was not greater than in the control, which was due to the same root N uptake rate (Supplementary Figure S6A). Also, no N leaching occurred during plant growth due to the use of weighing irrigation. The NH₃ emission were not measured, but it can be assumed to be <20% according to a previous study (Stumpe et al., 1984). Therefore, this suggests that most N added to the soil is absorbed by soil microorganisms. The study in the desert steppe also showed that N addition significantly increased the amount of microbial biomass N in soil when there was sufficient water (Zhu et al., 2022). It is also confirmed by a mate-analysis that N addition leads to a 10% increase of soil microbial N (Zhou et al., 2017). Soil inorganic N concentration in the 50 kg N ha⁻¹ yr.⁻¹ treatment was higher than that in the 30 kg N ha⁻¹ treatment yr.⁻¹ by 19.67 mg kg⁻¹, which is close to the difference in N addition rate between the two treatments. Soil residual N rate was 58.9% in the 50 kg N ha⁻¹ yr.⁻¹ treatment, suggesting 29.5 kg N ha⁻¹ yr.⁻¹ cannot be used by plants and soil microorganisms during growth period (Supplementary Figure S6B). This indicates that the soil is N-limiting, and 30 kg N ha⁻¹ yr.⁻¹ is a key change-point between N-limiting and non-limiting. However, it should be clarified that a short-term pot experiment cannot fully mimic natural conditions for soil microorganisms. This value can only be a reference which could possibly be modified under natural conditions.