The study was conducted at the Erguna Forest-Steppe Ecotone Ecosystem Research Station (N50°10′46.1″, E119°22′56.4″). It is located in HulunBuir City, Inner Mongolia Autonomous Region. The soil is classified as chernozem according to the FAO classification. The region has a cold temperate continental climate with an average annual temperature of −1.59°C and an annual rainfall of 336.5 mm (2000–2020). The rainfall of the sample site was 148.2 mm in 2015 and 139.8 mm in 2020 according to the measurement of the research station. The background value of N deposition in this area is relatively low (Jia et al., 2014), which is conducive to simulation experiments of N deposition. Two dominant species, the C₃ perennial rhizome grass L. chinensis and C₃ perennial legume T. lanceolata, which together account for more than 50% of the peak community aboveground biomass, were used as our model plant species.

The grassland of the experiment site has been fenced since 2013 to prevent large animals, such as cattle and sheep, from eating and trampling the vegetation. The N addition experiment was established in 2014. All experimental plots (10 m × 10 m) had similar topography and land-use history. A completely randomized block design was used, and a 1-m buffer area was established between the plots. Three N compounds were selected: ammonium nitrate (NH₄NO₃), ammonium sulfate [(NH₄)₂SO₄], and ammonium hydrogen carbonate (NH₄HCO₃). Each N fertilizer was applied at six concentrations: 0, 2, 5, 10, 20, and 50 g N m^–2 yr^–1. There were 18 treatments, with each treatment replicated eight times, resulting in a total of 144 plots. The N compounds were applied annually in late May. In order to ensure that the N compounds were spread evenly in each plot, they were combined with baked fine sand.

Plant sampling was conducted during the season of vigorous plant growth (August) in 2015 (short term) and 2020 (long term). These sampling dates corresponded to two years in the short-term experiment and seven years in the long-term experiment. Plant height (H) was obtained based on the mean of five healthy individuals at every site. Ten to 15 complete, healthy plants of each species were harvested, placed in sealed bags, and temporarily stored in an insulation bucket in iceboxes. After sampling was completed, the sample was immediately brought to the laboratory to reduce the loss of water dispersion from the plant and to avoid leaf shrinkage.

In the laboratory, all samples were soaked in water for 6 h to ensure full rehydration. Each leaf was then cut from the stem and gently dried with tissue paper before measurement. Water-saturated leaf and stem mass were weighed. The leaves used to measure the leaf area are scanned with a scanner after being weighed (Canon LiDE120), and then the leaf area was determined by the ImageJ software¹. All leaves were then dried for 48 h at 65°C and then weighed. The leaf dry matter content (LDMC), specific leaf area (SLA), and stem-leaf ratio (S:L) were calculated. The leaf total N content (LN) was determined by the Kjeldahl apparatus (K9860, Hanon, Dezhou, China) after extraction with sulfuric acid, and the leaf total P content (LP) was determined by a UV-spectrophotometer (UA-5500, METASH, Shanghai, China) with the wavelength set to 700 nm. The N to P ratio (N:P) was calculated as the ratio of the total N content of the leaf to the total P content. All trait measurements were based on previous standardized measurements (Cornelissen et al., 2003; Pérezharguindeguy et al., 2013).

The N compound type was used as a categorical variable with three levels: NH₄HCO₃ (AC), NH₄NO₃ (AN), and (NH₄)₂SO₄ (AS). The different sampling years were also categorical variables with two levels: 2015 (short term) and 2020 (long term). We used four-way ANOVA to test the effects of N addition rates, N compounds type, period, species, and their interactions on leaf chemical and morphological traits. The effects of different N addition rates on leaf chemical and morphological traits were tested using one-way ANOVA, followed by post hoc tests [Tukey’s honest significance difference (HSD)] in periods of treatment for each N compound type, and the effect of year was tested by t-tests in each N addition rate of each N compound type. When necessary, data were natural log-transformed to meet the normality assumption of ANOVA and t-tests (include SLA). The variation in plant leaf functional traits was analyzed based on differences by terms (N addition time), N compound type and N addition rates using PERMANOVA (999 permutations, Adonis function) and visualized by principal component analysis (PCA). In order to determine whether the relationship between the leaf functional traits of the two species in different years had changed, we used the ggpairs function in the Ggally package for plotting and calculated the correlation and significance between each two based on Pearson’s correlation analysis (Emerson et al., 2013). All statistical analyses were performed using R software 4.0.3 (R Core Team 2019²).

The results showed that the N addition rate, species and period had significant effects on the LN, LP and N:P of the two plants (P < 0.01), while N compound type had no significant effect (Table 1). The N addition rate and period had an interactive effect on the LN, LP and N:P of the two plants (P < 0.01, Table 1).

Nitrogen addition significantly increased the LN in both species. However, under high N addition (20 and 50 g N m^–2 yr^–1), the LN of the two species showed a slowing down or a slight downward trend with increasing N addition (Figure 1). The increasing trend of LN under long-term N addition was significantly higher than that under short-term N addition (Figure 1). Under both short- and long-term N addition treatments, the LN of the two species showed significant differences at each N addition rate of the two periods. Similar results were detected for all three N addition types.

The LP of the two species did not show a significant trend in N addition rate with the addition of various N compounds (Figure 2). The LP of L. chinensis under N addition was only marginally lower than the control. The LP of the two species showed significant differences at each N addition rate of the two periods. The N content under long-term N addition was significantly higher than under short-term N addition (Figure 2, P < 0.05).

The N:P of the two species differed significantly in response to the N addition rate and period (Figure 3). The N:P of the leaf of L. chinensis significantly increased under long-term N addition (Figure 3A, P < 0.05), while that was not significant under short-term N addition. Furthermore, the difference between the two periods was not significant (Figure 3A, P > 0.05). The change trend of the N:P in the T. lanceolata leaves was not obvious with N addition. The N:P under long-term N addition was significantly lower than under short-term N addition (Figure 3B).

According to four-way ANOVA, the N addition rate, species and period had significant effects on the LDMC, S:L, H and SLA of the two plants (P < 0.05), while N compound type had no significant effect, except for the LDMC (Table 1). The N addition rate and period had an interactive effect on the LDMC, S:L, H and SLA of the two plants (P < 0.01, Table 1).

With the increase in N addition, the H of L. chinensis showed a significant upward trend under long-term N addition (P < 0.05), but there was no significant change under short-term N addition (Figure 4A). With the exception of the AS treatment under long-term N addition, the response of T. lanceolata H to N addition rate was not significant (Figure 4B, P < 0.05). There was a significant inter-annual difference in the H of the two species. The plant height under long-term N addition was significantly higher than under short-term N addition in both species (Figure 4, P < 0.05).

The N addition rate significantly affected the LDMC of the two species under long-term N addition (Figure 5, P < 0.05), but there was no significant difference in short-term N addition. The LDMC showed a gradual decrease with increased N addition. However, under high N treatment (20, 50 g Nm^–2yr^–1), LDMC demonstrated a slowing-down or slight upward trend (Figure 5). In period of the response between years, the LDMC of plants under long-term N addition was significantly lower than under short-term N addition (Figure 5, P < 0.05).

The S:L of the two species demonstrated a significantly different response to the N addition rate and period (Figure 6). The S:L of L. chinensis showed an upward trend under long-term N addition (Figure 6A). The t-test showed that the response of L. chinensis S:L to period was significant (Figure 6A, P < 0.05). The S:L under long-term N addition was significantly higher than that under short-term N addition, while the response of T. lanceolata S:L to period was not significant (Figure 6B).

Nitrogen addition increased the leaf SLA of the two species, especially under long-term N addition (Figure 7). However, under high N addition (20 and 50 g Nm^–2yr^–1), the leaf SLA of the two species slowed down or decreased slightly with the increase in N addition (Figure 7). The leaf SLA under long-term N addition was significantly higher than under short-term N addition (Figure 7, P < 0.05).

The PCA of all variables showed that short- and long-term N addition resulted in obvious differences in the spatial aggregation of leaf functional traits, and the response of leaf functional traits to N addition rate and period differed (Figure 8). Compared with short-term N addition, L. chinensis had higher H, LN, LP, S:L, and SLA and lower LDMC under long-term N addition (Figure 8A), and T. lanceolata had higher H, LN, LP, and SLA and lower LDMC and N:P. There was a significant correlation between the leaf functional traits of L. chinensis and T. lanceolata under short- and long-term N addition (Supplementary Figures 1–4, P < 0.05). Compared with short-term N addition, long-term N addition had more pairs of traits that were significantly related (Supplementary Figures 1–4). Under long-term N addition, the H and LP, H and N:P, and LP and LN of both L. chinensis and T. lanceolata showed a significant change trend that was opposite to that under short-term N addition (Supplementary Figures 1–4, P < 0.05), which indicated that long-term N addition will alter the relationships of some plant functional traits.

Leaf N and P concentrations and the N:P stoichiometric ratios of the two species in the meadow steppe showed significant changes in response to the N addition rate and period. N addition led to an increase in the LN of the two species (Figure 1), which corroborates the results of previous studies (Huang et al., 2012; Lü et al., 2013). Nitrogen addition caused an increase in soil inorganic N concentrations, which subsequently enhanced the N concentrations in the leaves (Shi et al., 2020). Moreover, N addition increased the re-absorption rate of plant N (Lü et al., 2012; Cao et al., 2021), which in turn increased the plant N content. However, the LN did not always increase with the increase in the N addition rate. At high N addition rate (20 and 50 g Nm^–2yr^–1), the LN decreased slightly, which may be caused by the ammonium toxicity caused by excessive N fertilizer (Van den Berg et al., 2005; Phoenix et al., 2012). Ammonium toxicity effects mainly occur in plants that absorb ammonium ions more easily and exchange these ions with hydrogen ions and potassium plasma, thereby interfering with the cation balance in plants (Van den Berg et al., 2005).

Previous studies on the response of leaf P concentration to N addition obtained different conclusions, including negative (Menge and Field, 2007), neutral (Chen et al., 2015), and positive effects (Lü et al., 2013; Huang et al., 2018). In this work, N addition reduced the LP of L. chinensis leaves (Figure 2), resulting in a nutrient imbalance. N addition increases the biomass of the plant, and the plant will also grow higher under N addition, which leads to the dilution of the P concentration (Zhang et al., 2017). In order to maintain a stoichiometric balance, the demand for P increases as the N concentration of the leaves increases, and the release of available P from plants through soil formation is usually very slow (Houlton et al., 2008; Zhang et al., 2017). The T. lanceolata did not show a similar phenomenon, which may be due to the specific characteristics of the species (Mayor et al., 2014; Wan et al., 2019). The N and P concentrations were less influenced by N addition under short-term N addition than under long-term N addition, probably because of the timing of N application. Because as the time of nitrogen addition increases, the accumulated nitrogen in the soil increases (Sun et al., 2020).

The N-to-P ratio is a measure of plant growth limitation. Güsewell (2004) proposed an N:P > 20 and <10 as the evaluation index of plant N-P limitation. When N-to-P ratio > 20, this indicates that plant growth is limited by P, while N-to-P ratio < 10 indicates that plant growth is limited by N. N-to-P ratio of 10 – 20 indicates that plant growth was not limited or co-limited by N and P (Yan et al., 2017). In this study, the N:P of the L. chinensis leaves gradually increased with N addition rate in both periods, which indicated that growth gradually changed to P limitation (Figure 3A). The N:P of T. lanceolata under long-term N addition was significantly lower than under short-term N addition. According to the criteria proposed by Güsewell, the growth of T. lanceolata was mainly limited by P under short-term N addition, while it was limited by both N and P under long-term N addition. The results indicated that N addition could gradually change the elements involved in ecosystem limitations.

Nitrogen addition had a significant effect on SLA and LDMC in the plant leaves, which aligns with previous studies (Knops and Reinhart, 2000) and demonstrates that N addition influences these leaf functional traits. In this study, N addition increased the H of the plants, indicating that N addition promoted plant growth, the plant growth strongly affected by nitrogen limitation. The results showed that, N addition increased the SLA of the plant leaves and decreased the LDMC. From an inter-year perspective, long-term N application had higher SLA and lower LDMC compared with short-term N application. SLA and LDMC are key indicators of nutrient utilization strategies (Garnier et al., 2004). A high SLA and low LDMC represent rapid nutrient acquisition, which is conducive to the growth of plants in a nutrient-rich environment, while low SLA and high LDMC are suitable for plants in nutrient-poor environments. In the environment, plants adopt a conservative growth strategy (Wilson et al., 2010; Rose et al., 2013). Our results showed that N addition had a beneficial effect on the growth of L. chinensis and T. lanceolata. Taken together, our results suggest that the plasticity of plant leaf functional traits expressed allowed the plants to adapt to environment changes under increased N deposition (Agrawal, 2001).

Under the background of N addition, changes in plant leaf morphological traits represent a type of phenotypic plasticity. Although our results reflected that the trait plasticity of both plants changed in a favorable direction for plant growth, longer N deposition experiments are encouraged to understand whether N saturation of leaf traits would occur in future.

Our results suggest that the effects of different N compounds on leaf functional traits were not significant. As the N compounds selected in the experiment all have ammonium ions, this shows that different anions (SO₄^–, HCO₃^–, NO₃^–) have no significant effect on leaf functional traits.

We found that under long-term and short-term N application, the effect of N addition rates on plant traits was similar. Wright et al. (2004) proposed the “leaf economic spectrum” (LES) and found that the traits of plant leaves are arranged in an orderly manner along a continuously changing combination spectrum of functional traits. The survival strategies of plants are divided into a “”quick investment-return” strategy and “slow investment-return” strategy. In our study, compared with short-term N addition, long-term N addition made plant leaves having higher LN, higher LP, higher plant height, lower LDMC, and higher SLA, indicating that N addition would result in L. chinensis and T. lanceolata having a tendency toward larger and thinner leaves and a “quick investment-return” strategy. Therefore, meadow grassland plants adapt to the long-term nitrogen deposition environment mainly through the distribution and regulation of resources. In addition, a fast strategy may be beneficial in N deposition condition because it allows plants to take advantage of high resource, thus increasing their probability of survival. We also found that the correlation between plant leaf functional traits changed under long-term N addition conditions. This is similar to previous studies by Wright and Cannon (2001) and Pensa et al. (2010). Our results suggest that long-term nitrogen application tends to lead to closer relationships between traits. In addition, the correlation of traits involved in the “leaf economic spectrum” tended to prevail under N addition, but the magnitude of the correlation coefficients did not remain constant, and even opposite relationships were observed. This suggests that the relationships between plant traits are not absolutely stable, but vary with the environment. Therefore, when we try to use the general correlation of leaf traits to predict other relatively difficult-to-measure traits from some of the easily measured traits, we need to take into account the environmental changes such as local climate and geographical characteristics.