Leaf Photosynthetic and Functional Traits of Grassland Dominant Species in Response to Nutrient Addition on the Chinese Loess Plateau

Leaf photosynthetic and functional traits of dominant species are important for understanding grassland community dynamics under imbalanced nitrogen (N) and phosphorus (P) inputs. Here, the effects of N (N0, N50, and N100, corresponding to 0, 50, and 100 kg ha−1 yr−1, respectively) or/and P additions (P0, P40, and P80, corresponding to 0, 40, and 80 kg ha–1 yr–1) on photosynthetic characteristics and leaf economic traits of three dominant species (two grasses: Bothriochloa ischaemum and Stipa bungeana; a leguminous subshrub: Lespedeza davurica) were investigated in a semiarid grassland community on the Loess Plateau of China. Results showed that, after a three-year N addition, all three species had higher specific leaf area (SLA), leaf chlorophyll content (SPAD value), maximum net photosynthetic rate (PNmax), and leaf instantaneous water use efficiency (WUE), while also having a lower leaf dry matter content (LDMC). The two grasses, B. ischaemum and S. bungeana, showed greater increases in PNmax and SLA than the subshrub L. davurica. P addition alone had no noticeable effect on the PNmax of the two grasses while it significantly increased the PNmax of L. davurica. There was an evident synergetic effect of the addition of N and P combined on photosynthetic traits and most leaf economic traits in the three species. All species had relatively high PNmax and SLA under the addition of N50 combined with P40. Overall, this study suggests that N and P addition shifted leaf economic traits towards a greater light harvesting ability and, thus, elevated photosynthesis in the three dominant species of a semiarid grassland community, and this was achieved by species–specific responses in leaf functional traits. These results may provide insights into grassland restoration and the assessment of community development in the context of atmospheric N deposition and intensive agricultural fertilization.


Introduction
The semiarid Loess Plateau region in China is characterized by severe water scarcity, soil erosion, and nutrient-poor soils, which all greatly limit vegetation growth [1]. Grassland, with a size of~2.7 × 10 5 km 2 , accounts for ca. 43% of the regional total land area and is the dominant vegetation type on the Plateau [2]. It provides essential ecosystem functions and services such as carbon sequestration, soil and water conservation, and biodiversity [3,4]. Grassland management and restoration are of great ecological and economic significance in the region. Fertilization, as an effective management practice to responses to varying N and P fertilization conditions is seldom conducted on the semiarid Loess Plateau, which is needed to better understand grassland community development.
The dominant species occupy important ecological niches and play vital roles in maintaining community structure and function [40,41]. Biomass increases after N/P additions tend to be achieved by decreasing species diversity or increasing the biomass of dominant species [42]. Quantifying physiological and growth characteristics of dominant species under N and/or P addition could, thereby, be important for the evaluation of community productivity and dynamics, as well as provide valuable information for grassland management and restoration. Bothriochloa ischaemum (L.) Keng (a C 4 perennial grass), Stipa bungeana Trin. (a C 3 perennial grass), and Lespedeza davurica (Laxm.) Schindl. (a C 3 N-fixing subshrub) are co-dominant species in the natural/restored grasslands on the semiarid Loess Plateau of China [43]. The previous study on a regional grassland community, targeting these species, has shown that the addition of N and P combined improved grassland productivity and decreased species diversity, primarily via effects on tall clonal and annual species [44], which, once again, suggested species or functional-type-specific responses within a community, while the variation of leaf functional traits in these dominant species, after N and P fertilization, have not been fully assessed. Thus, we examined photosynthetic diurnal change, SPAD value, and leaf economic traits, including N mass , P mass , N mass /P mass ratio, SLA, and LDMC of the three dominant species, following a three-year N and P addition experiment in a typical semiarid grassland community on the Loess Plateau. We tested the hypotheses that: (1) N/P addition would increase photosynthetic rates and alter the photosynthetic diurnal dynamics of the three dominant species in the peak growing season, and these photosynthetic responses would be related to species-specific shifts in leaf functional traits; (2) addition of N and P combined would further promote photosynthesis compared with N/P additions alone.

Environmental Factors
Photosynthetically active radiation (PAR) and air temperature (T a ) showed a singlepeaked diurnal curve during the experimental period, and the maximum values appeared at 12:00 h and 14:00 h, with the values of 1854 µmol·m −2 ·s −1 and 30.2 • C, respectively ( Figure 1). The relative humidity (RH) remained stable during the daytime (~13%) (Figure 1). conditions, which should be systematically assessed [38]. Efforts have been made to quantify the variation of LES of grassland species under differed soil nutrient availability, which confirmed species-specific patterns [39], while the assessment of species-specific responses to varying N and P fertilization conditions is seldom conducted on the semiarid Loess Plateau, which is needed to better understand grassland community development.
The dominant species occupy important ecological niches and play vital roles in maintaining community structure and function [40,41]. Biomass increases after N/P additions tend to be achieved by decreasing species diversity or increasing the biomass of dominant species [42]. Quantifying physiological and growth characteristics of dominant species under N and/or P addition could, thereby, be important for the evaluation of community productivity and dynamics, as well as provide valuable information for grassland management and restoration. Bothriochloa ischaemum (L.) Keng (a C4 perennial grass), Stipa bungeana Trin. (a C3 perennial grass), and Lespedeza davurica (Laxm.) Schindl. (a C3 N-fixing subshrub) are co-dominant species in the natural/restored grasslands on the semiarid Loess Plateau of China [43]. The previous study on a regional grassland community, targeting these species, has shown that the addition of N and P combined improved grassland productivity and decreased species diversity, primarily via effects on tall clonal and annual species [44], which, once again, suggested species or functional-type-specific responses within a community, while the variation of leaf functional traits in these dominant species, after N and P fertilization, have not been fully assessed. Thus, we examined photosynthetic diurnal change, SPAD value, and leaf economic traits, including Nmass, Pmass, Nmass/Pmass ratio, SLA, and LDMC of the three dominant species, following a three-year N and P addition experiment in a typical semiarid grassland community on the Loess Plateau. We tested the hypotheses that: (1) N/P addition would increase photosynthetic rates and alter the photosynthetic diurnal dynamics of the three dominant species in the peak growing season, and these photosynthetic responses would be related to species-specific shifts in leaf functional traits; (2) addition of N and P combined would further promote photosynthesis compared with N/P additions alone.

Environmental Factors
Photosynthetically active radiation (PAR) and air temperature (Ta) showed a singlepeaked diurnal curve during the experimental period, and the maximum values appeared at 12:00 h and 14:00 h, with the values of 1854 μmol·m −2 ·s −1 and 30.2 °C, respectively (Figure 1). The relative humidity (RH) remained stable during the daytime (~13%) (Figure 1).

Diurnal Changes in Photosynthesis
The diurnal changes of net photosynthetic rate (P n ) and leaf instantaneous water use efficiency (WUE) of the three dominant species showed a double-peak curve under different N and P addition treatments. The first peak appeared at 10:00 h, the second at 14:00 h, and the midday depression of the photosynthesis (so-called "noon break") appeared at around 12:00 h ( Figure 2). The leaf transpiration rate (T r ) of B. ischaemum mostly showed a double-peak diurnal course. While diurnal changes of T r in S. bungeana and L. davurica showed a single peak. The diurnal changes of net photosynthetic rate (Pn) and leaf instantaneous water use efficiency (WUE) of the three dominant species showed a double-peak curve under different N and P addition treatments. The first peak appeared at 10:00 h, the second at 14:00 h, and the midday depression of the photosynthesis (so-called "noon break") appeared at around 12:00 h ( Figure 2). The leaf transpiration rate (Tr) of B. ischaemum mostly showed a double-peak diurnal course. While diurnal changes of Tr in S. bungeana and L. davurica showed a single peak. Compared with CK (i.e., N0P0), N addition alone and addition of N and P combined significantly increased the Pn values at 10:00 h and 14:00 h in the three species (except under N50P40 and N50P80 treatments in S. bungeana). The greatest Pn values appeared at 10:00 h under N and P combined treatments (i.e., N50P40 and N50P80) for the three species. The WUE of the three species significantly increased by N addition alone compared with CK (  Compared with CK (i.e., N0P0), N addition alone and addition of N and P combined significantly increased the P n values at 10:00 h and 14:00 h in the three species (except under N50P40 and N50P80 treatments in S. bungeana). The greatest P n values appeared at 10:00 h under N and P combined treatments (i.e., N50P40 and N50P80) for the three species. The WUE of the three species significantly increased by N addition alone compared with CK ( Table 1). The WUE of L. davurica increased significantly under all levels of P alone additions, while the WUE of the two grasses only significantly increased under N0P40 (Table 1). Under N and P combined addition, the maximum WUE values of B. ischaemum, S. bungeana, and L. davurica were 1.17, 1.09, and 1.47 µmol mmol −1 , respectively (Table 1).
N addition, alone, significantly increased the L s values of B. ischaemum and L. davurica (Table 1). P addition, alone, increased (p < 0.05) and decreased (p < 0.05) the L s of L davurica and S. bungeana, respectively, while only significantly increasing the L s of B. ischaemum under N0P80 treatment. N and P interaction significantly affected the L s values of the three species (Table 1). Under the addition of N50 combined with P, the L s of L. davurica increased, and those of B. ischaemum decreased (both p < 0.05). Under the addition of N100 combined with P, the L s of B. ischaemum significantly increased, while those of S. bungeana decreased significantly (p < 0.05; Table 1).  N addition, alone, increased the P Nmax values of the three species (p < 0.05), while there was no difference between them under N50 and N100 ( Figure 3). The increase in P Nmax in the two grasses was about two times larger than those of L. davurica under N addition alone ( Figure 3). P addition, alone, increased the P Nmax values of L. davurica and S. bungeana (p < 0.05; Figure 3). N and P combined addition only significantly affected the P Nmax values of S. bungeana and L. davurica (p < 0.05; Figure 3).   (Table 1). P addition, alone, increased (p < 0.05) and decreased (p < 0.05) the Ls of L davurica and S. bungeana, respectively, while only significantly increasing the Ls of B. ischaemum under N0P80 treatment. N and P interaction significantly affected the Ls values of the three species (Table 1). Under the addition of N50 combined with P, the Ls of L. davurica increased, and those of B. ischaemum decreased (both p < 0.05). Under the addition of N100 combined with P, the Ls of B. ischaemum significantly increased, while those of S. bungeana decreased significantly (p < 0.05; Table 1).
N addition, alone, increased the PNmax values of the three species (p < 0.05), while there was no difference between them under N50 and N100 ( Figure 3). The increase in PNmax in the two grasses was about two times larger than those of L. davurica under N addition alone ( Figure 3). P addition, alone, increased the PNmax values of L. davurica and S. bungeana (p < 0.05; Figure 3). N and P combined addition only significantly affected the PNmax values of S. bungeana and L. davurica (p < 0.05; Figure 3).  Values are mean ± SD. Different capital letters above the column indicate significant differences among P additions under each N addition rate, while different small letters indicate significant differences among N additions under each P addition rate. Numbers in parentheses are F-values, while '**', and '***' indicate p ≤ 0.01, and p ≤ 0.001, respectively. 'n.s.' indicates no significant difference.

Leaf SPAD Value
N addition, alone, significantly increased the SPAD values of the three species ( Figure 4). P addition, alone, significantly increased the SPAD values of B. ischaemum and L. davurica (p < 0.05), while SPAD values only increased under low levels of P addition alone (i.e., N0P40) in S. bungeana (p < 0.05). Addition of N and P combined significantly increased the SPAD values of both B. ischaemum and L. davurica (Figure 4).

Leaf SPAD Value
N addition, alone, significantly increased the SPAD values of the three species (Figure 4). P addition, alone, significantly increased the SPAD values of B. ischaemum and L. davurica (p < 0.05), while SPAD values only increased under low levels of P addition alone (i.e., N0P40) in S. bungeana (p < 0.05). Addition of N and P combined significantly increased the SPAD values of both B. ischaemum and L. davurica (Figure 4).

Figure 4.
Leaf SPAD values of the three species under different N and P addition treatments. Different capital letters above the column indicate significant differences among P additions under each N addition rate, while different small letters indicate significant differences among N additions under each P addition rate. Numbers in parentheses are F-values, while '*' and '***' indicate p ≤ 0.05 and p ≤ 0.001, respectively. 'n.s.' indicates no significant difference.

Leaf N and P Concentration (N mass and P mass ) and N mass /P mass Ratio
N addition, alone, significantly increased the N mass values of the two grasses, but it had no effects on N mass of L. davurica. The high level of P addition, alone (N0P80), significantly increased P mass of all species except L. davurica ( Figure 5). N and P interaction significantly affected N mass of the three species, while it only significantly affected P mass of the two grasses. Under the addition of N50 combined with P, the P mass of B. ischaemum and S. bungeana, as well as the N mass of L. davurica, increased significantly. Under the addition of N100 combined with P, the P mass of S. bungeana increased significantly, while those of L. davurica decreased significantly ( Figure 5). N addition, alone, significantly affected the Nmass/Pmass ratios of all three species. The low level of N addition alone (N50P0) significantly increased the Nmass/Pmass of the two grasses; the high level of N addition, alone (N100P0), only increased the Nmass/Pmass of S. bungeana grass (p < 0.05, Figure 5). N and P addition interaction significantly affects the Nmass/Pmass of the two grasses but has no effect on the subshrub. Under the addition of N50 combined with P, the Nmass/ Pmass of B. ischaemum and S. bungeana decreased significantly, Figure 5. Leaf nitrogen (N) and phosphorus (P) content (N mass and P mass ), as well as the N mass /P mass ratio, of the three species under different N and P addition treatments. Values are mean ± SD. Different capital letters above the column indicate significant differences among P additions under each N addition rate, while different small letters indicate significant differences among N additions under each P addition rate. Numbers in parentheses are F-values, while '*', '**', and '***' indicate p ≤ 0.05, p ≤ 0.01, and p≤ 0.001, respectively. 'n.s.' indicates no significant difference. N addition, alone, significantly affected the N mass /P mass ratios of all three species. The low level of N addition alone (N50P0) significantly increased the N mass /P mass of the two grasses; the high level of N addition, alone (N100P0), only increased the N mass /P mass of S. bungeana grass (p < 0.05, Figure 5). N and P addition interaction significantly affects the N mass /P mass of the two grasses but has no effect on the subshrub. Under the addition of N50 combined with P, the N mass / P mass of B. ischaemum and S. bungeana decreased significantly, while the N mass /P mass of L. davurica increased under N50P40 treatment (p < 0.05). Under the addition of N100 combined with P, the N mass / P mass of S. bungeana decreased significantly, while the N mass /P mass of B. ischaemum and L. davurica had no significant changes ( Figure 5).

Specific Leaf Area (SLA) and Leaf Dry Matter Content (LDMC)
N addition, alone, significantly increased the SLA values of the three species, and the low level of N addition alone (N50P0) decreased the LDMC of the two grasses (p < 0.05, Figure 6). P addition, alone, significantly increased SLA of L. davurica, while it had limited effects on the two grasses ( Figure 6). N and P addition interaction significantly affected both the SLA and LDMC values of all three species. Under the addition of N50 combined with P, the SLA of B. ischaemum and L. davurica increased significantly (p < 0.05), and the SLA of S. bungeana increased under N50P40, while it decreased significantly under N50P80 (p < 0.05); LDMC was comparable between different levels of P additions in the two grasses, but it decreased in L. davurica. Under the addition of N100 combined with P, the SLA values of B. ischaemum and L. davurica increased significantly, and the SLA of S. bungeana increased (p < 0.05) only under the high level of N addition (N100P80); LDMC, among different levels of P addition (i.e., N100P40 and N100P80), was comparable in B. ischaemum and L. davurica, while it significantly decreased under N100P80 in S. bungeana ( Figure 6).  PNmax was positively correlated with WUE, SPAD, Nmass, and SLA in all three species, while it was negatively correlated with LDMC in S. bungeana and L. davurica (Figure 7). PCA analysis showed that the variance explained by the first and second principal components was 37.5% and 23.5%, respectively, with a total value of 61% (Figure 8). The first principal component had a high correlation with PNmax, WUE, Pmass, and SLA; the second principal component had high correlation with LDMC. SPAD, Nmass, and Nmass/ Pmass ratio are correlated with both principal components (Figure 8). In a score plot of PCA analysis, under N addition alone, B. ischaemum and L. davurica gradually moved to the right with N addition level, and S. bungeana gradually moved to the upper right. Under P addition, Figure 6. Specific leaf area (SLA) and leaf dry matter content (LDMC) of the three species under different N and P addition treatments. Different capital letters above the column indicate significant differences among P additions under each N addition rate, while different small letters indicate significant differences among N additions under each P addition rate. Numbers in parentheses are F-values, while '*', '**', and '***' indicate p ≤ 0.05, p ≤ 0.01, and p ≤ 0.001, respectively. 'n.s.' indicates no significant difference. P Nmax was positively correlated with WUE, SPAD, N mass , and SLA in all three species, while it was negatively correlated with LDMC in S. bungeana and L. davurica (Figure 7). PCA analysis showed that the variance explained by the first and second principal components was 37.5% and 23.5%, respectively, with a total value of 61% (Figure 8). The first principal component had a high correlation with P Nmax , WUE, P mass , and SLA; the second principal component had high correlation with LDMC. SPAD, N mass , and N mass / P mass ratio are correlated with both principal components (Figure 8). In a score plot of PCA analysis, under N addition alone, B. ischaemum and L. davurica gradually moved to the right with N addition level, and S. bungeana gradually moved to the upper right. Under P addition, alone, L. davurica gradually moved to the right with P addition level, and B. ischaemum slightly moved to the upper right. Under the addition of N combined with P, with the increase in fertilizer application, all three species moved towards higher P Nmax , WUE, SPAD, and SLA values (Figure 9).

Discussion
The diurnal dynamics of photosynthesis reflect plants' sustained ability to carry out physiological metabolism and biomass accumulation throughout the daytime [45,46], which have been extensively studied in numerous dryland species (e.g., [19][20][21][22]). Our results, corroborated with others, showed that the diurnal course of photosynthetic rate showed a double-peaked curve in the three grassland dominant species, and they showed an evident "noon break" of photosynthesis. The noon break is a mechanism to avoid stresses such as excess light, high temperature, and water deficit during the midday [47], and it is a result of stomatal or non-stomatal restriction [48]. Our measurements showed that, during the period of 10:00-12:00 h, the Pn values gradually decreased while the Ls values increased (Figure 2 and Table 1), suggesting the "noon break" was likely caused by stomatal limitation [46,48]. In line with our first hypothesis and consistent with others (e.g., [49,50]), the Pn values of the three species considerably increased by nutrient addition, particularly during the peak photosynthetic period (~10:00 h). This is expected since environmental conditions (e.g., light and temperature) are relatively optimal for photosynthesis during this period, hence the promotion of nutrient addition would be most effective. In addition, during the noon time with high air temperature and light radiation, the Pn was slightly increased after fertilization, which may be due to increased stomatal conductance due to N and P addition, and this ostensibly alleviated the "noon break" [51]. Our study indicated that N and P fertilization could improve the photosynthetic ability of the three species at the diurnal scale and increase the daily accumulative carbon assimilation. However, we only focused on short-term responses during the peak growing season (i.e., July), so future studies should be taken to further assess intra-or interannual patterns of their photosynthesis to better understand the long-term effects of fertilization Both N and P are essential elements of key compounds involved in the photosynthetic process, and appropriate N and P additions would increase the content of these

Discussion
The diurnal dynamics of photosynthesis reflect plants' sustained ability to carry out physiological metabolism and biomass accumulation throughout the daytime [45,46], which have been extensively studied in numerous dryland species (e.g., [19][20][21][22]). Our results, corroborated with others, showed that the diurnal course of photosynthetic rate showed a double-peaked curve in the three grassland dominant species, and they showed an evident "noon break" of photosynthesis. The noon break is a mechanism to avoid stresses such as excess light, high temperature, and water deficit during the midday [47], and it is a result of stomatal or non-stomatal restriction [48]. Our measurements showed that, during the period of 10:00-12:00 h, the P n values gradually decreased while the L s values increased ( Figure 2 and Table 1), suggesting the "noon break" was likely caused by stomatal limitation [46,48]. In line with our first hypothesis and consistent with others (e.g., [49,50]), the P n values of the three species considerably increased by nutrient addition, particularly during the peak photosynthetic period (~10:00 h). This is expected since environmental conditions (e.g., light and temperature) are relatively optimal for photosynthesis during this period, hence the promotion of nutrient addition would be most effective. In addition, during the noon time with high air temperature and light radiation, the P n was slightly increased after fertilization, which may be due to increased stomatal conductance due to N and P addition, and this ostensibly alleviated the "noon break" [51]. Our study indicated that N and P fertilization could improve the photosynthetic ability of the three species at the diurnal scale and increase the daily accumulative carbon assimilation. However, we only focused on short-term responses during the peak growing season (i.e., July), so future studies should be taken to further assess intra-or interannual patterns of their photosynthesis to better understand the long-term effects of fertilization.
Both N and P are essential elements of key compounds involved in the photosynthetic process, and appropriate N and P additions would increase the content of these compounds and, subsequently, the photosynthetic rate [17,26,30]. This was observed in our study: the P Nmax , SPAD, and N mass of the three species increased significantly after N addition alone, and there were strong positive correlations between N mass , SPAD, and P Nmax (p < 0.05; Figures 3-5 and 7). The increase in P Nmax and WUE with the N addition level was much greater (larger regression slopes) in the two grass species than in the legume L. davurica (Table 2), with the greatest increase (the largest slope) of P Nmax and WUE, along with the N addition level, in C 4 grass B. ischaemum (Table 2). Together, this suggested that the two grasses were more sensitive to N addition alone than the legume. This is consistent with our previous study quantifying the plant biomass of B. ischaemum and L. davurica mixtures under varying soil moisture and nutrient supplies [52]. We suspect that the subshrub L. davurica may not be N-limited due to its N fixation ability and is, thereby, insensitive to exogenous N fertilization. On the other hand, the photosynthetic rate does not continuously increase with N addition amounts after passing a threshold [29,53], which was also observed, here, as the P Nmax values of the three species were not significantly different between N50 and N100 (Figures 3 and 6; Table 1). Fossil fuel combustion and extensive fertilization have greatly increased atmospheric N deposition globally in recent decades [54]. Chronic N input by long-term N deposition may, hence, alleviate N limitation and promote plant photosynthesis and growth of regional grassland species, but it may, meanwhile, intensify plant P limitation by increasing P demand [55,56]. Table 2. Regression slopes (SE) derived from the multiple linear regression analysis between photosynthetic characteristics (P Nmax and WUE) and N addition, P addition, and N and P addition interaction in the three species.

Species
Variable P Nmax WUE Here, the P addition, alone, had greater effects on P Nmax , SPAD, and SLA of the leguminous L. davurica among the three species (Figures 3, 4 and 6; Table 2). This may be ascribed to P, as it could promote the activity of nitrogenase in the root nodules of legumes and enhance their photosynthesis [34,57], and elevated leaf P content can also directly improve photosynthetic capacity by promoting ATP and NADPH synthesis, as well as regeneration of RuBP [33,58]. Compared with N or P addition alone, the three species had higher P Nmax , WUE, and SLA values under N and P combined additions, suggesting a synergetic effect of N and P on plant photosynthesis (Figures 5 and 6). This confirms our second hypothesis and suggests that appropriate N and P combined fertilization should be considered to maintain regional grassland productivity. A myriad of studies have documented this synergetic effect in grasslands worldwide (e.g., [59,60]). Previous studies in the Loess Plateau grasslands also reported the N and P combination had synergetic effects on community productivity [44]. A recent long-term (over 66 years) nutrient addition study in a mesic grassland in South Africa also highlighted that N and P combined addition promoted plant P acquisition and uptake (e.g., increased organic P storage, P recycling, and plant P utilization), which may contribute to the synergetic effect of N and P combined addition [59].
Drylands (e.g., the semiarid Loess Plateau) are often co-limited by water and nutrients [8], as well as characterized by frequent drought events, which greatly impact plant N and P uptake [61]. Nutrient addition, such as N, at an appropriate rate could improve post-drought recovery of grassland and increase the aboveground biomass production [62].
Contrarily, some studies reported that nutrient addition increased grassland drought sensitivity and constrained its recovery from drought events [10]. Besides, grass species with different photosynthetic pathways (C 3 vs. C 4 ) may respond differentially to drought and rewatering under nutrient addition conditions [63]. For the regional grassland, previous studies have quantified the photosynthetic responses of dominant species following rainfall events and reported species-specific patterns [64]. Nevertheless, the interaction of soil moisture (especially drought) and fertilization on dominant species performance remains less understood in the regional grassland and should be assessed, considering recurrent drought events, under future climate scenarios [65].
Leaf functional traits, particularly those so-called economic traits, are invoked to explain plant resource acquisition and utilization [36]. Among them, the leaf N mass / P mass ratio indicates environmental N and P availability where the plant grows [56]. In general, N mass / P mass ratio less than 10 indicates the N limitation, and greater than 20 indicates the P limitation [56]. The leaf N mass / P mass ratio of the three species, averaged across treatments, was 18.8 (B. ischaemum), 11.3 (S. bungeana), and 24.3 (L. davurica), respectively ( Figure 5), suggesting species-specific N and P limitations. The N mass /P mass ratio of the two grasses increased significantly with N addition, while no noticeable change was found in L. davurica ( Figure 5). This indicates that N addition may lead to P limitation in the two grass species. Meanwhile, increased soil N and P availability would release plants from nutrient competition to other resource competition, such as light and water [66]. Grassland dominant species may accordingly alter their leaf functional traits to maximize light harvesting to maintain dominance. According to the LES theory, plants with higher light capture, resource acquisition, and turnover capacity show higher SLA, N mass , and P mass in contrast to the slow-growth ones with higher LDMC and conservative nutrient resource use [36]. Similar to other studies (e.g., [66]), the three species studied here shifted to a fast-growth strategy after N addition with larger, thinner, and N-rich leaves (higher SLA, N mass, and SPAD), as well as higher assimilation rate per unit leaf area (higher P Nmax ). Though score plots from PCA analysis indicated that, under N addition, three species adopted different strategies to improve their light harvesting: C 4 grass B. ischaemum mainly by increasing SLA and P Nmax , while C 3 grass S. bungeana and C 3 subshrub L. davurica primarily increased leaf N content and SPAD (Figure 9), and only L. davurica had notable shifts in photosynthetic and leaf functional traits under P additions ( Figure 9B), which suggests that the three species had different trade-off strategies in photosynthetic performance and leaf economic traits in response to N and/or P addition [17,66], and these should be considered when assessing N and P fertilization effects on community structure and functions. The P Nmax values of the three species were mostly highest under the 'N50P40 treatment among all treatments, indicating that it could be considered an optimal fertilization measure for improving grassland production.
Totally, there were 9 treatments, including a control (N0P0), two N addition alone treatments (N50P0, N100P0), two P addition alone treatments (N0P40, N0P80), four N and P combined addition treatments (N50P40, N50P80, N100P40, N100P80), and three replicates per treatment. N and P additions were conducted once a year, on rainy days,

Leaf SPAD Value
Leaf SPAD value (a measure of leaf relative chlorophyll content) was measured on three newly fully-expanded healthy leaves per species per treatment using a chlorophyll meter (SPAD-502 model, Konica-Minolta, Osaka, Japan) on 20-22 July 2019.

Leaf Functional Traits
The 10-20 newly-fully expanded healthy leaves were randomly sampled per species per treatment, stored in zipped plastic bags, and quickly taken back to the laboratory, in an insulated box with ice packs, for leaf functional traits measurements. Leaves were weighed with an analytical balance (d = 0.0001 g). The fresh leaves were scanned (Epson duplex scanner, Epson, Tokyo, Japan), and the leaf area was derived using ImageJ (National Institutes of Health, Bethesda, MD, USA). Then, leaves were oven-dried at 75 • C for 24 h and ground with a high-throughput tissue grinder (MM-400, Retsch, Haan, Germany). Specific leaf area (SLA, m 2 g -1 ) was calculated as leaf area divided by leaf dry mass. Leaf dry matter content (LDMC, g g -1 ) was calculated as leaf dry mass divided by fresh mass. After digestion with H 2 SO 4 -HClO 4 , the mass-based leaf N concentration (N mass ) was obtained using a Kjeldahl N analyzer (FOSS-8400, Foss, Höganäs, Denmark). The massbased leaf P concentration (P mass ) was determined by a molybdenum blue colorimetry (UV-2600 ultraviolet-visible spectrophotometer, Shimadzu, Kyoto, Japan). N mass /P mass ratio was then calculated.

Statistical Analysis
All statistical analyses were performed with SPSS 20.0. One-way analysis of variance (ANOVA) was used to compare the differences in leaf photosynthetic characteristics (WUE, L s , and P Nmax ) and leaf functional traits (SPAD, N mass , P mass , N mass /P mass , SLA, and LDMC) of the three species under different N and P addition treatments. Tukey's HSD test was used for multiple comparisons. Two-way ANOVA was used to test the effects of N addition, P addition, and their interaction on P Nmax , SPAD value, N mass , P mass, SLA, and LDMC. Pearson correlation was used to explore the relationship between leaf photosynthetic characteristics (P Nmax , WUE) and leaf functional traits (SPAD value, N mass , P mass, N mass / P mass , SLA, and LDMC). Multiple linear regression was used to explore the relationship between N addition, P addition, and their interaction, as well as P Nmax and WUE. Principal component analysis (PCA) was conducted on photosynthetic characteristics and leaf functional traits. Graphing was performed with Origin 2021 (Origin Lab Software, Chicago, IL, USA).

Conclusions
Our three-year field fertilization study suggested that N addition-alone or combined with P-improved the photosynthesis of the three grassland dominant species on the semiarid Loess Plateau of China. All three species shifted to a fast-growth strategy with increased P Nmax , SLA, and N mass , as well as reduced LDMC under N and/or P addition. Furthermore, species-specific shifts in leaf functional traits were observed among the three species following N and/or P addition, of which C 4 grass B. ischaemum increased SLA and P Nmax , and C 3 grass S. bungeana and subshrub L. davurica mainly increased leaf N and SPAD. P addition seems to only effectively impact the P n of L. davurica. Evident N and P synergetic effects on the photosynthetic performance in all three species were observed, and a combination of 50 kg ha −1 yr −1 N and 40 kg ha −1 yr −1 P addition could be considered optimal fertilization for improving grassland productivity locally.

Data Availability Statement:
The data presented in this study are available on reasonable request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.