The impact of nitrogen enrichment on grassland ecosystem stability depends on nitrogen addition level

STOTEN-24206; No of Pages 10 Science of the Total Environment xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

The impact of nitrogen enrichment on grassland ecosystem stability depends on nitrogen addition level Decao Niu a,1, Xiaobo Yuan a,1, Arianne J. Cease b, Haiyan Wen a, Chunping Zhang a, Hua Fu a,⁎, James J. Elser c,d a

State Key Laboratory of Grassland Agro-ecosystems, College of Pastoral Agriculture Science and Technology, Lanzhou University, Lanzhou 730000, China School of Sustainability, Arizona State University, Tempe, AZ 85281, USA School of Life Sciences, Arizona State University, Tempe, AZ 85281, USA d Flathead Lake Biological Station, University of Montana, Polson, MT 32125, USA b c

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• The response of community stability to N enrichment was nonlinear. • Current N deposition level had no significant impact on community stability in semi-arid grassland, Loess Plateau. • Changes in community stability were driven by community structure and species composition.

a r t i c l e

i n f o

Article history: Received 18 July 2017 Received in revised form 28 September 2017 Accepted 29 September 2017 Available online xxxx Editor: Elena PAOLETTI Keywords: Community stability Loess Plateau Nitrogen deposition Plant diversity Portfolio effects Species asynchrony

a b s t r a c t Increasing atmospheric nitrogen (N) deposition may affect plant biodiversity, subsequently altering ecosystem stability. While a few studies have explored how simulated N deposition affects community stability and its underlying mechanisms, the experimental levels of N addition used are usually higher than current and future N deposition rates. Thus, their results could produce highly uncertain predictions of ecosystem function, especially if the responses to N deposition are nonlinear. We conducted a manipulative experiment that simulated elevated atmospheric N deposition with several N addition levels to evaluate the effect of N deposition on ecosystem stability and its underlying mechanisms in a semiarid grassland in northern China. Here we show that N addition altered community diversity, reducing species richness, evenness, diversity and dominance. In addition, we found that N addition at current N deposition levels had no significant impact on community stability. In contrast, N addition at levels from 4.6 to 13.8 g N m−2 yr−1 significantly decreased community stability, although community stability for the 13.8 g N m−2 yr−1 treatment was higher than that for the 4.6 g N m−2 yr−1 treatment. These results indicate that the response of community stability to N enrichment is nonlinear. This nonlinear change in community stability was positively correlated with species asynchrony, species richness, and species diversity as well as the stability of dominant species and the stability of the grass functional group. Our data suggest a need to re-evaluate the mechanisms responsible for the effects of N deposition on natural ecosystem stability across multiple levels of N enrichment and that additional experimentation with gradients of N loads more similar to future atmospheric N deposition rates is needed. © 2017 Elsevier B.V. All rights reserved.

⁎ Corresponding author at: No.768 Jiayuguan West, Chengguan District, Lanzhou 730000, Gansu, China. E-mail addresses: [email protected] (X. Yuan), [email protected] (A.J. Cease), [email protected] (H. Wen), [email protected] (C. Zhang), [email protected] (H. Fu), [email protected] (J.J. Elser). 1 These authors contributed equally to this work.

https://doi.org/10.1016/j.scitotenv.2017.09.318 0048-9697/© 2017 Elsevier B.V. All rights reserved.

Please cite this article as: Niu, D., et al., The impact of nitrogen enrichment on grassland ecosystem stability depends on nitrogen addition level, Sci Total Environ (2017), https://doi.org/10.1016/j.scitotenv.2017.09.318

2

D. Niu et al. / Science of the Total Environment xxx (2017) xxx–xxx

1. Introduction With rapidly increasing use of fossil fuels and artificial fertilizers, global ecosystems will continue to receive elevated atmospheric N deposition (Tilman, 1999; Dentener et al., 2006). Indeed, from now to the end of this century, N deposition is predicted to increase 2.5-fold (Galloway et al., 2008) over current atmospheric N deposition rates (worldwide) of 0.05 to 2 g N m−2 y− 1 (Galloway et al., 2008) and 0.042 to 2.42 g N m−2 y−1 (in China) (Liu et al., 2013). Given that N is often the limiting nutrient for plant growth in terrestrial ecosystems (Vitousek et al., 2002; Elser et al., 2007), changes in N availability may cause profound changes in plant productivity and community structure (Galloway et al., 2008; Bai et al., 2010; Hautier et al., 2014), subsequently influencing ecosystem stability (Grman et al., 2010; Yang et al., 2011; Yang et al., 2012; Isbell et al., 2013; Hautier et al., 2015; Zhang et al., 2016). Thus, a better understanding of shifts in plant community structure and composition and the response of ecosystem stability to N enrichment is critical for projecting the impacts of future global change scenarios. Rapid species losses and increases in productivity under N enrichment have consistently been reported (Clark and Tilman, 2008; Bai et al., 2010; Yang et al., 2011; Avolio et al., 2014) but only a few studies have reported the subsequent effects on ecosystem stability (Grman et al., 2010; Yang et al., 2011; Yang et al., 2012; Gao et al., 2014; Zhang et al., 2016). Accumulating evidence from field observations (Tilman and Downing, 1994; Yang et al., 2012), laboratory experiments (Romanuk et al., 2006; Hector et al., 2010), and theoretical models (Lehman and Tilman, 2000; Loreau and de Mazancourt, 2013; Gross et al., 2014) suggest that greater diversity is associated with greater ecosystem stability. Several underlying mechanisms likely explain these positive diversity–stability relationships, including: • The overyielding effect (how much more a species grows when grown with other species than it does in monoculture, including increasing the complementary effect if species interactions increase niche partitioning or facilitation and decreasing the selection effect if species interactions favor unproductive species; Tilman, 1999; Lehman and Tilman, 2000; Isbell et al., 2009). • Species asynchrony (in which decreases in the abundance of some species are compensated for by increases in the abundance of another species; Loreau and de Mazancourt, 2013; Hautier et al., 2014). • The portfolio effect (statistical averaging in which temporal variance in species abundances scales with mean species abundance with slope N 1; Loreau and Hector, 2001; Isbell et al., 2009; Gherardi and Sala, 2015). • The covariance effect (in which there is increasingly negative covariance in the abundance of competing species at higher species diversity; Hector et al., 2010). • Increases in effective diversity, including species dominance and evenness (Grman et al., 2010; Yang et al., 2012).

northern China. They found that higher species diversity was associated with greater temporal stability resulting from several mechanisms: population asynchrony, the portfolio effect, and the greater stability of component populations. The divergent role of various mechanisms for sustaining ecosystem stability under N enrichment in the above reports might reflect the influence of different geographic regions with distinct climate and vegetation. While previous studies have provided a detailed view of grassland ecosystem stability in response to N enrichment (Grman et al., 2010; Pan et al., 2010; Zeng et al., 2010; Yang et al., 2011; Mattingly and Reynolds, 2014), none these results can give a realistic projection of the impacts of N deposition on ecosystem stability because the majority of these reports were based on the experiments with high N addition levels that are unrealistic relative to current or even projected N deposition rates. For example, the minimum N addition doses in the studies summarized above (i.e. in Yang et al., 2011, Grman et al., 2010 and Yang et al., 2012) were 30 g N m−2 y−1, 12.5 g N m−2 y−1, and 10 g N m−2 y−1, respectively, which are, respectively, 15, 7, and 5 times the world ‘s highest atmospheric nitrogen deposition rate (2 g N m−2 y−1; Galloway et al., 2008). Gomez-Casanovas et al. (2016) mentioned that loads commonly used in experimental N additions were higher than levels these ecosystems are predicted to experience in the future, which renders highly uncertain predictions of ecosystem function (e.g. greenhouse gas emission), especially if the responses to N deposition are nonlinear. Humbert et al. (2016) also reported that impacts of N addition on plant biodiversity in mountain grasslands depend on dose, application duration, and climate. Thus, further investigation of grassland ecosystem stability to N enrichment should involve multiple N loads that are more consistent with current and future N deposition rates. Based on the reasoning just described, we hypothesized that the strength of N enrichment's influence on grassland ecosystem stability depends on the dose of N addition. If this is true, then the response of grassland ecosystem stability to low N addition levels would be different from that under high N addition levels. To test this and to evaluate underlying mechanisms, we conducted a manipulative experiment that simulated elevated atmospheric N deposition with several N addition levels (especially including low N addition levels) in a semiarid grassland on the Loess Plateau from 2009 to 2014. The Loess Plateau is located in northern China and is widely known for its serious soil erosion (Y.H. Zhang et al., 2014; C.P. Zhang et al., 2014). The soil is classified as Sierozem that is characteristic of the China Loess and dominated by clay (Li et al., 2010). This region is being subjected to increasing deposition of atmospheric N with ongoing development in China's western region (Wei et al., 2011). We predict that the impacts of low N addition level similar to current N deposition rates on grassland community stability will not be obvious because this region's soil has high buffering capacity. However, responses of community stability to high N addition are likely to be pronounced as soil buffering is overcome (Aerts and Chapin, 1999). 2. Material and methods 2.1. Site description

Thus, the general supported pattern is that stability is positively correlated with diversity but there is considerable discussion of the relative contributions of these mechanisms. Furthermore, the diversity-stability relationship itself, however, might be influenced by N enrichment. For instance, Yang et al. (2011) found that long-term N addition strongly reduced species richness but increased community stability by enhancing species dominance in an alpine meadow. Grman et al. (2010) found that, while loss of species diversity was observed after N enrichment, the stability of N-fertilized communities was unchanged, probably because of increased compensatory dynamics and increased abundance of stable dominant species in southwestern Michigan, USA. In addition, Yang et al. (2012) conducted a 7-year field experiment testing the effects of N addition on the relationship between diversity and temporal stability of herbaceous plant communities in a temperate steppe in

This study was conducted in a fenced (since 2005) grassland at the Semiarid Climate and Environment Observatory of Lanzhou University (SACOL) (35°57′N, 104°09′E; altitude 1966 m), about 40 km southeast of Lanzhou city, China (Huang et al., 2008). The region experiences a continental semi-arid climate with mean annual air temperature of 6.7 °C and mean annual precipitation of ~ 382 mm. The dominant grass is Stipa bungeana. The soil in this region is classified as Sierozem, a calcareous soil that is characteristic of the China loess (Li et al., 2010). 2.2. Experimental design The experiment used a complete block design with 30 4-m × 5-m blocks. Blocks were separated by walkways at least 0.5-m wide. There

Please cite this article as: Niu, D., et al., The impact of nitrogen enrichment on grassland ecosystem stability depends on nitrogen addition level, Sci Total Environ (2017), https://doi.org/10.1016/j.scitotenv.2017.09.318

D. Niu et al. / Science of the Total Environment xxx (2017) xxx–xxx

were five replicates for each of these six treatments including a control treatment and five N additions at 1.15, 2.30, 4.60, 9.20, and 13.80 g N m−2 yr− 1. Hereafter, these treatments will be denoted as: N0, N1, N2, N3, N4, and N5. Nitrogen additions were initiated in May 2009 and repeated twice annually with 50% of the annual N addition in late May and 50% in late June of each year. N was added as urea [CO(NH2)2] dissolved in 10 L water and applied to the plots on a rainy day with a portable sprayer. The control plot received 10 L of water without N fertilizer. The low levels of N addition-N1 (1.15 g N m−2 yr−1) and N2 (2.3 g N m−2 yr−1) were chosen to represent the current rate of atmospheric N deposition at this site, which was reported with highest rate of 2.2 g N m− 2 yr−1 and average 0.8 g N m−2 yr−1 (Lü and Tian, 2007). 2.3. Plant community composition monitoring Within each plot, a permanent quadrat of 1-m × 1-m was designated. From 2009 to 2014, plant density and species richness (S) were determined in each permanent quadrat when biomass reached its peak level. Plant cover was measured within the same quadrat by putting a 1-m × 1-m metal frame with 100 equally distributed grids above the canopy. The number of species in the measured quadrats was recorded as species richness of the plant community. Community composition was quantified by counting the occurrence of each species in the 100 grids. We used individual species frequency as the abundance of the species. To examine changes in functional groups over time following N addition, we classified each plant species on the basis of N-use strategies into grasses (G), N-fixing forbs (NF), and non-N fixing forbs (NNF). The abundances of functional groups were the sum of all the species abundance within the measured quadrats. Then, we summed the total cover of each functional type in each replicate for each year of study. 2.4. Plant aboveground productivity measurement In mid-August of each year, peak aboveground biomass was estimated by clipping at ground level within 1-m × 1-m quadrats randomly located in each plot and sorting the biomass into grasses, N-fixing forbs, and non-N fixing forbs. Dead biomass was also removed. Care was taken to not resample in previously clipped areas and the permanent quadrats were never clipped. After clipping, biomass was taken back to the laboratory and oven-dried at 65 °C for 48 h and then weighed to determine biomass. 2.5. Calculations Species evenness (E), Shannon-Wiener diversity (Di) and Simpson diversity (Do) were used to describe community diversity, and calculated as: E¼

−

P

Di ¼ −

P i ln P i ln S

X

Do ¼ 1−

P i ln P i

X

P 2i

ð1Þ ð2Þ ð3Þ

where Pi is the relative importance value of species i based on vegetation coverage, density and height; and S is the total number of species in each permanent quadrat. As plant biomass is correlated to plant cover in these sites (Yuan et al., 2016), we used plant cover instead of biomass to determine the temporal stability of communities and populations. By using this method, all the ecological information could come from the permanent quadrat, avoiding extra error due to spatial heterogeneity. Temporal stability (S = μ / δ; Eq. (5)) was quantified as the ratio of mean vegetation cover values to its

3

standard deviation from 2009 to 2014 (Tilman, 1999; Lehman and Tilman, 2000; Yang et al., 2011). Temporal stability was evaluated for total community cover and individual species cover and referred to as plant community stability [ST(S)] and population stability [Si(S)] and calculated, respectively, as: CiðSÞ Si ðSÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Var ½CiðSÞ

ð4Þ S

ST ðSÞ ¼

μ ∑i CiðSÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi S S δ ∑i Var ½CiðSÞ þ ∑i≠ j CovðCiðSÞ; CjðSÞÞ

ð5Þ

where S is the total number of species in each permanent quadrat. We denote Si(S) as temporal stability of species i coexisting in a community of S species and denote ST(S) as the community stability from 2009 to 2014. The stability values of three functional groups were calculated based on the functional group coverage values and their variance. It is difficult to determine species asynchrony in multispecies communities due to the strong mathematical constraints presented by negative covariations between species (Loreau and de Mazancourt, 2008). Using species synchrony is an alternative way to measure species asynchrony, which can indirectly reflect the variation in covariance (Loreau and de Mazancourt, 2008, 2013; Yang et al., 2011). Community-wide species synchrony (φb) in vegetation coverage (φb) was quantified via the following equation δ2 bT φb ¼  2 S ∑i¼1 δbi

ð6Þ

where δ2bT is the variance in community coverage and δbi is the standard deviation in coverage of species i in a community with S species over the years 2009–2014. From Eq. (6), φb fluctuates between 0 (perfect asynchrony) and 1 (perfect synchrony) (Loreau and de Mazancourt, 2008; Isbell et al., 2009; Yang et al., 2012). There is evidence for the portfolio effect when the temporal variance (δ2) in the coverage of a species scales with its mean coverage (m), following the power function: δ2 ¼ cmz

ð7Þ

where c is a constant and z is the scaling power (Tilman, 1999; Lehman and Tilman, 2000). Some studies suggest that diversity will increase plant community temporal stability when z N 1 (Tilman, 1999). To test for the portfolio effect at different N addition levels, temporal variance and mean cover of each species were determined in each permanent quadrat across the 6 years. The value z is the slope of the regression line of log (variance) versus log(mean) for each permanent quadrat. 2.6. Statistical analyses All statistical analyses were performed using SPSS 20.0 for Windows (USA), and statistical significance was set at α = 0.05. The total biomass, biomass of different functional groups, plant community diversity, and functional type abundance data were analyzed in two-way mixedmodel repeated-measure analysis of variance, using year as a between-subject effect and nitrogen treatment as the within-subject effect (Table S1). A General Linear Model (GLM) with a Duncan test was used to test the statistical difference in the mean values of different N addition treatments, simultaneously comparing the effects of N addition, year, and their interaction on the mean values. To determine the significance of differences between different levels of N addition, we performed ANOVA with Tukey post-hoc tests. We also performed ANOVA to test for the effect of N enrichment on temporal stability of community cover and species asynchrony. Temporal stability of

Please cite this article as: Niu, D., et al., The impact of nitrogen enrichment on grassland ecosystem stability depends on nitrogen addition level, Sci Total Environ (2017), https://doi.org/10.1016/j.scitotenv.2017.09.318

4

D. Niu et al. / Science of the Total Environment xxx (2017) xxx–xxx

population cover was analyzed using fertilization and species identity as fixed factors. To test the covariates that could explain the N addition effect on community temporal stability (including species richness, total cover, species evenness, species dominance, species diversity, and species asynchrony), we used simple linear regression. To determine the scaling power coefficient z, ANCOVA was used with N addition as a factor and the log(variance) of each species as the dependent variable and log(mean) as covariate in each permanent quadrat; the N addition versus log (mean) interaction was also tested. We used constrained linear ordination technique redundancy analysis (RDA) to analyze the response of community stability, population stability, and plant functional group stability to N addition. In this analysis, we used the common species that occurred frequently in all N addition plots, resulting in 9 species and 3 functional groups for ordination. The community stability data were used as the environmental variables and population stability and functional groups as the independent variable within the model. In this model, we used a restricted Monte Carlo permutation test (random permutations) to test the null hypothesis. RDA and Monte Carlo permutation tests were performed with the CANOCO 4.5 software. We also analyzed the relationships among stability in community cover, population cover, and functional group cover based on simple linear regression.

3. Results 3.1. Species richness, species diversity, species evenness and species dominance Across the six years, plant community biodiversity measures (S, E, Di and Do) showed modest declines with increasing N addition with sharp declines in treatments with high N addition (Fig. 1). For example, significant loss in species richness was found in the N5 (13.8 g N m−2 yr−1) treatment (P b 0.05; Fig. 1a), as well as for evenness, diversity and dominance (Fig. 1b–d). Across all years, the effect of N addition on species richness, evenness, diversity and dominance was relatively consistent as there was no significant interactive effect between N addition and year (Fig. 1; Table S1). 3.2. Plant biomass During our study, we grouped all plant species into three functional groups and assessed how these functional groups' biomass and aboveground total biomass (BAT) responded to the different N addition treatments over time (Fig. 2; Table S1). Overall, we found N addition produced different effects on functional group biomass and total

Fig. 1. Effects of N additions on plant community diversity indices [richness (S), evenness (J), Shannon diversity (Di), and dominance (Do)] across six years. Different bars indicate mean value (±SE) for each treatment (n = 5). N0, 0 g N m−2 yr−1; N1, 1.15 g N m−2 yr−1; N2, 2.30 g N m−2 yr−1; N3, 4.60 g N m−2 yr−1; N4, 9.20 g N m−2 yr−1; N5, 13.80 g N m−2 yr−1.

Please cite this article as: Niu, D., et al., The impact of nitrogen enrichment on grassland ecosystem stability depends on nitrogen addition level, Sci Total Environ (2017), https://doi.org/10.1016/j.scitotenv.2017.09.318

D. Niu et al. / Science of the Total Environment xxx (2017) xxx–xxx

biomass, respectively. Although N-fixing forbs biomass (BNF) and nonN-fixing forbs biomass (BNNF) showed non-significant responses to N addition, BNL showed a positive relationship with increasing N addition, an opposite trend than BL (Fig. 2c & d). Both BNF(F5,144 = 3.80, P b 0.05) and BNNF(F5,144 = 15.05, P b 0.05) varied significantly with year, but there was no significant N addition versus year interaction (Table S1). Grass biomass (BG) ranged from 39.70 to 154.54 g m−2 and increased significantly with N addition during the course of 6-year experiment (F5,35 = 14.22, P b 0.05; Fig. 2b) while BG also varied significantly with year (F5,144 = 37.78, P b 0.05; Table S1). No significant interactive effects of N addition versus year on BG biomass were detected (Table S1). BAT had a similar change as BG, increasing with the rate of N addition, and fluctuated from 47.67 to 213 g m−2 during the study period (Fig. 2a). In all plots, total biomass significantly increased with N addition compared to the control plot and reached a peak in the N4 treatment (9.6 g m−2) except 2009 and 2013 (Fig. 2a). Results of repeated-measures ANOVAs showed that BAT was significantly affected by N addition (F5,35 = 17.06, P b 0.05) and year (F5,144 = 48.14, P b 0.05) but no N versus year interaction was observed (Fig. 2a; Table S1). However, in 2011 the aboveground biomass was dramatically suppressed due to low precipitation (240.0 mm b annual precipitation 328 mm) (Fig. 2a; Fig. S1e). 3.3. Temporal stability pattern and underlying mechanisms The strength of N enrichment influence on grassland community stability depended on the dose of N addition (Fig. 3a). Relative to the control treatment (N0), there was no obvious influence on plant community stability in the N1 (1.15 g N m− 2 yr−1) and N2 (2.3 g N m−2 yr−1) treatments. However, when N addition increased from 4.6 to 13.8 N m−2 yr−1 (i.e. N3, N4, and N5), plant community stability declined significantly compared to the control treatment but then had a tendency to rise with further increase in N-addition level. In contrast, plant population stability was not significantly influenced by N enrichment (F5, 300 = 4.52; P N 0.05; Fig. 3b). Meanwhile, plant species synchrony was significantly affected by N addition but its response to increasing N dose was opposite of that for plant community stability (Fig. 3c), with its highest value in the N3 treatment (4.6 g N m−2 yr− 1) and its lowest in the N5 treatment (13.8 g N m−2 yr−1). Generally, the lowest value of species synchrony

5

indicates the highest value of species asynchrony. The effects of N enrichment on stability in mean population cover differed for various species (Table S2). For most species, temporal stability was not correlated with fertilization or some species disappeared completely with fertilization (Table S2). However, population cover for three species, Leymus secalinus, Leguminosae, and Artemisia annua, significantly increased with fertilization (all P b 0.05; Table S2). By linear regression analysis of plant community stability with six parameters as potential drivers, we found that mean cover, species evenness, and species dominance showed no significant relationship with plant community stability (Table 1; P N 0.05). However, there was a significantly negative relationship between plant species synchrony and plant community stability, which indicates that species asynchrony played a role in sustaining plant community stability in our experiment (Table 1; P b 0.05). We also found that species richness and species diversity were positively associated with plant community stability (Table 1; P b 0.05) and that there was a significant relationship between log-transformed variance and mean value of each plant species cover at each N addition level (P b 0.05; Fig. 4a–f). All the slopes of these relationships (scaling power z) were N 1, which also supports an inference that plant community temporal stability was sustained by plant species richness and diversity through a portfolio effect mechanism. Finally, N addition had a significant positive effect on the scaling power z (F5,229 = 9.44; P b 0.05; Fig. 4g–h), especially for high rates of N addition (N5; Fig. 4h) but had no significant effect on the species variance (F5,229 = 1; P N 0.05). An ordination diagram illustrates the divergent responses of plant species population stability and stability of three plant functional groups to N enrichment (Fig. 5), which may help explain the decline in community stability following N addition. Correlations of community stability, population stability, and functional stability showed that, among all species, only Stipa bungeana stability was positively correlated (P b 0.05) with community stability (except at N3 and N4 treatments (Table S3). The stability of grasses (G)′ was significantly correlated with community stability in the N1 and N5 treatments only while the stability of N-fixing forbs (NF) was associated with community stability only in the N2 treatment (Table S3). Most of the stability variations of various species as well as of functional groups had negative associations with community stability in the N3 treatment (Table S3).

Fig. 2. Responses of average aboveground biomass and different functional groups biomass to N addition from 2009 to 2014 based on two-way mixed-model repeated-measure analysis of variance. Different points indicate mean value (±SE, n = 5) for each treatment. See Fig. 1 for treatment abbreviations.

Please cite this article as: Niu, D., et al., The impact of nitrogen enrichment on grassland ecosystem stability depends on nitrogen addition level, Sci Total Environ (2017), https://doi.org/10.1016/j.scitotenv.2017.09.318

6

D. Niu et al. / Science of the Total Environment xxx (2017) xxx–xxx

Fig. 3. The effect of N addition on a) plant community stability, b) plant population stability, and c) plant species synchrony across six years (Mean ± SE, n = 5). Different letters above the error bar indicate significant differences among the N addition treatments based on Duncan's post hoc test when P b 0.05.

4. Discussion 4.1. Effect of N addition on plant biodiversity and community structure Both theoretical and empirical findings suggest that N inputs should impact the composition of different functional groups and as well as Table 1 Relationships between plant community stability and six parameters as potential drivers based on simple linear regression. The six parameters were included as covariates. F and R2 values were calculated using simple linear regression (n = 30). *: significant at P b 0.05, † not significant. Parameters

F1,29

R2

Mean community cover Species synchrony Species richness Species evenness Species diversity Species dominance

0.60† 14.37* 3.11* 0.51† 4.33* 2.68†

0.021 0.319 0.316 0.133 0.366 0.296

biodiversity in grassland ecosystems. In our study, species richness, evenness, and diversity as well as dominance tended to decline following N addition (Fig. 1; Table S1), congruent with previous studies. For instance, Avolio et al. (2014) found that richness decreased with N addition in a tallgrass prairie. Similarly, Bai et al. (2010) conducted a long-term N enrichment experiment in fenced temperate steppe and found a large loss in plant species richness at all N input rates. This richness loss was accompanied by increased dominance of annuals but decreased dominance of grasses and forbs. Y.H. Zhang et al. (2014) and C.P. Zhang et al. (2014) also carried out an experiment in temperate steppe with broader N addition doses and considered N addition frequency as a factor. They found that plant species richness decreased more quickly at high rates and with low frequency of N addition. Our results showed that the abundance and plant species biomass varied with N addition across years and that responses to N enrichment differed both for plant species and for plant functional groups (Fig. 2; Figs. S1 & S2; Table S1). These results are in agreement with previous studies from a temperate steppe (Yang et al., 2011), a subalpine grassland (Bassin et al., 2007), an alpine meadow (Yang et al., 2011) and a tallgrass prairie (Avolio et al., 2014). The divergence of response to N enrichment for different species and functional groups in our study may reflect several factors. First, dominant functional groups or species can exclude other species following enrichment of N due to their abilities to capture soil resources more quickly than others, ultimately influencing the dominance of such plants in the community (Díaz and Cabido, 2001; Grime, 2001; Lavorel and Garnier, 2002; Bai et al., 2004; Pakeman, 2004; Suding et al., 2005; Yang et al., 2011). Supporting this explanation in our study, G′ biomass, dominant functional group, increased with N addition (Fig. 2b) but biomass of NF and NNF declined with N addition (Fig. 2c & d). Second, increased plant productivity and canopy height due to N addition results in a shift in competition among plant species from belowground competition for nutrients to aboveground competition for light (Grime, 1973; Newman, 1973; Hautier et al., 2009; DeMalach et al., 2017). Light competition is sizeasymmetric, which means that faster-growing or taller species can capture much more light per unit size than lower-growing or shorter species, thus enhancing competitive exclusion among species (Hautier et al., 2009; DeMalach et al., 2017). In this study, the biomass of tall-stature species (e.g. grasses) increased with N addition, especially at high N enrichment levels (Fig. 2b; Figs. S1a, b & S2). This light interception then likely constrained the growth of shorter-stature species (e.g. N-fixing forbs) (Fig. 2; Figs. S1c & S2). Other mechanisms besides differences in individual plant species' competitive ability may also have contributed to inter-specific differences in response to N fertilization. For example, Suding et al. (2005) reported that, under the increased competition pressure at community-level, the smaller population sizes of rare species relative to the most abundant species can cause the random-loss of rare species. In line with the report of Suding et al. (2005), in our study we also found that the abundance of the NF and NNF functional groups (consisting of rare species) declined more strongly with high N addition (e.g. N5) than that for the grasses (G), which involved dominant species (Figs. S1 & S2). We also found that the cover of Gueldenstaedtia multiflora (an N-fixing forb) declined strongly with high N addition (e.g. N5) (Fig. S1c), similar to the result of Avolio et al. (2014). This response may indicate that the rhizobial symbiosis of these N-fixing forbs was disrupted, switching it from mutualism to parasitism following N addition and leading to a decline in N use efficiency (Johnson, 2010; Avolio et al., 2014). 4.2. Effect of N addition on ecosystem temporal stability Over six years, we found there was no obvious influence of N addition on plant community stability in the N1 (1.15 g N m−2 yr−1) and N2 (2.3 g N m−2 yr−1) treatments (Fig. 3a) that mimic current atmospheric N deposition rates in this region (0.2 to 2.2 g N m−2 yr−1; Lü and Tian, 2007). Current atmospheric N deposition rates are 0.05 to

Please cite this article as: Niu, D., et al., The impact of nitrogen enrichment on grassland ecosystem stability depends on nitrogen addition level, Sci Total Environ (2017), https://doi.org/10.1016/j.scitotenv.2017.09.318

D. Niu et al. / Science of the Total Environment xxx (2017) xxx–xxx

7

Fig. 4. Linear relationship between the log-transformed variance and mean value of plant species cover at each N addition level. The point shows each plant species' data collected in a field plot basing on 6 years of monitoring from 2009 to 2014. R2 and P value were estimated based on linear regression. (a) N0; (b) N1; (c) N2; (d) N3; (e) N4; (f) N5; (g) Coefficient (z) of meanvariance scaling relationships in control plot (solid line) and N addition (dashed line) plots; (h) Effects of the rate of N addition on the slope.

2 g N m− 2 y− 1 worldwide (Galloway et al., 2008) and 0.042 to 2.42 g N m− 2 y− 1 in China (Liu et al., 2013). Thus, the results from our study indicate that low levels of nitrogen enrichment should not destabilize community dynamics in semiarid grasslands. In our study, however, while community stability had a tendency to increase with increased N addition when the dose increased from 4.6 to 13.8 g N m− 2 yr−1 (i.e. N3, N4, and N5), plant community stability was significantly lower in high N addition treatments than in the control (N0) (Fig. 3a) and there was a strong decline in stability across low levels of N addition. Thus, our results are different from previous reports that long-term N addition strongly increases plant community stability (Grman et al., 2010; Pan et al., 2010; Zeng et al., 2010; Yang et al., 2011; Mattingly and Reynolds, 2014). It is likely that the response of grassland ecosystem stability to N enrichment depends on the dose of N addition, supporting the reports of Humbert et al. (2016) who found that the responses of greenhouse gas (GHG) fluxes to N addition were highly nonlinear. The vast majority of previous studies used minimum doses of N addition higher than 10 g N m−2 yr−1 (Grman et al., 2010; Yang et al., 2011; Yang et al., 2012), which reflects levels that are well into the high N addition range in our study. Interestingly, plant community stability tended to increase when N addition dose increased from 4.6 to 13.6 g N m−2 yr− 1 (Fig. 3a). In other words, community stability in our study may have been greater than the control treatment if we had

established treatments with even higher N addition similar to those in previous studies. To our knowledge, there has been only one other study (Zhang et al., 2016) considering the effects of rate of N addition on ecosystem stability that involved multiple N loads encompassing the low range of levels that we applied (1, 2, 3, 5, 10, 15, 20, and 50 g N m2 yr−1). Zhang et al. (2016) found a negative linear relationship between grassland ecosystem stability and the magnitude of N addition (Zhang et al., 2016). Humbert et al. (2016) also found that impacts of nitrogen addition on plant biodiversity in mountain grasslands depended on dose, application duration, and climate. Thus, the lack of an increase in stability at higher doses of N in Zhang et al.'s (2016) study might be attributed to differences in nitrogen addition dose, application duration, and climate from experiments reported elsewhere. In our study, an explanation for the lack of change in plant community stability in response to low N addition levels (b 2.3 N m−2 yr−1) may involve high soil buffering capacity to resource enrichment. The soil of Loess Plateau was developed from Quaternary aeolian soil (Wang et al., 2016). Thus, a main characteristic of the China Loess is clay dominance (Li et al., 2010). Aerts and Chapin (1999) noted that, because of the high buffering capacity of some soils' chemical properties, low levels of nutrient addition may not influence plant growth. We also found evidence in our study of neutral responses of plant species richness to low level N addition treatments (Fig. 1). Meanwhile, the

Please cite this article as: Niu, D., et al., The impact of nitrogen enrichment on grassland ecosystem stability depends on nitrogen addition level, Sci Total Environ (2017), https://doi.org/10.1016/j.scitotenv.2017.09.318

8

D. Niu et al. / Science of the Total Environment xxx (2017) xxx–xxx

Fig. 5. Ordination diagram showing the results of redundancy analysis (RDA) of community stability, plant species stability, and stability of three plant functional groups at each N addition level. ST denotes community stability, and G, NF and NNF denote stability of community, grasses, N-fixing forbs and non-N-fixing forbs, respectively. See Table S2 for abbreviations of species names. (a) N0; (b) N1; (c) N2; (d) N3; (e) N4; (f) N5.

strong changes in plant community stability for N addition levels from 4.6 to 13.6 N m−2 yr−1 may reflect stronger responses of plant diversity and productivity to increased resource enrichment as soil buffering was overcome (Figs. 1 & 2). In line with some previous studies suggesting that the effect of N addition on stability is mediated by its effect on diversity (Grman et al., 2010; Yang et al., 2012; Zhang et al., 2016), in our system we observed a link between modest losses in species diversity and community cover stability (Fig. 1a & c; Table 1). Regression analysis also revealed that species richness and species diversity were positively associated with temporal community stability (Table 1; P b 0.05). Meanwhile, in our study, the scaling exponent z was N 1 in all N addition treatments (Figs. 4 & 3), which also suggests a positive relationship between diversity and temporal stability (Doak et al., 1998; Tilman, 1999). In addition, the nonlinear response of plant community stability to N addition rates in this study also reflects its strong dependence on plant diversity, which decreased more slowly at low N enrichment than at high N enrichment (Fig. 1). Taken as a whole, the results presented here show that the relationship between plant community stability and N addition rate was nonlinear, which implies that assuming that the relationship is linear can either overestimate or underestimate plant community stability response to future N deposition. 4.3. Underlying mechanisms explaining the effects of N addition on community stability These nonlinear changes in community stability were positively correlated with species asynchrony, species richness, and species diversity as well as the stability of dominant species and of the grass functional group (Table 1; Tables S2 and S3). Previous studies have suggested that N addition decreases plant diversity and might then decrease community temporal stability via mechanisms involving species asynchrony, the portfolio effect, population stability, compensatory dynamics, the covariance effect, species dominance and species evenness (Tilman and Downing, 1994; Grman et al., 2010; Yang et al., 2011; Yang et al., 2012; Zhang et al., 2016). Species asynchrony means that different species have different responses to environmental fluctuations, such as rainfall events and nutrient pulses. This asynchrony in turn tends to generate compensatory

dynamics at the community level, ultimately yielding more stable ecosystem properties (Yang et al., 2011; Loreau and de Mazancourt, 2013). Grman et al. (2010) pointed out that species asynchrony might be driven by negatively correlated species responses to environmental drivers. Generally, the lowest value in species synchrony indicates the highest value in species asynchrony (Loreau and de Mazancourt, 2008). In our study, species asynchrony also changed nonlinearly with increased N addition (Fig. 3c) and a significantly positive relationship between plant species asynchrony and plant community stability was found (Fig. 3c; Table 1; P b 0.05). These findings indicate that species asynchrony likely played a key role in sustaining plant community stability. For instance, following increased N addition, a decrease in the cover of the sub-dominant species Cleistogenes squarrosa generated a negative effect on the grass functional group, which was compensated for by an increase in the cover of dominant species Stipa bungeana (Figs. S1a, b & S2). Finally, this species asynchrony increased the cover of the grass functional group and thus maintained plant community stability. A previous study conducted in this system (An et al., 2011) found that the shortage of plant available P was stronger than N with increased N addition, which may lead to an enhancement of the complementarity effect via the plants' nutrient utilization strategies (e.g. N-fixing species and non-N fixing species) and thereby maintain or increase community stability (Suding et al., 2005; Elser et al., 2007; Hautier et al., 2009, 2015; Isbell et al., 2013; Song and Yu, 2015). However, several recent studies on the effect of N addition on community stability have failed to find evidence of species asynchrony (Yang et al., 2011), perhaps because soil acidification following large N additions causes leaching of calcium and magnesium and activates mineral-associated aluminum, increasing vulnerability of species to environmental fluctuations or intensifying competition (Lucas et al., 2011; Wei et al., 2013). We also found a significant relationship between log-transformed variance and mean value of each plant species cover at each N addition level (P b 0.05; Fig. 4a–f), and all the slopes of these relationships (scaling power z) were N1. These results also support an inference that plant community temporal stability was sustained by plant species richness and diversity through a portfolio effect mechanism (Doak et al., 1998; Tilman, 1999). Isbell et al. (2009) reported that species interactions that favor unproductive species more than productive species can increase temporal

Please cite this article as: Niu, D., et al., The impact of nitrogen enrichment on grassland ecosystem stability depends on nitrogen addition level, Sci Total Environ (2017), https://doi.org/10.1016/j.scitotenv.2017.09.318

D. Niu et al. / Science of the Total Environment xxx (2017) xxx–xxx

stability by decreasing the selection effect and increasing the complementarity effect. In our study site, grasses are not only the dominant species but also the most productive species (e.g. Stipa bungeana; Leymus secalinus) and their biomass increased with N addition (Fig. 2; Fig. S1). This response might then modulate the selection effect and the complementarity effect and thus decrease community temporal stability. Furthermore, the results of RDA showed that population stability and the stability of the three functional groups tended to diverge following N enrichment (Fig. 5) and that perennial species (i.e. G) and annual species (i.e. NNF) were almost all negatively correlated with community stability in the N3 treatment. Annual species, which must regenerate each year from seed, might be more sensitive to climatic variability than perennials and thus have more compensatory dynamics (Grman et al., 2010). However, our results showed that annual species (e.g. Heteropappus altaicus; Fig. S1d; Table S2) were significantly lower in abundance in low N treatments, which may also contribute to low community stability in those low N addition treatments. In addition, Doak et al. (1998) reported that decreases in plant species evenness and increases in plant species dominance would reduce community stability while Isbell et al. (2009) noted that this effect would depend on whether the dominant species is the most stable. However, our results showed that species evenness and dominance were not significantly affected by N addition and that evenness and dominance were not associated with the N-induced decline in community temporal stability (Fig. 1c & f; Table S3). Our results showed that the scaling coefficient of mean-variance in N addition treatments was greater than control treatment (z = 1.76; Fig. 4g), which indicated that the response of dominant species that have a high mean biomass to N addition was more variable (Grman et al., 2010; Zhang et al., 2016) and thereby decreased the evenness effect and dominance effect. Thus, the nonlinear response of community temporal stability to N addition in our study is likely to be explained by species richness, species diversity, species asynchrony, and stability of dominant species. 5. Conclusions A six-year field experiment manipulating N addition significantly increased plant productivity but impaired plant diversity. In turn, this diversity loss decreased plant community temporal stability at moderately high levels of enrichment. Our study demonstrates that the response of community stability to N enrichment is nonlinear in this grassland ecosystem. Thus, approaches that assume that the relationship is linear can either overestimate or underestimate predicted plant community stability to future N deposition. At the mechanistic level, we found that nonlinear pattern in community stability following N enrichment was positively correlated with species asynchrony, species richness, species diversity, and the stability of dominant species and of grass functional group. These underlying mechanisms facilitate the maintenance of community stability and thus may contribute to future dynamics in regions experiencing amplified N deposition. Overall, our results suggest that predicting community stability of grasslands in response to changes in the atmospheric N deposition can be greatly improved by designing experiments with N loads comparable with future atmospheric N deposition rates. Acknowledgements This study was supported by National Basic Research Program of China (2016YFC0500506; 2014CB138703), the National Basic Research Program for Science and Technology in China (2012FY111900), National Natural Science Foundation of China (31572458, 31201837, 41671106, and 31602001) and Changjiang Scholars and Innovative Research Team in University (IRT13019). The authors are grateful to the Semi-Arid Climate and Environment Observatory of Lanzhou University (SACOL) for providing the meteorological data and supporting the fieldwork. We would like to thank Zhuo An, Yi Yang, Huige Han and Zeqin Teng for their help with sample collection and analysis.

9

Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.scitotenv.2017.09.318.

References Aerts, R., Chapin, F.S., 1999. The mineral nutrition of wild plants revisited: a re-evaluation of processes and patterns. Adv. Ecol. Res. 30, 1–67. An, Z., Niu, D.C., Wen, H.Y., et al., 2011. Effect of N addition on nutrient resorption efficiency and C:N:P stoichiometric characteristics in Stipa bungeana of steppe grasslands in the Loess Plateau, China. Chin. J. Plant Ecol. 35, 801–807. Avolio, M.L., Koerner, S.E., La Pierre, K.J., et al., 2014. Changes in plant community composition, not diversity, during a decade of nitrogen and phosphorus additions drive above-ground productivity in a tallgrass prairie. J. Ecol. 102, 1649–1660. Bai, Y.F., Han, X.G., Wu, J.G., et al., 2004. Ecosystem stability and compensatory effects in the Inner Mongolia grassland. Nature 431, 181–184. Bai, Y.F., Wu, J.G., Clark, C.M., et al., 2010. Tradeoffs and thresholds in the effects of nitrogen addition on biodiversity and ecosystem functioning: evidence from Inner Mongolia grasslands. Glob. Chang. Biol. 16, 358–372. Bassin, S., Volk, M., Suter, M., Buchmann, N., et al., 2007. Nitrogen deposition but not ozone affects productivity and community composition of subalpine grassland after 3 yr of treatment. New Phytol. 175, 523–534. Clark, C.M., Tilman, D., 2008. Loss of plant species after chronic low-level nitrogen deposition to prairie grasslands. Nature 451, 712–715. DeMalach, N., Zaady, E., Kadmon, R., 2017. Light asymmetry explains the effect of nutrient enrichment on grassland diversity. Ecol. Lett. 20, 60–69. Dentener, F., Stevenson, D.S., Ellingsen, K., et al., 2006. The global atmospheric environment for the next generation. Environ. Sci. Technol. 40, 3586–3594. Díaz, S., Cabido, M., 2001. Vive la difference: plant functional diversity matters to ecosystem processes. Trends Ecol. Evol. 16, 646–655. Doak, D.F., Bigger, D., Harding, E., et al., 1998. The statistical inevitability of stability-diversity relationships in community ecology. Am. Nat. 151, 264–276. Elser, J.J., Bracken, M.E., Cleland, E.E., et al., 2007. Global analysis of nitrogen and phosphorus limitation of primary producers in freshwater, marine and terrestrial ecosystems. Ecol. Lett. 10, 1135–1142. Galloway, J.N., Townsend, A.R., Erisman, J.W., et al., 2008. Transformation of the nitrogen cycle: recent trends, questions, and potential solutions. Science 320, 889–892. Gao, Y., He, N.P., Zhang, X.Y., 2014. Effects of reactive nitrogen deposition on terrestrial and aquatic ecosystems. Ecol. Eng. 70, 312–318. Gherardi, L.A., Sala, O.E., 2015. Enhanced interannual precipitation variability increases plant functional diversity that in turn ameliorates negative impact on productivity. Ecol. Lett. 18, 1293–1300. Gomez-Casanovas, N., Hudiburg, T.W., Bernacchi, C.J., et al., 2016. Nitrogen deposition and greenhouse gas emissions from grasslands: uncertainties and future directions. Glob. Chang. Biol. 22, 1348–1360. Grime, J.P., 1973. Competitive exclusion in herbaceous vegetation. Nature 242, 344–347 (UK). Grime, J.P., 2001. Plant strategies, vegetation processes and ecosystem properties. second ed. John Wiley & Sons, New York. Grman, E., Lau, J.A., Schoolmaster, D.R., et al., 2010. Mechanisms contributing to stability in ecosystem function depend on the environmental context. Ecol. Lett. 13, 1400–1410. Gross, K., Cardinale, B.J., Fox, J.W., et al., 2014. Species richness and the temporal stability of biomass production: a new analysis of recent biodiversity experiments. Am. Nat. 183, 1–12. Hautier, Y., Niklaus, P.A., Hector, A., 2009. Competition for light causes plant biodiversity loss after eutrophication. Science 324, 636–638. Hautier, Y., Seabloom, E.W., Borer, E.T., et al., 2014. Eutrophication weakens stabilizing effects of diversity in natural grasslands. Nature 508, 521–525. Hautier, Y., Tilman, D., Isbell, F., et al., 2015. Anthropogenic environmental changes affect ecosystem stability via biodiversity. Science 348, 336–340. Hector, A., Hautier, Y., Saner, P., et al., 2010. General stabilizing effects of plant diversity on grassland productivity through population asynchrony and overyielding. Ecology 91, 2213–2220. Huang, J.P., Zhang, W., Zuo, J.Q., et al., 2008. An overview of the semi-arid climate and environment research observatory over the Loess Plateau. Adv. Atmos. Sci. 25, 906–921. Humbert, J.Y., Dwyer, J.M., Andrey, A., et al., 2016. Impacts of nitrogen addition on plant biodiversity in mountain grasslands depend on dose, application duration and climate: a systematic review. Glob. Chang. Biol. 22, 110–120. Isbell, F.I., Polley, H.W., Wilsey, B.J., 2009. Biodiversity, productivity and the temporal stability of productivity: patterns and processes. Ecol. Lett. 12, 443–451. Isbell, F., Reich, P.B., Tilman, D., et al., 2013. Nutrient enrichment, biodiversity loss, and consequent declines in ecosystem productivity. Proc. Natl. Acad. Sci. 110, 11911–11916. Johnson, N.C., 2010. Resource stoichiometry elucidates the structure and function of arbuscular mycorrhizas across scales. New Phytol. 185, 631–647. Lavorel, S., Garnier, E., 2002. Predicting changes in community composition and ecosystem functioning from plant traits: revisiting the Holy Grail. Funct. Ecol. 16, 545–556. Lehman, C.L., Tilman, D., 2000. Biodiversity, stability, and productivity in competitive communities. Am. Nat. 156, 534–552. Li, X.D., Fu, H., Guo, D., et al., 2010. Partitioning soil respiration and assessing the carbon balance in a Setaria italica (L.) Beauv. Cropland on the Loess Plateau, Northern China. Soil Biol. Biochem. 42, 337–346.

Please cite this article as: Niu, D., et al., The impact of nitrogen enrichment on grassland ecosystem stability depends on nitrogen addition level, Sci Total Environ (2017), https://doi.org/10.1016/j.scitotenv.2017.09.318

10

D. Niu et al. / Science of the Total Environment xxx (2017) xxx–xxx

Liu, X.J., Zhang, Y., Han, W.X., et al., 2013. Enhanced nitrogen deposition over China. Nature 494, 459–462. Loreau, M., de Mazancourt, C., 2008. Species synchrony and its drivers: neutral and nonneutral community dynamics in fluctuating environments. Am. Nat. 172, E48–E66. Loreau, M., de Mazancourt, C., 2013. Biodiversity and ecosystem stability: a synthesis of underlying mechanisms. Ecol. Lett. 16, 106–115. Loreau, M., Hector, A., 2001. Partitioning selection and complementarity in biodiversity experiments. Nature 412, 72–76. Lü, C.Q., Tian, H.Q., 2007. Spatial and temporal patterns of nitrogen deposition in China: synthesis of observational data. J. Geophys. Res. Atmos. 112. Lucas, R.W., Klaminder, J., Futter, M.N., et al., 2011. A meta-analysis of the effects of nitrogen addition on base cations: implications for plants, soils, and streams. For. Ecol. Manag. 262, 95–104. Mattingly, W.B., Reynolds, H.L., 2014. Soil fertility alters the nature of plant–resource interactions in invaded grassland communities. Biol. Invasions 16, 2465–2478. Newman, E., 1973. Competition and diversity in herbaceous vegetation. Nature 244, 310. Pakeman, R.J., 2004. Consistency of plant species and trait responses to grazing along a productivity gradient: a multi-site analysis. J. Ecol. 92, 893–905. Pan, J.J., Widner, B., Ammerman, D., et al., 2010. Plant community and tissue chemistry responses to fertilizer and litter nutrient manipulations in a temperate grassland. Plant Ecol. 206, 139–150. Romanuk, T.N., Vogt, R.J., Kolasa, J., 2006. Nutrient enrichment weakens the stabilizing effect of species richness. Oikos 114, 291–302. Song, M.H., Yu, F.H., 2015. Reduced compensatory effects explain the nitrogen-mediated reduction in stability of an alpine meadow on the Tibetan Plateau. New Phytol. 207, 70–77. Suding, K.N., Collins, S.L., Gough, L., et al., 2005. Functional-and abundance-based mechanisms explain diversity loss due to N fertilization. Proc. Natl. Acad. Sci. U. S. A. 102, 4387–4392. Tilman, D., 1999. The ecological consequences of changes in biodiversity: a search for general principles. Ecology 80, 1455–1474.

Tilman, D., Downing, J.A., 1994. Biodiversity and stability in grasslands. Nature 367, 363–365. Vitousek, P.M., Hättenschwiler, S., Olander, L., et al., 2002. Nitrogen and nature. AMBIO J. Hum. Environ. 31, 97–101. Wang, X., Wei, H.T., Taheri, M., et al., 2016. Early Pleistocene climate in western arid central Asia inferred from loess-palaeosol sequences. Sci Rep 6, 20560. Wei, X.R., Shao, M.A., Fu, X.L., et al., 2011. The effects of land use on soil N mineralization during the growing season on the northern Loess Plateau of China. Geoderma 160, 590–598. Wei, C.Z., Yu, Q., Bai, E., et al., 2013. Nitrogen deposition weakens plant–microbe interactions in grassland ecosystems. Glob. Chang. Biol. 2013 (19), 3688–3697. Yang, Z.L., van Ruijven, J., Du, G.Z., 2011. The effects of long-term fertilization on the temporal stability of alpine meadow communities. Plant Soil 345, 315–324. Yang, H.J., Jiang, L., Li, L.H., et al., 2012. Diversity-dependent stability under mowing and nutrient addition: evidence from a 7-year grassland experiment. Ecol. Lett. 15, 619–626. Yuan, X.B., Niu, D.C., WU, S.J., et al., 2016. Estimation of aboveground biomass in the grassland of the Loess Plateau, Northern China. Acta Ecol. Sin. 36, 4081–4090. Zeng, D.H., Li, L.J., Fahey, T.J., et al., 2010. Effects of nitrogen addition on vegetation and ecosystem carbon in a semi-arid grassland. Biogeochemistry 98, 185–193. Zhang, Y.H., Lü, X.T., Isbell, F., et al., 2014. Rapid plant species loss at high rates and at low frequency of N addition in temperate steppe. Glob. Chang. Biol. 20, 3520–3529. Zhang, C.P., Niu, D.C., Hall, S.J., et al., 2014. Effects of simulated nitrogen deposition on soil respiration components and their temperature sensitivities in a semiarid grassland. Soil Biol. Biochem. 75, 113–123. Zhang, Y.H., Loreau, M., Lü, X.T., et al., 2016. Nitrogen enrichment weakens ecosystem stability through decreased species asynchrony and population stability in a temperate grassland. Glob. Chang. Biol. 22, 1445–1455.

Please cite this article as: Niu, D., et al., The impact of nitrogen enrichment on grassland ecosystem stability depends on nitrogen addition level, Sci Total Environ (2017), https://doi.org/10.1016/j.scitotenv.2017.09.318