Soil Microbial Community Composition and Diversity Remained Unchanged in a Semiarid Grassland in Northwestern China After 7 Years of Nitrogen Addition

Qian Guo Northwest A&F University https://orcid.org/0000-0001-7780-8452 Zhongming Wen Northwest A&F University Hossein Ghanizadeh Massey University Cheng Zheng Northwest A&F University Yongming Fan Planning and Design Institute of National Forestry and Grassland Administration Jinxin Lu Northwest A&F University Xue Yang Northwest A&F University Xinhui Yan Northwest A&F University Sihui Chen Northwest A&F University Wei Li (  liwei2013@nwsuaf.edu.cn ) Northwest A&F University

. Nitrogen deposition can also improve plant productivity (Avolio et al. 2014) and carbon (C) sequestration in some ecosystems (Lu et al. 2021). However, excess N inputs can negatively affect ecosystems by increasing soil acidi cation , reducing plant diversity , and diminishing soil microbial biomass and diversity (Liu et al. 2011;Wang et al. 2018a). In addition, excess N inputs have detrimental impacts on microbial community stability and activity , microbial community structures (Liu et al. 2017) and the utilization e ciency of microbial C and N . A meta-analytic assessment of 151 studies showed that excess N reduced total microbial biomass, fungal biomass, bacterial biomass, microbial respiration, and microbial community compositions in many terrestrial ecosystems (Zhang et al. 2018b). Also, it has been shown that excess N increased the relative abundance of copiotrophic phyla (Ascomycota, and Proteobacteria), but reduced that of oligotrophic phyla (Basidiomycota, Acidobacteria and Firmicutes) (  , and ecosystem sustainability (Toju et al. 2018). However, the global soil microbial community is threatened by N deposition, with N deposition negatively affecting soil microbial community structures and the function of ecosystems. Alterations in microbial community are associated with the differences in the community assembly process. Niche theory and neutral theory are two major theoretical models to explain community composition mechanisms in community ecology ). The niche theory posits that the shape of microbial communities is in uenced by deterministic abiotic and biotic factors such as environmental conditions and species interactions, whereas the neutral theory asserts that the microbial community assembly is a random diffusion process, and niche differences have no effects on it (Perronne et al. 2017). Several factors can affect microbial community assembly mechanisms, such as land use (Osburn et al. 2021), salinity ), type of rhizosphere minerals, and soil depth (Luan et al. 2020). How N deposition in uences microbial community assembly is a question of importance in designing appropriate management practices to minimize N deposition in the semiarid grasslands of northwest China.
The drivers changing soil microbial community are complex and di cult to disentangle as changes in soil microbial community can arise through alterations in soil properties (e.g. soil pH, soil dissolved organic carbon, available nitrogen) ( Changes in plant community and soil properties as a result of N addition can directly or indirectly affect the species diversity, community assembly process, community composition, microbial biomass and functional activities of soil microbial community (Craig et al. 2021). Here, we sought to determine whether N deposition altered the diversity, composition, and assembly processes of the bacterial and fungal communities in the semiarid steppe in China. For this, a high-throughput sequencing technology was used to assess the dynamic changes of soil microbial communities on the Loess Plateau in China in response to N addition. In addition, we studied the drivers affecting microbial community compositions to understand the links between N addition and microbial communities.

Study sites
This research was carried out at the national permanent scienti c research station in the NingXia Yunwu Mountain grassland ecosystem (106°21′-106°27′E, 36°10′-36°17′N), Ningxia Hui autonomous region, China. The region is dominated by a middle temperate semiarid climate, with an annual average temperature of 7.0°C and an annual average precipitation of 425 mm. The altitude of the area is 1800-2100 m, with the highest peak of 2148.4 m. According to the general soil classi cation system of China, the soil type is montane greycinnamon soil, which is equivalent to Haplic Calcisol in the FAO/UNESCO system. There are more than 297 plant species in the area, but the main species are Stipa grandis P.A.Smirn, Thymus mongolicus (Ronniger) Ronniger and Artemisia sacrorum Ledeb. (Cheng et al. 2016).

Experimental design and sampling
Thirty-six 6×10-m 2 plots composed of six N addition levels and six replicates were distributed in six columns and six rows. The plots were laid out in a randomized block design. Each plot was separated from the others by a 2-m buffer strip. The fertilization treatments consisted of different amounts of CO(NH2) 2 , namely 0, 5, 10, 20, 40 and 80 g m −2 yr −1 (hereafter referred to as N0, N1, N2, N3, N4 and N5), corresponding to 0, 2.34, 4.67, 9.34, 18.68 and 37.35 g N m 2 yr −1 , respectively. Each treatment had six replicate plots. The fertilizer was applied annually at the beginning of the growing season (usually in the end of April) from 2013 to 2019. The fertilizer was applied during a moderate rain event to avoid arti cial watering. The level of N was determined according to the local N deposition rate index and based on the international and domestic fertilization on the same type of grasslands. Each plot was divided into two subplots. One subplot of 4 m×6 m was used for vegetation monitoring, and the other subplot (6 m×6 m) was allocated for individual plant sampling. Vegetation monitoring and sampling were conducted using a 1 m × 1 m quadrat randomly placed in each plot in August 2019. In each subplot, the quadrat was placed at least 0.5 m away from the edge to avoid edge effects. The coverage, mean height, and abundance of each species were measured and recorded in each quadrat. Above ground material for all species was removed at ground level and placed in envelopes categorized by species. All samples were dried to a constant weight at 80°C and weighed.
In each quadrat, three soil samples were collected at soil depths of 0-10 cm using a 5 cm diameter cylinder auger. The samples of the same depth were pooled to form a single sample. Three quadrats were selected as replicates. Visible debris was removed from the pooled soil samples and each sample was divided into two subsamples. One soil subsample was immediately placed on ice and transferred to the laboratory where it was stored in a -80°C freezer prior to DNA analysis. The other subsample was air-dried and used for chemical analysis.

Leaf functional trait samples and measurements
Adjustments in plant functional traits signify the survival strategies of plants to changing environments. In this research, the response of four plant leaf functional traits, namely leaf carbon content (LC), leaf nitrogen content (LN), leaf phosphorus content (LP) and speci c leaf area (SLA) to N addition was evaluated. Leaf samples from 5 to 30 healthy dominant species individuals were collected in each plot. Leaf samples were collected from the upper canopy of each individual plant. The leaf samples collected from the same species were pooled to form a single leaf sample. The pooled samples (~ 20 g) were placed in paper bags and dried at 65°C for 48 h. The dried leaf samples were used to determine the contents of leaf carbon (LC), nitrogen (LN) and phosphorus (LP).

Determination index and method
The content of soil organic carbon (OC), and LC were determined using the dichromate oxidation method ). The content of soil total nitrogen (TN) and LN were determined using the Kjeldahl method ). The content of soil total phosphorus (TP) and LP were determined using the ammonium molybdate colorimetric method ). The content of soil alkali-hydrolyzable nitrogen (AN) and soil available phosphorus (AP) were determined using the ISNT (Roberts et al. 2011) and the Olsen methods ), respectively. The soil pH was measured by an automatic titrator (Metrohm 702, Swiss). Soil moisture content (SMC) was measured the gravimetric method (Liang and Wang, 2020). To assess SLA, images of the sampled leaves were obtained using a scanner (Yaxin-1242), and the images were analyzed using Image J software to estimate the leaf area (Guo et al. 2021). The leaf samples were then oven dried at 65°C for 48 h, and weighed. The SLA was estimated using Eq. 1.
The community weighted-mean trait values (CWM) was calculated using Eq. 2.
where p i represents the relative contribution of species i to the community, trait i donates the trait value of species i. Plant community richness was expressed by the number of species.

DNA extractions and high-throughput sequencing
The soil samples stored at −80°C were used for total genomic DNA extraction using the CTAB (cetyl trimethyl ammonium bromide) method (Robe et al. 2003). DNA concentration and purity was assessed on 1% agarose gels. The bacterial 16S rRNA genes of the V4 region were ampli ed following the method outlined previously using the primers 515F

Data analysis
Alpha diversity was applied to analyze the complexity of soil microbial diversity through Chao1, Shannon, and ACE indices. All of these indices were calculated using QIIME (v1.9.1, http://qiime.org/scripts/split_libraries_fastq.html) ( The effects of N addition on the components of species diversity and functional diversity were subjected to an analysis of variance (ANOVA), and the means were separated using the LSD test at 5% probability. All statistical analyses were performed using SPSS (Version 24.0).
The Pearson correlation analysis was performed to analyze the correlation between biotic and abiotic factors, and microbial diversity. The correlation heatmap was constructed using the OmicStudio tools at https://www.omicstudio.cn (Friedman and Alm 2012). The assembly processes (random vs. non-random) was assessed using a null model in the EcoSim R package, and the checkerboard score (C-score) was used to test the null hypothesis of the assembly process. The coe cient of niche width was calculated using the spaa package. The species contributing to dissimilarities in microbial community were determined using similarity percentage analysis (SIMPER) in the vegan package. Principal coordinates analysis (PCoA) and permutational multivariate analysis of variance (PERMANOVA) were used to determine the changes in microbial community compositions. The Kruskal-Wallis rank-sum test in the linear discriminant analysis effect size (LEfSe) method was used to identify the microbial taxa signi cantly affected by different N addition treatments following the method outlined previously (Segata et al. 2011). The linear discriminant analysis (LDA) (LDA > 4) was performed to estimate the effect size of each N level. The relationship between selected biotic and abiotic factors, and microbial community compositions was assessed using the piecewise structural equation model (SEM) (Lefcheck 2016). The SEM was performed using the piecewiseSEM package. Various criteria were used to determine the goodness of t of the SEM model, including Fisher's C statistic and AIC.
The representative sequences were then annotated using the Silva Database (http://www.arb-silva.de/) (Quast et al. 2013) based on the Mothur algorithm.

Result
The vegetation characteristics and soil properties across N addition treatments.
The N addition had a positive effect on LN CWM , and foliar N/P (Table 1). For instance, LN CWM and foliar N/P increased by 54% and 93% at N5 plots, respectively. The increase in foliar N/P of dominant species in grassland communities indicates that dominant plants gradually shifted from N-limited to P-limited. The N addition had a signi cantly negative in uence on diversity (S plant and H plant ). In addition, N addition had a positive effect on biomass, and but negatively affected LP CWM and SLA CWM , though the effect was not signi cant.
Increasing the level of N resulted in an inconsistent response in LC CWM and FDis; however, no signi cant effects were recorded for both parameters across all N addition treatments. The results of PCoA and PERMANOVA showed that N addition did not cause signi cant changes in plant community compositions (Fig. S2a, Table S1). An evaluation of soil properties showed that N addition had a small effect on soil properties ( Table 2). The results showed that increasing the level of N resulted in no signi cant changes in OC, TN, AN, TP, AP, pH, SMC, and these soil properties almost remained unaffected with N addition.  (Table 3). Also, there was no signi cant change in fungal α-diversity across all N addition treatments. Spearman's correlation coe cients showed that soil microbial α-diversity was weakly correlated with soil properties and vegetation characteristics (Fig. S1). Changes in soil bacterial richness (ACE, and Chao1 index) exhibited a negative correlation with SLA CWM (R 2 =-0.58, P=0.014; R 2 =-0.60, P=0.011) only (Fig. S1a). However, soil fungal diversity (ACE, Chao1 and Shannon indices) was weakly correlated with all factors (Fig. S1b). Composition of microbial communities.
LEfSe was used to determine the taxa that signi cantly differed in abundance under varied levels of N addition. For fungal communities, the signi cantly abundant taxa were Periconiaceae (at the family level) and Periconia (at the genus level) at N4 plots (Fig. S2b). However, Helotiales (at the order level) and Roesleria (at the genus level) were signi cantly abundant at N5 plots. The biomarkers of fungal community were associated with the Phylum Ascomycetes. However, no biomarkers were observed in the bacterial community across all N addition treatments (Fig. S2a).
PCoA was used to analyze changes in microbial community compositions across all N addition treatments, and PERMANOVA analysis was used to determine signi cant differences in microbial community compositions. PCoA showed that all plots were clustered together without signi cant separations (Fig. S3). PERMANOVA analysis showed that there was no signi cant change in bacterial (P=0.879) and fungal (P=0.060) community compositions across all N addition treatments (Table S1). Therefore, N addition did not change microbial community compositions.
Assembly and species turnover of microbial communities.
The niche width of microbial communities almost remained unchanged under different levels of N addition (Fig. S4). However, N altered the balance between random and deterministic processes in the microbial community (Fig. 2). C-score results showed that the value of standardized effect size (SES) changed signi cantly with increasing the level of N. Bacterial (N5, P<0.001) (Fig. 2a, Table S2) and fungal (N3, P<0.005; N4, P<0.001; N5, P<0.001) (Fig. 2b, Table S2) communities were transformed from random to deterministic processes in high levels of N.
Richness-based species exchange ratio (SERr) was used to quantify species turnover in microbial communities (Fig. 3a). The SERrs of bacterial and fungal communities ranged from 0.40-0.41 and 0.56-0.63, respectively. The SERrs of fungal communities were greater than those of bacterial communities, suggesting that the fungal communities were more susceptible to N addition. To explore the effects of species turnover on microbial community formation, the contribution of extinct and immigrated OTUs to the changes in microbial richness and community structure was assessed. In the bacterial communities, the immigrated OTUs accounted for more than 23% of the OTU richness under N addition, with the lowest (23.91%) and highest (29.59%) proportions recorded at N1 and N3 plots, respectively (Fig. 3b). In addition, 22.18 to 25.48% of the native OTUs were categorized as extinct OTUs, with the lowest and highest proportions recorded at N4 and N1 plots, respectively. In the fungal communities, the proportion of immigrated OTUs in the OTU richness under N addition varied from 28.04 to 52.00%, with the lowest and highest proportions recorded at N2 and N3 plots, respectively. The rate of OTUs extinction ranged from 30.17 to 52.93%, with the lowest and highest proportions recorded at N3 and N2 plots, respectively. These results indicated that N addition altered the OTU richness of microbial community. Also, it was noted that N addition led to higher proportions of immigrated and extinct OTUs in the fungal communities than the bacterial communities, indicating that fungal community compositions were more sensitive to N addition. However, the immigrated and extinct OTUs accounted for a small proportion of the relative abundance in microbial communities, and the proportions in bacterial and fungal communities varied from 1.25 to 1.85% and 2.13 to 5.56%, respectively. A high proportion of microbial richness along with a low proportion of community compositions indicate that species turnover had a greater contribution to the microbial richness than the microbial community structure. Similar results were also demonstrated by the SIMPER analysis, which showed that the immigrated and extinct OTUs contributed no more than 8% to the variation in microbial communities between the N treatments and control (Fig. 3c).
Drivers of microbial communities.
The structural equation model (SEM) illustrating the effects of soil properties and vegetation characteristics on soil microbial compositions and diversity is illustrated in Fig. 4. According to the results, 85 and 44% of the variance in bacterial and fungal community compositions, respectively, was explained by the tted model (Fig. 4a). The results showed that TN and plant community, respectively, had signi cantly positive and negative effects on the bacterial community, whereas pH and SLA CWM had signi cantly negative in uences on the fungal community. Overall, the results showed that plant community compositions (P < 0.05) and TN (P < 0.001) were important driving factors for bacterial community compositions, whereas fungal community compositions were mainly regulated by pH (P < 0.05) and SLA CWM (P < 0.05). A signi cant negative effect of SLA CWM on bacterial diversity was revealed by the model (Fig. 4b). However, none of the factors associated with soil properties and vegetation characteristics had a signi cant effect on fungal diversity, which was consistent with the results illustrated in Figure S1. The inconsistent response of microbial diversity to N addition recorded among several studies suggests that the effects of N addition on microbial diversity can be varied across different environments, ecosystem types, soil types and vegetation compositions. Our results showed that almost all key vegetation characteristics and soil properties were weakly correlated with changes in microbial diversity. However, there was a signi cantly negative correlation between SLA CWM and bacterial diversity. In contrast to our results, previous studies have reported a signi cant correlation between soil properties (e.g. soil pH) and microbial diversity (Zeng et al. 2016;Zhou et al. 2020), suggesting that the response of microbial diversity to soil properties depends on environmental habitats, soil types, the range of pH, and the size of samples (Li et al. 2016).
Both deterministic and random processes can drive the assembly of microbial communities, but the relative importance of each process can be determined by the environmental heterogeneity ). In the present study, a null model was used to assess the response of microbial community assembly to N addition. The results showed that the microbial community was initially dominated by the random processes, but the relative importance of deterministic processes increased with increasing the level of N. This result implies that microbial communities were controlled by random processes in the control group, whereas the deterministic processes dominated in Microbial community composition and its driving factors under N addition.
The results of the present research showed that the microbial community composition did not change signi cantly under N deposition. In addition, it was noted that the relative abundance of the dominant bacterial community phyla, including Proteobacteria, Acidobacteria, Actinobacteria, Gemmatimonadetes, Firmicutes, Chloro exi, and Rokubacteria, did not change signi cantly under N addition, which is inconsistent with the copiotrophic hypothesis (Fierer et al. 2012). According to the copiotrophic hypothesis, increasing N inputs can lead to a decline in the abundance of oligotrophic taxa while the abundance of copiotrophic taxa increases under N addition . Fungi compared to bacteria are more sensitive to N addition (Freedman et al. 2015). Although the composition of fungal community did not change, the relative abundance of dominant phyla and some taxa of fungal community changed signi cantly under N addition.
Compared with the control group, N addition increased the relative abundance of Ascomycota while the relative abundance of Basidiomycetes decreased with increasing N inputs. This observation is consistent with the prediction of copiotrophic hypothesis . The species of two genera, Periconia, and Roesleria whose relative abundance increased signi cantly with increasing N inputs were considered plant-pathogenic fungi (

Conclusion
In this research, the response of plants, soil and microbial community to N deposition was investigated. The results showed that plants compared to soil properties and microorganisms were more sensitive to N deposition. Nitrogen addition increased LN CWM and foliar N/P ratio, and as a result, it led to P limitation. Limitation of P has a detrimental impact on ecosystem productivity; thus, in grasslands, P needs to be supplied along with N application to improve the availability of soil P to plants, and to maintain the dynamic balance between P and N. The results of this research also showed that N addition only affected microbial assembly, while no signi cant effects on microbial diversity and compositions were observed under N addition. In addition, it was noted that the composition of microbial community was regulated by both vegetation characteristics and soil properties. Based on the SEM, plant community composition and TN determined the bacterial community compositions, while the composition of fungal community was mainly regulated by SLA CWM and soil pH. These results imply that at our study sites, both plant community composition and TN were better predictors of bacterial community compositions, whereas both SLA CWM and soil pH were better predictors of fungal community compositions. Studying the relationship between microbial community and N deposition improves our understandings of the response of grassland ecosystems to N deposition, as well as grassland management, utilization and conservation under N deposition in the future.

Declarations
Declaration of competing interest The authors declare that they have no known competing nancial interests or personal relationships that could have in uenced the work reported in this paper.

Figure 1
Changes in the relative abundances of bacteria (a) and fungi (b) across the N addition treatments (at the phylum level). The average relative abundance was estimated as a ratio between the abundance of the sequence type and the total number of sequences (n = 3) using normalized data. Different letters above bars within each taxon indicate signi cant differences (P < 0.05). The ns indicates no signi cant difference (P > 0.05).

Figure 2
Ecological processes shaping the bacterial (a) and fungal (b) community assembly. The checkerboard score (C-score) was used to test the null hypothesis of the assembly process No signi cant differences between the values of observed C-scores (C-scoreobs) and simulated C-scores (C-scoresim) indicate a random co-occurrence pattern. The signi cant difference between the values of C-scoreobs and C-scoresim indicates a non-random co-occurrence pattern. A standardized effect size (SES) of < -2 and > 2 represents aggregation and segregation, respectively. For more details on the analysis of C-score in these communities, see Supplementary Table S2.