Depression of Groundwater Table and Reduced Nitrogen Application Jointly Regulate the Bacterial Composition of nir S-Type and nir K-Type Genes in Agricultural Soil

: Despite the known inﬂuence of nitrogen fertilization and groundwater conditions on soil microbial communities, the e ﬀ ects of their interactions on bacterial composition of denitriﬁer communities have been rarely quantiﬁed. Therefore, a large lysimeter experiment was conducted to examine how and to what extent groundwater table changes and reduced nitrogen application would inﬂuence the bacterial composition of nir K-type and nir S-type genes. The bacterial composition of nir K-type and nir S-type genes were compared at two levels of N input and three groundwater table levels. Our results demonstrated that depression of groundwater table, reduced nitrogen application and their interactions would lead to drastic shifts in the bacterial composition of nir S-type and nir K-type genes. Structural equation models (SEMs) indicated that depression of groundwater table and reduced nitrogen application not only directly altered the species composition of denitriﬁer bacterial communities, but also indirectly inﬂuenced them through regulating soil nutrient and salinity. Furthermore, the variation in soil NO 3 − –N and electrical conductivity caused by depression of groundwater table and reduced nitrogen application played the most important role in altering the community composition of denitriﬁer bacterial communities. Together, our ﬁndings provide ﬁrst-hand evidence that depression of groundwater table and reduced nitrogen application jointly regulate the species composition of denitriﬁer bacterial communities in agricultural soil. We highlight that local environmental conditions such as groundwater table and soil attributes should be taken into account to enrich our knowledge of the impact of nitrogen fertilization on soil denitriﬁer bacterial communities, or even biogeochemical cycles.


Introduction
Global changes have caused substantial shifts in terrestrial ecosystem structure, processes and functioning via elevated greenhouse gas emissions and extensive nitrogen fertilization. As an essential component of greenhouse gases, nitrous oxide (N 2 O) has a warming potential almost 300 times larger than that of carbon dioxide [1]. Soil microbes play a crucial role in controlling ecosystem processes and functioning and regulating important biogeochemical cycles, such as nitrogen and carbon cycles [2][3][4]. Denitrification is one of the most crucial stages of the nitrogen (N) cycle which leads to nitrogen (N) losses in ecosystems through nitrogenous gases, including: N 2 , nitrous oxide (N 2 O) and nitric oxide (NO) [5]. It is notable that the denitrification process is principally regulated by two key denitrifying bacteria of nirK and nirS genes which can convert NO 2 − -N to NO [6,7]. Agricultural soil, the largest source of N 2 O, has been widely regarded as the main anthropogenic source of greenhouse gases [8,9]. Therefore, improving knowledge of the fundamental processes shaping the community composition of nirK-type and nirS-type denitrifier communities in agricultural soil can help us better understand the effect of global changes on biogeochemical cycles. Nitrogen is the most important restrictive factor for the growth of soil microbes, so nitrogen fertilizer application has been widely recognized as a key driving factor causing substantial changes in soil microbial composition in agricultural soil [10,11]. However, in order to increase crop yield, over-application of nitrogen fertilizers has caused a series of severe environmental problems, such as groundwater pollution, soil nitrate contamination and salinization [12][13][14][15][16]. More importantly, nitrogen form release will directly change soil fertility and the salt content of soil and underground water, thereby affecting the growth of soil microorganisms [17,18]. This may cause significant shifts in the community assembly of bacterial denitrifier communities in agricultural soil [8,19]. Nitrogen application can indirectly affect soil microbial composition by regulating soil properties and filed crops [20], or even directly influence soil microbes [11,17,21]. Therefore, over-application of nitrogen fertilizers also may exert profound negative influences on biogeochemical cycles (e.g., greenhouse gas emissions) through altering the abundance and composition of soil denitrifier communities [2][3][4][5]8,15]. It is therefore important to explore the appropriate nitrogen input amount to maintain sustainable agricultural development. In fact, the effect of nitrogen application on soil microbes has been well examined [14,20], but the impact varies greatly [22,23]. In addition, the response of bacterial composition to nitrogen input may differ between nirS and nirK genes [24]. In brief, the associated difference in the response of nirS and nirK genes to reduced nitrogen application still remains largely unclear.
The North China Plain is located in the eastern part of China, which is one of the most vital crop-producing regions of China. However, with the increase of population, the expansion of urban size, and the development of industry and agriculture, groundwater consumption has increased sharply, and a series of environmental geologic problems have occurred such as regional groundwater table fall, water quality worsening, surface subsidence and so on, especially in the North China Plain [25]. Some studies have found that groundwater table also has an important influence on soil microbes [18,20,26]. Groundwater table changes can cause obvious shifts in soil environmental conditions such as O 2 supply, salinity, pH and NO 3 − -N [27][28][29], in turn, altering the bacterial community structure of denitrifier communities [26]. For example, some studies have pointed out that groundwater table fluctuations have substantial impacts on the abundance for nirK-type, nirS-type and nosZ-type denitrifier communities [8,26]. Recent studies have examined the impact of groundwater table on soil denitrifier bacterial communities, but the results vary among different studies. Moreover, groundwater conditions and reduced nitrogen application are also likely to interact and determine the abundance and community structure of denitrifier bacterial communities [8]. For instance, over-release nitrogen forms will cause serious groundwater contamination [15,16,30]. However, rare tests have focused on the influence of depression of groundwater table, reduced nitrogen application and their interactions on the soil denitrifier bacterial communities in the agricultural soil of the North China Plain.
To explore how the depression of groundwater table, reduced nitrogen application and their interactions shape the community structure of denitrifier communities in agricultural soil, we conducted a large lysimeter experiment. The composition of nirS-type and nirK-type denitrifier communities was compared at two levels of N input (reduced N of 240 kg N ha −1 year −1 and conventional N of 300 kg N ha −1 year −1 ) and three groundwater table (2, 3 and 4 m). Soil nirK-type and nirS-type denitrifier communities were assessed based on the high throughput sequencing data of nirS-type and nirK-type genes on an Illumina. Specifically, we primarily attempt to examine the following questions-(1) Does the bacterial community composition of nirK-type and nirS-type genes vary remarkably between conventional N and reduced N levels? (2) Does groundwater table change lead to drastic shifts in the bacterial community composition of nirK-type and nirS-type genes? (3) How do depression of groundwater table and reduced nitrogen application affect nirK-type and nirS-type denitrifier communities?

Field Site and Experimental Description
The field site was established at the Xinxiang Agricultural Water and Soil Environment Field Scientific Observation and Experiment Station, Chinese Academy of Agricultural Sciences (35 • 1 N, 113 • 53 E), Xinxiang City, Henan Province, People's Republic of China. The mean annual air temperature and annual precipitation is 14.1 • C, 588.8 mm, respectively. Almost all precipitation is concentrated during July, August and September. The mean annual evapotranspiration is approximately 2000 mm. In 2019, this study was conducted on a large infiltration lysimeter platform, and all lysimeters were filled with silt loam soil obtained from the neighboring cropland. The site conditions are shown in Figure 1. The field site was sown with maize (Zea mays L.), where we examined six treatments produced by the interactions between two levels of N input (local conventional N of 300 kg N ha −1 year −1 (N300), reduced N of 240 kg N ha −1 year −1 (N240)) and three groundwater table (GW2, 2 m; GW3, 3 m and GW4, 4 m). Local conventional nitrogen level (300 kg N ha −1 year −1 ) was used as a control with 100% of normal nitrogen input. Excessively shallow groundwater table will limit the growth of crops and root aerobic respiration [31], and lead to soil waterlogging [32]. A previous study has observed that 2.5 m is a crucial groundwater depth to control soil salinization [28]. When the groundwater table is shallow, there are obvious two-ways of water exchange at the interface under the root layer, the upper soil moisture can infiltrate into deep soil and ground water, and crop evapotranspiration can partially utilize the groundwater. When the groundwater table becomes deeper, the soil moisture at the interface mainly migrated downward, and the critical groundwater depth for crop growth in the North China Plain is generally 4 m. Therefore, we used 2, 3 and 4 m as the treatments of groundwater depth in our tests. The experiment followed a random block design with four replicates. Furthermore, 120 K 2 O ha −1 year −1 and 150 kg P 2 O 5 ha −1 year −1 were applied in six treatments in the form of potassium sulphate and monocalcium phosphate, respectively. Seeding, fertilization and tillage operations were carried out at the same time. Furthermore, ground irrigation was conducted when the soil moisture was lower than 55% of the field water holding rate. A RS-XAJ-100 probe was buried in the 20 cm soil layer to monitor soil moisture and guide irrigation.

Soil Sampling and Physicochemical Properties Analysis
We randomly collected 5 soil subsamples from the 0-20 cm soil layer in each lysimeter and mixed together into one sample, so finally 24 soil samples were collected in total in this study. After that, each composite sample was sieved using a 2 mm screen and divided into two parts: one portion was stored in room temperature and air-dried for measuring the soil attributes, and the other was stored at −80 • C to extract DNA. For soil conditions, we measured soil organic carbon content, soil pH and total nitrogen content following the procedures described by Wang et al [33]. In addition, the conductivity meter was used to determine soil electrical conductivity, and then soil NO 3 − -N and NH 4 + -N contents were also measured using the methods of Li et al [34].

DNA Extraction, PCR Amplification and Illumina-Based Sequencing
E.Z.N.A. soil DNA kits (OMEGA, Norcross, Georgia, United States) was used to extract the Genomic DNA from the 0.5 g fresh soil samples according to the manufacturer's instructions. A 1% agarose gel electrophoresis and spectrophotometry were used to check the quality of extracted DNA. All the extracted DNA was stored at −20 • C for following analysis.
The denitrification functional gene was amplified with the universal primers of the forward nirS cd3a-F (5'-GTSAACGTSAAGGARACSGG-3') and the reverse R3cd-R According to the manufacturer's instructions, amplicons were extracted from 2% agarose gels, the purity was determined using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA) and the concentration were determined using QuantiFluor™-ST (Promega, Madison, WI, USA). The purified amplicon was sequenced isometrically and paired on an Illumina MiSeq platform.

Amplicon Sequencing and Sequencing Data Processing
High-quality sequences were extracted firstly by the QIIME package (Quantitative Insights Into Microbial Ecology) (v1.2.1, Gregory Caporaso, Flagstaff, AZ, USA). The raw sequences were selected according to the sequence length, quality, primer and tag, and the low-quality sequences were eliminated. Under the threshold of 97% identity, the unique sequence set was classified into operational taxonomic units (OTUs) performed with UCLUST. Usearch (version 8.0.1623, Robert Edgar, United States) was used to identify and remove chimeric sequences. UCLUST was used to analyze the taxonomy of each denitrification functional gene sequence against the nt database. To eliminate the effect of different sequencing depths on the analysis, a subset of 34,463 sequences per sample for nirK-type gene and 27,347 sequences per sample for nirS-type gene was randomly selected for the subsequent analysis, respectively.

Statistical Analyses
Bray-Curtis community dissimilarity matrices of nirK-type and nirS-type genes were evaluated (Hellinger-transformed the abundance data) and standardized environmental Euclidean distances of each variable were evaluated. Two-way ANOVA was used to examine the influence of groundwater table conditions, reduced nitrogen application and their interactions on soil attributes. Permutational analysis of variance (PERMANOVA) was performed to assess the differences in species composition (Bray-Curtis distance) of nirK-type and nirS-type genes among groundwater depth and nitrogen-treatment groups, and those differences in community composition were visualized through Nonmetric multidimensional scaling (NMDS).
Mantel tests (10,000 permutations) were carried out to test the significance of the influence of soil attributes on bacterial composition of nirS and nirK genes. Structural equation models (SEMs) were constructed to further test the indirect and direct impacts of groundwater depth changes, reduced nitrogen application and soil attributes on the soil bacterial composition of nirS and nirK genes. Before analysis, priori structural equation models were constructed in accordance with the associated theory ( Figure S1). The SEM model was fitted using the following indexes: comparative fit index (CFI), goodness of fit index (GFI), low root square mean error of approximation (RMSEA) and χ2 test [35]. SEM was performed with the "lavaan" package [36].

Soil Environmental Parameters
Two-way ANOVA showed that soil NO 3 − -N, NH 4 + -N, pH, EC, SOC and TSN were significantly changed by reduced nitrogen application and groundwater depth conditions (Table 1). Specifically, reduced nitrogen application remarkably decreased soil NO 3 − -N, EC, SOC and TSN, but increased soil NH 4 + -N and pH (Table 1 and Figure S2). Notably, soil NO 3 − -N, NH 4 + -N, EC, SOC and TSN were positively related to the depression of groundwater table (Table 1 and Figure S3). In addition, the interactions between groundwater table conditions and reduced nitrogen application significantly influenced soil pH, EC and SOC, but did not affect NO 3 − -N, NH 4 + -N and TSN (Table 1).

Soil Bacterial Composition of nirS-Type and nirK-Type Genes
Across six treatments, we identified a total of 827,112 and 656,328 high-quality bacterial sequences, which were classified into 3887 nirK-type OTUs and 3356 nirS-type OTUs, respectively. The dominant phyla of nirK-type bacterial denitrifier communities (relative abundance greater than 0.1%) across 24 samples were Proteobacteria, Actinobacteria, Chloroflexi and Planctomycetes, which occupied more than 55% of the total bacterial sequences (Table S1). nirS-type bacterial denitrifier communities were mainly dominated by Proteobacteria and Actinobacteria, which occupied 9.13% of the total bacterial sequences (Table S1). Furthermore, the unclassified nirK-type and nirS-type OTUs occupied 42.04% and 90.80% of bacterial sequences in total.
A total of nine dominant genera (relative abundance greater than 1.0%) were identified from all nirK-type OTUS ( Table 2 and Table S2). Principal component analysis (PCA) based on these dominant genera profiles demonstrated the significant separation across six treatments (Figure 2a). Moreover, the distribution of these dominant genera under six treatments showed that Bradyrhizobium and Devosia were dominant in N240GW2, Sinorhizobium was dominant in N300GW3, and Defluviimonas, Mesorhizobium and Rhizobium were dominant in N300GW2 and N300GW4 (Figure 2b). The results of Two-way ANOVA showed that reduced nitrogen application significantly affected the relative abundance of Mesorhizobium, Xanthomonas, Devosia, Rhizobium and Ensifer, but did not influence that of Sinorhizobium, Bosea, Bradyrhizobium and Defluviimonas ( Table 2 and Table S2). Groundwater table conditions significantly altered the relative abundance of Sinorhizobium, Mesorhizobium, Bosea, Devosia, Bradyrhizobium and Ensifer, but had no effect on that of other dominant genera. Besides Rhizobium and Ensifer, the interactions between groundwater table conditions and reduced nitrogen application could remarkably affect the abundance of the other seven dominant genera.  We used taxonomic abundances data from 9 nirK-type dominant genera (relative abundance greater than 1.0%) as quantitative variables, which were used to perform the PCA.
Only four dominant genera (relative abundance greater than 1.0%) were identified from all nirS-type OTUS (Table 3). PCA based on four dominant genera profiles indicated the significant separation between N300GW4 and other five treatments (Figure 3a). Meanwhile, the distribution of these dominant genera under six treatments showed that Azoarcus was dominant in N300GW4 (Figure 3b). Two-way ANOVA indicated that the relative abundance of four dominant genera were significantly affected by groundwater table conditions. Reduced nitrogen application meaningfully influenced the relative abundance of Pseudomonas, Azoarcus and Magnetospirillum. However, the interactions between groundwater table conditions and reduced nitrogen application only could change the relative abundance of Azoarcus.  We used taxonomic abundances data from 4 nirS-type dominant genera (relative abundance greater than 1.0%) as quantitative variables, which were used to perform the PCA.

Influences of Depression of Groundwater Table, Reduced Nitrogen Application and Soil Environmental Parameters on the Bacterial Composition of nirS-Type and nirK-Type Genes
The results of PERMANOVA and NMDS indicated that the variation in soil bacterial community composition of nirK-type and nirS-type genes was significantly regulated by reduced nitrogen application (nirK-type: F = 1.52, P < 0.05; nirS-type: F = 1.61, P < 0.05), groundwater table levels (nirK -type: F = 1.75, P < 0.01; nirS-type: F = 2.10, P < 0.01), and their interactions (nirK-type: F = 1.96, P < 0.01; nirS-type: F = 2.11, P < 0.01; Table 4 and Figure 4).  Mantel test showed that the bacterial community composition of nirK-type genes was significantly related to NO 3 − -N, EC, SOC and TSN, but not related to NH 4 + -N, pH ( Table 5). The species composition of nirS-type genes only had a correlation with NO 3 − -N and EC (Table 5). SEM further demonstrated that the bacterial community structure of nirK-type genes was mainly determined by groundwater table conditions together with reduced nitrogen application, NO 3 − -N, EC, SOC and TSN ( Figure 5). Groundwater table conditions, reduced nitrogen application, NO 3 − -N and EC jointly shaped the bacterial community structure of nirS-type genes ( Figure 5). Groundwater table conditions could indirectly and directly influence the species composition of nirK-type genes, but only had an indirect impact on that of nirS-type genes. In contrast, reduced nitrogen application had both direct and indirect influences on the species composition of nirS-type genes, but only could indirectly regulate that of nirK-type genes. Figure 5. SEM describing the direct and indirect effects of depression of groundwater table, reduced nitrogen application and soil attributes on the species composition (two axes from nonmetric multidimensional scaling) of nirK (a) and nirS genes (b). The solid red arrows indicate the significant direct paths (P < 0.05), and the dashed grey arrows represent the significant indirect paths (P < 0.05). N-input, reduced nitrogen application; GWD, groundwater table condition treatments; KC1, NMDS1 for nirK genes; SC1 and SC2, NMDS1 and NMDS2 for nirS genes.

Effects of Depression of Groundwater Table and Reduced Nitrogen Application on Soil Attributes
It is widely believed that reduced nitrogen application and increased groundwater table may have very important influences on soil physico-chemical conditions [37][38][39]. Our results showed that reduced nitrogen application led to a certain degree of decline in soil fertility, but it also could significantly reduce the soil salt content, thereby reducing the risk of soil salinization (Table 1 and Figure S2). We also found that both soil nutrient and salt contents obviously increased with the increase of groundwater table (Table 1 and Figure S2). Excessively shallow groundwater table will limit the growth of crops and root aerobic respiration [31]. An appropriately increased groundwater table can alleviate waterlogging and improve crop growth status [40,41]. However, an excessively lowered groundwater table will increase the difficulty of water absorption from groundwater and reduce crop water utilization, which would adversely influence the nutrient absorption of crops from the soil [42][43][44]. Furthermore, a previous study has demonstrated that groundwater salinity increases with the depression of groundwater table, and soil salt content is positively related to groundwater salinity [28]. Therefore, both soil nutrient and salt contents were positively related to groundwater table. More importantly, we found that the interactions between depression of groundwater table and reduced nitrogen application also had important influences on some soil physico-chemical properties ( Table 1). Taken together, these findings may suggest that future cropland management should simultaneously develop appropriate nitrogen input and groundwater table levels to improve soil physico-chemical conditions.

Effects of Altered Groundwater Table and Reduced Nitrogen Application on Bacterial Composition of Denitrifier Communities
We observed that depression of groundwater table, reduced nitrogen application and their interactions significantly affected the abundance of the dominant genera of both nirK-type and nirS-type genes (Tables 2 and 3), which may confirm the viewpoint that different groundwater table and nitrogen fertilisation levels cause significant changes in the abundance of denitrifier bacteria [8,26]. Our results further confirmed that the bacterial composition of nirK-type and nirS-type genes differed significantly between conventional N and reduced N levels (Tables 4 and 5), indicating that reduced nitrogen application will alter the community structure of denitrifier communities in agricultural soil [20]. Meanwhile, we also observed that depression of groundwater table would lead to drastic shifts in the bacterial composition of nirS-type and nirK-type genes. SEM showed that reduced nitrogen application and groundwater table changes could not only directly affect the species composition of denitrifier communities, but also indirectly influence it through causing significant variations in soil nutrient and salinity ( Figure 5). Furthermore, the bacterial composition of nirK-type and nirS-type genes was obviously changed by the interactions between depression of groundwater table and reduced nitrogen application.
Nitrogen input will directly change the soil fertility and salt content, thereby affecting the growth of soil microorganisms [17,18]. Groundwater table conditions can affect soil salt content, which in turn affects soil microorganisms and crop growth [28,32,45]. In addition, changes in groundwater table can also regulate soil nutrient conditions by affecting soil nutrient absorption of crops [43,44], thereby affecting soil microbial communities. Therefore, depression of groundwater table and reduced nitrogen application act simultaneously and synergistically to shape the species composition of bacterial denitrifier communities. Taken together, our results provide robust evidence that nitrogen fertilisation and groundwater table conditions jointly regulate soil denitrifier bacterial communities. More local environmental conditions, such as groundwater table and soil conditions should be taken into account to enrich our understanding of the impact of nitrogen fertilization on biogeochemical cycles.

Influences of Soil Attributes on Bacterial Composition of Denitrifier Communities
As the carrier of microbes, soil properties directly determine the distribution and composition of microbial communities [46][47][48]. In the past few decades, plenty of studies have assessed the impact of soil attributes on microbial communities in different ecosystems [46,49,50], they found that the relative influence of different soil factors varied among microbial taxa [51,52]. In our study, the bacterial composition of nirS-type and nirK-type genes was affected by soil factors (Figure 5), implying that the variations in soil properties caused by depression of groundwater table and reduced nitrogen application would lead to important shifts in community composition of soil bacterial denitrifier communities [17,26]. However, we also found that the composition of nirK-type genes was mainly determined by NO 3 − -N, EC, SOC and TSN ( Figure 5). In contrast, that of nirS-type genes was only shaped by NO 3 − -N and EC. This may indicate that the response of bacterial denitrifier communities to soil factors differs between nirS-type and nirK-type genes. More importantly, both NO 3 − -N and EC had the most important effects on the bacterial composition of nirS-type and nirK-type genes, which suggests that depression of groundwater table and reduced nitrogen application mainly regulate the composition of denitrifying bacterial communities by affecting soil nutrient and salinity. In addition, soil pH is widely regarded as the foremost driver shaping soil microbial community structure [46,53], yet results from our study showed that soil pH had no substantial effect on soil denitrifying bacterial communities. This may be partly because the soil samples collected in this study were all alkaline. These soil denitrifying communities had no response to soil pH, probably because of the absence of acidic soil in all samples [54], since the intracellular pH of most bacteria is close to neutral, and the optimum pH range for bacterial growth is usually narrow [55]. When soil pH is outside the adaptive range, microbes will grow slowly to enhance stress resistance [55]. Therefore, those slow-growing bacterial species may have higher resistance and adaptability to high soil pH. As a result, both bacteria of nirS-type and nirK-type genes were not sensitive to the relatively slight changes in pH (ranged from 9.02 to 9.65). In brief, future research also should take into account soil physicochemical properties among agricultural ecosystems to enrich our understanding of the impact of global changes on biogeochemical cycles.

Research Limitation and Future Perspectives
Together, our study provides first-hand evidence that nitrogen fertilization and groundwater table conditions jointly shape the bacterial composition of denitrifier communities. We highlight that local environmental conditions, including groundwater table and soil attributes should be taken into account to enrich our knowledge of the effect of nitrogen fertilization on soil denitrifying bacteria communities, or even biogeochemical cycles. It is notable that our study was a short-term experiment and only selected two levels of N input and three groundwater table levels, which may result in probable biased results. Therefore, longer observation period and more nitrogen input and groundwater table should be considered in further research to more precisely elucidate the intrinsic influence of nitrogen fertilization on agricultural ecosystems, and test the appropriate nitrogen input amount to maintain sustainable agricultural development.

Conclusions
Our research conducted a comprehensive evaluation about the influence of depression of groundwater table, reduced nitrogen application and their interactions on bacterial community structure of denitrifier communities in agricultural soil. Our results found that the bacterial composition of nirS-type and nirK-type genes was significantly changed by depression of groundwater table, reduced nitrogen application and their interactions. SEM demonstrated that reduced nitrogen application and groundwater table changes could not only directly affect the species composition of denitrifier communities, but also indirectly influence it by causing significant variations in soil nutrient and salinity. Furthermore, the variations in soil NO 3 − -N and EC caused by depression of groundwater table and reduced nitrogen application played the most important role in altering the community composition of denitrifier bacterial communities. Taken together, our findings provide first-hand evidence that nitrogen fertilization and groundwater table conditions jointly shape the bacterial composition of denitrifier communities.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2073-4441/12/12/3459/s1, Figure S1: Priori structural equation models including direct and indirect effects of depression of groundwater table, reduced nitrogen application and soil attributes on the soil bacterial composition of nirS and nirK genes, Figure S2.
Influence of reduced nitrogen application on soil physicochemical condition, Figure S3. Influence of depression of groundwater table on soil physicochemical condition, Table S1. Observed phyla from nirS-type and nirK-type denitrifier communities, Table S2. Effects of depression of groundwater table, reduced nitrogen application and their interaction on the relative abundance of dominant genera in nirK-type denitrifier communities.
Author Contributions: F.B., X.Q., P.L. and D.Q. designed this study; F.B. and Y.S. performed the field investigation and collected the data; F.B., X.Q. and Z.D. developed the methods; F.B., X.Q., J.W., W.G. and H.L. wrote the paper; J.W. conducted the language editing. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement:
The bacterial raw sequences of this study have been submitted to SRA of NCBI database, with the accession number PRJNA669295.

Conflicts of Interest:
All authors declare no conflict of interest