Effects of Nitrogen Input on Community Structure of the Denitrifying Bacteria with Nitrous Oxide Reductase Gene (nosZ I): a Long-Term Pond Experiment

Excessive nitrogen (N) input is an important factor influencing aquatic ecosystems and has received increasing public attention in the past decades. It remains unclear how N input affects the denitrifying bacterial communities that play a key role in regulating N cycles in various ecosystems. To test our hypothesis—that the abundance and biodiversity of denitrifying bacterial communities decrease with increasing N—we compared the abundance and composition of denitrifying bacteria having nitrous oxide reductase gene (nosZ I) from sediments (0–20 cm) in five experimental ponds with different nitrogen fertilization treatment (TN10, TN20, TN30, TN40, TN50) using quantitative PCR and pyrosequencing techniques. We found that (1) N addition significantly decreased nosZ I gene abundance, (2) the Invsimpson and Shannon indices (reflecting biodiversity) first increased significantly along with the increasing N loading in TN10–TN40 followed by a decrease in TN50, (3) the beta diversity of the nosZ I denitrifier was clustered into three groups along the TN concentration levels: Cluster I (TN50), Cluster II (TN40), and Cluster III (TN10–TN30), (4) the proportions of Alphaproteobacteria and Betaproteobacteria in the high-N treatment (TN50) were significantly lower than in the lower N treatments (TN10–TN30). (5) The TN concentration was the most important factor driving the alteration of denitrifying bacteria assemblages. Our findings shed new light on the response of denitrification-related bacteria to long-term N loading at pond scale and on the response of denitrifying microorganisms to N pollution.


Introduction
Nitrogen (N) is a key element sustaining life and is one of the most abundant elements on earth [1]. In 2050, 60% of global production of N is expected to come from anthropogenic sources [2], such as industrial fertilizer production, biological fixation of N in agricultural systems, and combustion. The increases in N production have resulted in large fluxes of N in nature [3]. Unintentional N enrichment has a variety of negative environmental impacts, including soil acidification [4], reduction of global terrestrial biodiversity [5], increased nutrient runoff to aquatic ecosystems, causing eutrophication [6], and potential risk of toxicological stress on aquatic changes [7][8][9][10] can ultimately impact ecosystem services and human well-being [11].
In aquatic ecosystems, bacteria-mediated denitrification is potentially an important pathway for N removal by converting nitrate/nitrite into gaseous products (N 2 , NO, N 2 O) [12]. Mitigation of anthropogenic N 2 O release is important because it is greenhouse gas and an important contributor to ozone layer depletion [9]. N fertilizer has reduced N 2 O emission from field crops [10], and N addition increased the denitrification potential in the Broadbalk wheat experiment [13]. High N 2 O emissions could be the legacy effect from previous N addition to cropland or to an interactive effect of N addition and climate change [14]. However, the change of N 2 O-producing bacteria at different nitrogen loadings is unclear.
N addition increases microbial biomass [15], decreases fungal diversity, and alters the fungal community composition [16] in the soil. In lake sediments, N input enhanced the relative abundances of the genera Flavobacterium, Pseudomonas, Arenimonas, Novosphingobium, Massilia, Aquabacterium, and Bacillus but inhibited those of Sporacetigenium, Gaiella, Desulfatiglans, Nitrospira, and Haliangium [17]. As ecological functions tie up closely with microbial communities, the changes of microbial communities inevitably impact functions [18]. Soil studies have revealed that N addition decreased the population of microbial nitrogen fixers [19], weakened the biological nitrogen fixation capacity [20], and led to loss by denitrification [21,22]. However, Kramer et al. [23] found that the application of N in orchards significantly enhanced the activity and efficiency of soil denitrifiers and reduced nitrate leaching. The ultimate effect of N addition on denitrifiers, however, remains unclear and needs to be elucidated at contrasting N loadings in different ecosystems.
In this study, the clade I group of nosZ gene was selected to quantify the abundance and communities of denitrifiers, because most microbes with nosZ clade I are complete denitrifiers and play an important role in the N 2 O reduction in aquatic systems [24], while the nosZ clade II is more relevant in soils [25]. In order to investigate the direct influence of N on the abundance and composition of nosZ I denitrifying bacteria, we collected sediments (0-20 cm) from the long-term (4 years) experimental ponds of five nitrogen fertilization (TN10, 10 kg NH 4 Cl per month; TN20, 20 kg NH 4 Cl per month; TN30, 30 kg NH 4 Cl per month; TN40, 40 kg NH 4 Cl per month; TN50, 50 kg NH 4 Cl per month) in the part of Lake Bao'an in Wuhan, China. We hypothesize that (a) high-N concentration will reduce the abundance of denitrifying bacteria, (b) denitrifying bacteria communities will cluster according to the N gradient, and (c) the differences of nosZ I communities will reflect adaptive shifts by the microbial communities to the N concentration that they face. Relevant findings are expected to further the understanding on the response of denitrifying microorganisms to N pollution.

Experimental Design and Samples Collected
We performed the experiments in 10 ponds with a depth of 1.8 ± 0.2 m (mean ± standard error) and an area of ca. 0.08 ha for each in a warm and humid subtropical climate on the south bank of the middle reaches of the Yangtze River and northeast of Lake Bao'an (N 30° 17′ 17″, E 114° 43′ 45″) in China. At the time of the experiment, the ponds had received different loadings of nitrogen for about 4 years (since January 2016). The sediments and water were introduced from Lake Bao'an with the aim to create a natural lake system [26]. The experimental treatments consist of five fertilization regimes, with additions of NH 4 Cl of 10 (TN10), 20 (TN20), 30 (TN30), 40 (TN40), and 50 (TN50) kg per month. NH 4 Cl fertilizer (NH 4 Cl, ertilizer (NH NHofGuoyao Chemical Reagent Co., Ltd.) was added to ponds every month. Phosphate fertilizer was not added. In October 2019, samples from the upper approx. 20 cm of the sediment were collected using an in situ sediment gravity sampler (Rigo, Φ = 0 cm). We randomly collected three sediment samples from each pond and added the sediment to sampling tubes (LaMotte, #1055) that were stored in a portable refrigerator until being transported back to the laboratory. We collected a total of 30 sediment samples (3 individuals * 2 sample sites * 5 ponds). In the lab, one part of the samples was immediately stored at -80 imr microbial investigation, while the other part was used for nutrient analyses.

Physico-chemical Parameters
Sediment samples were naturally air dried at room temperature and sieved through a 1-mm screen. Subsequently, pH was determined in deionized water at a soil/solution ratio of 1:2.5 using HI 99121 pH meter (Hanna instruments Inc., Italy) [27]. The concentrations of KCl extractable nitrate (NO 3 − ) and ammonium (NH 4 + ) were determined [28]. Total nitrogen (TN) was determined by cadmium reduction method [29]. Available phosphorus (AP) was extracted with 0.5 M NaHCO 3 and determined with the molybdenum blue method, and available potassium (Avail K) was analyzed after extraction at a sediment-to-ammonium acetate ratio of 1:10 [30]. Organic matter (OM) was determined by potassium dichromate volumetric method [31]. All sediment properties were measured by the Analysis and Testing Center of the Chinese Academy of Agricultural Sciences (IEDA, CAAS, Beijing, China).

DNA Extraction and Quantitative PCR (qPCR) Analysis
The Fast DNA SPIN Kit (MP Biomedicals, OH, USA) was used to extract total microbial DNA from the sediment [16]. A standard curve with 10-fold serial dilution was generated by a plasmid containing a copy of randomly selected nos Z I PCR products with primer nosZ1840-F / nosZ2090-R [32]. The calculated equation was: abundance of gene copy number/μL= (amount/ μL*6.022*10 14 )/ (length * 324). The qPCR was conducted in a volume of 20 μL, which consisted of 1 μL DNA template, 8.4 μL sterile and DNA-free water, 10 μL premix (2×SG qPCR MasterMix), 0.3 μL forward primer (10 μM), and reverse primer (10 μM). The condition of qPCR was set as 95°C for 3 min, followed by 40 cycles of 95°C for 10 s, 55°C for 10 s, and 72°C for 20 s. The correlation coefficients of standard curves (R 2 ) were higher than E e ff cts of Nitrogen Input on Community Structure of the Denitrifying Bacteria with Nitrous… 455 0.99. Finally, the absolute abundance of gene copy numbers was calculated and normalized to 1μL DNA.

Pyrosequencing, Bioinformatics Processing, and Statistical Analysis
The nosZ I gene was amplified by a touchdown PCR using TransGene high-fidelity enzyme with primer nosZ1840-F / nosZ2090-R [32]. The raw sequences were submitted in the NCBI Sequence Reading Archive (SRA) with the registration number of PRJNA734490. The nucleotide sequences of nosZ I were analyzed with the QIIME-1.9.1 pipeline. In short, low-quality sequences were discarded, and the remaining sequences were converted to amino acid sequences using the Fun-Gene Pipeline of the Ribosomal Database Project [19]. The sequences that did not match the nitrous oxide reductase gene sequence were removed and the remaining high-quality sequences were subsampled based on the lowest number of reads in a sample. Operational Taxonomical Units (OTUs) were classified with a similarity of 95%. For alpha diversity, Shannon, Chao, and Shannon Even indices were calculated for each sample using the UPARSE pipeline [33]. For beta diversity, Bray-Curtis dissimilarities were calculated for the microbiota of all 30 samples using the R package vegan [20]. The Bray-Curtis dissimilarities were used for principal coordinate analysis (PCoA) and hierarchical clustering [34]. Analysis of variance was performed with a randomized complete block design using one-way analysis of variance (ANOVA) with Tukey's HSD test using IBM SPSS Statistics 21 [35]. We used the term "Differential genera" to refer to genera that are found in significantly different proportions among five groups. A phylogenetic tree was inferred of these differential genera using the neighbor joining method in MEGA v.6.1 and displayed using the iTOL (Interactive Tree Of Life, https:// itol. embl. de/) together with data on the average relative abundances of genera [20]. The correlations between nosZ I communities (at genus level) and physicochemical parameters were determined with redundancy analysis (RDA) using CANOCO 5.0.

Physico-chemical Characteristics of Sediments
Data on sediment characteristics are provided in Table 1. pH in TN40 and TN50 was significantly lower than in TN20 and TN30. The content of NH 4 + and TN increased with the increase of N levels. The content of AK in TN50 was significantly lower than at the other N levels. The contents of NO 3 − and AP in TN30 were significantly higher than in the other four treatments. However, the OM content did not differ significantly among the five groups

Alpha and Beta Diversity
The Invsimpson and Shannon indices in T10-T40 increased gradually and significantly with increasing N loading but decreased again in T50 (P < 0.05, Fig. 2C, D). There was no obvious difference among the five treatments in the richness indices (Ace and Chao) (P > 0.05, Fig. 2A, B). Correlation analysis showed that Ace was positively correlated with TN (Table 2, R=0.92, P < 0.05) and AK negatively with Ace (R=0.98, P < 0.01) and Chao (R=0.96, P < 0.05).
Principal coordinate analysis showed a close correlation between TN and the nosZ I denitrifier communities (Fig. 3A). The community structures in TN30 were similar to those in TN10 and TN20, while in TN40 it differed significantly from that of the first three groups, and TN50 was separated significantly from the other four treatment groups. Along the PC1 axis, the nosZ bacterial communities in TN50 were separated from the other four treatment groups.
The results from the hierarchical clustering analyses were consistent with those of the PCoA analyses. The nosZ I communities in TN50 (Cluster I) were separated from the other four groups. In the remaining four treatment groups, most samples of TN10 and TN20 (8 out of 12 samples, Cluster II) were not in the same branch as TN30 and TN40 (Cluster III) (Fig. 3B).

Taxa Composition
The sequencing results (raw data and high-quality reads) are shown in Table S1. A total of 3912958 raw data and 3652792 clean sequences were obtained from the 30 sediment samples, with an average of 121759 sequences per sample (Table S1). The coverage ranged from 98.5 to 99%, indicating that the depth of this sequencing is sufficient to reflect the denitrifying microorganisms in the samples (Table S1). A large number of the nosZ I gene sequences were affiliated with Proteobacteria, accounting for 95.6-98.3% of the denitrifying bacteria in the sediment samples, followed by Terrabacteria (0.12-0.27%). At the class level, the four dominant classes showed different trends in the different treatments. The proportions of Alphaproteobacteria in TN40 and TN50 were lower than in TN10, TN20, and TN30 (Fig. 4A). Similarly, the class Betaproteobacteria in TN50 was significantly lower than in the other four groups (Fig. 4B). A different pattern was found for Gammaproteobacteria, which was significantly higher in TN50 than in the other four groups (Fig. 4C). There was no significant difference in the proportion of Deltaproteobacteria among the five groups (Fig. 5D).
At the order level, the proportions of Rhizobiales (Fig. 5A) and Nevskiales (Fig. 5B) in the high TN groups were significantly lower than in the low TN content treatments, while the trend was opposite for Pseudomonadales, Alteromonadales, Rhodospirillales, and Oceanospirillales ( Fig. 5E, F, G, H, respectively). The proportions of Nitrosomonadales (Fig. 5C) and Neisseriales (Fig. 5D) increased until TN30 and TN40, followed by a decrease in TN50.
In total, there were 26 differential significantly among the five groups, and they all belonged to the classes Alphaproteobacteria, Betaproteobacteria, or Gammaproteobacteria ( Fig. 6 and Table S2). For the most dominant genus Alcanivorax, the proportion increased significantly with increasing TN loading, and the proportion of TN50 increased to 48.6%, being 416 times higher than that of TN10. Pseudomonas, Marinobacter, and Rhodospirillum showed the same trend. However, the relative abundances of Mesorhizobium, Ralstonia, and Massilia decreased with the increase of TN (marked with orange shading in Table S2). The proportion of the 14 genera (marked with gray shading in Table S2) first increased and then decreased significantly with the increase of TN.

Discussion
We found that N addition in the ponds had a marked effect on the community and abundance of denitrifying bacteria, as assessed by marker gene analysis of sediment samples collected about 4 years after the initiation of the experimental treatments with contrasting N loadings (in the interval 10-50 kg NH 4 Cl per month) of the ponds.
Response of Physico-chemical Properties and nosZ Gene Abundance of Sediment to N Input Several studies have shown that the addition of N can change soil acidity due to the consumption or generation of H + [20,36,37]; for example, NO 3 − -based fertilizer can increase soil pH due to the consumption of H + [38]. We found that N addition decreased sediment pH ( Table 1), reflecting that we used a NH 4 + -based fertilizer. Furthermore, acidic conditions inhibited the expression of the nosZ gene and activity of the N 2 O reductase, which would increase N 2 O emission and the N 2 O/(N 2 O+N 2 ) ratio [39]. This suggests that the acidification of the environment caused by N addition may exacerbate climate warming. Final integral ecological consequences of N pollution can become more complicated under the increasing risks of climate warming and eutrophication in the future [40][41][42][43].
With the increase of NH 4 + addition, the NO 3 − content in the sediment increased and peaked in TN30, followed by a sharp decrease (TN40 and TN50) ( Table 1). These likely demonstrate that low doses of NH 4 + (TN10 and TN20) promote the metabolic activity of nitrifying bacteria [44] and increase the yield of NO 3 − ; when NH 4 + continues to increase, a shift occurs to inhibition of the metabolic activity and abundance of these bacteria. The high number of nosZ I genes in TN20 and TN30 and the lower copy number in TN40 and TN50 concur with this view. These results showed that more N decreased nosZ I genes, and Li reported that N 2 O emission from sediments or soils was closely negatively correlated to nosZ abundance [45]. Therefore, N will increase N 2 O emissions from aquatic environment, which may give us a warning that higher N addition (>40 kg NH 4 Cl per month) may increase the potential for ozone layer destruction in climate warming [46]. However, it remains to be verified by subsequent metatranscriptomics whether the nosZ I gene copies were transcribed into a corresponding active enzyme that can catalyze NH 4 + to produce NO 3 − . We found that the decrease of pH, caused by N addition, was significantly positively correlated with the copy number of nosZ I gene, which is consistent with the results of previous studies on the copy number of bacteria [47], fungi [14], ammonia-oxidizing prokaryotes [48], and N-fixing bacteria [16] in soil, and ammonia-oxidizing bacteria in eutrophic lake [49]. Harter et al. [50] also found that high pH (8.4) induced a reduction of N 2 O emissions and influenced the gene expression by regulating the N 2 O enhancement from soil. However, denitrifier-carrying nirK genes become more abundant with increasing N in soils, while denitrifier-carrying nirS may be at an advantage at low N [51].  Table S2 Friedl et al. [52] found that an increase in NO 3 − availability due to fertilization promoted the reduction of NO 3 − and suppressed nosZ I abundance in agricultural soils, while we found that there was a significant negative correlation between TN (not NO 3 − ) and nosZ abundance. These findings suggest that the suppression of the N 2 O reductase and increased N substrate availability may cause large pulses of N 2 O from the sediments as observed after N addition to soils [53].

Response of Alpha and Beta Diversity to N Input
The Invsimpson and Shannon indices showed similar pattern, both increased with N from TN10 to TN40 but decreased again in TN50. N addition may thus be a key factor controlling the biodiversity and heterogeneity of denitrifiers, as also found in a study of denitrification across a N gradient in a western U.S. watershed [53]. NosZ diversity has been shown to be positively associated with N 2 O emission during composting in agricultural soil [54,55]. Thus, the decreased diversity of the nosZ I gene observed here supports this conclusion and suggests that lower N addition (10-40 kg NH 4 Cl per month) increases, while high N (50 kg NH 4 Cl per month) decreases the functional capacity to remove N 2 O. The lower nosZ I copy number and diversity indicate that high-N addition will increase the risk of N 2 O emission. The beta diversity of nosZ I denitrifier was strongly influenced by N and clustered according to the concentration of N, as also seen for other microorganisms related to N cycling, including ammonia-oxidizing archaea [56] and N-fixing bacteria [51] in black soil using experimental plots.
We found that the changes of some denitrifying groups exhibited a certain regularity with increasing N at class, order, and genus levels. Alphaproteobacteria and Betaproteobacteria within nosZ gene, which indicate the genetic potential of such organisms to consume N 2 O [57], were lower in the high-N treatment. This result reflects that the addition of high N increases the possibility of N 2 O emission. Rhizobiales, which are known to dominate the denitrification process in wastewater treatment systems [58], soils [59], and lake sediments [60], have been shown to be significantly influenced by elevated CO 2 [59] in wheat roots, and isolates of this groups carry the nosZ gene in their genome and are able to affect N 2 O emissions during wheat growth [61]. In these studies, N had a pronounced effect on the structure of the N 2 O-reducing bacterial community. In the present study, Rhizobiales sharply decreased between TN30 and TN40 and remained low at TN50 (Fig. 4A), which is consistent with findings by Ling et al. [62] in soil. The genus Mesorhizobium, belonging to the order Rhizobiales, is reported to be able to denitrify under both aerobic and anaerobic conditions [63], and it was also significantly negatively affected by increasing N (Fig. 5). Bradyrhizobium (in the order Rhizobiales), which is known to be a major contributor to denitrification [64], reached the highest proportion in TN30 but then decreased markedly in TN40 and TN50 (Fig. 6), which corresponds well with changes of NO 3 − in sediments. Wang et al. [65] also reported the proportion of Bradyrhizobium to be positively correlated with sediment NO 3 − in a eutrophic lake. Thus, we propose that Bradyrhizobium (containing nosZ I gene) may use substantial amounts of available N to support their growth, whereas denitrification and denitrifiers will become increasingly less important when NO 3 − is low. Accordingly, the decreased NO 3 − in TN40 and TN50 might have restricted the growth of Bradyrhizobium, leading to increasing TN concentrations in the sediments [65].
Rhodospirillales, Pseudomonadales, Alteromonadales, and Oceanospirillales were more abundant in the high-N environment (Fig. 4E-H). The proportions of the genera Marinobacter (order Alteromonadales), Pseudomonas (order Pseudomonadales), Rhodospirillum (order Rhodospirillales), and Alcanivorax (order Oceanospirillales), were 2, 32, 34, and 416 times higher at the highest N loading than at the lowest loading, respectively ( Fig. 5 and Table S2). Species belonging to the genus Marinobacter can perform reduction of nitrate to N 2 by respiratory processes in anaerobiosis [66]. Species belonging to the genus Alcanivorax were isolated from organically enriched marine sediments and dominated in the bacterial consortium responsible for N removal [67]. The genus Pseudomonas has been reported as capable of reducing nitrate and nitrite, and the species P. stutzeri is a strong denitrifier and possesses the entire complement of denitrifying enzymes [64]. Thus, the significant increase in the proportion of Marinobacter Alcanivorax and Pseudomonas in the high-N treatment indicates an enhanced ability of eliminating nitrate or nitrite in the high-N treatment environment.

Conclusions
In conclusion, the physico-chemical characteristics, nosZ I gene abundance, and the composition of denitrifying bacterial communities of sediments in the experimental ponds in Lake Bao'an were clearly affected after 4 years of N application. The microbial community data presented support the idea that high concentrations of N application have negative effects on denitrifying bacteria in lake sediments, expressed as low diversity and abundance, which may diminish the denitrification capacity of N 2 O reduction to N 2 in lake sediments. The community structure of nosZ I denitrifiers was closely negatively related with the level of N addition as well. However, we only conducted research at the DNA level, and future research is needed to determine the impact of N on the functional diversity of denitrifiers