Soil Microbial Community Involved in Nitrogen Cycling in Rice Fields Treated with Antiozonant under Ambient Ozone

ABSTRACT

Ethylenediurea (EDU) can effectively mitigate the crop yield loss caused by ozone (O₃), a major, phytotoxic air pollutant. However, the relevant mechanisms are poorly understood, and the effect of EDU on soil ecosystems has not been comprehensively examined. In this study, a hybrid rice variety (Shenyou 63) was cultivated under ambient O₃ and sprayed with 450 ppm EDU or water every 10 days. Real time quantitative polymerase chain reaction (RT-qPCR) showed that EDU had no significant effect on the microbial abundance in either rhizospheric or bulk soils. By applying both metagenomic sequencing and the direct assembly of nitrogen (N)-cycling genes, EDU was found to decrease the abundance of functional genes related to nitrification and denitrification processes. Moreover, EDU increased the abundance of genes involved in N-fixing. Although the abundance of some functional genes did not change significantly, nonmetric multidimensional scaling (NMDS) and a principal coordinates analysis (PCoA) suggested that the microbial community structure involved in N cycling was altered by EDU. The relative abundances of nifH-and norB-harboring microbial genera in the rhizosphere responded differently to EDU, suggesting the existence of functional redundancy, which may play a key role in sustaining microbially mediated N-cycling under ambient O₃.

IMPORTANCE Ethylenediurea (EDU) is hitherto the most efficient phytoprotectant agent against O₃ stress. However, the underlying biological mechanisms of its mode of action are not clear, and the effects of EDU on the environment are still unknown, limiting its large-scale application in agriculture. Due to its sensitivity to environmental changes, the microbial community can be used as an indicator to assess the environmental impacts of agricultural practices on soil quality. This study aimed to unravel the effects of EDU spray on the abundance, community structure, and ecological functions of microbial communities in the rhizosphere of rice plants. Our study provides a deep insight into the impact of EDU spray on microbial-mediated N cycling and the structure of N-cycling microbial communities. Our findings help to elucidate the mode of action of EDU in alleviating O₃ stress in crops from the perspective of regulating the structure and function of the rhizospheric soil microbial community.

INTRODUCTION

Although the concentration of tropospheric ozone (O₃) has been decreasing in some countries in Europe and in the USA, the concentration of O₃ in East Asia has shown an increase (1), especially in China and India (2–4). As a greenhouse gas and a major air pollutant, O₃ is detrimental to crops and, thus, to crop production (5, 6), including rice (Oryza sativa L.), which is a major food crop that is widely grown and consumed in Asia (7, 8). O₃ directly causes crop damage through O₃-induced oxidative stress (9). After being absorbed into the apoplast through the stomata, O₃ rapidly degrades and produces reactive oxygen species (ROS), thereby interfering with various physiological processes (10, 11). As a result of O₃ stress, it has been estimated that about 4% of the global rice yield and more than 10% of the regional rice yield are lost (12, 13). Recently, it was estimated that the ambient O₃-caused relative yield loss for inbred and hybrid rice cultivars in China reached 19% and 42%, respectively, based on dose-response relationships that were derived from Asian field experiments (5). Therefore, it is important to protect rice and other crops against O₃ phytotoxicities.

Ethylenediurea (EDU) is the most efficient phytoprotectant against O₃ oxidative stress that is currently known (14). It has been widely used for the evaluation of the effects of O₃ on plants under both ambient and elevated O₃ concentrations (i.e., differences between no-EDU and EDU-treated plants are attributed to O₃ stress) (15–18). According to previous studies, EDU can increase photosynthetic pigments (19), improve photosynthetic capacity (20), and alter the concentrations of antioxidants (21) as well as the activities of antioxidant enzymes (16, 22). However, although EDU has been applied in research for more than 4 decades (23), its protective mechanism remains unclear (24–26). The main potential modes of action of EDU include: (i) directly degrading O₃ on the leaf surface (26), (ii) reducing O₃ stomatal uptake flux by decreasing stomatal conductance (27), (iii) removing reactive oxygen species (ROS) by increasing enzymatic and nonenzymatic antioxidants (28), (iv) disrupting homeostasis and producing a rebound response as a xenobiotic within a framework (24, 29), and (v) acting as a kind of nitrogen (N) fertilizer at high concentrations (800 to 1,600 mg L⁻¹) in N-demanding plants grown in N-impoverished soils (18, 30). However, whether EDU acts as an N fertilizer remains controversial. Although treatment with EDU at low concentrations (no more than 400 mg L⁻¹) had no effect on the N content in leaves (31, 32), it did increase the activities of enzymes involved in N-metabolism (33). Moreover, Shang et al. (31) recently showed that EDU ameliorated the yield loss of hybrid rice cultivar SY63, while it enhanced the allocation of leaf N to photosynthesis as well as the grain N accumulation.

Diverse and abundant microorganisms inhabit soils and play important roles in the cycling of nutrient elements, soil fertility, and crop productivity (34, 35). In the rhizosphere, plant roots preferentially stimulate or inhibit specific taxa, thereby resulting in the formation of a microbial community with higher microbial activity and fast biomass turnover that is distinct from that of bulk soil (36). Of particular note are plant growth-promoting rhizobacteria (PGPR) that can actively colonize the rhizosphere and have the ability to suppress various diseases (37). They are also key drivers of rhizosphere N turnover (38). Microorganisms (especially bacteria) are sensitive to environmental changes (39). Therefore, the microbial community can be used as an indicator of soil quality to evaluate the ecological impacts of environmental changes, including agrochemicals, while also shedding light on the potential involvement of soil microbes in the mode of action of EDU against O₃ phytotoxicity (40).

EDU is commonly applied as a soil drench or a foliar spray. A preliminary assessment suggested that a soil drench could directly influence the composition and functions of the soil microbial community (41). Plant growth-promoting bacteria applied as a foliar spray not only enhance plant growth but also cause successions in rhizosphere microbial activities (including both metabolism and community composition) indirectly though regulating the amount and quality of exudates (42, 43), which account for about 11 to 40% of the total carbon (C) fixed via photosynthesis (44). In return, the feedback of rhizospheric microbiota determine overall plant fitness (45). Generally, the rhizosphere processes and functions are regulated by multiple interactions among root exudates, soil nutrients, and soil microbes (46).

There have been some studies focusing on the effects of elevated O₃ on paddy soil microbial communities. Elevated O₃ was also found to negatively affect the biomass of bacteria, methanogenic archaea, and nematodes, as well as extracellular enzyme activities of paddy soil (47–50). The increased structure complexity of soil bacterial and fungal communities caused by elevated O₃ suggested a microbial survival strategy in response to limited resources (51, 52). It was noteworthy that these responses were dependent on crop cultivars. For example, the response of an O₃-sensitive cultivar was stronger than that of an O₃-tolerant cultivar (47, 53). Elevated O₃ decreased the availability of soil inorganic N and dissolved organic carbon, which serve as metabolized substrates or energy sources for microbial organisms, thereby resulting in reduced rates of soil nitrification and denitrification as well as decreasing N₂O emission (48, 54). In addition, elevated O₃ can reduce N-fixation rates or impact nodulation in legumes as well as disrupt N-fixing symbioses in nonlegumes (55). The reduction of N-fixation rates caused by elevated O₃ is due to the reduced availability and transportation of C assimilates to roots (56), which may be associated with the reduced root biomass (57). However, there has been no study focusing on the effects of elevated O₃ on soil microbiota-mediated N cycling for rice plants. Whether an EDU foliar spray can negate or even overcompensate the negative effects of elevated O₃ on N-fixation in agro-ecosystems is also still unknown.

Several studies attempted to explore the correlations between N fertilization and the structure of microbial communities (58–60). However, most of these studies focused on a limited number of genes by applying polymerase chain reaction (PCR)-based methods, resulting in possibly biased results caused by unequal amplification (61). Metagenomic sequencing can avoid the bias caused by PCR and provides a more accurate means by which to understand the roles of the microbial community in ecosystems (62). In studies using metagenomic sequencing, high-quality metagenomic assembly is essential for accurately mining the composition and function of microorganisms in particular environments. Gene-targeted metagenomic assembly combines annotation steps to provide an improved strategy for the discovery of the information of functional genes (63). Although the need to reveal the role of soil microorganisms in determining the effects of EDU on plants has been pointed out before (24), there is still no published study addressing it. Hence, the use of gene-targeted metagenomic technology is a promising approach by which to study the effect of EDU, a synthetic chemical of which approximately 22% is N, on soil microbial communities.

In the present study, Xander (63), a gene-targeted program, was applied to assemble N-cycling-related functional genes in order to evaluate the effects of EDU on the microbial communities involved in N cycling in the rhizospheric soil of rice plants under ambient O₃ conditions. The O_3-sensitive hybrid rice cultivar Shanyou 63 (64), which is widely cultivated in China, was selected as the test plant. We hypothesized that the relative abundance of N-cycling-related functional genes and their taxonomic distribution would be affected by EDU. We also hypothesized that the microbial communities that were involved in N cycling in the rhizosphere would be more sensitive to EDU than were those in the bulk soil.

RESULTS

Effects of EDU on the rhizospheric soil biochemical characterization of rice plants.

As shown in Fig. 1, the soil biochemical characterizations were significantly affected by EDU, except for the content of nitrate (NO₃^–−N). For the rhizosphere, EDU significantly decreased the pH as well as the contents of total nitrogen (TN), ammonium (NH₄⁺−N), available potassium (AK), and available phosphorus (AP) by 4%, 34%, 35%, 25%, 26%, and 44%, respectively, compared to the water treatment. EDU remarkably increased the content of organic matter (OM) by 36%, compared to the water treatment. For the bulk soil, EDU decreased the content of TN and increased the content of OM. Generally, rhizospheric soil was more sensitive to EDU, compared to bulk soil.

FIG 1.

Effects of ethylenediurea (EDU) and soil compartment on the pH (A), contents of organic matter (OM) (B), total nitrogen (TN) (C), ammonium (NH₄⁺–N) (D), nitrate (NO₃⁻−N) (E), available potassium (AK) (F), and available phosphorus (AP) (G), as well as the number of gene copies of the 16S rRNA gene (H) in soils. The error bars denote the standard error (n = 3). Different letters above bars indicate significant differences among treatments (P < 0.05).

The soil compartment also significantly affected the soil biochemical characteristics, except for the contents of TN and NH₄⁺–N (Fig. 1A–G). The pH value of the rhizosphere soil was significantly lower than that of the bulk soil across the two EDU treatments. The contents of TN and NO₃^–−N of the rhizospheric soil were significantly lower than those of the bulk soil (P < 0.01) under the EDU treatment, whereas the content of OM of the rhizosphere was higher than that of the bulk soil. The contents of NH₄⁺–N and AK of the rhizosphere were significantly higher than those of the bulk soil under the water treatment. In addition, the interaction of EDU and the soil compartment significantly affected all of the tested biochemical characteristics (Fig. 1A–F), except for the content of AP. (Fig. 1G).

Effects of EDU on the microbial abundance of rhizospheric soil.

Based on metagenome sequencing and annotation, bacteria were found to be the dominant microbial group, accounting for 99.5%. The 16S rRNA gene copies were used to estimate the bacterial abundance by applying quantitative PCR (qPCR). The data showed that the number of the 16S rRNA gene copies ranged from 2.89 × 10⁹ to 4.96 × 10⁹ copies g⁻¹ soil in rhizospheric and bulk soils across different treatments. However, no significant differences were found between the EDU and water treatments or between the rhizospheric and bulk soils (Fig. 1H).

Effects of EDU on the abundance of functional genes involved in N cycling.

As shown in Fig. 2A, EDU decreased the abundance of some N-cycling related functional genes of the rhizospheric soil, including amoA-AOA, amoA-AOB, nirK, and norB. However, EDU significantly increased the abundance of the nifH gene. The results suggested that EDU led to succession toward the rhizospheric microbial community, with a decreased abundance of microbes being involved in nitrification and denitrification processes and an increased abundance of microorganisms being involved in the N fixation process. Notably, EDU significantly decreased the abundance of nirK that encodes heme-containing nitrite reductases in the rhizospheric soil. The abundance of nirS, which encodes copper-containing nitrite reductase, showed little variation in the rhizosphere. For the bulk soils, EDU also increased the abundance of the nifH gene. Unlike the rhizospheric soils, EDU led to a significant increase of the nirS gene and a notable decrease of the nosZ gene in the bulk soils (Fig. 2A).

FIG 2.

The abundance of functional genes involved in N cycling (gene number per 10,000 cells) under different treatments (A). ER, rhizospheric soil under EDU treatment; WR, rhizospheric soil under water treatment; EB, bulk soil under EDU treatment; WB, bulk soil under water treatment. The error bars denote the standard error (n = 3). Different letters above bars indicate significant differences among treatments (P < 0.05). The taxonomic distribution of N-cycling-related functional genes (at the phylum or subphylum level) (B).

Significant differences were found between the rhizospheric and bulk soils in the abundances of amoA-AOA, nifH, nirS, and nosZ genes (Table 1), especially the abundance of the amoA-AOA and the nosZ gene of the rhizospheric soil, which were lower than those of the bulk soil across all treatments (Fig. 2A). The interaction of EDU and the soil compartment significantly affected the abundances of amoA-AOA, nirK, nirS, and nosZ (Table 1).

TABLE 1.

Individual and interactive effects of EDU and soil compartment on the abundance of N-cycling-related functional genes, based on the results of an ANOVA

Genes	EDU	Soil compartment	EDU soil compartment
amoA-AOA	0.3786	<0.0001 a	0.0373 a
amoA-AOB	0.0774	0.8511	0.0518
nifH	<0.0001 a	0.0016 a	0.6436
nirK	0.8033	0.2781	0.0172 a
nirS	0.0136 a	0.016 a	0.0024 a
norB	0.1324	0.1670	0.1208
nosZ	0.0411 a	0.0001 a	0.0015 a

^aResults (Bolded) with P < 0.05.

Due to the nonnormal distribution of our data, Spearman’s correlation coefficient was used to explore the possible correlation between the abundance of functional genes involved in N cycling and soil properties. The relative abundance of amoA-AOA was significantly positively correlated with pH, with an R² of 0.674 (Table 2). The abundance of amoA-AOB was positively correlated with the contents of TN and NO₃⁻−N. Significant negative correlations were found between the abundance of nifH and the contents of TN, NH₄⁺−N, AK, and AP. The abundance of norB was negatively correlated with the content of OM while being positively correlated with the contents of TN and AK. A significant positive correlation was found between the abundance of nosZ and pH (Table 2).

TABLE 2.

The Spearman’s correlations (r) between soil chemical characterizations and the abundance of N-cycling-related functional genes

Genes	Value	Spearman’s correlationa
		pH	OM	TN	NH4+−N	NO3–−N	AK	AP
AmoA-AOA	r	0.821a,b	−0.517	0.336	−0.063	0.175	0.182	−0.294
	P	0.001	0.085	0.286	0.846	0.587	0.571	0.354
AmoA-AOB	r	0.374	−0.441	0.727b	0.559	0.692b	0.543	0.441
	P	0.234	0.152	0.007	0.059	0.013	0.068	0.152
nifH	r	0.091	0.441	−0.699b	−0.846b	−0.503	−0.637b	−0.881b
	P	0.778	0.152	0.011	0.001	0.095	0.029	0.001
nirK	r	0.207	−0.203	0.028	−0.126	0.350	0.168	−0.252
	P	0.519	0.527	0.931	0.697	0.265	0.601	0.430
nirS	r	0.379	−0.105	−0.175	−0.287	0.224	−0.151	−0.287
	P	0.224	0.746	0.587	0.366	0.484	0.640	0.366
norB	r	0.439	−0.776b	0.615b	0.252	0.378	0.669b	0.070
	P	0.154	0.003	0.033	0.430	0.226	0.017	0.829
nosZ	r	0.691b	−0.336	−0.070	−0.294	−0.490	−0.298	−0.350
	P	0.013	0.286	0.829	0.354	0.106	0.347	0.265

^an = 12.

^bSignificant correlation.

Effects of EDU on the taxonomic distributions of the N-cycling functional genes.

Four phyla (subphyla) were identified from the seven functional genes (Fig. 2B) after taxonomic assignment. Genes, such as nifH and norB, were detected in Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria. However, others such as amoA-AOA, amoA-AOB, nirK, and nirS were detected in only one phylum (subphylum). The results indicated different taxonomic distributions of functional genes. Both nonmetric multidimensional scaling (NMDS) and a principal coordinates analysis (PCoA) suggested that both EDU and the soil compartment were crucial factors in the shaping of the N-cycling community, based on Bray-Curtis distance analyses of the microbial communities that are involved in N cycling (Fig. 3).

FIG 3.

An NMDS plot (A) and a PCoA plot (B) of the microbial communities under different treatments. ER, rhizospheric soil under EDU treatment; EB, bulk soil under EDU treatment; WR, rhizospheric soil under water treatment; WB, bulk soil under water treatment.

At the genus level, amoA-AOA, amoA-AOB, nirK, and nirS were detected in single genera, including Nitrososphaera, Nitrosospira, Sinorhizobium, and Paracoccus. It was also found that microbes that harbor the same functional gene but are taxonomically distinct coexisted (Fig. 4). For example, nifH was harbored by Acidithiobacillus, Azosppirillum, Bukholderia, Gluconacetobacter, Magnetospirillum, Methylobacterium, Rhizobium, and Rhodopseudomonas. Furthermore, Brucella, Sphingomonas, Rhodospirillum, Phenylobacterium, Neisseria, Alicycliphilus, Achromobacter, Acidovorax, Cupriavidus, Janthinobacterium, Ralstonia, Kangiella, and Pseudomonas harbored the norB gene. Those taxa that harbored the same functional gene responded differently to EDU. For the nifH gene, Gluconacetobacter was only detected in the rhizosphere under the EDU treatment at a relative abundance of 0.4%. Methylobacterium was only found in the rhizosphere, and more abundant Methylobacterium was detected under the EDU treatment, compared to the water treatment. EDU significantly increased the abundance of Magnetospirillum in both the rhizospheric and bulk soils by 31 to 795%. A significant decrease in the abundance of Rhizobium caused by EDU was found in the bulk soil, whereas no remarkable change was found in the rhizospheric soil. In the rhizospheric soil, EDU decreased the abundances of Burkholderia and Acidithiobacillus by 7.4% to 14%. The relative abundance of Azospirillum was not significantly changed. All of these results suggested the existence of functional redundancy.

FIG 4.

Changes in the abundance of microbial genera involved in N cycling.

DISCUSSION

EDU applied as a foliar spray protected the hybrid rice yield loss under an ambient O₃ concentration and increased the root biomass as well as the leaf N allocation for rice plant growth in our previous study in this experimental setting (31). Shang et al. (31) suggested that the increased root biomass may enhance the uptake and transport of N, which plays an important role in photosynthesis. The results were consistent with the reduced contents of TN and NO₃^–−N in the rhizosphere soil in this study. Additionally, in the present study, although EDU foliar application had no significant effect on the microbial biomass, it led to decreased abundances of microorganisms involved in nitrification and denitrification processes and increased abundances of microbes involved in N fixation in the rhizospheric soil of rice plants, based on the analysis of the relative abundance of N-cycling-related functional genes. The opposite results were found in a study focusing on the effects of long-term (20 years) fertilization on N-cycling-related microbial communities in the North China Plain under a gradient of N fertilization levels (65), suggesting that the effect of EDU on the microbial-mediated N-cycle was inconsistent with that on N fertilizers.

Biological N-fixation, reducing N from the atmosphere to ammonia and supporting plant growth, is one of the most important ecosystem processes that is driven by microbes (66). The short-term addition of N fertilizers leads to an enrichment of fast-growing diazotrophs (67). These groups may support their vegetative growth by utilizing fertilizer resources (68) instead of fixing N, as the latter is an energy-expensive process (69). By using the ¹⁵N₂-labeling method, Fan et al. (70) found that long-term (4 decades) N fertilization led to a drastic reduction in N fixation (about 50%) that was associated with a reduction in the relative abundance of N fixers (such as Geobacter spp). The process of N fixation and some keystone diazotrophs may be more suppressed with the continued application of increased soil fertilization in the future. In this study, EDU applied as a foliar spray could significantly increase the abundance of genes involved in N fixing and could decrease the abundance of genes involved in nitrification and denitrification processes. Based on these results, it can be inferred that the foliar application of EDU may not only protect against crop yield loss caused by O₃ pollution but also counteract the reduction in N fixation caused by N fertilization without affecting rhizosphere microbial abundance.

In this study, EDU led to the increased relative abundance of some nitrogen-fixers (including Magnetospirillum spp. and Azospirillum spp.) (71, 72) as well as reductions in the relative abundances of rhizospheric nifH-harboring Burkholderia spp. that play an important role in the nodulation of plant hosts (73, 74). The nifH-harboring Gluconacetobacter spp., which contain N fixation-related enzymes (75), were only detected in the rhizosphere under the EDU treatment, compared to the water treatment. Gluconacetobacter diazotrophicus is a plant growth-promoting, nonnodule-producing, endophytic N-fixing bacterium that is less crop-specific. Generally, EDU may not only increase the microbial functional genes involved in N fixation but also alter the pathways of N fixation by recruiting special microbial groups.

In this study, different taxa were found to catalyze the same process (such as the taxonomic distribution of nifH- and norB-harboring bacteria) (Fig. 2, 4). These taxa may show different activities and adaptabilities to environmental stresses. For example, Nitrosolobus sp. TCH716, which is an ammonia-oxidizing bacterium (AOB), was isolated from alkaline soil and showed adaption to an alkaline environment (76). Another AOB strain, namely, TAO100, that was assigned to Nitrosococcus was well-adapted to acidic soil (77). Additionally, different responses to nitrite, a highly toxic end product of ammonia oxidation, were found in three closely related AOB strains, indicating that the adaptive and regulatory strategies of AOB strains were inconsistent (78). All of these results suggested that. In the present study, EDU affected some functional genes (such as nifH and norB) not only in terms of variation in the abundance of functional genes but also regarding changes of composition of microbial communities carrying functional genes. Therefore, changes in environmental adaptation among populations with the same ecological function can lead to critical response patterns for specific functional groups, an aspect of dynamics that was overlooked in previous studies that were based solely on analyses of gene abundance.

Microbiota harboring the same functional gene presented distinct taxonomic differences in the present study. These taxa showed different responses to EDU, indicating the existence of functional redundancy. By applying comparative metagenomic methods, several studies found that the assembly of microbial communities was based on functional genes, rather than on species (79–81), suggesting that different microbial assemblages can exhibit similar community gene profiles that are defined by specific environments (82). Based on a meta-analysis, Biggs et al. (83) found that functional redundancy may enhance the stability and resilience to a disturbance of microbial communities. Therefore, the existence of extensive environmental adaptabilities among different microbial groups may play a key role in maintaining N cycling under environmental changes.

Previous studies have documented that the rhizospheric bacterial communities showed higher biomass (84), displayed increased activities (85), and contributed more significantly to nutrient cycling (86), compared to the bulk soil. Consistent results were found in this study. For example, the abundance of the nifH gene in the rhizospheric soil was higher than that observed in the bulk soil across all treatments. In addition, some bacterial groups that are endemic to the rhizosphere were also found in the present study. First, the nifH-harboring Methylobacterium spp. were only detected in the rhizosphere soil across all treatments. Pure cultures of Methylobacterium spp. have been reported to possess nitrogenase activity via nodA genes and can colonize rice roots, thereby resulting in numerous lateral roots (87). Acidovorax is another rhizosphere-specific, norB-harboring bacterial genus. Some strains belonging to Acidovorax have been demonstrated to have the ability to carry out the nitrate-dependent oxidation of ferrous ions to form ferric precipitates during the process of microbial nitrate reduction (88, 89). Among them, the facultatively anaerobic Acidovorax sp. NO1 has been reported to be capable of oxidizing arsenite and reducing nitrate (90). Due to its special growth habit, most of the growth period of rice is under a flooded environment, in which iron (Fe) oxyhydroxide plaque is usually formed on the surfaces of rice roots. Ferrous oxidation and ferric oxide precipitation on root surfaces lead to the formation of Fe plaque, which acts as a sink or source of several nutrients (including iron, phosphorus, zinc, manganese, arsenic, and copper) (91). In addition, Fe (II)-oxidizing bacteria have been documented to be crucial for elements cycling in the rhizosphere (92). Generally, except for N cycling, these Acidovorax spp. may play important roles in the cycling of other nutrients in the rice rhizosphere.

In this study, the entire foliage of each plant was sprayed with 450 ppm EDU until runoff, which means that some of the sprayed liquid ends on the soil. On one hand, the indirect effects of the EDU foliar spray on the rhizosphere microbial community involved in N cycling occur though changing root responses (including increased root biomass, activity, unexplored root exudate, and so on) as a result of the plant stress status. In this study, the increased root biomass enhanced the N uptake from the rhizosphere, resulting in a reduction in the contents of TN and NH₄⁺–N, which significantly negatively correlated with the abundance of the nifH gene. On the other hand, EDU dripping might directly change the nutrient status and physiochemical characterizations of microhabitats of microbial communities in both the rhizospheric soil and in the bulk soils. Therefore, two treatments, namely, a foliar spray with a plastic pad to avoid dripping and a soil drench of EDU, should be set up to evaluate the indirect and direct effects of EDU foliar application. Meanwhile, the determination of nitrogen oxide emissions from rice fields can further supplement the effect of EDU on the soil N cycle, which is mediated by microorganisms.

In conclusion, a distinct impact of EDU foliage spray on the structure of microbial communities that take part in N cycling and the significant changes in taxa with EDU application under an ambient O₃ concentration were found in this study by applying PCR-unbiased metagenomic sequencing combined with the assembly and annotation of functional genes involved in N cycling. These taxon-specific response patterns provide a deeper insight into the effects of EDU on the belowground ecosystem, especially on soil N cycling and the related microbial population. These results will also provide new insights into the mechanism by which EDU alleviates plant stress caused by O₃.

MATERIALS AND METHODS

Experimental site and plant culture.

This experiment was performed in Sanwudou Agricultural Production Comprehensive Service Professional Cooperative Union (32°23’N, 119° 33’E) in Yangzhou City, Jiangsu Province, China. This area has a subtropical monsoon climate and a hot and rainy summer. The annual mean temperature and rainfall of this region in 2020 were 16.3°C and 1,214.8 mm, respectively.

Shanyou 63 was grown in 30 pots filled with local farmland soil, containing 14 ± 0.8 g kg⁻¹ organic matter, 1.48 ± 0.01 g kg⁻¹ total N, 40 ± 2.1 mg kg⁻¹ AK, and 17 ± 4.8 mg kg⁻¹ AP. These pots were randomly placed in 3 open-top chambers (OTCs), with 10 pots in each OTC. The average concentration of O₃ from 8:00 to 18:00 was 39.6 ± 1.9 ppb from July 25 to October 15. In each OTC, 5 pots of rice plants were sprayed with 450 ppm EDU, and the other 5 pots were sprayed with an equal volume of water every 10 days, beginning on August 10, as suggested in previous studies (22, 24, 26). In total, EDU was sprayed 9 times during the whole experimental period.

Soil collection and chemical analysis.

One pot was randomly selected from 5 pots for each treatment, per OTC. Soil within 2 mm of the root surface was considered to be rhizospheric soil (84). The bulk soil samples were aseptically collected (>1 cm from a root), and the root system of a rice plant was then extracted. The rhizosphere was obtained by vortexing and sonicating the roots in phosphate buffer (93). Portions of the rhizospheric and bulk soils were stored at −80°C for DNA extraction, and the rest were stored at 4°C for chemical analysis. The pH, content of organic matter, total nitrogen (N), nitrate (NO₃^––N), ammonium (NH₄⁺−N), available potassium, and available phosphorus (AP) were measured according to the methods described by Sun et al. (94).

Microbial DNA extraction and evaluation of 16S rRNA gene copies.

For each sample, 0.5 g fresh soil were used to extract the microbial DNA, using a Fast DNA SPIN Kit (MP Biomedicals, Santa Ana, CA), as suggested by Sun et al. (94). Then, the DNA was verified via agarose gel electrophoresis and qualified using a NanoDrop 2000 spectrophotometer (Thermo Scientific, USA). Bacterial abundance was determined by applying a qRT-PCR method with the primer set 536F/907R (95). Each PCR was performed in triplicate, and the details of the amplification conditions are described in Sun et al. (94).

Metagenome sequencing and annotation.

A sequencing library was constructed, and the template was prepared by using library preparation kits (New England Biolabs, Inc., Beverly, MA, USA). A paired-end library with an insert size of about 500 bp was generated for each sample and then sequenced on a HiSeq 2500 sequencing system (Illumina, CA, USA), performed by Magigene (Gunagzhou, China).

The raw sequences were filtered using PRINSEQ (version 0.20.4) with the default parameters (96). The adapter sequences were removed, and sequences with ambiguous bases (N) or quality score means below 30 were removed. In total 845,257,632 clean reads (about 150 bp) were obtained from 12 samples (Table S1). High-quality reads were used for the metagenomic analysis. Functional genes involved in N cycling were assembled using Xander (63). The reference sequences and hidden Markov models (HMMs) for rplB as well as bacterial amoA, amoB, nifH, nirS, nirK, norB, and nosZ were provided by Xander. De Bruijn graphs were constructed using a kmer size of 30. The assembled contigs were filtered with a length cutoff of 15 and an HMM score cutoff of 50. The contigs were clustered at 99% amino acid identity, and the longest contigs were chosen as the representative contigs. Then, chimeras were removed, and the nearest reference match of each cluster was determined using FrameBot with the default parameters, except that the parameter “percent identity cutoff” was increased from 0.4 (default) to 0.6 so as to enhance the reliability of the taxonomic identification. For the statistical analysis, the abundance of each functional gene was normalized to 10,000 cells against the total rplB counts.

Statistical analysis.

The statistical analysis was performed using R (version 3.4.3), based on the normalized abundance of each functional gene involved in N cycling. NMDS and a PCoA that was based on the Bray-Curtis distance were performed to study the community patterns using the “vegan” R package. The correlation between the relative abundance of functional genes and soil properties was measured using Spearman’s rank correlation and visualized using Cytoscape (version 3.6.0) (97). The effects of EDU, the soil compartment, and their interaction on soil chemical properties, the copies of 16S rRNA genes, and the relative abundance of functional genes involved in N cycling were analyzed using a split-plot analysis of variance (ANOVA). Tukey’s honestly significant difference (HSD) test was applied to evaluate post hoc differences. All of the statistical analyses were conducted at a level of significance of α = 0.05 in JMP 10.0 (SAS Institute Inc., Kerry, USA).