Active Soil Nitrifying Communities Revealed by In Situ Transcriptomics and Microcosm-Based Stable-Isotope Probing

The role of manipulated microcosms in microbial ecology has been much debated, because they cannot entirely represent the in situ situation. We collected soil samples from 20 field plots, including 5 different treatments with and without nitrogen fertilizers for 22 years, in order to assess active nitrifying communities by in situ transcriptomics and microcosm-based stable-isotope probing. The results showed that chronic N enrichment led to competitive advantages of Nitrosospira cluster 3-like AOB over N. viennensis-like AOA in soils under field conditions. Microcosm labeling revealed similar results for active AOA and AOB, although an apparent discrepancy was observed for nitrite-oxidizing bacteria. This study suggests that the soil microbiome represents a relatively stable community resulting from complex evolutionary processes over a large time scale, and microcosms can serve as powerful tools to test the theory of environmental filtering on the key functional microbial guilds.

ABSTRACT

Long-term nitrogen field fertilization often results in significant changes in nitrifying communities that catalyze a key step in the global N cycle. However, whether microcosm studies are able to inform the dynamic changes in communities of ammonia-oxidizing bacteria (AOB) and archaea (AOA) under field conditions remains poorly understood. This study aimed to evaluate the transcriptional activities of nitrifying communities under in situ conditions, and we found that they were largely similar to those of ¹³C-labeled nitrifying communities in the urea-amended microcosms of soils that had received different N fertilization regimens for 22 years. High-throughput sequencing of 16S rRNA genes and transcripts suggested that Nitrosospira cluster 3-like AOB and Nitrososphaera viennensis-like AOA were significantly stimulated in N-fertilized fresh soils. Real-time quantitative PCR demonstrated that the significant increase of AOA and AOB in fresh soils upon nitrogen fertilization could be preserved in the air-dried soils. DNA-based stable-isotope probing (SIP) further revealed the greatest labeling of Nitrosospira cluster 3-like AOB and Nitrosospira viennensis-like AOA, despite the strong advantage of AOB over AOA in the N-fertilized soils. Nitrobacter-like nitrite-oxidizing bacteria (NOB) played more important roles than Nitrospira-like NOB in urea-amended SIP microcosms, while the situation was the opposite under field conditions. Our results suggest that long-term fertilization selected for physiologically versatile AOB and AOA that could have been adapted to a wide range of substrate ammonium concentrations. It also provides compelling evidence that the dominant communities of transcriptionally active nitrifiers under field conditions were largely similar to those revealed in ¹³C-labeled microcosms.

IMPORTANCE The role of manipulated microcosms in microbial ecology has been much debated, because they cannot entirely represent the in situ situation. We collected soil samples from 20 field plots, including 5 different treatments with and without nitrogen fertilizers for 22 years, in order to assess active nitrifying communities by in situ transcriptomics and microcosm-based stable-isotope probing. The results showed that chronic N enrichment led to competitive advantages of Nitrosospira cluster 3-like AOB over N. viennensis-like AOA in soils under field conditions. Microcosm labeling revealed similar results for active AOA and AOB, although an apparent discrepancy was observed for nitrite-oxidizing bacteria. This study suggests that the soil microbiome represents a relatively stable community resulting from complex evolutionary processes over a large time scale, and microcosms can serve as powerful tools to test the theory of environmental filtering on the key functional microbial guilds.

INTRODUCTION

Nitrification is a critical link in global nitrogen cycling that is typically characterized through a two-step microbial process: the oxidation of ammonia to nitrate via nitrite by ammonia oxidizers and the oxidation of nitrite to nitrate by nitrite oxidizers. Soil ammonia oxidizers consist of canonical ammonia-oxidizing bacteria (AOB), ammonia-oxidizing archaea (AOA), and recently discovered complete ammonia oxidizers (comammox). AOA often outnumber AOB in soil (1), and a recent study showed an active role for comammox in nitrification of agricultural soils under intensive nitrogen fertilization (2). The comammox live an oligotrophic lifestyle, and the relative contribution of distinct ammonia oxidizers to nitrification under field conditions was poorly understood (3–5). Numerous studies have shown that the dominant activity of either AOA or AOB in different soils might be progressively selected by various environmental factors. For instance, AOA frequently dominate ammonia oxidation in acidic soils (6–8), with two representative acidophilic strains isolated from such soils (9). In contrast, AOB grow better in higher-pH soils, although they also occur in acidic soils (10–12). Ammonia source and concentration are also important in controlling ammonia oxidizer activity, with AOA and AOB preferring a slow supply of ammonia from organic N mineralization and a high concentration of inorganic ammonium, respectively (13–16). Consequently, it has long been assumed that different environmental conditions will lead to niche differentiation of ammonia oxidizer phylotypes with distinct physiological traits, such as substrate affinities and pH adaptation capabilities, between AOA and AOB (17–19).

Agricultural soils represent a human-influenced ecosystem with an intensive N input from ammonium-based fertilizers, and nitrification is estimated to be responsible for approximately 70% of ammonia-based fertilizer N loss (20). This also leads to significant environmental concerns, including reduced nitrogen utilization efficiency and groundwater nitrate contamination through nitrate leaching (21), increased N₂O emissions (14), and soil acidification (22). All of these negative impacts are closely associated with the diversity and activities of nitrifying communities that have been constantly selected and shaped by the long-term application of N fertilizers. For example, long-term ammonium-based fertilization significantly changes the abundance and composition of AOB but has no or only slight effects on those of AOA (23–25). Intriguingly, there are also reports showing either positive (26, 27) or negative (28, 29) impacts of long-term chronic chemical N application on AOA population size. These results suggest that AOB and AOA respond differently to long-term N addition under field conditions, and a microcosm study on soil nitrifying composition and activity might be site specific.

Since both archaea and bacteria are capable of ammonia oxidation and commonly coexist in soils, it is always challenging to quantify the relative contributions of AOA and AOB in agricultural soils, especially under field conditions. The quantification of ammonia oxidizer abundance, using either real-time quantitative PCR (qPCR) of ammonia monooxygenase (amoA) or high-throughput sequencing of 16S rRNA genes, can provide the first clue about their activity, with positive nitrification often being accompanied by an increase in gene abundance over time (6, 30, 31). The relative contributions between AOA and AOB, however, might not be able to be confidently determined using these DNA-based quantifications, as some ammonia oxidizer phylotypes might be present but remain dormant in soils (32). ¹³C-labeled-DNA-based stable-isotope probing (DNA-SIP) is a powerful tool used to track the autotrophic nitrifiers through the recovery and analysis of [¹³C]DNA produced by cells growing on ¹³C-labeled CO₂ and, more importantly, to distinguish active and dormant nitrifier phylotypes across various soil habitats (29, 33, 34). However, DNA-SIP usually relies on the artificial supplementation of high substrate levels in lab microcosms to ensure successful labeling, and the results might not entirely represent the in situ situation. Monitoring of microbial gene transcript abundance provides another strategy for estimation of in situ nitrifier activity due to the higher sensitivity of RNA than of DNA in reflecting the metabolic state of microbial communities under field conditions (35, 36). DNA and RNA constitute microbial nucleic acids, and both of them allow the phylogenetic characterization of microbial groups (rRNA genes and their transcripts). The presence of RNA is indicative of protein synthesis potential, while the existence of DNA is the consequence of cell replication. Thus, microbial RNA in situ is more sensitive than DNA for reflecting the metabolic state of microbial assemblies under field conditions. Nevertheless, DNA-based molecular surveys have been used more frequently than RNA-based tools, in part due to their ease of use. Recent findings suggested that extracellular DNA derived from dead cells accounts for approximately 40% of microbial DNA in soils, which leads to an overestimation of microbial abundance and a biased estimate of the relative abundance of different taxa; this might further obscure the variation between treatments (37). An air-drying procedure can reduce the extracellular DNA (38) and potentially increase the sensitivity of DNA-based approaches to decipher the adaptation strategies of ammonia oxidizers in intensively fertilized soils (39).

In the present study, we collected soil samples from 20 field plots that had been subjected to five different fertilization regimens for 22 years. In situ activity of nitrifying communities was assessed via high-throughput sequencing of 16S rRNA gene transcripts of samples from fresh soils, and DNA-SIP was employed to reveal the taxonomic identities of active nitrifiers in microcosms amended with high-concentration urea. Furthermore, the abundance and composition of nitrifying communities were also assessed in air-dried soils and compared using RNA and DNA-SIP strategies.

RESULTS

Changes in nitrifying community abundance under field conditions.

The absolute abundance of 16S rRNA genes ranged from 1.85 × 10¹⁰ to 2.56 × 10¹⁰ per gram (dry weight) of soil (g⁻¹ d.w.s) in all freshly sampled soils and showed no significant differences between non-N-fertilized soils (CK [no fertilizer] and PK [phosphorus and potassium fertilizer]) and N-fertilized soils (NK, NP, and NPK). It suggested no evident effect of nitrogen fertilizations on the whole abundance of soil prokaryotes, which was in line with previous findings (40). Thus, the relative abundances of 16S rRNA genes (or transcripts) were calculated by dividing the number of sequence reads affiliated with AOA, AOB, and nitrite-oxidizing bacteria (NOB) by the total sequence reads in each replicate, to assess the response of nitrifying communities to chronic N enrichment under field conditions. A total of 263,101 high-quality sequence reads were obtained, with AOA, AOB, and NOB accounting for 1.53%, 0.55%, and 3.39% of total reads (Table S1), respectively.

Long-term nitrogen fertilization lead to significantly increased AOB transcriptional activity and abundance, as indicated by both RNA- and DNA-based surveys. AOB-related 16S rRNA gene transcripts accounted for 0.09 to 0.18% of the sequences in non-N-fertilized soils but significantly increased to 1.24 to 1.60% in N-treated soils (Fig. 1a; Table 1). At the DNA level, the relative abundances of AOB were 0.01 to 0.05% in non-N-fertilized soils but significantly increased to 0.38 to 0.92% and 0.35 to 0.67% in fresh N-fertilized soils and after air drying, respectively. Intriguingly, chronic N enrichment resulted in higher increments of AOB transcripts (1.06 to 1.42%) than of AOB gene abundances (0.31 to 0.89%).

FIG 1.

Population changes of ammonia-oxidizing archaea (AOA) and bacteria (AOB) in an agricultural soil under field conditions after different 22-year fertilization regimens. (a) Relative abundance of 16S rRNA transcripts of AOA and AOB in fresh soil; (b) 16S rRNA genes in fresh soil; (c) 16S rRNA genes in air-dried soil. (d and e) amoA gene abundances in fresh soil (d) and air-dried soil (e). The whole microbial community was pyrosequenced using the universal primer pair 515F-907R, and the 16S rRNA gene sequence reads of nitrifying communities were selected by the Ribosomal Database Project taxonomic classifier. The relative abundance of 16S rRNA genes or transcripts was then calculated by dividing the number of sequence reads affiliated with AOA and AOB by the total 16S rRNA gene sequences in soil, respectively. “no N” fertilization includes CK (no fertilizer application) and PK (P and K application without N); “+ N” fertilization includes NP (fertilizer application with N and P), NK (fertilizer application with N and K), and NPK (fertilizer application with N, P, and K). Error bars represent the standard errors of the means (n = 4). The significant differences among treatments are shown in Table S3.

TABLE 1.

Proportional changes in population dynamics of AOB, AOA, and NOB in soils that received no nitrogen fertilization (CK and PK) and nitrogen fertilization (NP, NK, and NPK) for 22 years^a

Soil	High-throughput sequencing of 16S rRNA gene/transcriptb						Quantitative PCR of amoA gene: AOB/AOAc
	AOB/AOA ratio			NOB/(AOB+AOA) ratio
	RNA (fresh soil)	DNA (fresh soil)	DNA (dry soil)	RNA (fresh soil)	DNA (fresh soil)	DNA (dry soil)	DNA (fresh soil)	DNA (dry soil)
CK	0.88	0.01	0.01	8.82	3.16	0.97	0.05	0.04
PK	1.38	0.02	0.01	12.6	5.05	1.31	0.07	0.06
NP	5.81	0.45	0.10	2.61	3.44	0.62	0.47	0.41
NK	30.5	0.57	0.14	2.44	1.78	0.64	0.73	0.36
NPK	11.4	0.23	0.07	2.18	1.57	0.61	0.47	0.29

^aValues are means of four replicates from each treatment.

^bCalculated based on the relative abundance, i.e., by dividing the number of sequence reads affiliated with AOA, AOB, and NOB by the total number of 16S rRNA gene sequences.

^cCalculated based on absolute abundance of AOA and AOB by real-time quantitative PCR.

AOA transcriptional activity was less affected by N fertilization than that of AOB. At the RNA level, the relative abundance of AOA transcripts (0.06 to 0.23%) was not significantly different between soils with and without N fertilization histories (Fig. 1a; Table 1). However, at the DNA level, the relative abundance of AOA showed an increasing trend in soils with an N fertilization history except for NP soils (Fig. 1b; Table S3), and statistically significant increases were observed in air-dried soils, from 1.88 to 2.59% in non-N-fertilized soils to 4.45 to 4.86% in soils with an N fertilization history (Fig. 1c). The stimulation by N fertilizers of AOA and AOB was demonstrated to a greater extent in air-dried than fresh soils, with the observation of significant increases in copy numbers of AOA and AOB amoA genes under N fertilization conditions in air-dried soils (Fig. 1e). In fresh soils, bacterial rather than archaeal amoA gene abundances differed significantly between non-N-fertilized and N-fertilized soils (Fig. 1d; Table 1), and both bacterial and archaeal amoA gene abundances were higher in fresh than air-dried soils.

Both high-throughput sequencing of 16S rRNA genes and real-time quantitative analysis of amoA genes suggested that AOA outnumbered AOB in all soils (except in the fresh NK soil) (Fig. 1d). However, the transcriptional activity of AOA was much lower than that of AOB in N-fertilized soils and showed no significant difference from that of AOB in non-N-fertilized soils (Fig. 1a). This indicated that AOA communities possessed activity that was similar to or much lower than that of AOB in these soils under field conditions, despite having a higher cell abundance. The ratio of 16S rRNA transcripts to 16S rRNA genes for AOB was indeed 13 to 67 and 41 to 350 times higher than AOA in the five soils when they were fresh and after air drying (Table S2), respectively, implying a much lower cell activity of AOA than AOB in the alkaline soils.

Nitrification activity and ammonia oxidizer abundances in soil microcosms.

SIP of active nitrifying communities was conducted in CK and NPK microcosms, representing non-N-fertilized and N-fertilized soils, respectively. Soil nitrification activity was assessed according to the temporal change in soil NO₃⁻ concentrations in each microcosm during the 63-day incubation period. NO₃⁻ significantly accumulated over time following urea amendment. Soil nitrification activity ranged from 11.35 to 11.56 μg NO₃⁻-N day⁻¹ g⁻¹ d.w.s, with no significant differences observed between CK and NPK soils (Fig. 2a). Soil NO₃⁻-N concentrations remained very low in the presence of acetylene, which completely abolished ammonia oxidation (Fig. 2a). In the presence of acetylene, the NH₄⁺ concentration was equivalent to the total urea concentration applied to the soils (800 μg N g⁻¹ d.w.s) (Fig. 2b), suggesting the stoichiometric conversion of ammonium to nitrate by nitrifying communities.

FIG 2.

Changes in the concentration of soil NO₃⁻-N (a) and NH₄⁺-N (b) and bacterial (c) and archaeal (d) amoA gene abundance in a nonfertilized (CK) and an N-fertilized (NPK) soil. The SIP microcosm was incubated with either ¹²CO₂, ¹³CO₂, or ¹³CO₂ in the presence of C₂H₂ for 63 days. Error bars represent standard errors of the means of three replicates. Different letters above the bars indicate significant differences (P < 0.05) among treatments for each soil based on a one-way analysis of variance.

Following microcosm incubation, bacterial amoA genes increased from 4.41 × 10⁷ to 1.79 × 10⁸ g⁻¹ d.w.s in CK soils but decreased from 7.09 × 10⁸ to 4.06 × 10⁸ g⁻¹ d.w.s in NPK soils (Fig. 2c). Archaeal amoA genes did not change after incubation in either soil type (Fig. 2d). AOB abundances were significantly lower in acetylene-amended soils than those without acetylene supplementation, but AOA abundance was unchanged by acetylene amendment (Fig. 2c and d).

Autotrophic nitrifiers identified by ¹³C-labeled DNA.

Quantification of amoA genes across the buoyant density of the DNA gradient was used to assess the labeling efficiencies of AOB and AOA following ¹³CO₂-DNA-SIP microcosm incubation. The highest amoA gene abundances of both AOB and AOA in the control microcosms (¹²CO₂ and ¹³CO₂+C₂H₂) were detected in the ‘light’ fractions, i.e., those with a buoyant density of 1.710 to 1.725 g ml⁻¹ (Fig. 3a to d). A peak shift of the highest bacterial amoA gene abundance toward the “heavy” fractions was observed in the ¹³CO₂-amended microcosms of both CK (Fig. 3a) and NPK (Fig. 3b) soils. Of note, higher abundances of archaeal amoA genes were also detected in the heavy fractions from ¹³CO₂-amended microcosms in both CK (Fig. 3c) and NPK (Fig. 3d) soils than the control microcosms, while CK soil showed much stronger labeling than NPK soils.

FIG 3.

Quantitative distribution of amoA gene abundances of bacteria (a and c) and archaea (b and d) across the buoyant-density gradients retrieved from nonfertilized (CK) and N-fertilized (NPK) soils, following SIP microcosm incubation with either ¹²CO₂, ¹³CO₂, or ¹³CO₂ in the presence of C₂H₂ for 63 days. The ratio of maximum quantities of the amoA gene represents the ratio of amoA gene copies in each DNA fraction to the maximum gene quantity for each treatment. Error bars represent standard errors of the means of three replicates. Different letters above the bars indicate a significant difference (P < 0.05) between CK and NPK soils based on a one-way analysis of variance. Red bars represent the ¹³C-labeled heavy DNA fractions; blue bars represent the unlabeled light DNA fractions.

The abundances of ¹³C-labeled amoA genes in AOA and AOB were calculated from fractions 4 to 7. These abundances in bacteria were 1.57 × 10⁸ and 3.01 × 10⁸ g⁻¹ d.w.s in CK and NPK soils (Fig. 4d), respectively, while those of archaeal amoA genes were 3.67 × 10⁷ and 1.58 × 10⁷ g⁻¹ d.w.s in CK and NPK soils (Fig. 4e), respectively. The higher AOB and lower AOA activity in N-fertilized soils (NPK) than non-N-fertilized soils (CK) was also supported by the higher ratio of ¹³C-labeled AOB amoA gene copies to ¹³C-labeled AOA amoA gene copies in NPK soils (19.0) than that in CK soils (4.3). High-throughput sequencing of the 16S rRNA gene resulted in similar results, with ratios of ¹³C-labeled AOB to AOA of 3.3 in CK soils and 7.2 in NPK soils.

FIG 4.

Comparative analysis of active nitrifying communities in soils by in situ transcriptomics and microcosm-based SIP. (a to c) Relative abundance of distinct phylotypes of AOB, AOA, and NOB, respectively, in soils under field conditions after different 22-year fertilization regimens. The calculation of 16S rRNA gene abundance of nitrifying communities was the same as for Fig. 1. Values are the means of four replicates. (d and e) Taxonomic dynamics of AOB and AOA, respectively, reflected by [¹³C]DNA, based on amoA gene pyrosequencing (insets show the relative abundance of 16S rRNA gene phylotypes of AOB and AOA, respectively, in [¹³C]DNA). (f) Relative abundance of 16S rRNA gene-based NOB phylotype in [¹³C]DNA. Designations are the same as those in Fig. 1. The significant differences among treatments (P < 0.05) are shown in Table S3. Designations in parentheses represent the recent taxonomy of AOA determined by global phylogenetic analysis of the known AOA diversity based on amoA and 16S rRNA genes (76).

Nitrifier compositions under in situ and microcosm incubation conditions.

AOB, AOA, and NOB compositions were identified under both field conditions (for all soils) and microcosm incubation (CK and NPK) based on 16S rRNA analysis. AOB under field conditions were dominated by phylotypes closely related to Nitrosospira cluster 3, Nitrosospira sp. strain Nsp65, Nitrosomonas communis, and Nitrosomonas oligotropha (Fig. 4a), with transcripts and gene abundances of all phylotypes being significantly increased by the field N fertilization regimen, except for N. oligotropha-related gene abundance (Table S3). The ¹³C-labeled AOB in CK and NPK samples were dominated by Nitrosospira cluster 3-related AOB (>91.2%), followed by Nitrosospira sp. strain Nsp65- and N. communis-related phylotypes (Fig. 4d). The phylogenetic analysis of AOB amoA genes confirmed that Nitrosospira cluster 3-related and Nitrosospira sp. strain Nsp65-related AOB were the main drivers of nitrification in both CK and NPK soils in response to urea amendment, even though interspecies differences existed (Table S7). Although ¹³C-labeled AOB were composed of similar phylotypes in CK and NPK soils, the absolute abundance of Nitrosospira cluster 3- and Nitrosospira sp. strain Nsp65-related AOB were 1.9- and 2.1-fold higher, respectively, in NPK than CK soils (Fig. 4d) in terms of ¹³C-labeled AOB amoA gene abundance under urea-amended microcosm conditions.

AOA in field samples were mainly affiliated with four clusters within group 1.1b lineage, including Nitrososphaera viennensis-, fosmid clone 54d9-, and “Candidatus Nitrosocosmicus franklandianus”-related phylotypes and a phylotype related to an uncultured Nitrososphaera organism (Fig. 4b). Intriguingly, “Candidatus Nitrosotalea devaniterrae”-related AOA within the group 1.1a-associated lineage were also detected in NPK soils (Fig. 4b). 54d9- and “Candidatus Nitrosocosmicus franklandianus”-related AOA were the most abundant in soils (0.72 to 4.07% of total reads) and significantly increased with nitrogen fertilization history compared to samples with no N fertilization (Table S3). However, 54d9- and “Candidatus Nitrosocosmicus franklandianus”-related transcripts accounted for no more than 0.05% of total reads and did not differ between non-N-fertilized and N-fertilized soils (Table S3). In addition, N. viennensis-related transcripts significantly increased in N-fertilized soils (Table S3). N. viennensis- and “Candidatus Nitrosocosmicus franklandianus”-related AOA composed 94.0% and 6.0%, respectively, of ¹³C-labeled active AOA in CK soils, whereas N. viennensis-related AOA was the only identified ¹³C-labeled AOA phylotype in NPK soils (Fig. 4e). Based on the phylogenetic analysis of archaeal amoA genes, N. viennensis-related AOA accounted for 86.2 to 94.8% of the AOA community in the ¹³C heavy fractions (Table S7). Integrating the ratio with copy number of AOA amoA genes in ¹³C-labeled DNA, the absolute abundances of N. viennensis-related AOA in CK were 3.0 times higher than those in NPK soils under urea-amended microcosm conditions (Fig. 4d).

Nitrospira-like rather than Nitrobacter-like NOB were the dominant nitrite oxidizers in all soils under in situ conditions at both the RNA and DNA levels (Fig. 4c). However, Nitrobacter-like NOB emerged as the most abundant ¹³C-labeled NOB following urea amendment, representing 54.8 to 72.6% of NOB sequences (Fig. 4f; Table S3).

DISCUSSION

Nitrifying activity under in situ versus microcosm incubation conditions.

The present study combined DNA-SIP and RNA-based approaches to assess the phylogenetic landscape of active nitrifying communities under in situ conditions relative to those in microcosms with excessive N input. AOB transcriptional activity was exceptionally high in field soils that received N fertilizers compared to those without N input, while AOA stimulation was observed only in air-dried soils. AOB appeared to dominate the ammonia oxidization in all soils under both field conditions and microcosm conditions. Following urea fertilization, AOB in the microcosm were significantly stimulated to a markedly greater extent than AOA, as the ratios of actively growing (¹³C-labeled) AOB to AOA (4.3 to 19.0) were considerably higher than the ratios of total AOB to AOA abundance (<1) (Fig. 3e and f). Under in situ conditions, both AOA and AOB transcripts were detected in all soils, but AOB had higher transcriptional activity and likely dominated the ammonia oxidation in soils with an N fertilization history (NP, NK, and NPK). The relative contribution of AOA and AOB to nitrification in soils without N fertilization history (CK and PK) remains uncertain (Fig. 1a). Our transcriptomics results lend strong support to previous DNA-based observations for the important role of AOB in alkaline agricultural soils (11, 12) and also indicate that the metabolic spectrum of AOA is much more complicated than previously appreciated.

Nitrosospira cluster 3 showed the highest transcriptional activity in soils under field conditions (Fig. 4a) and accounted for the vast majority of the active (¹³C-labeled) AOB following urea-amended incubation (Fig. 4d). Members of Nitrosospira cluster 3 constitute the most ubiquitous AOB in soils and can be abundantly detected in a wide range of ecosystems with varied environmental conditions (40–43). This group might represent the “generalist” soil nitrifiers (44, 45) and continuously support soil nitrification processes under perturbed N conditions. Our study showed the kinetic versatility of Nitrosospira cluster 3 under different ammonium concentrations.

The dominantly active AOA groups, however, differed between microcosm and in situ conditions. After SIP microcosm incubation with urea amendment, the active (¹³C-labeled) AOA were dominated by the N. viennensis-related lineage (Fig. 4e). N. viennensis is considered a neutrophilic AOA with an optimal pH of 7 to 8 (46). This physiological trait supports the active cell proliferation of N. viennensis-related AOA in our soils (pH ranging from 8.0 to 8.4). However, our N. viennensis-related AOA showed less activity under field conditions, particularly in soils without an N fertilization history. In such soils, 54d9, “Candidatus Nitrosocosmicus franklandianus,” and an uncultured cluster codominated the actively transcribing AOA (Fig. 4b), similar to previous results (23, 25, 47, 48). Although delayed or absent N. viennensis growth is observed at concentrations above 10 mM NH₄⁺ in pure culture (46), the N. viennensis-related lineage showed the highest AOA activity following a weekly addition of urea equivalent to 12.3 mM NH₄⁺, indicating that this lineage could be the most competitive AOA for ammonia oxidation under high-N conditions.

Nitrospira organisms were the most abundant and likely the most active NOB in soils under field conditions (Fig. 4c), but Nitrobacter also grew fast during urea-amended microcosm incubation (Fig. 4f). These two genera were the main nitrite oxidizers in soils (47) but possessed distinct kinetic strategies for growth. Batch culture studies and ecological investigation have characterized Nitrobacter as an r-strategist, with a higher growth rate and specific activity but lower N substrate affinity than Nitrospira, a k-strategist (49–52). The dominant presence of Nitrospira in field conditions and the increased growth of Nitrobacter following urea addition were highly consistent with their physiological and kinetic traits, indicating that Nitrospira and Nitrobacter could be responsible for the dominant nitrite oxidation activity in N-limited conditions and after N fertilization, respectively, in alkaline agricultural soils. It is worth mentioning that we could not distinguish the activity of canonical nitrite-oxidizing and comammox Nitrospira by using the 16S rRNA gene as a biomarker in the present study (53, 54). However, both transcriptomics and SIP indicated the important role of Nitrospira in nitrification in the alkaline agricultural soils tested in this study. This might be explained, in part, by both canonical and comammox Nitrospira being able to perform nitrite oxidation. Intriguingly, several studies appeared to suggest the absence of comammox activities in nitrifying soils (16, 55, 56). The combined application of a nitrite-oxidizing inhibitor and SIP would provide a powerful means to differentiate canonical nitrite oxidizers from comammox Nitrospira in complex soil.

Effect of different fertilization histories on ammonia oxidizers.

Although both AOA and AOB were present and active in all soils, their abundance, activity, and composition were affected by different fertilization regimens. Long-term N fertilization might have a profound effect on the ammonia oxidizer assemblage. All soils had similar low ammonium concentrations (<2.7 mM) at the time of sampling, but the AOB abundances were significantly higher in soils with an N fertilization history than in soils that had never received N fertilizers (Fig. 1b to e). Similarly, AOB were more transcriptionally active in fields with an N fertilization history than in soils that had never been N fertilized (Fig. 1a). Previous studies often linked AOB to nitrification at high NH₄⁺ concentrations due to the much lower substrate affinity of AOB than AOA (57, 58). Our results demonstrate that the occurrence of AOB activity, even in limited NH₄⁺ concentrations, is consistent with a recent proposal that niche differentiation is due not to an inability of AOB to grow under low concentrations but rather to differences in the ability of AOB and other ammonia oxidizers to compete for NH₄⁺ (16). Regular N fertilization history constantly changes soil N conditions, which could have selected for specific AOB phylotypes that can adapt to a wide range of ammonia concentrations or redundant AOB phylotypes that adapt to high and low ammonia concentrations (59). This is supported by the distinct AOB compositions between N-fertilized and non-N-fertilized soils (Fig. 4).

AOA abundance and transcriptional activity seemed less affected by the fertilization regimen than those of AOB (Fig. 1). However, following microcosm incubation with urea and ¹³CO₂ amendments, AOA appeared to grow faster and were labeled to a greater extent in CK soil, which had no N fertilization history, than in NPK soil, which received regular N fertilizers (Fig. 3b and d). This demonstrated that AOA can grow well under high NH₄⁺ concentration conditions, as previously observed (43, 60, 61), and the higher AOA activity in CK than NPK soils might be due to decreased competition from AOB activity (16, 60). It was noteworthy that the AOA population was apparently greater in fertilized soils than nonfertilized soils after air drying. The constant proliferation of AOA with N input might have occurred slowly under field conditions, although the competitive disadvantage of AOA relative to AOB was progressively strengthened over the course of different 22-year fertilization regimens.

Advantages of using air-dried soils for ecological investigation.

There is much debate over the use of air-dried soils for ecological study, especially for soil microorganisms. The air-drying process could potentially increase soil osmotic stress and decrease the abundance and activity of microbial communities, including nitrifiers (62, 63). However, the air-drying process can diminish extracellular DNA residues from dead cells (38, 64), which might result in the misestimate of microbial abundance. We performed a DNA analysis on both fresh and air-dried soils to evaluate the potential use of dry soil for studying nitrifier ecology. Our results showed significantly decreased nitrifier abundance (Fig. 1d and e), but the compositions were not fundamentally altered by the air-drying process (Fig. 4). The changed abundance of certain AOB and AOA phylotypes in response to long-term field N fertilization in fresh soils was still observed in air-dried soils. Intriguingly, the AOA abundance showed clearer changes in response to long-term N fertilization in air-dried soil than in fresh soils. This might be attributable to the elimination of extracellular DNA: air-dried soils could more accurately reflect the differences between treatments. Meanwhile, AOB abundance following the air-drying process decreased to a greater extent than that of AOA. In N-fertilized soils, the proportional decrease in abundance of both AOA and AOB in air-dried soils was much lower than that in nonfertilized soils. These results suggested that the differences in DNA recovery efficiency might be associated with the distinct responses of specific microbial groups to air drying and long-term N fertilization regimens (39).

Taking our results together, DNA-SIP and/or RNA-based approaches showed that both AOB and AOA actively performed ammonia oxidation in all alkaline agricultural soils. Phylogenetic analysis showed that Nitrosospira cluster 3 and N. viennensis clusters were the dominant active AOB and AOA, respectively, following urea-amended microcosm incubation. These two clusters also represented significantly enriched AOB and AOA phylotypes in situ after 22 years of N fertilization. While AOB were consistently dominated by the Nitrosospira cluster 3-related lineage in various soils and conditions, the compositions of active AOA were more greatly affected by different fertilization regimens and soil N conditions. Fosmid clone 54d9- and “Candidatus Nitrosocosmicus franklandianus”-related AOA exclusively dominated the in situ AOA abundance and transcriptional activity in soils with different N fertilization histories. However, only N. viennensis-related AOA showed increased transcriptional activity in soils following 22 years of N fertilization history and became the predominantly growing AOA cluster in urea-amended soil microcosms, irrespective of previous fertilization history. Our results suggest that these agricultural soils might select for, and rely on, a large variety of phylogenetically distant AOA clusters with distinct kinetics for ammonia oxidation under different N situations but that bacterial ammonia oxidization might depend on a few select AOB clusters that adapt to a wide range of N conditions.

MATERIALS AND METHODS

Site description and soil sampling.

The long-term field experiment site was established in 1989 and contained plots that underwent different fertilization regimens. It was located in the Fengqiu State Key Agro-Ecological Experimental Station, Fengqiu County, Henan Province, China (35°01′N, 114°34′E). This site represents the most important cropping regions for wheat and maize production in China (i.e., the Huang-Huai-Hai plain). The soil is derived from alluvial sediments of the Yellow River and is classified as a calcaric fluvisol (FAO), with a sandy loam texture (approximately 9% clay, 21.8% silt) and a pH of >8 in the plough layer. Soils undergoing five different treatments were selected, including: (i) CK, no fertilizer application; (ii) PK, phosphorus (P) and potassium (K) application without N fertilizer; (iii) NP, fertilizer application with N and P; (iv) NK, fertilizer application with N and K; and (v) NPK, fertilizer application with N, P, and K. The annual application rates of N (as urea), P (as superphosphate), and K (as potassium sulfate) were 150, 32.7, and 124.5 kg ha⁻¹ for winter wheat (Triticum aestivum L.) and 150, 26.2, and 124.5 kg ha⁻¹ for summer maize (Zea mays L.), respectively (65). Each treatment was set up in a 9.5-m by 5-m plot with four replicates.

Surface soil at a depth of 0 to 15 cm from each replicate plot was collected in October 2011 (at the maize harvesting stage) and then mixed and homogenized by passing through a 2-mm sieve to remove plant debris and stones. For each replicate, a 0.5-g fresh subsample was immediately mixed with 700 μl RNAlater (Ambion, Life Technologies, USA) in a sterile 2-ml centrifuge tube to stabilize cellular RNA and stored at −80°C until RNA extraction. Approximately 200 g fresh subsamples was stored at −20°C before DNA extraction; 200-g subsamples were air-dried to a constant weight at room temperature to remove extracellular DNA and then stored at −20°C before DNA extraction. Soil DNA and RNA were extracted immediately when prepared. The rest of the soil was kept at 4°C for several days before microcosm construction.

DNA and RNA extraction.

DNA was extracted using a FastDNA spin kit for soil (MP Biomedicals, Cleveland, OH, USA), according to the manufacturer’s instructions. RNA was extracted as previously described (66–68) and purified using recombinant DNase I (RNase free) (TaKaRa Biomedical Technology, Beijing, China) and an RNeasy minikit (Qiagen, Germany) to remove DNA. DNA contamination was ruled out by PCR amplification by employing the universal 16S rRNA gene primers (515F and 907R) of the RNA extracts. The quantities of the DNA and RNA extracts were determined using a NanoDrop ND-1000 UV-visible light spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).

Real-time quantitative PCR.

Prokaryotic 16S rRNA gene and bacterial and archaeal amoA gene abundances in soil DNA extracts were determined using qPCR on a CFX96 optical real-time detection system (Bio-Rad Laboratories, Inc., Hercules, CA, USA), using the primers and conditions shown in Table 2 (69, 70). The qPCR standard was generated using plasmid DNA from one representative clone containing target genes; a dilution series of a standard template over 7 to 8 orders of magnitude per assay was used. The negative control had sterile water used as the template instead of soil DNA extract. Amplification specificity was verified by using melting curve analysis and by agarose gel electrophoresis.

TABLE 2.

Primers and conditions used in this study

Primer	Primer sequence (5′–3′)	Target gene	Thermal profile	Molecular analysis	Reference
amoA-1F	GGGGTTTCTACTGGTGGT	Bacterial amoA gene	95°C, 3.0 min; 40 cycles (95°C, 10 s; 56°C, 30 s; 72°C, 30 s; 80°C, 5 s with plate read); melt curve, 65.0°C to 95.0°C; increment, 0.5°C; 5 s + plate read	Real-time PCR	69
amoA-2R	CCCCTCKGSAAAGCCTTCTTC	Bacterial amoA gene	94°C, 5.0 min; 35 cycles (95°C, 30 s; 56°C, 30 s; 72°C, 30 s); 72°C, 10.0 min; hold at 4°C	Pyrosequencing and clone library	69
Arch-amoAF	STAATGGTCTGGCTTAGACG	Archaeal amoA gene	95°C, 3.0 min; 40 cycles (95°C, 10 s; 55°C, 30 s; 72°C, 30 s; 80°C, 5 s with plate read); melt curve, 65.0°C to 95.0°C; increment, 0.5°C; 5 s + plate read	Real-time PCR	70
Arch-amoAR	GCGGCCATCCATCTGTATGT	Archaeal amoA gene	94°C, 5.0 min; 35 cycles (95°C, 30 s; 55°C, 30 s; 72°C, 45 s); 72°C, 10.0 min; hold at 4°C	Pyrosequencing and clone library	70
515F	GTGCCAGCMGCCGCGG	16S rRNA gene	95°C, 3.0 min; 40 cycles (95°C, 10 s; 55°C, 30 s; 72°C, 30 s; 80°C, 5 s with plate read); melt curve, 65.0 to 95.0°C; increment, 0.5°C; 5 s + plate read	Real-time PCR	71
907R	CCGTCAATTCMTTTRAGTTT	16S rRNA gene	94°C, 5.0 min; 35 cycles (95°C, 30 s; 55°C, 30 s; 72°C, 30 s); 72°C, 10.0 min; hold at 4°C	Pyrosequencing	71

High-throughput sequencing.

High-throughput sequencing was conducted by analyzing the V4 regions of 16S rRNA gene for in situ soil nucleic acid samples. The 515F-907R primer pair (with sample-specific barcode sequences attached to the 5′ end of the 515F primer) was used for 16S rRNA gene amplification (71). PCR for DNA samples was conducted in a total volume of 50 μl using 0.25 μl TaKaRa Ex Taq HS (5 U μl⁻¹), 5 μl 10× Ex Taq buffer (Mg²⁺ Plus), 4 μl deoxynucleoside triphosphate (dNTP) mixture (10 mM), 1 μl of each primer (20 mM), and 1 μl DNA template. The PCR conditions for DNA are shown in Table 2. PCR for RNA samples was performed in a 50-μl volume by using a One Step RNA PCR kit (TaKaRa Biotech, Dalian, China) and following the manufacturer's instructions. For each sample, PCR was conducted with three technical replicates, and the resultant PCR products were pooled and purified using a Mini BEST DNA fragment purification kit, version 3.0 (TaKaRa Biotech, Dalian, China). Finally, the purified PCR amplicons were combined in equimolar ratios in a single tube in preparation for pyrosequencing analysis. The required adapters were added to the PCR amplicon fragments before they were run on a Roche FLX 454 pyrosequencing machine (Roche Diagnostics Corporation, Branford, CT, USA).

Soil microcosms for SIP.

DNA-SIP microcosms were constructed in triplicate for CK and NPK treatments, which represented soils without and with a nitrogen fertilization history, respectively, as previously described in detail (10). Briefly, fresh soil equivalent to 6.0 g (dry weight) of soil in 120-ml serum bottles tightly capped with butyl stoppers were incubated at 28°C in the dark. The three microcosm treatments included a supplement of 5% ¹²CO₂, 5% ¹³CO₂, or 5% ¹³CO₂ in the presence of 100 Pa acetylene, a suicide inhibitor of ammonia oxidation. Either [¹²C]urea or [¹³C]urea was added on a weekly basis for the first 8 weeks in ¹²CO₂- and ¹³CO₂-amended microcosms, respectively, to achieve a concentration of 100 μg N g⁻¹ d.w.s weekly and a total concentration of 800 μg N g⁻¹ d.w.s throughout the incubation period. CO₂ and acetylene were renewed every week. The ¹³CO₂ (99 atom%) was purchased from Sigma-Aldrich. The headspace CO₂ concentration was measured as previously described (10, 11). Before SIP of each microcosm, a 3-day preincubation of soil at 40% maximum water holding capacity was carried out to reactivate microbial activity.

Destructive sampling was performed after 63-day incubation. About 2.0 g soil was sampled from each microcosm and frozen at −20°C for DNA extraction. The rest of the soil replicate was used for inorganic N extraction using 2 M KCl and determination of NH₄⁺-N, NO₂⁻-N, and NO₃⁻-N concentrations using a Skalar SAN Plus segmented flow analyzer (Skalar, Inc., Breda, Netherlands). All NO₂⁻-N, concentrations were below the detection limit, and the nitrification rates were determined by the temporal change in NH₄⁺-N and NO₃⁻-N concentrations in the soil during incubation.

Identification of [¹³C]DNA.

Density gradient centrifugation was performed to separate the ¹³C-labeled DNA from the [¹²C]DNA in triplicate SIP microcosms as previously described (11). DNA fractionation was carried out by displacing the gradient medium with sterile water from the top of an ultracentrifuge tube using an NE-1000 single-syringe pump (New Era Pump Systems, Inc., Farmingdale, NY, USA) with a precisely controlled flow rate of 0.4 ml min⁻¹. A total of 15 DNA gradient fractions were generated with equal volumes of about 350 μl, and a 65-μl aliquot of each fraction was used for a refractive index measurement with an AR200 digital hand-held refractometer (Reichert, Inc., NY, USA) to determine buoyant density as specified previously (72). Nucleic acids were recovered from CsCl solution by precipitation using 2 volumes of polyethylene glycol (PEG) 6000 at 37°C for 1 h, followed by centrifugation at 13,000 × g for 30 min. The fractionated DNA was further purified by washing with 70% ethanol and then dissolved in 30 μl Tris-EDTA (TE) buffer and stored at −20°C for further analysis.

Real-time quantitative PCR of archaeal and bacterial amoA genes was performed on DNA extracts from each fraction to verify the autotrophic AOA and AOB efficiency of ¹³C incorporation using the primers and conditions mentioned above (Table 2). High-throughput sequencing of 16S rRNA genes was conducted using the DNA recovered from the heavy buoyant density fractions (fractions 4 to 7) from the ¹³CO₂ and ¹³CO₂+C₂H₂ microcosms after 63 days of incubation. High-throughput sequencing was carried out and clone libraries of amoA genes of bacteria and archaea were also constructed from the pooled PCR products of the ¹³C-labeled heavy and light DNA fractions of ¹³CO₂-labeled microcosms.

Bioinformatics analysis.

Raw sequences were processed by QIIME (Quantitative Insights Into Microbial Ecology; version 1.3.0) (73). Poor-quality sequences (quality score < 25) and short sequences (<200 bp) were removed. The high-quality sequence reads of 16S rRNA genes were then clustered into operational taxonomic units (OTUs) using a 97% identity threshold with the UPARSE algorithm (74). A representative sequence for each OTU was picked for taxonomic identification using a Ribosomal Database Project (RDP) classifier (http://rdp.cme.msu.edu/). Sequences classified as AOA (Thaumarchaea), AOB (Nitrosomonas and Nitrosospira) (Tables S1 and S2), and NOB (Nitrobacter and Nitrospira) (Table S1) were then selected to calculate the relative abundances, and the representative sequences were extracted with mothur software for phylogenetic analysis (Fig. S2 to S4). Phylogenetic trees were constructed using the neighbor-joining method in MEGA 7.0 with 1,000 replicate bootstraps (75). For amoA gene data sets, OTUs were picked using 80% and 85% identity thresholds for AOA and AOB, respectively. A representative sequence for each OTU was also selected for the phylogenetic analysis. Finally, the composition and relative abundance were estimated for different nitrifier phylotypes according to the similarity of 16S rRNA and amoA genes on the phylogenetic trees, respectively (Fig. S2 to S4; Tables S3 to S7).

Statistical analysis.

One-way analysis of variance (ANOVA) was performed to assess the effects of different fertilization histories on soil nitrification rate, 16S rRNA and amoA gene absolute abundance, and the relative abundance of different nitrifier-related 16S rRNA genes, followed by a Tukey post hoc test to determine the significant differences in mean values. All analyses were conducted using SPSS 16.0 (IBM Corp., Armonk, NY, USA), and P values of <0.05 were considered statistically significant.

Data availability.

The raw amplicon sequence data sets for 16S rRNA genes (or transcripts) and amoA genes have been deposited in the National Center for Biotechnology Information database under the accession number PRJNA628867. All amoA gene sequences obtained from clone library sequencing from heavy DNA fractions have been deposited in GenBank with accession numbers MT416025 to MT416062.