Introduction

Nitrification is the only oxidative process that links the reduced and oxidized pools of inorganic nitrogen to sustain the global nitrogen cycle (Leininger et al. 2006). Aerobic ammonia oxidation, the first and rate-limiting step in nitrification, thus plays an integral role for the global nitrogen balance. Ammonia-oxidizing bacteria (AOB) as well as recently discovered ammonia-oxidizing archaea (AOA) (Francis et al. 2005; Konneke et al. 2005; Venter et al. 2004) are two distinct ammonia-oxidizing microorganisms (AOMs) that catalyze the oxidation of ammonia to hydroxylamine and further to nitrite. Both AOMs demonstrate wide distribution in terrestrial (Adair and Schwartz 2008; Le Roux et al. 2008; Wessen et al. 2011), freshwater (Liu et al. 2013; Wu et al. 2010), and marine environments (Beman et al. 2010; Bouskill et al. 2012; Mosier and Francis 2008).

Oceans cover 71 % of the earth’s surface and contain enormous biomass and large biodiversity; thus, marine microorganisms are a crucial component in global nutrient cycles (Arrigo 2005). Influenced by many environmental factors, such as salinity (Caffrey et al. 2007), dissolved oxygen (DO) (Bouskill et al. 2012; Molina et al. 2010), and ammonia concentration (Wuchter et al. 2006), the abundance and activity of AOA and AOB varied in different marine environments. Among these factors, ammonia has been identified as one of the most significant factors structuring AOM ecosystems. A time series study in the North Sea revealed that the abundance of AOA amoA gene was negatively correlated with ammonium concentration (3–12.7 μM) (Wuchter et al. 2006). Physiological study of the first isolated AOA strain demonstrated that the ammonia substrate threshold for AOA was considerably lower than that for AOB (Martens-Habbena et al. 2009). Therefore, an acceptable hypothesis could be proposed that AOA may contribute more to nitrification in environments with low ammonia concentration but AOB may be more favorable in environments with high ammonia concentration.

The same phenomenon of AOA and AOB influenced by ammonia concentration was found in soil environment (Glaser et al. 2010). More importantly, pieces of evidence for the above hypothesis have been provided by Di et al. (Di et al. 2010; Di et al. 2009), Verhamme et al. (Verhamme et al. 2011), and Pratscher et al. (Pratscher et al. 2011) in lab simulating microcosms for soil environment. Unfortunately, in aquatic environments, the hypothesis has not been confirmed yet in laboratory cultivation system, and the dynamics of phylogenetic diversity of active AOA and AOB under different ammonia concentrations is still unclear. Hence, the influence of ammonia concentration on the activities of aquatic AOA and AOB needs to be further verified, especially as terrestrial and marine species mainly belong to different lineages.

By setting up estuarine sediment-coastal water microcosms with a series of ammonia concentrations, this study aims to address the critical knowledge gaps in understanding the transcriptional activity of AOA and AOB and the dynamics of phylogenetic diversity of active AOA and AOB at different ammonia concentrations in aquatic environments. Molecular technologies of quantitative PCR (qPCR), quantitative reverse transcription PCR (RT-qPCR), and 454 pyrosequencing were applied to track precisely the archaeal and bacterial amoA genes. On the basis of chemolithotrophic ammonia oxidation carried out by AOA and AOB (Jia and Conrad 2009; Pratscher et al. 2011; Xia et al. 2011; Zhang et al. 2010), DNA-stable isotope probing (SIP) was also adopted to reveal potential active AOA and AOB who absorbed inorganic carbon in the coastal microcosms.

Materials and methods

Sediment sampling and characteristics measurement

The sediment and coastal water samples used for microcosms’ incubation were collected from the Hangzhou Bay at the site of N 30° 12.780′, E 120° 51.144′, using a sediment sampler (Van Veen, HYDRO-BIOS, Germany). The sediment properties were as follows: water content 31.6 %, pH 8.53, total organic carbon (TOC) 361.4 mg g−1 dried sediment, ammonia nitrogen (NH4 +-N) 22.69 μg g−1 dried sediment, nitrite nitrogen (NO2 −-N) no detection, nitrate nitrogen (NO3 −-N) 2.89 μg g−1 dried sediment, and total phosphorus (TP) 0.53 mg g−1 dried sediment. The water quality was as follows: pH 7.80, DO 5.82 mg L−1, salinity 5.36 psu, chemical oxygen demand (COD) 3.07 mg L−1, NH4 +-N 0.17 mg L−1, NO2 −-N 0.03 mg L−1, NO3 −-N 2.89 mg L−1. The methods adopted for the detection of samples’ characteristics were listed in Table S1.

Microcosm incubation

The laboratory microcosms were constructed in a series of 120 mL Erlenmeyer flasks, each with 10 g sediment and 50 mL coastal water sample sealed with sterile sealing film. The microcosms were incubated under five different coastal water NH4 +-N concentrations of 0.1, 2, 6, 12, and 20 mg L−1, maintained by supplementing ammonium sulfate to the designed concentration after determining the demands every morning. For example, as the ammonium in microcosms with NH4 +-N concentrations of 0.1 mg L−1 could be consumed completely in 24 h, 0.1 mg NH4 +-N L−1 was supplemented into the microcosms everyday. The microcosm without additional ammonium sulfate was constructed as the control. The total amounts of NH4 +-N supplemented into the microcosms are listed in Table S2. Triplicate microcosms were constructed for each NH4 +-N concentration. Six groups of the microcosms with NH4 +-N concentration from 0 (the control) to 20 mg L−1 were coded as A0, A0.1, A2, A6, A12, and A20. Microcosms were incubated at 25 °C in dark and mixed by shaking every morning for air exchange to maintain aerobic condition. At the day 28, approximate 4 g of the cultivated sediment in each microcosm were sampled and frozen at −70 °C immediately for further analysis. After 56-day cultivation, all the cultivated sediments were collected, parts of each sample were used to RNA extraction immediately, and the remaining sediment was frozen at −70 °C for further analysis. The aqueous phases after 56-day cultivation were used for water quality analysis.

Stable isotope probing of potential active ammonia oxidizers

The DNA-SIP microcosms were constructed in 120-mL serum bottles with 10 g sediment and 50 mL coastal water sample sealed with rubber stoppers and plastic caps. Low (0.1 mg L−1) and high (20 mg L−1) NH4 +-N concentrations were designed for DNA-SIP study. The NH4 +-N concentrations were maintained by supplementing ammonium sulfate into the microcosms every morning. The total amounts of NH4 +-N supplemented into the microcosms are also listed in Table S2. The air in the headspace of each bottle was replaced with synthetic air (80 % N2 and 20 % O2) to remove CO2 from the microcosms, and external bicarbonates were added to the microcosms as the additional inorganic carbon (IC) sources. At each NH4 +-N concentration, two groups of microcosms were supplemented respectively by two IC sources, 100 mg L−1 NaH12CO3 and 100 mg L−1 NaH13CO3 (Sigma-Aldrich Co., St Louis, MO, USA). The initial pH of the microcosms was shown in Table S3. The sodium bicarbonates were complemented into the corresponding microcosms every 2 weeks. Four groups of the SIP microcosms were coded as 12C-0.1, 12C-20, 13C-0.1, and 13C-20. Duplicate microcosms of each group were incubated at 25 °C in the dark. Sediments after 28 and 56-day cultivation were sampled and frozen at −70 °C immediately for further analysis. The aqueous phases after 56-day cultivation were collected for water quality analysis.

Nucleic acid extraction and reverse transcription

DNA was extracted from 0.5 g of sediment samples using the DNA PowerSoil® Total DNA Isolation Kit (Mo Bio, Carlsbad, CA). The concentration of extracted DNA was measured by a NanoDrop 2000 UV-Vis spectrophotometer (Thermo Fisher, Wilmington, MA, USA). RNA was extracted from 3 g of samples using the RNA PowerSoil® Total RNA Isolation Kit (Mo Bio, Carlsbad, CA). The extracted RNA solution was treated by DNase to eliminate the residual DNA using a RTSTM DNase Kit (Mo Bio, Carlsbad, CA) immediately, followed by reverse transcription with a PrimeScriptTM RT reagent Kit (Takara, Dalian, China). The extracted DNA and the obtained complementary DNA (cDNA) were stored at −70 °C for further analysis.

SIP DNA fractionation

The DNA samples from the duplicate microcosms were mixed prior to fractionation. Density gradient centrifugation of the mixed DNA in CsCl gradients was performed as described by Neufeld et al. (Neufeld et al. 2007). Briefly, approximate 2 μg extracted DNA was combined into a 7.163 M CsCl gradient buffer to achieve an initial buoyant density of 1.72 g mL−1. The ultracentrifugation of the mixed solution was performed in 5 mL OptiSeal polyallomer tubes with a VTi 90 vertical rotor (Beckman Coulter, Palo Alto, CA, USA) at 177,000 gav for 40 h at 20 °C. Fourteen equal volumes were fractionated from centrifuged gradient by slowly loading air into the top of the tubes with needles and an Ismatec@ peristaltic pump (IDEX, Wertheim, Germany). After fractionation, the density of each fraction was determined by weighing method using an electronic analytical balance (XS105DU, METTLER TOLEDO, Switzerland) and a micropipette (Eppendorf, Germany). Nucleic acids were precipitated from CsCl gradients by polyethylene glycol 6000 (Wako, Japan) and glycogen (Fermentas, Canada) overnight and re-dissolved in 30 μL sterile water after washing with 70 % ethanol.

qPCR

Quantitative PCR was performed on an iCycler IQ5 thermocycler (Bio-Rad, CA, USA) based on SYBR Green I method. The primers and thermal programs for quantitative amplification of bacterial and archaeal amoA genes were listed in Table S4. Each reaction was performed in 20 μL, which consisted of 10 μL of SYBRs Premix Ex TaqTM (Takara, Dalian, China), 1.0 μL of template DNA, and 0.4 μL of primers. The standard plasmids for quantitative analysis were selected from the positive clones of the target genes amplified from the sediment sample, as described previously (Bai et al. 2012b). Tenfold serial dilutions of the plasmids DNA were subjected to a quantitative PCR assay in triplicate to generate a standard curve and to check the amplification efficiency. The correlation coefficients of the standard curves (R 2) were all >0.99, and the amplification efficiencies ranged from 94.8 to 104.7 %.

Pyrosequencing and phylogenetic analysis of amoA genes

The archaeal and bacterial amoA genes in the cDNA samples from non-SIP microcosms and the DNA samples from the heavy fractions of SIP microcosms were analyzed by pyrosequencing. The pyrosequencing was performed on a Roche GS-FLX Titanium Sequencer (Roche Diagnostics Corporation, Branford, CT) using a described process (Bai et al. 2012a). Briefly, the amplifications were achieved on an ABI9700 thermocycler (ABI, Foster City, USA) with barcoded primers, using TransStart Fastpfu DNA polymerase (TransGen, Beijing, China). The primers and thermal programs were listed in Table S4. After purifying the PCR products, the amplicon libraries were generated by emulsion PCR and sequenced on the Roche GS-FLX Genome Sequencer using the 454/Roche B sequencing primer kit. The pyrosequencing flowgrams were converted to sequence and analyzed with standard pipeline adopting QIIME software (Caporaso et al. 2010). Initially, low quality and ambiguous reads, i.e., the reads with ambiguous base >0, sequence lengths <440 bp, or average sequence quality <25, were eliminated by filtering and denoising after barcodes were removed from the sequence reads. Subsequently, putative chimeras were identified and removed from the pre-treated sequences with Chimera-uchime. Applying Mothur software (Schloss et al. 2009), the remained sequences were clustered into different OTUs with 97 % of sequence identity threshold for bacterial amoA gene (Gao et al. 2013; Wang et al. 2012) and 95 % for archeal amoA gene (Beman et al. 2008; Mosier and Francis 2008). Then, the rarefaction and diversity statistics were calculated. The representative sequences in main OTUs were selected to blast with the National Center for Biotechnology Information (NCBI) database and obtain the closest sequences published in the NCBI. The phylogenetic trees of the selected OTUs and the closest sequences were constructed using the neighbor-joining method with the MEGA 5.2 software (Kumar et al. 2008). All original 454 sequences were archived at the NCBI Sequence Read Archive (SRA) under accession SRP035514.

Results

Microcosms’ incubation with different ammonia concentrations

After incubation for 56 days, a significant increase of nitrate in the aqueous phase (Fig. 1) indicated the stimulation of nitrification in all microcosms; however, nitrite accumulated in the three microcosms supplemented with the higher ammonia concentrations, especially in A20 where final concentration of NO2 −-N (163.16 ± 13.29 mg L−1) was higher than that of NO3 −-N (63.20 ± 5.79 mg L−1) (P = 0.000). Together, the total amount of accumulated nitrite- and nitrate-nitrogen increased from 6.69 ± 0.14 to 226.36 ± 7.92 mg L−1 as the NH4 +-N concentration improved from control (less than 0.1 mg L−1) to 20 mg L−1.

Fig. 1
figure 1

Nitrite-N and nitrate-N in the aquatic phases of the microcosms after 56-day cultivation. Group “Ax” represents the microcosm with ammonia concentration of “x” mg L−1. Error bars represent the standard deviations of the triplicate microcosms. Different letters above the bars (lowercase letters for nitrite-N, capital letters for nitrate-N) indicate a significant difference (P < 0.05) by ANOVA analysis using Student-Newman-Keulstest method

Full size image

In the beginning of the microcosm incubation, AOA (8.94 × 106 ± 1.65 × 106 copies g−1) was nearly one order of magnitude higher than AOB (1.12 × 106 ± 2.11 × 105 copies g−1) (P = 0.001) in the sediments, as shown in Fig. 2. After 28-day cultivation, the copy numbers of archaeal amoA gene dropped in all the sediments, fluctuating in a range from 8.23 × 105 ± 2.2 × 105 to 3.28 × 106 ± 3.18 × 105 copies g−1 sediment, while the copy numbers of bacterial amoA gene dropped only in the sediments of group A0 and A0.1 but maintained in group A2, A6, and A12 and even rose to 2.36 × 106 ± 2.35 × 105 copies g−1 sediment in group A20. After 56-day cultivation, the copy numbers of archaeal amoA gene recovered to 3.26 × 106 ± 8.44 × 105 to 4.49 × 106 ± 1.30 × 106 copies g−1 sediment in A0, A0.1 and A6, even exceeded the original quantity in A2, A12, and A20, while the copy numbers of bacterial amoA gene maintained in A0.1 and A20 and increased significantly in the other groups, even far exceeding the original quantity in group A2, A6, and A12.

Fig. 2
figure 2

Changes in the abundance of amoA genes in the sediments cultivated under different ammonia concentrations. Error bars represent standard errors of triplicate microcosms. Different letters above the bars indicate a significant difference (P < 0.05). a Archaeal amoA and b bacterial amoA

Full size image

Regression analysis (Fig. 3) showed that the archaeal amoA gene abundance had slightly negative correlation and moderate positive correlation with the NH4 +-N concentration in day 28 and day 56, respectively. However, significant positive correlations emerged between the bacterial amoA gene abundance and the NH4 +-N concentration. In day 56, if excluding the data of group A20, such positive correlation became evident again.

Fig. 3
figure 3

Relationship between NH4 +-N concentration and amoA gene copy numbers in the microcosms. Vertical error bars represent respectively the standard errors of amoA gene copies from triplicate microcosms

Full size image

The abundance of transcribed amoA genes measured by RT-qPCR was shown in Fig. 4. Archaeal amoA gene had more transcripts than bacterial at lower ammonia concentrations of group A0 (P = 0.018) and A0.1 (P = 0.074) but less transcripts than bacterial at the other higher ammonia concentrations. The abundance of transcribed bacterial amoA gene in group A0 (144 ± 4 copies g−1 sediment) and A0.1 (281 ± 103 copies g−1 sediment) were much lower than that in other groups (8.9 × 103 ± 3.5 × 103 to 3.5 × 105 ± 3.9 × 104 copies g−1 sediment) (P = 0.049 and 0.051 for A0 and A0.1 comparing to A20, respectively). The transcribed archaeal amoA gene increased from group A0 to group A2 but decreased subsequently from group A2 to group A12 and finally was not detected in all three microcosms of group A20.

Fig. 4
figure 4

Changes in the abundance of transcribed amoA genes in mRNA of the sediments cultivated under different ammonia concentrations after 56-day cultivation. Error bars represent the standard errors of triplicate microcosms. Different letters above the bars (lowercase letters for archaeal amoA, capital letters for bacterial amoA) indicate a significant difference (P < 0.05)

Full size image

DNA-SIP incubation

During the incubation of microcosms adding different C isotopes as the IC source, the changes of nitrite, nitrate concentrations, and the abundance of amoA genes in the 13C-NaHCO3 microcosm were consistent with that in the 12C one, as Fig. 5 indicated.

Fig. 5
figure 5

DNA-SIP microcosms’ incubation. a Nitrite-N and nitrate-N in SIP microcosms after 56-day cultivation; b change of the abundance of archaeal amoA gene; and c change of the abundance of bacterial amoA gene. Error bars represent the standard errors of duplicate microcosms. Different letters above the bars indicate a significant difference (P < 0.05), lowercase letters for nitrite-N, capital letters for nitrate-N in a. Group “12C-x” and “13C-x” represent, respectively, the microcosms adding 12C or 13C as the inorganic carbon source with ammonia concentration of “x” mg L−1

Full size image

Nitrification achieved in all SIP microcosms, especially under high ammonia concentration (Fig. 5a). Nitrate accumulated much more than nitrite under both low and high ammonia concentrations. The change of the abundance of archaeal amoA gene was different under the different ammonia concentrations (Fig. 5b). After 56-day cultivation, the archaeal amoA gene slightly increased in microcosms with 0.1 mg L−1 NH4 +-N but decreased evidently in microcosms with 20 mg L−1 NH4 +-N. The abundance of bacterial amoA gene increased in all groups of microcosms during the cultivation (Fig. 5c). The increment of bacterial amoA gene in the 20 mg L−1 NH4 +-N microcosm was significantly higher than that in the microcosm with 0.1 mg L−1 NH4 +-N.

The relative proportions of both amoA genes in the fractions after density gradient centrifugation of DNA were illustrated in Fig. 6. Before the cultivation, the archaeal and bacterial amoA genes reached the maximum proportions in light fractions of buoyant densities around 1.68 and 1.69 mg L−1, respectively. In the 12C-NaHCO3 treatment, the abundance peaks of archaeal and bacterial amoA genes did not shift throughout the whole incubation at both 0.1 and 20 mg L−1 NH4 +-N concentrations. In the 13C-NaHCO3 treatment, a clear shift of archaeal amoA gene abundance was observed only at 0.1 mg L−1 NH4 +-N concentration that the peak moved to heavy fractions with a final buoyant density around 1.71 mg L−1 at both day 28 and 56; while under the same low NH4 +-N condition, the shift of bacterial amoA gene abundance was not evident at day 28 but became clear at day 56 that the peak moved to the heavy fraction of buoyant density around 1.72 mg L−1, and under the high NH4 +-N condition of 20 mg L−1, only the peak of bacterial amoA gene abundance exhibited a clear-cut shift to heavy fractions with a buoyant density around 1.72 mg L−1 at both days 28 and 56.

Fig. 6
figure 6

Distribution of the relative abundance of archaeal and bacterial amoA genes in CsCl gradient for the DNA-SIP microcosms. Fourteen fractions of genomic DNA extracted from centrifuge tubes covered a range of buoyant density from 1.66 to 1.76 g mL−1. The vertical axis reveals the proportion of AOA or AOB amoA gene in each fraction to the total abundance of a gradient set. Vertical and horizontal error bars represent, respectively, the standard errors of the relative abundance and buoyant density from triplicate detection of each fraction sample

Full size image

Diversities of active AOA and AOB in different microcosms

After 56 days incubation under different ammonia concentrations, the archaeal and bacterial amoA genes in the cDNA samples as well as in the DNA samples of the heavy fractions in the 13C-NaHCO3-treated microcosms were analyzed by pyrosequencing.

For AOA, owning to the low abundance of archaeal amoA in the cDNA sample from group A12 and in the DNA-SIP sample of the heavy fraction from group 13C-20, no target product was obtained by the PCR amplification. Thus, five pyrosequencing libraries of archaeal amoA gene from A0, A0.1, A2, A6, and 13C-0.1 were constructed. The estimations and the rarefaction curves of the five libraries were shown in Table S5 and Fig. S1, respectively. A total of 126 OTUs were obtained from 8916 sequences. The main OTUs of archaeal amoA gene (Fig. S2) from different microcosms were assorted by phylogenetic tree. As shown in Fig. S3, all detected AOA OTUs were placed into four clusters of Nitrosopumilus maritimus. The distribution and relative abundance of the detected AOA clusters were further analyzed (Fig. 7a). The main active AOA in A0 were N. maritimus C12 sister and N. maritimus SCM1 clusters, accounted for 48.1 and 44.0 % of the total sequences, respectively. With ammonia increased, N. maritimus C12 sister gradually increased to over 80 %; however N. maritimus SCM1 cluster decreased sharply and finally disappeared. Different composition of AOA was found in 13C-0.1, where N. maritimus A10, N. maritimus C12, and N. maritimus C12 sister clusters accounted for 66.9, 24.9, and 1.9 %, respectively.

Fig. 7
figure 7

Distribution and relative abundance of the phylogenetic AOA and AOB groups. The group without identification consists of the OTUs that accounted for less than 1 % of the total sequences in the pyrosequencing library. a Phylogenetic AOA groups and b phylogenetic AOB groups

Full size image

For AOB, owning to the low abundance of bacterial amoA gene in A0 and A0.1, no target product was obtained by the amplification. Thus, six pyrosequencing libraries of bacterial amoA gene from A2, A6, A12, A20, 13C-0.1, and 13C-20 were constructed by pyrosequencing. The estimations and the rarefaction curves of the six libraries were shown in Table S6 and Fig. S4, respectively. A total of 168 OTUs were obtained from 15,129 sequences. The main OTUs of bacterial amoA gene (Fig. S5) from different microcosms were also assorted by phylogenetic tree. As shown in Fig. S6, the detected AOB OTUs were placed into five clusters of genus Nitrosomonas and one cluster of genus Nitrosospira. The distribution and relative abundance of the detected AOB clusters were also analyzed (Fig. 7b). In A2, Nitrosomonas oligotropha and Nitrosomonas nitrosa clusters accounted for 61.2 and 34.5 % in the total sequences, respectively. However, both clusters decreased equally in A6 and A12 and almost disappeared in A20. Nitrosomonas communis cluster became the main active AOB under high ammonia conditions, which accounted for 42.2, 48.5, and 95.1 % in A6, A12, and A20, respectively. In DNA-SIP microcosms, the composition of 13C-labeled AOB amoA gene were markedly different, as N. nitrosa cluster was absolutely the dominant AOB in both 13C-0.1 and 13C-20.

Discussion

In this study, a series of coastal sediment-water microcosms with different ammonia concentrations were incubated. The nitrite and nitrate in the non-SIP microcosms (Fig. 1) and SIP microcosms (Fig. 5a) demonstrated the achievement of nitrification. The total amount of nitrite and nitrate increased with the improvement of ammonia concentration, indicating that the ammonia oxidation capacity of the sediment was enhanced in ammonia-rich environments. However, nitrite accumulation implied that nitrification was incomplete in the high ammonia groups, which might be due to the insufficient IC source according to the chemolithotrophy of nitrite oxidizer (Xia et al. 2011). The less accumulation of nitrite in the SIP microcosms (Fig. 5a), which were supplemented with external bicarbonate, supported this deduction.

A strong evidence for ammonia concentration effecting on the activities of AOA and AOB in aquatic environments was provided by this study. First, the positive correlations between the abundance of AOB amoA gene, not AOA amoA gene, and ammonia concentration were revealed in this study (Fig. 3). These indicated that ammonia may stimulate the growth of AOB in the NH4 +-N concentration range studied. However, the linear regressions were obstructed by the lower abundance of AOB amoA gene in group A20 at day 56 (Fig. 2), which was probably a result of the inhabitation of AOB growth by highly accumulated nitrite (Fig. 1), as nitrite exerts inhibitory effect on the respiration rate of AOB (Contreras et al. 2008). The negative correlation between the archaeal amoA gene abundance in day 28 and the NH4 +-N concentration indicated that in the NH4 +-N concentration range studied, higher NH4 +-N concentration may inhibit the growth of AOA. However, a moderate positive correlation was found between AOA abundance and the NH4 +-N concentration in day 56. The recovered AOA was related with the accumulated nitrite in the microcosms (Fig. 1). We deduced that the growth of AOA might be promoted by the highly accumulated nitrite in the microcosm through “nitrifier denitrification” using copper-containing nitrite reductase (nirK), which has been proved widespread in AOA (Blainey et al. 2011; Lund et al. 2012; Walker et al. 2010).

Second, the ratio of the transcribed AOB/AOA amoA gene increased from 0.1 in low ammonia microcosms (group A0 and A0.1) to 43 in high ammonia microcosm (group A12), and the transciption of AOA amoA even was not detected in the highest ammonia microcosm (group A20) (Fig. 4).

Third, the incorporation of 13C into genomic DNA revealed inconsistent growth and activity between AOA and AOB at both low and high ammonia concentration (Fig. 5). In the microcosms with 0.1 mg L−1 NH4 +-N, both AOA and AOB involved in nitrification as the peaks of archaeal and bacterial amoA genes shifted to a heavy fractions of buoyant densities (Fig. 6). Nevertheless, the slower shift rate of the peak of AOB amoA gene was consistent with the less AOB amoA gene in cDNA (Fig. 4), indicating that AOA exerted higher ammonia oxidation activity than AOB at the lower ammonia concentration. In the microcosms with 20 mg L−1 NH4 +-N, the stability of the peak of AOA amoA gene (Fig. 6) indicated the inhibited nitrifying activity of AOA, which was in good agreement with the undetected AOA amoA gene in cDNA (Fig. 4). Consistently, the abundance of AOA amoA gene in SIP microcosms was found decreased at the higher ammonia concentration (Fig. 5b). Therefore, AOA plays a more important role in ammonia oxidation in the ammonia-limited water environments (0.1 mg L−1), while AOB thrives in the ammonia-rich ones (20 mg L−1).

Pure culture experiments of AOA species also support the verdict, as AOA have low half-saturation constant and strong tolerance in minimal ammonia environments. The kinetic study of N. maritimus SCM1 obtained the lowest half-saturation constant (Km = 133 nM) and the minimum ammonia concentration (less than 20 nM), which was over 100-fold lower than that required by AOB (Martens-Habbena et al. 2009). The growth of AOA strains were inhibited at certain thresholds of environmental ammonia concentration, e.g., N. maritimus SCM1, 2 mM (Martens-Habbena et al. 2009); Nitrososphaera gargensis, 3.08 mM (Hatzenpichler et al. 2008); Nitrososphaera viennensis EN76, 20 mM (Tourna et al. 2011), Nitrosotalea devanaterra Nd1, 20 mM (Lehtovirta-Morley et al. 2011), and AOA strain MY1, 50 mM (Jung et al. 2011). Those ammonia thresholds were significantly lower than the maximum ammonia tolerance (50–1000 mM) of the majority of AOB.

Although AOA may play a more important role in ammonia-limited aquatic environments, another discovery in this study is that the abundance of AOA and the transcribed AOA amoA gene increased as ammonia concentration was increased from 0.1 to 2 mg L−1 (Fig. 2a and Fig. 4). We consider that the ammonia oxidation activity of AOA would be inhibited only when the environmental ammonia exceeded a certain concentration, for example, 6 mg L−1 in this study.

Regarding the AOM species in the coastal microcosms, all detected AOA were N. maritimus. In contrast with the AOA strains isolated from soils (Lehtovirta-Morley et al. 2011; Tourna et al. 2011) and hot springs (de la Torre et al. 2008; Hatzenpichler et al. 2008), our study shows that AOA in different niches mainly belong to different lineages.

The composition of AOB was a little more complicated than that of AOA. The dominant active AOB gradually shifted from N. oligotropha and N. nitrosa in A2 to N. communis in A20 (Fig. 7b). According to previous researches, the growth of N. oligotropha could be inhibited by low ammonia concentration similar to the AOA strain Candidatus Nitrososphaera sp. EN76 (Tourna et al. 2011); the half-saturation constant of Nitrosomonas europaea was 10 times higher than that of N. oligotropha (Martens-Habbena et al. 2009), and the activity of hydroxylamine-cytochrome c reductase (HCR, transforming hydroxylamine to nitrite) of N. communis strain YNSRA was twice higher than that of N. europaea ATCC25978T (Tokuyama et al. 2004). These explained the shift of the active AOB in our study.

Comparing the SIP microcosms with the non-SIP microcosms (Fig. 7), the potential active ammonia oxidizers detected in the heavy fractions were quite different from the active ammonia oxidizers measured in the RNA from non-SIP microcosms. This phenomenon may attribute to the influence of the supplement of inorganic carbon. In a recent study (Fukushima et al. 2013), N. nitrosa lineage was identified as the dominant AOB in the high-IC bioreactor (100 mg L−1 IC), while N. europaea increased in the low-IC bioreactor (15 mg L−1). Likewise in the SIP microcosms in this study, the initial IC concentration was 100 mg L−1, the dominant AOB cluster detected in heavy fractions was N. nitrosa lineage. In addition, nitrifier denitrification and other metabolic abilities of AOM undiscovered may be the reasons.

Taken together, our study provides solid evidence for the higher activity of AOA in lower ammonia concentration (NH4 +-N ≤ 0.1 mg L−1) and AOB in higher ammonia concentration (NH4 +-N ≥ 2 mg L−1) and showed the shift of diverse lineages of the active AOM with NH4 +-N increasing from less than 0.1 to 20 mg L−1. The results confirmed the effect of ammonia concentration on the AOM ecosystem and revealed the structures of AOM under different ammonia concentrations. The observation of inconformity between the transcribed and 13C-labeled amoA genes suggests the complex metabolic pathways of the ammonia oxidizers, for which further research is need.