Responses of community structure of amoA-encoding archaea and ammonia-oxidizing bacteria in ammonia biofilter with rockwool mixtures to the gradual increases in ammonium and nitrate



Tomoko Yasuda, Institute of Livestock and Grassland Science, National Agriculture and Food Research Organization, 2 Ikenodai, Tsukuba, Ibaraki 305-0901, Japan.




To investigate community shifts of amoA-encoding archaea (AEA) and ammonia-oxidizing bacteria (AOB) in biofilter under nitrogen accumulation process.

Methods and Results

A laboratory-scale rockwool biofilter with an irrigated water circulation system was operated for 436 days with ammonia loading rates of 49–63 NH3 g m−3 day−1. The AEA and AOB communities were investigated by denaturing gradient gel electrophoresis, sequencing and real-time PCR analysis based on amoA genes. The results indicated that changes in abundance and community compositions occurred in a different manner between archaeal and bacterial amoA during the operation. However, both microbial community structures mainly varied when free ammonia (FA) concentrations in circulation water were increasing, which caused a temporal decline in reactor performance. Dominant amoA sequences after this transition were related to Thaumarchaeotal Group I.1b, Nitrosomonas europaea lineages and one subcluster within Nitrosospira sp. cluster 3, for archaea and bacteria, respectively.


The specific FA in circulation water seems to be the important factor, which relates to the AOB and AEA community shifts in the biofilter besides ammonium and pH.

Significance and Impact of the Study

One of the key factors for regulating AEA and AOB communities was proposed that is useful for optimizing biofiltration technology.


Considerable amounts of ammonia (NH3) are emitted from the livestock manure composting process. Ammonia emissions from the swine manure composting, for example, amounted to 11–14% of the total nitrogen in manure (Kuroda et al. 1996; Fukumoto et al. 2003) and even more than 70% under aerobic decomposition conditions from poultry excreta (Kirchmann and Lundvall 1998). Ammonia emissions from the agricultural animal waste were estimated to account for as much as 40% of global atmospheric NH3 emissions in 1993 and followed an increasing trend (Galloway et al. 2004). These NH3 emissions pose a potentially serious problem for the environment, such as soil acidification and eutrophication of soils and water, in addition to the odour problem.

Biological air purification technology is widely used on a livestock farm to treat NH3-loaded gases because of its relatively low energy requirement and operational costs and high-efficiency removal compared with other chemical or physical technologies (Williams and Miller 1993; Deshusses 2005). Among the three major configurations of biological technology (biofilters, biotrickling filters and bioscrubbers), biofilters are usually more simple and cost-effective than others (Deshusses 2005) and are suitable for the application on livestock farms. Nitrification plays an important role in the biofiltration systems as a critical step in NH3 removal, and NH3 absorption is accelerated in the resultant acidity (Joshi et al. 2000). The products of nitrification, together with the residual ammonium (NH4+), therefore tend to be accumulated in the packing materials and drain water (Kanagawa et al. 2004; Chen et al. 2005; Yasuda et al. 2009).

One of the concerns in the nitrogen accumulating condition is the potential damage to nitrification activity, which is attributed to the high concentrations of NH4+ and nitrite/nitrate (NO2/NO3). Ammonia oxidation is the first step of chemoautotrophic nitrification, which is mainly catalysed by ammonia-oxidizing bacteria (AOB). The adverse effects of the high salt concentrations on the NH3 oxidation activity of Nitrosomonas europaea, due to high osmotic pressure, have been reported (Hunik et al. 1992). Another study observed the upper limit of the cumulative NH3 removal in the biofilter under a certain condition (Smet et al. 2000). On the contrary, an AOB isolate from packed tower deodorization plants of chicken farms reportedly showed a preference for high concentrations of NH4+, as much as 303 mmol l−1 (Hatayama et al. 1999). Although community shifts of AOB have been observed during biofilter operation (Sakano and Kerkhof 1998; Jun and Wenfeng 2009), the effects of the community shifts on biofilter performance remain to be investigated.

A certain archaeal group, recently categorized as thaumarchaeota, is known to involve autotrophic ammonia oxidation (Könneke et al. 2005; Brochier-Armanet et al. 2008). The occurrence of the archaeal amoA gene in nitrifying wastewater treatment plants (Park et al. 2006; Wells et al. 2009; Zhang et al. 2009), as well as in natural environments such as oceans, soils and sediments has been widely described (Francis et al. 2005; Leininger et al. 2006; Mosier and Francis 2008). The amoA-encoding archaea (AEA) are considered to play an important role in ammonia oxidation (Nicol and Schleper 2006; Francis et al. 2007), although not all the AEA might be autotrophic ammonia oxidizers as pointed out recently by Mußmann et al. (2011). Archaeal amoA genes were also detected in a full-scale biofilter packed with rockwool mixture (Yasuda et al. 2010). The effects of the high strength of NH4+ and NO2/NO3 on the community structure of AEA in the air purification biofilter are totally unknown. How the AEA and AOB communities respond to the gradual changes in the reactor environment under the nitrogen accumulation process must be determined in order to achieve long-term NH3 removal performance.

In this study, a laboratory-scale biofilter with rockwool mixtures was operated with an NH3 loading rate of 49–63 NH3 g m−3 day−1. The reactor was basically designed as a scaled down rockwool biofilter except that irrigated water was circulated. Water circulation types of biofilters are often used in livestock farms, because wastewater volume can be minimized. Supplied NH3 is considered to be converted into another gaseous and organic nitrogen in addition to NH4+ and NO2/NO3 (Joshi et al. 2000; Smet et al. 2000; Chen et al. 2005). However, the forms of gaseous nitrogen compounds have not been well investigated. Denitrification is possibly responsible for the gaseous nitrogen losses, but the possible contribution of an anaerobic ammonium oxidation (anammox) pathway cannot be excluded, considering that anammox activity appeared in livestock wastewater treatment facilities (Waki et al. 2010). The contribution of either denitrification or anammox to the gaseous nitrogen loss was determined using a 15N tracer experiment as a means for better understanding of the nitrogen elimination mechanisms. Changes in the community structure of AEA and AOB were investigated using denaturing gradient gel electrophoresis (DGGE) and sequencing methods. The abundance of amoA genes was determined by real-time PCR.

Materials and methods

Description of laboratory-scale NH3 deodorization reactor

A schematic diagram of a laboratory-scale biofilter was shown in Fig. 1. Rockwool packing materials, a mixture of mainly rockwool, urethane, zeolite and dried chicken faeces, were obtained from a full-scale biofilter (Yasuda et al. 2009), and its aggregates were broken down by sieving (4·0-mm mesh). Separated particles were subsequently mixed except for urethane and stored at 4°C until use. Rockwool mixtures were packed within a transparent plastic cylinder (φ = 8 cm, 30 cm height) with a volume of 0·8 l (223 g dry weight) and placed in the incubator (28°C). The reactor was operated 5 days before the experiment with only air and water supply for adjusting water content at approximately 60%. Ammonia gas was prepared by mixing NH3 standard gas (0·5% in N2) and air using an oil-less air compressor (Amadera Pneumatics Co., Ltd, Tokyo, Japan) and a gas mixture device (Kofloc GMP-1, Kyoto, Japan) and supplied to the reactor from the bottom via sterilized 0·2-μm filter (Millex FG50; Millipore, Billerica, MA, USA). Inlet gas diverged; one was connected to the reactor, and the other, to a 1·5 mol l−1 H2SO4 solution (150 ml) to collect total amounts of nitrogen as NH4+, which was loaded throughout the experimental period. The outlet gas of the reactor was also collected in the 1·5 mol l−1 H2SO4 solution (70 ml). Gas flow rates were measured and controlled using a variable area flow meter with a valve (RK1250 series; Kofloc) connected between the gas mixture device and either the H2SO4 trap or the reactor. Gas flow rates were 0·39 ± 0·03 (SD) and 0·40 ± 0·01 l min−1 for the H2SO4 trap and the reactor, respectively. The gas flow rate was also checked using a dry gas meter (DC-1; Shinagawa, Tokyo, Japan) at the start and end of the operation and at each sampling time. Circulated water was added intermittently with the timer controller for 1 min every 4 h at the flow rate of 8 ml min−1 in a counter-current mode of gas flow. Leachates and condensation were collected and returned automatically to a water container with valves and time controller. Sterilized distilled water was added to maintain the water volume (approximately 66 ml). The reactor was operated for 436 days in a constant room temperature (20°C).

Figure 1.

A schematic diagram of the reactor. 1, NH3 gas cylinder; 2, air compressor; 3, gas mixture device; 4, area flow meter; 5, sampling ports; 6, rockwool packing materials; 7, supporting materials (small stones); 8, circulation water reservoir; 9, distilled water; 10, condensation water; 11, solenoid valves; 12, pumps; 13, incubator; 14, refrigerator; 15, H2SO4 traps.

Sampling and chemical analyses

Ammonia concentration was measured with a detection tube and a gas sampler (Gastec Co., Ltd, Kanagawa, Japan). Inlet and outlet gases were collected in a 1-l Tedlar bag (GL Sciences Inc., Tokyo, Japan), and N2O was measured by a gas chromatograph equipped with a 63Ni-electron-capture detector (GC-14A; Shimadzu, Kyoto, Japan).

0·1 ml of the circulation water was periodically sampled for analysing NH4+-N, NO2-N, NO3-N, pH and electrical conductivity (EC). NH4+-N was measured by an ion chromatograph (DX-120; Dionex, Osaka, Japan), while NO2-N and NO3-N were measured by a different ion chromatograph (HIC-VP super; Shimadzu) (Waki et al. 2008), respectively. pH and EC were measured by a compact pH meter (Twin pH; Horiba Ltd, Kyoto, Japan) and EC meter (Twin Cond; Horiba Ltd), respectively. Total nitrogen was measured using an NC analyser (NC-220F; Sumika Chemical Analysis Service, Ltd, Osaka, Japan). The circulation water (10 ml) was also sampled at days 317 and 436 for molecular analysis.

The packing materials were sampled at approximately 1 cm below the surface on days 0, 32, 150, 317 and 436 for chemical and molecular analyses. During the sampling, NH3 gas supply was stopped. On day 436, the packing materials were sampled from two other depths besides the surface: the middle (4·5–5·5 cm depth) and bottom (9–10 cm depth). Because of the sampling of materials, the volume of the filter bed had decreased by 21% on a dry-weight basis on the last sampling day. The moisture contents, pH, amounts of NH4+-N, NO2-N and NO3-N in the packing materials were measured as described in the previous study (Yasuda et al. 2009). Nitrogen contents were measured based on the method described previously with some modification (Terada 2001). Briefly, an NC analyser was used after drying at 55°C for 48 h with 5% HCl to prevent NH3 volatilization and grinding with a mortar and pestle. Data are shown as an average of duplicate or triplicate analyses for the chemical analyses of the packing materials.

15N tracer experiments

Denitrification and anammox activities of the packing materials sampled at 150 and 317 days were measured by a 15N tracer technique according to the method previously reported (Yoshinaga et al. 2011). About 4 g of wet samples was suspended in the 10-fold diluted mineral medium to wash NH4+ and NO3 by centrifugation three times at 4400 g for 5 min. The washed samples were resuspended in the 200 ml medium and purged with Ar to eliminate dissolved oxygen. Portions (15 ml) of the suspension were transferred into 25-ml vial anaerobically, and headspace gas was displaced with He. Vials were preincubated overnight at 25°C with stirring at 160 rev min−1 in the dark. Subsequently, NH4+ and 15NO2 were supplied at the final concentrations of 1 mmol l−1 for the determination of denitrification activity, and 15NH4+ and NO2, for the determination of anammox activity, respectively. Vials were again incubated with stirring at 160 rev min−1 at 25°C. The N2 isotopologues (28N2, 29N2, 30N2) in the headspace gas in each vial were quantified using a GC–MS as described previously (Yoshinaga et al. 2011) for the determination of denitrification activity or GC–IRMS as described previously (Yasuda et al. 2011) for the determination of anammox activity.

DNA extraction, PCR and DGGE analysis

DNA was extracted from the total of 1·0–3·1 g (wet basis) of the rockwool packing materials by the methods described in the previous study (Yasuda et al. 2010). Circulation water was centrifuged at 3300 g for 20 min to collect suspended biofilms, and subsequently, the same protocols used for the packing materials were applied. The quality of the extracted DNA was checked using a UV spectrophotometer (UV-1650PC; Shimadzu), and the A 260/280 ratio was 1·6–1·8. Extracted DNA was quantified using Fluorescent DNA Quantification Kit (Bio-Rad, Hercules, CA, USA) according to the manufacturer's instruction.

Primers used for the amplifications of amoA were the same used in the previous study (Yasuda et al. 2010): amoA19F1-GC (CGC CCG CCG CGC CCC GCG CCC GTC CCG CCG CCC CCG CCC GAT GGT CTG GCT TAG ACG; modified from Leininger et al. 2006; underlined sequence denotes the GC clamp) and amo247Ry1 (CAA ACC ATG CGC CTT TTG CGA CCC A; modified from Treusch et al. 2005) for archaea, and amoA-1F-GC and amoAR1 (Avrahami et al. 2003) for bacteria, respectively. DGGE analysis of archaeal amoA was also conducted using primer set CrenamoA23f and CrenamoA616r (Tourna et al. 2008) to examine whether similar results of the changes in the band patterns among samples were obtained. Archaeal amoA gene amplification with amoA19F1-GC/amo247Ry1 was performed in a 25-μl reaction mixture containing 3–11 ng template DNA, 0·2 μmol l−1 of each primer, 2·5 mmol l−1 MgCl2, 0·2 mmol l−1 of each dNTP, 1 U AmpliTaq Gold (Applied Biosystems, Foster City, CA, USA), 10 μg bovine serum albumin and 2·5 μl of 10× PCR Gold Buffer. The PCR with CrenamoA23f/CrenamoA616r was performed in a 50-μl reaction mixture containing 3–11 ng template DNA, 0·5 μmol l−1 of each primer, 2·5 mmol l−1 MgCl2, 0·2 mmol l−1 of each dNTP, 1 U AmpliTaq Gold, 20 μg bovine serum albumin and 5 μl of 10× PCR Gold Buffer for primer set. Bacterial amoA gene amplification was performed in a 50-μl reaction mixture containing 6–22 ng template DNA, 0·5 μmol l−1 of each primer, 2·5 mmol l−1 MgCl2, 0·2 mmol l−1 of each dNTP, 1 U AmpliTaq Gold, 20 μg bovine serum albumin and 5 μl of 10× PCR Gold Buffer. Amplification was performed using an iCycler (Bio-Rad) with the following thermal profile: initial denaturation at 95°C for 10 min, then 35 cycles of denaturation at 95°C for 1 min, annealing at 55°C for 1 min, extension at 72°C for 40 s and final extension at 72°C for 10 min for archaeal amoA with amoA19F1-GC/amo247Ry;, initial denaturation at 95°C for 5 min, then 35 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s, extension at 72°C for 1 min and final extension at 72°C for 10 min for archaeal amoA with CrenamoA23f/CrenamoA616r; initial denaturation at 94°C for 5 min, then 40 cycles of denaturation at 94°C for 45 s, annealing at 57°C for 30 s, extension at 72°C for 1 min and final extension at 72°C for 7 min for bacterial amoA, respectively. The PCR products were checked by electrophoresis on 2% agarose gels.

Denaturing gradient gel electrophoresis was performed as described previously (Yasuda et al. 2010). In brief, PCR products (c. 80 ng for 32-well gel) were separated at 60°C on a 6% polyacrylamide gel along a denaturant gradient of 35–55% (100% denaturant contained 7 mol l−1 urea and 40% formamide) at 120 V for 8 h for archaeal amoA with amoA19F1-GC/amo247Ry1, denaturant gradient of 20–60% at 100 V for 15 h for archaeal amoA with CrenamoA23f/CrenamoA616r and denaturant gradient of 45–65% at 100 V for 17 h for bacterial amoA, respectively. DGGE bands were stained with SYBR Green I (Molecular Probes, Eugene, OR, USA) for 30 min and were visualized with a UV illuminator (Epi-Light UV FA500; Taitec, Saitama, Japan).

Phylogenetic and statistical analysis

The sequences of the specific DGGE bands with high intensity were determined as described previously (Yasuda et al. 2010). Multiple alignments and phylogenetic analyses were conducted using mega ver. 5.0 (Tamura et al. 2011). DGGE gel image was scanned and analysed using a Luminous Imager (Aisin Cosmos R&D Co., Ltd, Aichi, Japan), and intensities of the bands were calculated. The intensities of the each band were transformed to Pi, where Pi is the importance probability of a band in a gel lane (Eichner et al. 1999). Pi was calculated as described previously (Hanajima et al. 2009). To express the diversity of microbial community structures, the Shannon index of general diversity (H) was calculated based on the band importance probability data as = −Σ Pi log Pi (Eichner et al. 1999). Principal component analysis was performed using the princomp procedure of Sas (SAS 2008). Correlations between environmental parameters and the microbial community structure were analysed via Spearman's rank correlation coefficient. The amoA gene sequences determined in this study have been deposited in the DDBJ database, under the following accession numbers: DGGE bands excised from the gel of archaeal amoA, AB702707AB702714, and DGGE bands excised from the gel of bacterial amoA, AB702697AB702706, respectively.

Real-time PCR

The primer pairs used for the amplification of the amoA gene were amoA19F (Leininger et al. 2006) and amo247Ry1 for archaea and amoA-1F and amoAR1 for bacteria, respectively. The reaction was performed in a 10-μl reaction mixture containing 1·2–4·4 ng template DNA, 0·5 μmol l−1 of each primer, 5 μl of 2× mix (SsoFast EvaGreen; Bio-Rad) and 2 μg bovine serum albumin (20 mg ml−1), for archaea; and 20-μl reaction mixture containing 1·2–4·4 ng template DNA, 0·5 μmol l−1 of each primer, 10 μl of 2× mix and 4 μg bovine serum albumin, for bacteria. The PCRs were performed using MyiQ2 real-time PCR systems (Bio-Rad) with the following cycling conditions: enzyme activation at 98°C for 2 min, 45 cycles of denaturation at 98°C for 5 s and then annealing/extension at 55°C for archaea or 57°C for bacteria for 10 s, respectively. Melt curve analysis was performed with 95°C for 1 min and 50°C for 1 min, from 55 to 95°C with a reading made every 0·5°C, and the samples held for 10 s per step. Reaction measurements were replicated at least twice. A standard curve was prepared using serial dilutions of a known copy number of the plasmid pGEM-T Easy vector (Promega, Madison, WI, USA) containing the amoA gene of clones RW-1 (accession number AB525377; 6·68 × 100–6·68 × 104copies) for archaea and Nit. europaea (NBRC14298; 3·13 × 102–3·13 × 106 copies) for bacteria, respectively, with r2 > 0·98. The amplification efficiencies were in the ranges of 70–97%.


Reactor performance and characteristics of circulation water

Ammonia was effectively removed by the reactor, and average NH3 removal efficiency was 99·5% (Fig. 2a). Concentrations of NH3 in the outlet gas increased up to 8 ppm at day 158. Changes in the concentrations of NH4+-N and NO3-N, and in the pH values in circulation water are shown in Fig. 2b. NH4+-N and NO3-N were gradually accumulated in the water and finally increased to 9567 and 15 792 mg l−1, respectively. The average ratio of the concentrations of NO3-N to those of NH4+-N was 2·5 ± 0·8 (SD) throughout the operational period. The water pH was 7·8 on day 0 and was maintained near neutral values and then gradually decreased to 5·1 from day 188. NO2-N was accumulated with peak height on day 9 of 271 mg l−1 and gradually decreased with the second peak at day 182 of 131 mg l−1 (Fig. 2c). Free NH3 (FA) and free HNO2 (FNA) in the circulation water were calculated as described previously (Anthonisen et al. 1976). FA increased up to 27·6 mg l−1 on day 168, when the removal efficiency dropped and secondary accumulation of NO2-N had occurred (Fig. 2a,c). After day 189, FA concentration did not exceed 10 mg l−1, which was the lowest value of the ranges where inhibition of NH3 oxidation starts according to Anthonisen et al. (1976). From day 186 and day 253, the FNA concentrations increased more than 0·2 mg l−1, the lowest value that the inhibition of nitrification initiated. After that period, the FNA concentration was kept zero except on day 322 (Fig. 2c). EC values in the circulation water gradually increased up to about 11 S m−1. There was a correlation between EC values and either NO3-N or NH4+-N (R2 = 0·97, 0·87, respectively, < 0·01, n = 64). Water contents of the packing materials were kept at 59 ± 1·9% (SD) during the operational period.

Figure 2.

(a) Ammonia removal performance of the reactor. Concentrations of NH3 in the inlet (●) and the outlet (○) of the reactor, and removal efficiencies (+); (b) Time courses of concentrations of NH4+-N (▲), NO3-N (▵) and pH (×) in the circulation water; and (c) Time courses of concentrations of NO2-N (dotted line), the calculated free ammonia (FA) (grey circles) and free HNO2 (FNA) (thick line) in the circulation water.

Nitrogen mass balance of the reactor and 15N tracer experiment

Nitrogen mass balance of the reactor is shown in the Table 1. 14·9 g of nitrogen was introduced to the reactor throughout the operational period. Large portions of nitrogen were accumulated in the packing materials as NH4+-N and NO3-N. Residual nitrogen, which was estimated by the subtraction of the sum of inorganic nitrogen from the total nitrogen, gradually decreased after day 32 in the packing materials. The ratios of NO3-N to NH4-N were higher in circulation water than in the packing materials: 2·5, 2·4 and 1·6 for water and 0·8, 1·6, 1·1 and 1·4 for the packing materials, on days 32, 150, 317 and 436, respectively. Percentage of N2O in the outlet gas was 0·4–0·6% of the inlet nitrogen (the sum of inflow NH3-N and the initial amounts of the total nitrogen on day 0).

Table 1. Nitrogen mass balance and pH of reactor
 Day 0Day 32Day 150Day 317Day 436
  1. nd, not detected.

  2. a

    Sum of inflow NH3-N.

  3. b

    Circulation water in tube and reservoir only. Water in packing materials is included in packing materials category.

  4. c

    Sum of released NH3-N when valve opened for collecting water remaining in tubes and withdrawn nitrogen as water and packing material samples.

  5. d

    Unrecovered nitrogen expressed as a percentage of inlet nitrogen (sum of inflow NH3-N and the initial amounts of the total nitrogen at day 0).

Ammonia loading rate (g NH3 m−3 day−1)4952555963
In (NH3-N; g)a0·001·245·1110·8814·89
Out (NH3-N; mg)0·005·9910·7737·1043·89
Packing materials (PM) total; g1·072·213·795·627·96
PM NH4+-N; g0·00250·531·092·363·37
PM NO2-N; mg0·5214·090·540·360·34
PM NO3-N; g0·0180·451·742·674·59
Circulation water (water) total; gb0·00190·120·340·991·64
Water NH4+-N; g0·000530·0330·0980·280·62
Water NO2-N; mg0·596·981·36ndnd
Water NO3-N; g0·000820·0820·240·701·02
N2O-N; mg0·0012·1135·0495·15121·1
Sampling loss; gc0·000·00150·170·541·45
Unknown (%)d−1·829·639·129·7
Packing materials pH (KCl)6·17·06·44·44·6
Circulation water pH7·87·57·44·85·1

Approximately 29·6–39·1% of the inlet nitrogen was unrecovered after day 150 (Table 1). Therefore, batchwise 15N tracer experiments were conducted to examine for the presence of anammox and denitrification activity in the packing materials. The amounts of 30N2 in the headspace gas of the vials increased linearly (R2 = 0·91–0·99) for 9–11 h of anaerobic incubations after the substrate addition (NH4+ and 15NO3) (Fig. S1). On the contrary, 29N2 did not increase after approximately 1 month of anaerobic incubation with 15NH4+ and NO2 (Fig. S2). The 29N2 increased parallel to the increase in 30N2 when 15N-labelled NO3 was used as a substrate, because the suspension contained nonlabelled NOx as well. Denitrification activities were determined as the production rate of N2, which was estimated from the net production of 29N2 and 30N2. Denitrification activities of the packing materials sampled on days 150 and 317 were similarly estimated to be 0·24 and 0·22 mg N2 (g dry samples)−1 day−1, respectively.

Analysis of archaeal and bacterial amoA diversity and abundance

Denaturing gradient gel electrophoresis band profiles of PCR products targeting the archaeal and bacterial amoA gene are shown in Fig. 3a (for amoA19F1-GC/amo247Ry1) and 3b, respectively. The DGGE profiles clearly showed the changes in banding patterns along with the reactor operation for both amoA genes. For archaeal amoA gene, the intensity of band RA2 became weaker after 150 days. On the other hand, band RA1 increased in relative intensity especially after day 317 and appeared also in circulation water. Despite these changes in band intensity and appearance, approximately 60% of the bands (seven of 12 bands) commonly appeared in every sample of packing materials. On the contrary, only two bands appeared in the circulation water of a total nine bands on day 317 and two out of 10 bands on day 436, respectively. In contrast, the band patterns of bacterial amoA gene changed from day 0 to day 150, but they were relatively constant after that. Bands RB2 and RB4, which appeared with high intensity on day 0, disappeared on day 32. There were no common bands among the packing material samples. Unlike the archaeal amoA profiles, many bands commonly appeared between the packing materials and in the circulation water: seven of 13 bands on day 317 and 10 of 15 bands on day 436, respectively. These changes in band patterns were also shown in the results of the principle component analysis (PCA) of DGGE bands (Fig. 4a,b). For archaeal amoA, days 0 and 32 were plotted closely, and the band patterns changed greatly between days 32 and 150. The patterns of circulation water were plotted separately from those of the packing materials. For bacterial amoA, the band patterns on days 0, 32 and after 150 were all different from each other. For both microbial communities, stratified differences in the community structure were relatively small on day 436. The DGGE profile of archaeal amoA gene amplified with the different primer set CrenamoA23f/CrenamoA616r, and its PCA also exhibited the above-mentioned profile changes among samples obtained with amoA19F1-GC/amo247Ry1 (Fig. S3).

Figure 3.

Denaturing gradient gel electrophoresis patterns of archaeal amoA amplified with amoA19F1-GC/amo247Ry1 (a) and bacterial amoA (b). Numbers above the band profiles indicate the sampling days: from the packing materials of the reactor on days 0, 32, 150, 317, 436; and from the circulation water on days 317 and 436. Letters above the gel image, u, m and l represent the upper (u), middle (m) and lower (l) depth, respectively, of the packing materials sampled on day 436. Arrows indicate sequenced band names except for Band RA9*.

Figure 4.

Principal component analysis of denaturing gradient gel electrophoresis bands of archaeal (a) and bacterial (b) amoA fragments from the packing materials (○) and circulation water (●). Numbers beside the symbols indicate the sampling days. Letters, u, m and l represent the upper (u), middle (m) and lower (l) depth, respectively, of the packing materials sampled on day 436.

The specific DGGE bands shown in Fig. 3a,b were sequenced, and phylogenetic relationships are shown in Fig. 5a,b. Obtained DGGE band sequences were all affiliated with those of the DGGE bands of the rockwool packing materials sampled from the full-scale biofilter for both archaeal and bacterial amoA genes (Yasuda et al. 2010). The packing materials from the same full-scale biofilter were used as seed culture in this study. Apart from the biofilter clones, the nucleotide sequences of archaeal amoA had 89–100% similarity with environmental clones from soils, deep-sea sediments, aquaria and wastewater treatment plant, and those of bacterial amoA had 96–99% similarity with soil clones. For archaeal amoA, from bands RA3 to RA8 were placed within the Group I.1b. Band RA1 was located in a different cluster from the above bands, but also were categorized within the Group I.1b. Band RA2 was affiliated with clone RW-2 and fell into Group I.1a, which decreased in relative intensity after day 150. The sequence of band RA2 had 100% similarity with sequences from wastewater treatment plants, especially biofilm of biological aerated filter. Band RA9* appeared in the packing materials throughout the operation. The position of Band RA9* in a DGGE gel was identical to that of Clone RW-3, which had 100% nucleotide similarity with the enrichment culture of ammonia-oxidizing archaeon, Ca. Nitrososphaera sp. JG1 from agricultural soil (Kim et al. 2012). For bacterial amoA, the specific bands that appeared with high intensities after 150 days of operation were affiliated with the Nitrosospira multiformis cluster within the Nitrosospira sp. cluster 3 (RB6, 7, 9, 10) and Nit. europaea lineage (RB1). In contrast, bands RB2 and RB4, which were specific to day 0 sample, were placed in a different subcluster within Nitrosospira sp. cluster 3. Bands RB3, RB4 and RB8 all appeared in the circulation water of 317 and 436 days. Those bands were categorized within the same cluster with Nitrosospira briensis in the Nitrosopira sp. cluster 3.

Figure 5.

Neighbour-joining phylogenetic trees of archaeal (a) and bacterial (b) amoA gene sequences (207 and 306-bp fragment, respectively) retrieved from denaturing gradient gel electrophoresis bands shown in Fig. 3. Scale bar indicates 2 per 100 nucleotide positions. Bootstrap values (%) obtained with 1000 resamplings are shown at branch points (when more than 70%). The tree is outgrouped with Ca. Nitrosocaldus yellowstonii (Z97861) (a) and Nitrosococcus oceani (CP000127) (b), respectively. Sequences obtained from the full-scale biofilter with rockwool packing materials are underlined.

Changes in the abundance of amoA gene copy numbers in the packing materials during the reactor operation are shown in Fig. 6. The amoA gene copy numbers increased on day 32 for both archaea and bacteria and then decreased afterwards. The abundance of bacterial amoA gene decreased on day 317 but increased again on day 436. The amoA gene copy of bacteria outnumbered that of archaea except on day 317. The abundance of amoA gene copy numbers in the circulation water was also determined on days 317 and 436. The copy numbers per mL of water were 5·2 × 105 and 1·1 × 106 on days 317 and 436 for bacterial amoA, and 7·4 × 103 and 1·3 × 104 on days 317 and 436 for archaeal amoA, respectively. The ratios of the amoA gene copy numbers of the circulation water to those of the packing materials were higher for bacteria than archaea on both day 317 and day 436: 0·315 and 0·028 for bacteria on days 317 and 436, and 0·004 and 0·007 for archaea on days 317 and 436, respectively.

Figure 6.

Changes in abundance of amoA gene copy numbers of archaea (open columns) and bacteria (closed columns) in packing materials and circulation water during reactor operation. Upper sample was analysed for day 436.


amoA-encoding archaeal and bacterial community shifts under nitrogen accumulation process in the ammonia deodorization biofilter

The increases in the amounts of NO3 indicated that nitrification continuously occurred throughout the reactor operation. NH3 removal efficiency did not decrease in parallel with a slight increase in ammonia loading rate over the whole reactor operation. Concentrations of NH4+-N and NO3-N in circulation water gradually increased up to as much as 9567 mg l−1 (683 mmol l−1) and 15 792 mg l−1 (1128 mmol l−1), on day 436, respectively. The NH4+-N in the packing materials was also highly accumulated up to 17·36 g kg DM−1, but eventual failure of NH3 removal did not occur as in a previous study (Smet et al. 2000). Such high NH4+-N accumulation is not common in natural environments where nitrification is taking place. In anthropogenic wastewater treatment plants or animal waste management systems, concentrations of NH4+-N are still lower than those found in the present study, for example, 1·2–3·2 mmol l−1 in activated sludge bioreactors (Park et al. 2006), 27–360 mmol l−1 in animal wastewater treatment plants (Otawa et al. 2006) and 0·2–10·3 g kg−1 in livestock manure and composts (Yamamoto et al. 2012). Even though NH4+ concentrations in the surrounding environments were low, as much as 1 mol l−1 of NH4+ could be accumulated by Nitrosomonas cells (Schmidt et al. 2004). High concentrations of NH4+ itself reportedly did not inhibit nitrification (Mahne et al. 1996). Therefore, continuous NH3 removal could be achieved during the nitrogen accumulation process.

The community structure and abundances of AEA and AOB changed along with the reactor operation, although NH3 was effectively removed and continuous nitrification occurred. Microbial community oscillation is often observed when stable nitrification occurs (Egli et al. 2003; Layton et al. 2005; Wells et al. 2009). On the other hand, the ammonia removal efficiency dropped when FA increased (Fig. 2). Dynamic changes in community structures of AOB and AEA were observed along with this event. Community shift of the specific bacteria could be related to the operational failure (Wittebolle et al. 2005). Drastic microbial community structural change might lead to an operational failure, which should be repaired through reactor operation.

The above-mentioned community shifts happened in a different manner between AEA and AOB. There were many common bands for archaeal amoA DGGE profile, whereas the bacterial community changed drastically in the early stage of the operation, and no band commonly appeared among the packing material samples. The amoA gene abundance also fluctuated more for bacteria than for archaea. The copy numbers of archaeal amoA (1·5 × 106–5·2 × 107 g−1) were similar to those of particular soils and animal manure composts (106–107 g−1) (Nicol et al. 2008; Jia and Conrad 2009; Schauss et al. 2009; Yamamoto et al. 2012). Archaeal amoA genes were detected more abundantly in some agricultural waste compost, 2·8 × 108–1·1 × 109 g−1 (Zeng et al. 2011a). Bacterial amoA copy numbers in the reactor (1·7 × 106–1·6 × 109 g−1) were similar to those detected from the composts (Maeda et al. 2010; Zeng et al. 2011a; Yamamoto et al. 2012) and relatively higher than the reported values for soils, 4·2 × 103–5·2 × 107 g−1 (Leininger et al. 2006; Jia and Conrad 2009). Higher fluctuation of the amoA abundance of bacteria than that of archaea during agricultural waste composting was also reported previously (Zeng et al. 2011a). Bacterial amoA genes decreased in numbers from days 32 through 317. Both FA increase until day 168 and pH decrease especially between sampling day 150 and 317 were the possible reasons for this phenomenon because the previous studies showed that AOB activity was inhibited by high FA concentration (Anthonisen et al. 1976; Kim et al. 2006), and its abundance decreased due to the decrease in pH (Yao et al. 2011). On the other hand, AEA decreased only from day 32 to 150. The effect of pH on the abundance of ammonia-oxidizing archaea (AOA) was different from AOB, and AOA did not decrease in size at acidic condition (Nicol et al. 2008). This might have caused the slightly greater number of archaea than bacteria on day 317. Moreover, AEA seems to be more associated with the packing materials than the circulation water. According to the previous study, affinity to the dissolved oxygen is higher for AOA than AOB (Hatzenpichler 2012, Kim et al. 2012). AOA may be present in the inner side of the biofilm.

Dominant bacterial amoA sequences were associated with both Nitrosospira sp. cluster 3 and Nit. europaea lineage. Similar N. multiformis-like and Nit. europaea lineage amoA were detected in ammonia biofilter with perlite, and Nitrosospira-like amoA became dominant during the reactor operation (Sakano and Kerkhof 1998). Compost is also a source of both Nitrosospira-like amoA and Nit. europaea lineage species, although the latter seemed to be more dominant (Yamamoto et al. 2012). On the other hand, Nitrosomonas oligotropha lineage in addition to the N. europaea lineage was often detected in wastewater treatment plants (Otawa et al. 2006; Limpiyakorn et al. 2007; Wells et al. 2009). Concentration of NO2-N could be the selective factor among Nitrosomonas sp. (Limpiyakorn et al. 2007). The Nit. oligotropha group was possibly absent due to the temporal appearance of NO2-N more than 30 mg l−1 in the ammonia biofilter.

Responses of Nitrosospira sp. and Nitrosomonas sp. to the environmental variables such as dissolved oxygen, nitrite and metals were reportedly different from each other (Wells et al. 2009). The existence of both Nitrosospira sp. and Nitrosomonas sp. may guarantee the functional redundancy with the variation of environmental parameters in ammonia biofilter. Furthermore, dominant sequences changed within Nitrosospira sp. cluster 3 along with the reactor operation as described above. Maximum ammonium oxidization activities per cell were different even within the Nitrosospira sp. cluster 3, more than five times higher for N. multiformis than N. briensis (Belser 1979). Substrate affinity could be lower for N. multiformis. This might be one of the reasons why major bands of the packing materials after day 150 of operation were associated with the N. multiformis cluster within Nitrosospira sp. cluster 3.

As for AEA, sequences belonging to both Thaumarchaeal Groups I.1b and I.1a were detected in day 0 sample. The Group I.1b became dominant along with the operation. Three of five wastewater treatment plants harboured both types of AOA and strains within the Group I.1b seemed to be widely distributed in the wastewater treatment plant according to the previous results (Park et al. 2006). Ca. Nitrososphaera gargensis-like sequences within Group I.1b were also detected as a dominant group in cattle manure compost (Yamamoto et al. 2012). The affinity for ammonia of Ca. Nitrososphaera sp. JG1 within the Group I.1b was lower than that of strains in the Group I.1a according to Kim et al. 2012; although Ca. Nitrososphaera gargensis and Ca. Nitrosotalea devanaterra have been previously reported to prefer a low NH3 concentration (Hatzenpichler 2012). Likewise with AOB, it is possible that phylogenetic shifts among AEA species were related to the substrate affinity.

According to a previous study (Prinčič et al. 1998), the AOB community structure in nitrifying culture of a wastewater treatment reactor drastically changed between 1000 and 3000 mg NH4+-N/L. pH is another important factor, which regulates microbial community change. The different pH caused different community structures of both AOA and AOB in soils after long-term field management (Nicol et al. 2008). In the present study, however, increase in NH4+-N concentration accompanied the decrease in pH. It is difficult to explain how AEA and AOB communities shift over the whole operation as a function of the single effects of neither NH4+-N concentrations nor pH. Therefore, FA, which is a function of NH4+-N, pH and temperature, was introduced as another factor.

To explore the relationships between community structural changes and the environmental parameters, Spearman's rank correlation coefficients between Shannon index (H) or 1st and 2nd principle component (PC) scores and environmental parameters were calculated (Table 2). The H and PC scores were used as indicators of the microbial community structural diversity or similarities because the community diversity was represented by H and similarities among samples were shown by PCA plots. Spearman's rank correlation coefficients were used to assess correlations between microbial abundance and environmental parameters (Wells et al. 2009; Herrmann et al. 2011). The FA was expressed as the basis of amoA copy numbers as the specific FA, that is, biomass basis because inhibition is affected by the concentration of nitrifying micro-organisms as described previously (Villaverde et al. 2000).

Table 2. Spearman's rank correlation coefficients between microbial community structure determined by Shannon index (H′) or principle component (PC) scores and environmental parameter
H 1st PC scores2nd PC scores H 1st PC scores2nd PC scores
  1. AEA, amoA-encoding archaea; AOB, ammonia-oxidizing bacteria; FA, free ammonia.

  2. a

    < 0·05, n = 5.

Specific FA (mg l−1 [AEA amoA gene copy no.]−1)0·600·200·90a
Specific FA (mg l−1 [AOB amoA gene copy no.]−1)−0·90a−1·00a0·00
NH4+-N (mg l−1)0·30−0·400·70−0·10−0·500·00
NO2-N (mg l−1)0·210·82−0·150·210·360·67
NO3-N(mg l−1)0·30−0·400·70−0·10−0·500·00
N2O-N (ppm)0·300·500·000·500·300·70

For AEA, there was positive correlation between the specific FA and 2nd PC scores. The specific FA negatively correlated with the diversity index H and 1st PC scores for AOB. No significant correlation was found for other parameters. The results indicated that the AOB community structural diversity decreases when the specific FA increases, which is consistent with the result in the previous report that the high FA concentration was related to lower bacterial diversity in a nitrifying sequencing batch reactor (Zeng et al. 2011b). The statistical analysis also indicates that the specific FA could somehow affect AEA community structural change during the reactor operation. As mentioned above, the community structure changed drastically from day 32 to 150 for AEA and from day 0 to 32 and 32 to 150 for AOB, respectively (Fig. 3). This stage corresponded to the period when FA concentrations in circulation water were increasing, mainly due to the rise in NH4+-N concentration despite the initiation of the pH reduction. Weckhuysen et al. (1994) demonstrated that the NH3 elimination efficiency is also affected by the FA concentration in the biofilter. It could be hypothesized, therefore, that the specific FA concentration is another important factor affecting the community shifts of AOB and AEA in the present biofilter in addition to the NH4+-N concentration and pH. A proper balance between NH4+ concentration and pH is important for achieving long-term NH3 removal.

Nitrogen balance

Unknown nitrogen accounted for 29·6–39·1% of the total nitrogen (the sum of inflow NH3-N and the initial amounts of the total nitrogen in the packing materials) on days 150, 317 and 436. This unknown nitrogen was considered to be mainly denitrified and released into the air as N2, because denitrification activity of the rockwool packing materials was confirmed by 15N tracer experiment. Denitrification could occur using indigenous organic matter and cell debris of nitrifying microbes, because the amounts of residual nitrogen gradually decreased during the operation (Table 1). Zhu et al. (2001) demonstrated that oxygen penetration in the biofilm in water was more restricted than that in air. It could be hypothesized that the aerated biofilter has less anaerobic space. The higher oxygen partial pressure might depress the denitrification activity in the biofilter. It is not clear why residual nitrogen increased, and there was no unknown nitrogen on day 32.

N2O emissions accounted for only a small part of the gaseous emissions. It was estimated that much larger percentages of the removed NH3 could have been converted to N2O (Clemens and Cuhls 2003). Probably, N2O emissions were small in the present study because not much NO2 had accumulated in the reactor (Table 1).


The authors wish to thank Dr. E. Mikami for his valuable advice and Ms. N. Akasaka and Ms. K. Sumiya for their skilful assistance. This research was partly supported by the Ministry of Agriculture, Forestry and Fisheries of Japan through the research project study. We also thank two anonymous reviewers for their cooperative and valuable comments on this manuscript.