An Analysis of Operation Conditions and Microbial Characteristics in Swine Wastewater Treatment Plants with Spontaneously Enriched Anammox Bacteria

The spontaneous enrichment of anammox bacteria has been reported in swine wastewater treatment facilities. However, their causative conditions and microbial characteristics, which this study aims to explain, are poorly understood. We discovered eight treatment facilities where the collected red biofilms exhibited high anammox activity levels at 57–843 μmol-N2/g-ignition loss (IL)/h and anammox DNA concentrations of 4.3 × 108–1.6 × 1012 copies/g-IL. The facilities used various wastewater treatment methods—six of them employed a multi-stage continuous reactor, whereas aeration tanks were continuously aerated at another combination of six facilities. Levels of dissolved oxygen (DO) in these tanks were fairly low at ≤1 mg/L. Pyrosequencing of the biofilms indicated the presence of 3–62.5% Planctomycetes, and the dominant anammox in each biofilm comprised three operational taxonomic units (OTUs) similar to Candidatus Jettenia asiatica, Ca. Brocadia fulgida, and Ca. B. caroliniensis. This suggested that some particular species of anammox bacteria naturally thrive when operating a swine wastewater treatment facility at low DO levels. The frequent enrichment of anammox biofilms at the sampled sites indicated that these treatment facilities were good seed sources of anammox; therefore, anammox treatment would be a viable method for the removal of nitrogen from swine wastewater.


Introduction
Nitrogen (N) pollution in water systems due to livestock excrement has become a serious problem owing to the worldwide increase in livestock populations [1]. In Japan, pig farming, in particular, generates a large amount of wastewater containing nitrogen; therefore, an appropriate treatment method is required. The activated-sludge process is the most widely used method for wastewater treatment in swine farms in Japan. Untreated swine wastewater is a mixture of urine, feces, and service water; this is first pre-treated by removing suspended solids using screens, gravity sedimentation, or mechanical separation with chemicals, and then treated with the activated-sludge process. Various activatedsludge processes, such as a sequencing batch reactor and a single or multi-stage continuous reactor with or without a membrane, are employed. In most cases, the temperature and aeration conditions in the process are not controlled; hence, they fluctuate with time and are highly variable. Moreover, owing to the low ratio of biochemical oxygen demand (BOD) to N in the influent, N is inadequately removed in the activated-sludge process using traditional nitrogen removal processes, such as denitrification following nitrification, resulting in effluent containing an average total nitrogen (TN) concentration of approximately 430 mg/L [2]. Namely, nitrogen removal by the conventional process in the swine wastewater treatment process is limited; therefore, using anaerobic ammonium oxidation (anammox) technology for nitrogen removal is desirable.
Anammox is a bacterial process in which NH 4 + is oxidized to N 2 with NO 2 − as an electron acceptor under strictly anaerobic conditions [3,4]. Compared to the conventional method-nitrification followed by denitrification-this process is a low-cost method to remove nitrogen from wastewater as anammox does not require BOD to metabolize N and, through partial nitrification, reduces total aeration costs. Anammox organisms were discovered in 1995 in the wastewater treatment plant (WWTP) of a baker's yeast production facility [3]. Anammox exists naturally in many ecosystems (including seas, rivers, lakes, paddy fields, and wetlands [5,6]) and also in conventional WWTPs [7], though not dominantly. Owing to glacial growth rates with a doubling time longer than a week [8,9], these bacteria take a long time to reach high concentrations. Anammox sludge enrichment typically requires several months in a pilot plant. Therefore, during the start-up of a full-scale anammox reactor, maintaining the supply of anammox seed sludge is difficult [10].
However, spontaneous anammox bacteria enrichment in biofilms was recently reported at full-scale swine-wastewater activated-sludge plants at three farms in Japan [11] and two in China [12,13]. These biofilms were red in color, as is often observed in anammox sludge enriched in artificial wastewater; they also had high activity levels (up to 295 µmol/g-IL/h) and 16S rRNA gene copy numbers (up to 1.35 × 10 12 copies/g-IL) [11] for anammox. The naturally enriched anammox sludge in swine farms can be used to improve nitrogen removal in the facility by customizing the operation to increase suitability. It can also be used as a seed sludge to introduce the anammox process into other farms, though this requires additional biosafety considerations. Those naturally enriched in the swine wastewater treatment process are expected to adapt more readily to other instances of the same process and hence, are considered suitable for seeding in other swine WWTPs. The naturally enriched anammox can potentially contribute to solve the treatment problems of nitrogen-containing wastewater.
These facilities include a sequencing batch reactor, a single-stage continuous reactor, a multi-stage continuous aeration tank reactor [11], and combinations of continuous aerobic and anoxic tanks [12,13]. Various processes are considered acceptable in the treatment of swine wastewater for the natural enrichment of anammox; however, further research is required to accurately determine suitable processes from among them. Detailed information on the process facilities that promote natural enrichment and the microbial characteristics of the biofilms is required to effectively utilize anammox cultures in swine wastewater treatment. Therefore, the survey of swine wastewater treatment facilities for spontaneously enriched anammox was expanded to an additional five farms where anammox biofilms were found. In this study, the operating conditions and microbial characteristics of eight swine wastewater treatment facilities were investigated in detail.

Red Biofilms in Full-Scale Swine Wastewater Treatment Plants
We investigated the anammox biofilms from eight swine farms in Ibaraki, Saitama, Shizuoka, and Miyazaki prefectures in Japan (Table 1, Figure 1). The Sai-1, Shizu-2, Shizu-3, Miya-1, and Miya-2 farms were newly discovered to have anammox biofilms during this study, while the Iba-1, Iba-2, and Shizu-1 farms were the same facilities examined in a previous study [11]. Samples were collected from all the farms in the study, except for Shizu-1; the data for this farm were the same as that used in a previous study [14]. The pre-treatment methods and types of activated-sludge processes are shown in Table 1. In some facilities, the red biofilm was found only in a few aeration tanks; therefore, data regarding the number of affected aeration tanks were also recorded in this study. Moreover, Processes 2021, 9, 1010 3 of 12 biofilm was present on the walls of the activated-sludge aeration tanks, in the wastewater effluent from them, and in sedimentation tanks; hence, samples were also collected from these points. Only one sample collected at Shizu-1 was washed to remove the black sludge.

Anammox Activity Measurement
Samples of these biofilms were analyzed for anammox activity using the tracer incubation method [15,16]. The tested biofilms were then washed to remove any attached inorganic nitrogenous compounds, following which they were centrifuged and suspended in an inorganic medium. The centrifuged sludge was re-formulated to yield 100 mg/L volatile suspended sludge (VSS) or ignition loss (IL) in the medium. Ten milliliters of the suspended mixture was anaerobically batch-incubated in 25-mL glass bottles at 25 °C; inorganic nitrogen compounds were added to these mixtures to obtain 2.5 mmol/L of 15 NO2 -/ 14 NH4 + in each bottle. Anammox activity was then estimated based on 29 N2 generation in the bottles periodically measured during the incubation period for several hours.

Real-Time Quantitative Polymerase Chain Reaction (qPCR) Analysis of Anammox Bacteria
Total deoxyribonucleic acid (DNA) was extracted twice from each biofilm using FastDNA SPIN Kit for Soil (MP Biomedicals, Tokyo, Japan), bulked, and then purified with QIAEX II Gel Extraction Kit (Qiagen, Hilden, Germany). The abundance of anammox bacterial 16S rRNA genes was determined in triplicates by qPCR amplification using

Anammox Activity Measurement
Samples of these biofilms were analyzed for anammox activity using the tracer incubation method [15,16]. The tested biofilms were then washed to remove any attached inorganic nitrogenous compounds, following which they were centrifuged and suspended in an inorganic medium. The centrifuged sludge was re-formulated to yield 100 mg/L volatile suspended sludge (VSS) or ignition loss (IL) in the medium. Ten milliliters of the suspended mixture was anaerobically batch-incubated in 25-mL glass bottles at 25 • C; inorganic nitrogen compounds were added to these mixtures to obtain 2.5 mmol/L of 15 NO 2 -/ 14 NH 4 + in each bottle. Anammox activity was then estimated based on 29 N 2 generation in the bottles periodically measured during the incubation period for several hours.

Pyrosequencing Analysis
Pyrosequencing of the 16S rRNA genes was conducted as described by Ishimoto et al. (2020) [14]. These genes were amplified using the forward primer 563F and four mixed reverse primers, R1-4 [18]; the resulting amplicons were purified and paired-end sequenced on an Illumina MiSeq platform (Illumina, San Diego, CA, USA) according to standard protocols. Quality-filtered paired-end reads were then deposited in the DDBJ database under BioProject accession numbers PRJDB8118 and PRJDB8910 and BioSample accession numbers SAMD00165691, SAMD00165692, SAMD00165697-SAMD00165700, SAMD00191392, and SAMD00191393.

Analytical Methods
Analyses of wastewater characteristics were conducted according to standard methods at the NARO laboratory in Tsukuba (Japan). The measured parameters in liquid samples were pH, suspended solids (SS), VSS, IL, BOD, total nitrogen (TN), ammonium (NH 4 + ), total nitrite (NO 2 − ), nitrate (NO 3 − ), and DO as described in the previous study [11]. Electrical conductivity (EC) was measured using an EC meter (LAQUAtwin EC-33, Horiba, Kyoto, Japan). Furthermore, concentrations of N 2 isotopomers ( 28 N 2 , 29 N 2 , and 30 N 2 ) produced by tracer incubation were detected by a gas chromatograph-mass spectrometer under the same conditions as Suto et al. [11].

Anammox Activity and DNA Copy Numbers in the Biofilm
The eight biofilm samples showed anammox activity levels of 57-843 µmol-N 2 /g-IL/h, and anammox bacterial DNA copy numbers of 4.3 × 10 8 -1.6 × 10 12 copies/g-IL (Supplementary Table S1). The anammox activity and DNA copy numbers of the biofilm and SS samples in the aeration tanks from this study and a previous study [11] (Figure 2) showed a positive correlation. Barring a few cases, both values were higher in biofilms than in the SS samples. The highest copy number (1.6 × 10 12 copies/g-IL) in this study was as large as that of the anammox sludge enriched in artificial wastewater (6.23 × 10 12 copies/gvolatile SS) [19]. These results showed that red biofilms contained high concentrations of living anammox bacteria.

Facilities with Anammox Biofilms
Anammox biofilms were observed in various types of activated treatment processes, including a sequencing batch reactor, a single continuous reactor, and six multi-stage reactors (Table 1). Aeration was continuous in six reactors and intermittent in two, such that the minimum duration of a single cycle of intermittent aeration was 12 h. Membranes were used in two reactors. Pre-treatment methods included screening and mechanical separation with a flocculating agent in four facilities; the latter method causes greater declines in the SS concentration than the former owing to the flocculation process. The multistage continuous reactor and aeration tanks conspicuously occupied six facilities. In multi-stage reactors, sludge in the aeration tank moves progressively downstream; however, the biofilms tend to form on the walls of the first aeration tank.

Facilities with Anammox Biofilms
Anammox biofilms were observed in various types of activated treatment processes including a sequencing batch reactor, a single continuous reactor, and six multi-stage re actors (Table 1). Aeration was continuous in six reactors and intermittent in two, such tha the minimum duration of a single cycle of intermittent aeration was 12 h. Membrane were used in two reactors. Pre-treatment methods included screening and mechanica separation with a flocculating agent in four facilities; the latter method causes greater de clines in the SS concentration than the former owing to the flocculation process. The mu tistage continuous reactor and aeration tanks conspicuously occupied six facilities. I multi-stage reactors, sludge in the aeration tank moves progressively downstream; how ever, the biofilms tend to form on the walls of the first aeration tank.
As the quality of influent wastewater in livestock farms varies widely across and within the farms, the values of BOD, NH4 + -N concentration, and the BOD/N ratio wer found to vary from 400-5435 mg/L, 465-4931 mg/L, and 0.2-3.4 (Appendix Table A2), re spectively. The percentage removal of nitrogen showed wide variations, ranging from 48−99%. Despite BOD/N ratios as low as 0.2 or 0.9, some facilities exhibited adequate ni trogen removal percentages in the range of 93-98%. This can likely be attributed to th contribution of anammox in nitrogen removal (Iba-1, 2 November 2014).
In most swine farms, the DO levels in the aeration tanks were not regulated wit sensors and, therefore, fluctuated. However, the concentrations observed in facilities wit anammox biofilms were significantly lower at values ≤1 mg/L (Appendix Table A3) tha those found in conventional facilities where DO concentrations ranged 0-5.8 mg/L [15].

Microbial Characteristics of the Biofilms
Pyrosequencing analysis of biofilms showed the phyla composition, which com prised 3-62.5% Planctomycetes, 11-38% Proteobacteria, and 4-31% Chloroflexi (Figure 3a). This was in agreement with findings from previous studies, which show th Proteobacteria in the biofilm to include microorganisms with heterotrophic denitrificatio abilities and the Chloroflexi to scavenge organic compounds derived from anammox bac terial cells [20,21].  Table S2), respectively. The percentage removal of nitrogen showed wide variations, ranging from 48−99%. Despite BOD/N ratios as low as 0.2 or 0.9, some facilities exhibited adequate nitrogen removal percentages in the range of 93-98%. This can likely be attributed to the contribution of anammox in nitrogen removal (Iba-1, 2 November 2014).
In most swine farms, the DO levels in the aeration tanks were not regulated with sensors and, therefore, fluctuated. However, the concentrations observed in facilities with anammox biofilms were significantly lower at values ≤1 mg/L (Supplementary Table S3) than those found in conventional facilities where DO concentrations ranged 0-5.8 mg/L [15].

Microbial Characteristics of the Biofilms
Pyrosequencing analysis of biofilms showed the phyla composition, which comprised 3-62.5% Planctomycetes, 11-38% Proteobacteria, and 4-31% Chloroflexi (Figure 3a). This was in agreement with findings from previous studies, which show the Proteobacteria in the biofilm to include microorganisms with heterotrophic denitrification abilities and the Chloroflexi to scavenge organic compounds derived from anammox bacterial cells [20,21].
Most species in the Planctomycetes phylum were identified as Ca. Jettenia or Ca. Brocadia (Figure 3b). Of the Planctomycetes, 99% were Ca. Jettenia in the sample from Shizu-1, and Ca. Brocadia in the samples from Iba-2 and Miya-2. Other samples contained Ca. Jettenia and Ca. Brocadia at relatively low percentages of 0.1-78% and 17-91%, respectively. The remaining Planctomycetes included 0-0.8% of Brocadiaceae, while other genera accounted for 2-48% of the total Planctomycetes. Ammonium-oxidizing bacteria and nitrite-oxidizing bacteria existed very slightly; the Nitrosomonadaceae family, which includes ammonium-oxidizing bacteria and the Nitrospira genus which are nitrite-oxidizing bacteria, were detected infrequently in the biofilm, accounting for <1% of the total reads. This agrees with the distribution of ammonium-oxidizing bacteria in floccular fraction than in granules in nitrification-anammox reactors [22].  Ammonium-oxidizing bacteria and nitrite-oxidizing bacteria existed very slightl the Nitrosomonadaceae family, which includes ammonium-oxidizing bacteria and the N trospira genus which are nitrite-oxidizing bacteria, were detected infrequently in the bi film, accounting for < 1% of the total reads. This agrees with the distribution of amm nium-oxidizing bacteria in floccular fraction than in granules in nitrification-anamm reactors [22]. Figure 4 shows the phylogenetic tree for the OTUs obtained from the biofilm sam ples, which were based on 97% sequence similarity and were affiliated with anammo Three out of eight displayed immense similarities with known anammox species; OTU OTU2, and OTU3 were close to Ca. Jettenia asiatica (DQ301513), Ca. Brocadia fulgid (DQ459989), and Ca. B. caroliniensis (JF487828) with 99.0%, 99.0%, and 100% similarit respectively. Notably, each biofilm sample included one of these three OTUs as the mo dominant microbe (Table 2)-OTU1 (Ca. J. asiatica) corresponded to samples from Shiz  Figure 4 shows the phylogenetic tree for the OTUs obtained from the biofilm samples, which were based on 97% sequence similarity and were affiliated with anammox. Three out of eight displayed immense similarities with known anammox species; OTU1, OTU2, and OTU3 were close to Ca. Jettenia asiatica (DQ301513), Ca. Brocadia fulgida (DQ459989), and Ca. B. caroliniensis (JF487828) with 99.0%, 99.0%, and 100% similarity, respectively. Notably, each biofilm sample included one of these three OTUs as the most dominant microbe (Table 2)-OTU1 (Ca. J. asiatica) corresponded to samples from Shizu-1, Shizu-2, and Miya-1; OTU2 (Ca. B. fulgida) to samples from Iba-1, Iba-2, and Sai-1; and OTU3 (Ca. B. caroliniensis) to samples from Miya-2. OTU3 was also relatively high in the sample from Miya-1. In Shizu-3, samples affiliated with anammox were lower among the total reads; however, its dominant microbe OTU1 possessed the highest read counts. sample from Miya-1. In Shizu-3, samples affiliated with anammox were lower among the total reads; however, its dominant microbe OTU1 possessed the highest read counts.

Discussion
In Japan, swine WWTPs employ various techniques at pre-treating and at aerationtreating that constitute aeration tanks, aeration time optimization, and sludge retention. In this study, naturally enriched anammox biofilms were found at every type of treatment facility generally used in swine farms (Table 1), indicating that spontaneous anammox

Discussion
In Japan, swine WWTPs employ various techniques at pre-treating and at aerationtreating that constitute aeration tanks, aeration time optimization, and sludge retention. In this study, naturally enriched anammox biofilms were found at every type of treatment facility generally used in swine farms (Table 1), indicating that spontaneous anammox enrichment is not restricted to a particular type of facility. However, among the eight facilities surveyed, six included multi-stage continuous reactors, while six performed continuous aeration. Therefore, these conditions were thought to be suitable for anammox enrichment, although they also were observed in reactors including anoxic tanks in China [12,13]. The multi-stage reactor allows for different conditions in each aeration tank; reducing the BOD and accumulating nitrites would facilitate conditions suitable for anammox enrichment to be maintained there. The sequencing batch reactor changes the conditions in the tank, whereas the continuous reactor, which conducts continuous aeration and maintains constant conditions, would be suitable for anammox growth.
During the activated-sludge treatment process, BOD was reduced, and nitrogen was removed inside the aeration tanks (Supplementary Table S2). Simultaneous partial nitrification, anammox, and denitrification (SNAD) were observed in them [23]. SNAD progresses at low DO levels; laboratory and full-scale SNAD studies have utilized swine waste, its anaerobic digestion liquor, and other wastewaters as landfill leachate and sludge digester liquor, with the DO concentration set at various levels, such as below 0.5 mg/L [24], 0.3 mg/L [25], 0.3-0.5 mg/L [26], <0.2 mg/L [12], and 0.19-0.54 mg/L [13]. A two-year study by Ishimoto et al. [14] demonstrated the importance of limiting DO levels below 0.3 mg/L among the DO fluctuations to improve nitrogen removal in the Shizu-1 facility. Similarly, in this study, the DO levels in the aeration tank were observed to be ≤1 mg/Lbarring one case, most of them were <0.5 mg/L. Thus, we could conclude that low DO levels in WWTPs promote the simultaneous enrichment of anammox in swine wastewater treatment facility.
The low DO levels in aeration tanks would be an indicator of the presence of naturally enriched anammox in swine WWTPs. The newest facility in this survey was constructed five years ago (Sai-1 in Table 1); however, the duration for which it has been operating at low DO levels is uncertain. Nitrogen removal by the SNAD process under low DO concentration is thought to occur due to the following reason. In most cases, ammonium oxidizing bacteria have higher oxygen affinity than nitrite-oxidizing bacteria; therefore, the low DO condition carries partial nitrification [27]. Anammox and denitrification prefer anoxic conditions; however, they would function at low DO concentration, especially in biofilm. Even if anammox enrichment does not occur in the facility, continuous aeration at low DO levels can effectively remove nitrogen from wastewater by shortcut nitrificationdenitrification pathway via nitrite [28]. This carries effective nitrogen removal at a low BOD/N ratio and supplies substrate as nitrite for anammox. Therefore, low DO levels are recommended for the treatment of swine wastewater to enable the effective removal of nitrogen by conventional nitrification-denitrification and SNAD; this would also be accompanied by anammox enrichment. Since anammox is ubiquitous in conventional WWTP, although in a very low concentration [7], it is expected that prolonged maintenance of operating tanks at low DO levels enables the spontaneous enrichment of anammox in swine farms. However, this requires further investigation for confirmation.
Moreover, in this study, anammox biofilm was observed on the walls of the aeration tanks sedimentation tanks and as floating granules peeled from there. In these facilities, the equipment used to hold biofilm has not been installed. Therefore, nitrogen removal would be improved by maintaining low DO concentration in the aeration tanks and keeping the biofilm in the aeration tank using some devices like hydrocyclone [29] or carrier materials [12,13].
Resulting from SNAD processes, Ca. Kuenenia stuttgartiensis, Ca. B. caroliniensis, and Ca. B. fulgida were reported as the dominant anammox species [30,31]. In this study, Ca. B. caroliniensis, Ca. B. fulgida, and Ca. J. asiatica predominated the biofilm (Figure 4, Table 2). This agrees with previous suggestions that the growth of Ca. Kuenenia stuttgartiensis could be inhibited by the chemical oxygen demand (COD) of the SNAD process [30], and that conditions were more conducive to the growth of Ca. Brocadia in the biofilm than that of Ca. Kuenenia [32]. Ca. Brocadia was observed in various anammox reactors, while Ca. Jettenia was not as prevalent [21]. One common feature among the three species observed in this study was their ability to utilize organic compounds. For example, Ca. B. caroliniensis was spontaneously enriched in a full-scale glycerol-fed digester liquid treatment plant; therefore, it is suggested to have the potential for glycerol utilization [33]. Ca. B. flugida showed enrichment in the presence of acetate and outperformed other anammox bacteria (Ca. B. anammoxidans and Ca. Kuenenia stuttgartiensis) in terms of oxidizing formate, acetate, and propionate compounds [34]. Ca. J. asiatica predominated in a granule-treated effluent from an anaerobic reactor with a low COD concentration [35], and uptake of acetate and propionate compounds occurred [36]. The presence of organic compounds (such as short-chain fatty acids) in specific concentrations could lead to the predomination of certain anammox bacterial species over others. Anammox is versatile not only as a means of metabolizing organic compounds but also in terms of its dissimilatory reduction of nitrates to ammonium [33], which is very useful when conducting swine wastewater treatment.
Possible factors impacting the distribution of these three species are discussed in this paragraph. Iba-1, Iba-2, and Sai-1, where Ca. B. fulgida was predominant, are located in eastern Japan and include two intermittent aeration processes (Table 1); Shizu-1-3, where Ca. J. asiatica was predominant, is located in central Japan and uses pre-treatment system without flocculating agents, while Miya-1 and Miya-2, with a relatively high proportion of Ca. B. caroliniensis, are located in southern Japan and use pre-treatment systems with flocculating agents. Therefore, aeration processes and pre-treatment methods that affect the BOD of the influent might be two of the determining factors. Since the influent wastewater pre-treated without using a flocculating agent tends to have higher BOD concentrations than that pre-treated with a flocculating agent, Ca. J. asiatica possibly has better chances of survival than Ca. Brocadia during BOD stress. This finding is consistent with the study by Chini et al. [37], which showed a shift from Ca. Brocadia to Ca. Jettenia with an increase in organic carbon concentration in the influent during the treatment of artificial wastewater or digestate from swine sludge. Additionally, biofilm samples from Shizu-1, where Ca. J. asiatica (OTU1) was observed to be the predominant species in this study, had Ca. Brocadia as the predominant microbe in October 2014 [11]. Therefore, the distribution of these three species might depend on the sampling site, operating condition, or temperature, and requires further investigation.

Conclusions
Spontaneously enriched anammox biofilms were found in conventional activatedsludge facilities at eight swine farms in Japan. Although various types of wastewater treatment methods had been employed in these facilities, continuous aeration in multistage continuous reactors at low DO levels under 0.5 mg/L were the most common. Three anammox species-Ca. J. asiatica, Ca. B. fulgida, and Ca. B. caroliniensis-proliferated in these biofilms. This study showed the occurrence of such biofilms to be increasing in frequency in swine WWTPs and the growth and predominance of the specific bacterial species in them. Swine WWTPs may be utilized as an optimal seed source of anammox bacteria, and anammox treatment would be an effective and inexpensive option for nitrogen removal from swine wastewater.