Comparison of the microbial communities in anaerobic digesters treating high alkalinity synthetic wastewater at atmospheric and high-pressure (11 bar)

A R T I C L E I N F O


Introduction
Biogas produced by conventional anaerobic digestion (AD) is primarily composed of methane (CH 4 , 50~70%) and carbon dioxide (CO 2 , 30~50%), and the ratio mainly depends on the substrate and pH of the fermentation process in the AD reactors (Wahid et al., 2019). The application possibilities of biogas are limited due to its low calorific value, which is caused by the low ratio of CH 4 /CO 2 (Lemmer et al., 2017;Lindeboom et al., 2011). Before injecting biogas into the natural gas grid or used in other high-grade applications (CH 4 > 90%, CO 2 < 8%), upgrading technologies are necessary (Omar et al., 2018;Wahid et al., 2019). External upgrading technologies, such as cryogenic, chemical absorption, membrane separation, pressure swing adsorption, water scrubbing, are applicable only for biogas flows higher than 100 m 3 /h Lindeboom et al., 2011). For small-scale digesters mounted in the so-called Decentralized Sanitation and Reuse (DeSaR) system, the upgrading equipment is neither available nor cost-effective Lindeboom et al., 2011). Thus, there is a demand for the development of new technologies that improve the biogas quality at a small scale.
Higher OLR becomes unstable.

Methanosaeta, Methanospirillum
J. Zhao, et al. Bioresource Technology 318 (2020) 124101 Lindeboom et al. (2011) proposed a novel AD concept: high-pressure anaerobic digestion (HPAD), which integrates biogas generation, upgrading and pressurisation in one step and hence it can condiserably reduce capital expenditures (Budzianowski and Postawa, 2017). Based on Henry's law, CO 2 dissolves better than CH 4 in the liquid phase when the pressure increases. Hence, the CH 4 content in the gas phase increases under pressure. The Henry's constant for CH 4 , CO 2 , H 2 S, and NH 3 are 0.0016, 0.0318, 0.115 and 62 mol/L/bar (0°C and 1 atm), respectively (Lindeboom et al., 2011). In the last decades several studies have been published investigating single stage HPAD reactors or two-stage presssurized reactors to upgrade biogas (Table 1). Specifically, the CH 4 content could reach more than 90% at pressures above 3 bar, but the pH drops at the same time, that could influence methanogens activity (Lindeboom et al., 2011(Lindeboom et al., , 2012. Moreover, the pressure itself may impose an influence on methanogenic activity as well. It was previously shown that some methanogens that were present in atmospheric AD could maintain their activity at pressures up to 90 bar (Lindeboom et al., 2011). These archaeal OTUs were closely related to Methanosaeta concilii, Methanosarcina acetivorans, Methanobacterium formicicum and Methanobacterium beijingense (Lindeboom et al., 2016). In another study, Methanosaeta (47%) and Methanospirillum (32.4%) dominated at 17 bar in a two-stage biofilm reactor, and the pressure was negatively correlated with the microbial diversity . In spite of these harsh conditions, these studies revealed that some methanogens managed to survive and served as the main methane contributor.
On the other hand, the decline of pH caused by the dissolution of CO 2 prevented stable operation of these high-pressure AD reactors (Table 1). Therefore, some alternative operations are suggested to alleviate the impact of CO 2 in HPAD. In previous studies, water scrubbing and silicate addition were used to remove CO 2 from the liquid phase, and could also further increase the CH 4 content (Lemmer et al., 2015b;Lindeboom et al., 2013a), but these approaches require extra investment. Intriguingly, in some studies using atmospheric AD treating a waste stream containing high alkalinity (high pH) some problems were encountered. Particularly, alkaline pretreament is commonly used to enhance biogas production from lignocellulosic biomass, but the hydrolysate with high alkaline (7.5% NaOH) content inhibited the subsequent methanogenesis stage in ambient AD (Zhu et al., 2010). These problems were also experienced when brewery wastewater/wastes (two thirds of the wastewater from the brewery is alkaline, pH 9~12) (Rao et al., 2007), or excess sludge (Li et al., 2018) was treated in ambient AD. It suggests that HPAD might be an alternative to treat these kind of alkaline waste streams because a suitable pH for operation can be maintained through the dissolved CO 2 . Under such circumstances, the microbial community within the HPAD remains poorly documented and requires more investigation.
Thus, in this study, a mesophilic HPAD single-stage bioreactor was operated at moderate high pressure, not exceeding 11 bar, and fed with high alkaline synthetic wastewater containing acetic acid, sodium acetate or glucose. The aims were to (1) evaluate the process stability to produce high calorific biogas in an HPAD system, (2) illustrate the microbial community using metagenomics genetic tools (16S-rDNA gene sequencing) and (3) analysis of the VFAs content in the reactors in relation to the total alkalinity. Hereto the results were compared with the methane yield and microbial diversity in a normal-pressure anaerobic digester (NPAD) reactor.

Reactor set-up
For the high-pressure experiment, an HPAD reactor (Suurmond, Switzerland) with a total volume of 2.0 L was used, and the liquid phase was manually controlled at 1.2~1.5 L. The reactor was designed for operation up to 300 bar in a temperature range of 0°C~350°C. This HPAD reactor was equipped with a pressure meter, an HPLC pump (Thermo, SpectraSystem P4000, Netherlands) for substrate feeding, a sample valve controlled by a computer, and a mechanical safety valve (Fig. 1A, B). The liquid and gas samples were taken using the same sample valve. When the sample tube was below the liquid phase, liquid samples were taken; when the sample tube was above the liquid phase, gas samples were collected using a gas bag.
For normal-pressure (atmospheric pressure) experiments, a glass NPAD reactor with a total volume of 2.0 L was used, and the volume of the liquid phase was the same as the HPAD reactor (1.2~1.5 L) (Fig. 1C). The temperature of the reactors was maintained at 37 ± 1°C with a stirring speed of 100 rpm.

Experimental start-up and operation
The reactors were inoculated with anaerobic sludge from a full-scale Notes: For the HPAD reactor, an HPLC pump was used for feeding, a sample valve used for liquid and gas samples collection, was controlled by a computer (When the tube was inside the liquid phase, liquid samples were taken; when the tube was above the liquid, gas samples were taken, which is a technical set up mentioned previously.) For the NPAD reactor, the water displacement method was used for the determination of the biogas production. One tube was used for taking liquid samples and the other one was used for feeding.
anaerobic digester treating aerobic sludge from a wastewater treatment plant (Garmerwolde, The Netherlands). The composition of synthetic wastewater was as follows: NH 4 Cl 2.0 g/L, K 2 HPO 4 ·3H 2 O 3.6 g/L, KH 2 PO 4 2.8 g/L, NaHCO 3 5.0 g/L, 1.0 mL/L of trace element solution, 1.0 mL/L of Wolfe's vitamin solution, and 1.0 mL/L cysteine-sulfide reducing agent dissolved in deionized water (Ronald, 2004). The components, salts and solvents were obtained from Sigma-Aldrich (St. Louis, MO, USA), Fisher scientific (Acros organics, USA) and Merck (KGaA, Germany). The synthetic wastewater was used as substrate with the same concentration (75 g/L) of acetic acid, sodium acetate, or glucose as carbon source during different phases, respectively ( Table 2).
The overall experiment was divided into seven phases (Table 2). Phase 1: cultivation phase, the sludge produced biogas and adapted to autogenerated pressure; phase 2-3: adaptation to autogenerated pressure conditions on acetic acid (pH of the synthetic wastewater was adjusted to 7 using 5 M NaOH and the total alkalinity was 18.6 g/L as CaCO 3 ); phase 4-6: pressure operation on sodium acetate/glucose (pH of the synthetic wastewater was adjusted to 7 using 37% HCl and the total alkalinity was 14.4 g/L as CaCO 3 ); phase 7: pressure operation on glucose.
Synthetic wastewater (10 mL) was daily added using an HPLC pump and a syringe for the HPAD and NPAD reactors, respectively. The effluent samples were taken at the end of each phase. For the HPAD reactor, the sample valve was controlled by a computer using a '10 s closed, 1 s open' procedure until the required sample volume was collected. From phase 2 on, when the pressure increased up to 10~11 bar, the samples were taken until the pressure decreased to 8 bar. The volume of the effluent samples was dependent on the pressure. For the NPAD reactor, the same sample volume was obtained using a syringe.

Gas production calculation
The volume of daily biogas production (DBP) of NPAD was measured by the water displacement method, and the volume of daily methane production (DMP) was calculated by DBP × CH 4 %.
For the HPAD reactor, it is hard to precisely calculated DBP and DMP, due to the lack of daily biogas composition data. Therefore, the estimated total CH 4 production in phase 2-5 was calculated using the total COD fed into the reactor during the corresponding phase (1 g COD yields 0.35 L CH 4 ; 80% COD is converted into biogas, the rest 20% is used for cell growth (Filer et al., 2019)). The estimated total CH 4 production in phase 6-7 was calculated based on CH 4 content (CH 4 production = pressure increased × headspace × CH 4 %).

Physical and chemical analyses
Total solids (TS) and volatile solids (VS) were determined by the standard methods (APHA, 2005). The pH was measured using a digital pH meter (H160, Hach, Germany). The total volatile fatty acids (VFAs) and total alkalinity (TA) were analyzed by titration with 0.1 N H 2 SO 4 to the endpoints of pH 5.0 and 4.4 with an auto-titrator (AT1000, Hach, Germany). The VFAs were analyzed with high-performance liquid chromatography (Agilent Technologies 1200 series) equipped with a Bio-Rad Aminex HPX-87H 300 × 7.8 mm column at 60°C using 5 mM H 2 SO 4 as eluent (0.5 mL/min) and detected using a UV-detector at 210 nm. The biogas composition was analyzed by gas chromatography (C2V-200 Micro GC, Thermo Scientific) with a GCC200-U-BND cartridge and a thermal conductivity detector. The temperature of the column, injector, and the detector was 60°C, 120°C, and 120°C, respectively. Helium was used as carrier gas.

High-throughput 16S-rDNA gene sequencing and analysis
The samples (2 mL) were thawed at room temperature and total DNA was extracted using the FastDNA® Spin Kit for Soil (MP Biomedicals, USA) according to the manufacturer's protocol. The extracted genomic DNA was used as the template in the PCR reactions. Part of the 16S-rDNA genes was amplified using the primers 27F (5′-AGR GTT TGA TCM TGG CTC AG-3′) and 1492R (5′-GGG TTA CCT TGT TAC GAC TT-3′ for Bacteria, and multiplexed with Arch21F (5′-TTC CGG TTG ATC CYG CCG GA-3′) for Archaea. The amplified DNA was "-" means below the detection limit of the HPLC method. Propionic and butyric acid in the HPAD reactor were below the detection limit of the HPLC method. a In the HPAD reactor, the estimated total CH 4 production in phase 2 to 5 was calculated using the total COD fed into the reactor during that phase (1 g COD yields 0.35 L CH 4 ; 80% COD is converted into biogas, the other 20% is used for cell growth (Filer et al. 2019)). The estimated total CH 4 production in phase 6 and 7 was calculated by CH 4 content (CH 4 production = pressure increased × headspace × CH 4 %). b The methane content of the biogas in the HPAD reactor was measured at the end of phase 6 and 7 due to the technology set up. c In the NPAD reactor, the actual total CH 4 production was calculated using the CH 4 content in the biogas. "ND" means not determined, because of technology set up. The HPAD reactor has one combined outlet for liquid and gas. When the tube was inside the liquid phase, liquid samples were taken; when the tube was above the liquid, gas samples were taken.
sequenced, using bTEFAP PacBio Sequel sequencing technology (MR DNA, Shallowater, Texas, USA). Diversity profiles were obtained by plotting the number of 16S-rDNA sequences of a species as a percentage of the total 16S-rDNA sequences in a sunburst graph.

Autogeneration of biogas pressure
The profiles of the pressure in the HPAD reactor and DBP/DMP in the NPAD reactor are plotted in Fig. 2. In phase 1, the increase of pressure in the HPAD reactor and the production of biogas in the NPAD reactor came from the residual biomass in the inoculum sludge. When the pressure became stable in the HPAD reactor and DBP/DMP became zero in the NPAD reactor, the synthetic wastewater was added into both reactors.
In phase 2, acetic acid solution was added into the reactors and was directly utilized by the methanogens. During this phase, the HPAD and NPAD reactors both ran steadily. The pressure in the HPAD reactor increased from 1.9 to 10.1 bar. The DMP of the NPAD reactor reached around 135 mL/d and declined slightly afterwards. The total methane produced in phase 2 for the HPAD and NPAD reactor was around 3.14 L and 2.17 L, respectively (Table 2). At the end of phase 2, the pressure of the HPAD reactor dropped to 8 bar due to the extraction of liquid samples.
In phase 3, for the HPAD reactor, the pressure increased from 8.0 to 10.6 bar. While the DMP of the NPAD reactor decreased from 141 to 65 mL/day (Fig. 2B), which was due to the inhibition by a high TA (10 g/L). Therefore, in this context, the HPAD reactor was advantageous over the NPAD reactor when dealing with high alkaline acetic acid-containing synthetic wastewater.
From phase 4 to phase 6, the substrate was changed to a sodium J. Zhao, et al. Bioresource Technology 318 (2020) 124101 acetate solution for both reactors. The TA of the sodium acetate synthetic wastewater (14.4 g/L) was lower than that of the acetic acid synthetic wastewater (18.6 g/L). The adjustment of the substrate was necessary because a high pH value (8.2~8.4) and high TA occurred at the end of phase 2 and 3, especially in the NPAD reactor (Fig. 2). The pressure profiles from phase 4 to phase 6 followed a similar pattern in the HPAD reactor, except for the time needed to reach a pressure of 11 bar. The gas production rate was 0.20, 0.19 and 0.21 L/d•L -1 , respectively for phase 4, 5 and 6. More time was needed for the pressure to reach 11 bar because of the increase of the headspace and the decrease of the liquid volume ( Table 2). As a consequence, the pressure inside the reactor increased slower (more biogas is required to fill the remaining headspace). For the NPAD reactor, although the substrate was changed to sodium acetate solution, the DMP was not restored. Furthermore, it completely ceased at the end of phase 5 (Fig. 2B), indicating that the microbial community, especially methanogens, were inhibited. Presumably, the inhibition of the methanogens was caused by the high TA, which will be discussed in the next Section 3.2. From phase 6 on, the substrate in the NPAD reactor was changed to glucose, a substrate that does not require pH adjustment before use. The DMP of the NPAD reactor recovered to around 49 mL/d. But the cumulative methane production of the NPAD reactor (0.39 L) in phase 6 was still much lower than that of the HPAD reactor (1.86 L).
In order to compare the HPAD reactor with the NPAD reactor, the substrate of the HPAD reactor was also changed to glucose in phase 7. The HPAD reactor continued to produce biogas, leading to the increase of the pressure. In contrast, the DMP in the NPAD reactor decreased again. Perhaps, the inhibition of the methanogens in the NPAD reactor was too harsh for them to recover.
Apart from the methane yield, the methane content is another important parameter. Due to the technology set up, the biogas samples from the HPAD reactor were taken only at the end of phase 6 and 7. In a separate preliminary experiment with glucose as the substrate, the biogas in the headspace reached 76.2% CH 4 and 20.4% CO 2 at 3 bar. When the pressure was further increased to 5 bar, the biogas content changed to 79.5% CH 4 and 15.3% CO 2 , respectively (data not shown). In this study, at the end of phase 6 and 7 (11 bar), the CH 4 concentration in the headspace ranged from 87.4% to 88.5%, while the CO 2 content was between 11.5% and 12.6%. Hence, at the pressure of 11 bar, the CO 2 partial pressure was around 1.3 bar. Based on Henry's law, the theoretical CH 4 content should have been around 97% at 11 bar, which is higher than 88% measured in the high pressure reactor. However, the CH 4 content in the gas phase of the HPAD reactor is not solely dependent on Henry's law (water as a liquid) but also determined by other factors (i.e., ions in the liquid and CO 2 saturation). The gas composition in this study was similar to another study, where the CH 4 content was 75~86%, and CO 2 content was 14~25% at 20 bar (Lindeboom et al., 2016). To note, the pH in the reactors had the strongest impact on the gas composition. The decrease of pH will lead to a significantly lower absorption of CO 2 in the liquid phase, which could further influence the CH 4 content in the gas phase (Lemmer et al., 2015b).
The CH 4 content in the biogas from the NPAD reactor ranged from 55% to 64% during phase 2-5, which was higher than the stoichiometric yields. The difference might be explained because part of the CO 2 formed during the decomposition of organic matter (i.e., acetate or glucose) remains in solution as (bi)carbonate, resulting in a higher CH 4 content in the biogas of the NPAD reactor than the expected stoichiometric value (Nolla-ardèvol et al., 2015). In phase 6, the CH 4 content of the NPAD reactor decreased to 35%~42%, suggesting that the methanogens were at least partly inhibited under the applied conditions.

pH, alkalinity, and volatile fatty acids
The operational status of the AD process is reflected by pH, alkalinity (TA), and VFAs. The pH and TA of the HPAD reactor showed a rising trend (Fig. 2). The pH value of the HPAD reactor was lower than that of the NPAD reactor during phase 1-5, which was due to carbonic acid formation from a higher amount of dissolved CO 2 in the HPAD reactor. The pH of the samples was measured externally at atmospheric pressure and was not the real pH inside the HPAD reactor since part of the CO 2 escaped immediately when the samples were taken. Therefore, the pH inside the HPAD reactor was lower than the pH of the samples at atmospheric pressure. During phase 6 and 7, the pH of the NPAD reactor was lower than that of the HPAD reactor, which was ascribed to the high concentration of VFAs accumulated in the NPAD reactor (Fig. 2E). At the end of phase 4 and 5, only acetic acid could be detected in the NPAD reactor, which was 143 mg/L and 2285 mg/L in phase 4 and phase 5, respectively (Table 2). At the end of phase 6, the concentration of acetic acid, propionic acid, and butyric acid was 3212 mg/ L, 178 mg/L, and 18 mg/L, respectively. The optimal pH range for methanogens is between 6.8 and 7.8. Thus, a pH of 8.2~8.4 caused by a high TA (phase 2-5 of the NPAD reactor) was too high for methanogens (Chen et al., 2015). The inhibited methanogens may fail to further convert acetic acid into biogas, resulting in the decline of pH in phase 6 and 7.
TA is related to pH in a complex manner and a suitable TA concentration is necessary to maintain optimal conditions in the AD process (Chen et al., 2015). And in this study, high TA was the main reason leading to the higher pH and VFAs concentration in the NPAD reactor. The substrates fed into the HPAD and NPAD reactors were the same in phase 2-5, but TA in the NPAD reactor was much higher than in the HPAD reactor (Fig. 2D). An optimal TA level ranges from 1.0 to 5.0 g/L. However, the TA of 10.0 g/L in the NPAD reactor was much higher than the upper limit of alkalinity (Osman and Sponza, 2005). Likewise, in a previous study, where Zhu et al. (2010) used 5% (w/w) NaOH to pretreat lignocellulosic biomass for subsequent AD. Although a better degradation of lignocellulose was achieved, high initial TA of the pretreated biomass (21.8 g/L) strongly inhibited the subsequent methanogenesis stage and led to VFAs accumulation (2000 mg/L acetic acid). Hence, choosing HPAD reactors instead of NPAD reactors to deal with wastewater containing high alkalinity could be an alternative. This idea is further illustrated below: The TA of the samples is defined as its acid-neutralizing capacity (Lemmer et al., 2017;Lindeboom et al., 2012) and is expressed as: Under pressure, CO 2 dissolves into the liquid phase and is partly converted into carbonic acid. Henry's law describes the partitioning of CO 2 between the gas and liquid phases. In the calculations of TA, the K H -value for hydration and dehydration reactions of CO 2 are important (log K H = -1.6, all constants at 35°C). H 2 CO 3 first dissociates into HCO 3 -, when the pH value is above 8.5, it further dissociates into CO 3 2-(K 1 is first dissociation constant, log K 1 = −6.3; K 2 is second dissociation constant, log K 2 = −10.3) (Lemmer et al., 2015a;Lindeboom et al., 2012). The pH of the HPAD reactor is below 8, and therefore, the dissociation of HCO 3 into CO 3 2can be neglected (Lemmer et al., 2015a). The constant K H is much higher than K 1 in HPAD, which means that the carbonic acid concentration is higher than the bicarbonate concentration in the HPAD reactor, and therefore the TA is lower than in the NPAD reactor.

Comparison of community diversity
The alpha diversity measurements of Bacteria and Archaea are represented using the number of OTUs, Margalef index (d), Shannon diversity index (H'), and Simpson diversity index (1-D) (Table 3). An improved digester performance is correlated with high microbial richness (reflected by a high Margalef index) and evenness (reflected by a high Shannon or Simpson diversity index) (Tao et al., 2020;Venkiteshwaran et al., 2015). The community analysis was performed on samples taken from the inoculum sludge and at the end of phase 7 from the HPAD and NPAD reactors. The bacterial community in the HPAD reactor had slightly higher indices than those of the NPAD reactor. Samples from the NPAD reactor had lower indices for Archaea compared with the indices obtained from the HPAD reactor. This was consistent with its corresponding poor digester performance (i.e., low cumulative methane yield: 0.45 L, high TA: 6.5 g/L, and high VFAs: 7455, 451 and 52 mg/L for acetic acid, propionic acid and butyric acid, respectively, at the end of phase 7). In contrast, the HPAD reactor showed better performance, reflected by its higher indices for Margalef, Shannon and Simpson (4.19, 1.71, and 0.68, respectively).

Phylogenetic analysis of bacteria at the phylum level
Fourteen major phyla, each represented > 0.1% of the total bacterial sequences in at least one sample, were identified (Fig. 3A). Several phyla were found ubiquitous and predominant at the end of phase 7 and included Chloroflexi, Bacteroidetes, Proteobacteria, and Firmicutes. These phyla are widely present in AD reactors, and their existence is thought to be due to the role that they play in the degradation of complex polymeric organic compounds and fermentation of monomeric sugars in AD systems, and their ability to survive under high alkalinity conditions (Alirio et al., 2018;Sun et al., 2014;Venkiteshwaran et al., 2015).
Being the most predominant phylum, Chloroflexi accounted for 33.1% in the inoculated sludge. The relative abundance (RA) of Chloroflexi decreased slightly in both the HPAD and NPAD samples (24.7% and 24.3%, respectively). Chloroflexi and Proteobacteria (Alpha-, Beta-, Gamma-, and Delta-) are bacteria that can utilize glucose and VFAs (Ros et al., 2017). A lower RA of Proteobacteria in the HPAD reactor (16.9%) than in the NPAD reactor (20.3%) indicated that high pressure might influence the activity of this phylum. The members of the Delta-proteobacteria are dominant (12%) in the HPAD reactor, and they can syntrophically degrade propionate (i.e., Syntrophus and Geobacter species) (Ariesyady et al., 2007;Ito et al., 2012). In the NPAD reactor, Delta-proteobacteria accounted for only 1.2%, explaining the propionic acid accumulation in this reactor. The highest RA (18.9%) among the Proteobacteria in the NPAD reactor was due to the Gammaproteobacteria. These bacteria usually show high hydrolytic activity, and they have a great adaptive capacity in an alkaline environment (Li et al., 2018;Tian et al., 2017).
The second-largest phylum, Bacteroidetes, accounted for 19.7% in the inoculated sludge. The RA of Bacteroidetes increased further to 26.1% and 28.8% in the samples from the HPAD and NPAD reactors, respectively. Bacteroidetes are dominant in mesophilic AD, and they can produce various lytic enzymes and acetic acid during the degradation of organic materials (Ros et al., 2017). Some members of Bacteroidetes are known for their survival in a high alkaline environment (soda lakes) (Lin et al., 2017;Nolla-ardèvol et al., 2015). Additionally, together with Proteobacteria and Chloroflexi, Bacteroidetes, are typical glucose degraders, which may explain their profusion in samples from the HPAD and NPAD reactors (Ito et al., 2012). Firmicutes had a much higher RA in the HPAD and in the NPAD reactors, reaching 24.5% and 21.9%, respectively. Some members of the Firmicutes are butyrate-oxidizing bacteria, while other Firmicutes species are capable of fermenting sugars, and are involved in homoacetogenesis and syntrophic oxidation of acetate (Guo et al., 2015;Nolla-ardèvol et al., 2015;Yi et al., 2014).

Phylogenetic analysis of bacteria at the genus level
There are more apparent differences between the HPAD and NPAD reactors at the bacterial genus level. The distribution patterns of 35 major bacterial genera in samples of the HPAD and NPAD reactors (each genus represented ≥ 0.5% of the total bacterial sequences) are presented in Fig. 3B. In the sample of the HPAD reactor, Dehalogenimonas (17.2%), Cytophaga (14.2%), Syntrophus (11.2%), Acetivibrio (7.4%), and Clostridium (5.9%) were found with a high RA. The presence of these five bacterial genera may suggest their robustness in a high pressure (11 bar) environment. The predominant genera in the NPAD reactor shared nothing in common with the genera obtained in the HPAD reactor except for Cytophaga (15.3%). The major bacteria genera in the NPAD reactor were Pseudomonas (16.5%), Longilinea (12%), Bacillus (7.8%), Bellilinea (6.5%), and Proteiniphilum (6.1%).
Group Ⅰ, VFAs producing group. Among all the known 127 bacterial strains in these samples, 64 strains can generate acetic acid, 35 strains can produce propionic acid, and 14 strains are butyric acid producers (Tao et al., 2020). Moreover, 8 kinds of archaeal genera can produce propionic acid through the succinate pathway. In the last phase, when glucose was used as a substrate, the NPAD reactor contained the highest RA of VFAs producing bacteria (68.76%) and had more acetic acidproducing and butyric acid-producing bacteria than the HPAD reactor ( Fig. 3D and E). Citrobacter was only present in the HPAD reactor and Proteiniphilum, Psychrobacter, and Bacillus, on the other hand, were only found in the NPAD reactor. Longilina and Bacillus are both carbohydrate-utilizing bacteria that were present in a high RA in the NPAD reactor (Alirio et al., 2018). Pseudomonas may use the valine degradation pathway to generate butyric acid via isobutyryl-CoA during glucose fermentation and could survive under high alkalinity conditions (Wainaina et al., 2019). That may explain why Pseudomonas dominated in the NPAD reactor. Cytophaga was present in a high RA in both the HPAD and NPAD reactors; they are common cellulolytic bacteria with some species capable of rapidly degrading cellobiose or glucose and producing VFAs (Tao et al., 2020). The RA of the genera Gracilibacter was higher in the HPAD reactor (5.53%) than in the NPAD reactor. In a previous study, it was reported that Gracilibacter was only functional in the acidogenesis stage and might produce short-chain fatty acids and H 2 (Usman et al., 2019).
Group Ⅱ, VFAs utilizing group. Syntrophic bacteria are VFAs utilizing bacteria and were present in a higher RA in the HPAD reactor (14.62%) than in the NPAD reactor (3.00%) (Fig. 3D). Probably, the enrichment of syntrophic bacteria in the HPAD reactor could partly answer the question why there were almost no VFAs detected in the HPAD reactor throughout phase 1-7. Although syntrophic bacteria account for a relatively small percentage in AD, their presence is critical and they can perform the rate-limiting step to maintain a rapid and stable AD process (Hao et al., 2020;Venkiteshwaran et al., 2015). Thus, the syntrophic relationships between bacteria and methanogens are necessary for a stable AD (Hao et al., 2020;Tao et al., 2020;Usman et al., 2019). In our study, Syntrophus was found to be the dominant syntrophic bacterium in the HPAD reactor (11.19%), and could syntrophically work with H 2 , formate and/or acetate utilizing methanogens to degrade butyrate, or other VFAs (Grabowski et al., 2005). The lack of Syntrophus in the NPAD reactor (0.91%) may partly lead to the accumulation of VFAs. Previously, Grabowski et al. (2005) used fluorescent in situ hybridization to demonstrate the formation of a close spatial association between Syntrophus and the methanogenic archaea (Methanocalculus and Methanosaeta). The high RA of Syntrophus  Zhao, et al. Bioresource Technology 318 (2020) 124101 (11.19%) and Methanosaeta (56.3%) in the HPAD reactor indicated a higher degradation rate of intermediate products (VFAs). The genus Clostridium could contribute to the hydrogen production because various well-defined syntrophic hydrogen-producing bacteria have been affiliated with this genus; they could syntrophically work with hydrogenotrophic methanogens (HM) (Lin et al., 2017;Ros et al., 2017). The correlation between a high RA of Clostridium and an efficient and stable AD performance was reported previously (Ros et al., 2017). In this study, it was shown that Clostridium was found more abundant in the HPAD reactor (5.9%) than in the NPAD reactor (1.2%). Candidatus Cloacimonas (Candidatus Cloacimonas acidaminovorans) is a hydrogenproducing syntrophic bacterium (Guo et al., 2015). Similarly, a slightly higher RA of Candidatus Cloacimonas was found in the HPAD reactor (2.8%), than in the NPAD reactor (1.4%). Hence, compared with the normal pressure conditions, high pressure favors the growth of syntrophic bacteria. Sporomusa is a homoacetogenic bacterium and was only detected in the HPAD reactor, accounting for 1.9%. Homoacetogens are H 2 consumers that might receive hydrogen produced by bacteria from Group Ⅰ. If methanogens are inhibited, then the accumulation of H 2 might result in the consumption of H 2 by the homoacetogens (Wahid et al., 2019). The high pressure resulted in an increase of P H2 in the HPAD reactor, which might explain the presence of Sporomusa in the HPAD reactor. This result was comparable to a previous study showing that the addition of H 2 in an AD reactor resulted in the enrichment of homoacetogens (Wahid et al., 2019). Besides, the higher concentration of H 2 in the HPAD reactor could not only be used by HM to produce CH 4 , but also by the acetoclastic methanogens (AM) to degrade acetate that was produced by the homoacetogens. Group III, function unknown group. Dehalogenimonas from the phylum Chloroflexi was identified to be a genus of organohalide-respiring bacteria, which could couple dehalogenation to a respiration process required for growth (Molenda et al., 2016). Dehalogenimonas had the highest RA in the sludge sample (31.7%) but was not the highest abundant bacterium in the samples of the HPAD and NPAD reactors.
To summarize, the distribution patterns of Group Ⅰ and Ⅱ genera were most likely the result of VFAs production and VFAs degradation. In particular, the enriched bacteria in the HPAD reactor were categorized in Group Ⅱ, and they could increase the conversion of VFAs into acetate, H 2 and CO 2 . Whereas the NPAD reactor contained more bacteria from Group Ⅰ, which could accelerate the degradation of glucose to VFAs. Due to the lack of a sufficient number of bacteria from Group Ⅱ, the NPAD reactor was vulnerable to VFAs accumulation.

Phylogenetic analysis of Archaea at the genus level
A total of 18 genera were identified in the inoculum sludge and in the two reactors. Five Archaea, i.e., Methanobrevibacter, Methanosaeta, Methanoculleus, Methanothermobacter, and Haloarcula were dominant (Fig. 3C), together accounting for 71.56~93.48% of the total Archaea. However, the distribution pattern was different in the three samples and was consistent with the alpha diversity of Archaea. Methanobrevibacter (26.8%) and Methanosaeta (56.3%) prevailed in the HPAD reactor while Methanoculleus (65.3%) and Methanothermobacter (24.3%) were dominant in the NPAD reactor.
Methanosaeta is one of the few Archaea that can metabolize acetic acid to produce methane in mesophilic AD, and also dominated in previous HPAD studies Lindeboom et al., 2016). The substrate during phase 2-6 was acetate in the HPAD reactor, which laid the foundation for the increase of Methanosaeta in the reactor. Methanosaeta has a competitive advantage over other acetate-metabolizing microorganisms due to its high affinity for acetate Ros et al., 2017;Venkiteshwaran et al., 2015). Moreover, the filamentous morphology of Methanosaeta may be of importance for the formation of the microbial floc structure that helps to stabilize the HPAD systems further (Venkiteshwaran et al., 2015). Methanosaeta concilii was abundantly present (50.9%) in the HPAD reactor. This was in line with Lindeboom et al. (2016), who found that this species was abundantly present in an HPAD fed with glucose. M. concilii has been detected in high-pressure subsurface gas, oil reservoirs, and aquatic sediments (Borrel et al., 2012;Grabowski et al., 2005;Lindeboom et al., 2016), indicating that M. concilii is a pressure-tolerant methanogen. Presumably, Methanosaeta works with Syntrophus and homoacetogens, i.e., Sporomusa, in the HPAD reactor to convert acetate into biogas, as explained before (bacterial Section 3.3.3). Methanosarcina is an important Archaea that can generate CH 4 through different metabolic pathways. But it did not prevail among these three samples, especially in the sample from the HPAD reactor.  also found that the RA of Methanosarcina decreased at elevated pressure. It is likely that Methanosarcina was not able to compete with Methanosaeta for acetate in the HPAD reactor, and also may be inhibited by the high alkalinity/VFAs in the NPAD reactor, which was in agreement with previous studies (Sun et al., 2014;Wahid et al., 2019). The lack of Methanosaeta and Methanosarcina strongly reduced the efficiency of acetic acid consumption (Lv et al., 2020), and resulted in VFAs accumulation in the NPAD reactor.
The second-largest genus, Methanobrevibacter has a share of 26.8% among the Archaea in the sample of the HPAD reactor, but its RA was only 1.9% in the sample of the NPAD reactor. This genus can use H 2 / CO 2 or formate to produce CH 4 , suggesting that Methanobrevibacter is one of the few HM that can survive under high pressure. This methanogen can also survive in acidic conditions. Some species, such as Methanobrevibacter acididurans were found in a reactor treating slurry from an acidogenic digester at pH 5.0 (Savant et al., 2002). It complied well with the situation in our case since the pH of the HPAD reactor was lower than the pH in the NPAD reactor because of the formation of carbonic acid in the liquid phase. Hence, these HM (Methanobrevibacter, Methanolinea, Methanoculleus, Methanospirillum, and Methanothermobacter) and their role in the compensation metabolism for AM in the HPAD reactor cannot be overlooked, especially Methanobrevibacter Zhao et al., 2018). These HM converted CO 2 and H 2 into CH 4 , which decreased the P H2 to maintain a stable HPAD reactor.
A high RA of Methanoculleus and Methanothermobacter in the NPAD reactor was observed in this study. These genera are HM and could survive in conditions where VFAs accumulate (Hori et al., 2006;Lin et al., 2017;Zhao et al., 2018), and Methanoculleus was also found in a high alkalinity AD process (Sun et al., 2014). The high VFAs, caused by high alkalinity, was most likely the main factor in changing the methanogenic communities in the NPAD reactor. It has been reported before that Methanothermobacter dominated in a reactor that was operated at thermophilic conditions (Lin et al., 2017). However, in this mesophilic AD experiment, Methanothermobacter thermautotrophicus was also detected in the NPAD reactor, accounting for 24.3% of the total Archaea. In the HPAD reactor, Methanosarcina and Methanococcus were not detected, although they were present in the inoculating sludge and in the NPAD reactor. This suggests that these two HM cannot tolerate high pressure, but this hypothesis needs further investigation.
Compared with the HPAD, there were more HM (93.4%) and less AM (1.7%) in the NPAD reactor. To date, HM tend to be more abundant in reactors operating at special conditions (i.e. low pH, high VFAs concentration, highly alkaline environments) (Nolla-ardèvol et al., 2015;Town et al., 2014;Zhao et al., 2018). Besides, in the NPAD reactor, there were more acetic acid-producing bacteria (67.8%), and almost no AM (1.7%) present; thus, the balance between these acetic acid-producing and acetic acid-consuming microorganisms was broken, which further led to acetic acid accumulation in this reactor. Previous reports also illustrated that AD reactors were stable and healthy when AM were present in a high RA Town et al., 2014;Usman et al., 2019). Moreover, the co-existence of both acetoclastic and hydrogenotrophic methanogens could provide a more robust microbial community Town et al., 2014;Usman et al., 2019).
To summarize, both acetoclastic and hydrogenotrophic methanogens were abundantly present in the HPAD reactor. The NPAD reactor was vulnerable to VFAs accumulation, most likely due to the low abundance of AM.

Conclusions
In the HPAD reactor, the CH 4 content was 88.5% when the pressure reached up to 11 bar. Compared with the NPAD reactor, the HPAD reactor was better able to deal with the high alkalinity acetate-containing synthetic wastewater. The advantage of HPAD was reflected by stable operation indexes and functional microbial guilds such as Syntrophus, Methanosaeta (M. concilii) and Methanobrevibacter. The HPAD reactor can be used in the future to treat alkaline industrial wastewater and upgrade biogas to green gas in a single step.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.