Effects of slurry reux on the stability and microbial community structure of corn stalk anaerobic digestion system

Anaerobic digestion(AD) of corn stalks with slurry reux and non-reux was compared and evaluated,to clarify the effects of slurry reux on AD. The reux of slurry increased methane production by 45.8% and also improved the buffer capacity and adaptability to high organic load rate(OLR). The results of high-throughput sequencing showed that slurry reux increased the richness of bacterial and archaea community, and decreased the diversity of the microbial community. Slurry reux also reduced the tendency of the enrichment of Chloroexi and the relative abundance of Methanothrix. The increase of OLR changed the main methanogenic pathway in the anaerobic digestion system, slurry reux could slow down this trend. pH and ORP were maintained in the suitable range, which showed signicant increase in the system stability of AD. The concentration of ammonia nitrogen decreased gradually. Results from high-throughput sequencing technique for microbial analysis indicated slurry reux increased the richness of bacterial and archaea community and decreased the diversity of community. The dominant bacteria did not change signicantly, the increase of organic load changed the main methanogenic pathway in the AD system, slurry reux could slow down this trend.


Introduction
China is rich in crop stalk resources, with an annual output of approximately 850 million tons (Li et al., 2016a); however, the utilization rate of these resources is relatively low (Du et al., 2017). Nearly 50% of the crop stalk are used for fuel and incineration annually, which caused environmental pollution as well as considerable wastage of resources. As an engineering solution to degrade organic waste and produce biogas, AD is considered a sustainable and effective approach to treat waste, generate clean energy, and recycle organic matter by utilizing stalk energy (Yanjin et al., 2020). However, the production of methane by AD also produced a large amount of slurry, which was rich in nitrogen, phosphorus, potassium, and other nutrients (Lu et al., 2010). If the slurry was not properly utilized and disposed of, it would inevitably cause secondary pollution (Chen et al., 2013;Qiu et al., 2016). Slurry re ux was an e cient and inexpensive treatment for slurry, which effectively reduced the amount of slurry that was discharged, as well as the subsequent disposal cost. Slurry re ux was therefore an e cient and inexpensive way to reduce emissions from the anaerobic digestion system, which in turn, can effectively reduce the slurry discharge and the subsequent disposal cost (Wu et al., 2018). Compared with an AD system where animal manure was used as the substrate, an anaerobic digestion system with crop stalk as the primary substrate lacks nutrients and buffering capacity (Zheng et al., 2020). Through slurry re ux, the basicity of the slurry could be used to balance the acidity of the digestion liquid, to maintain the pH value within an appropriate range. Additionally, slurry re ux can improve the number and activity of microorganisms in an AD system (Hu et al., 2014;Yang et al., 2016).
The biogas-production e ciency and system stability of the AD system depend largely on the activity of functional ora, while the diversity and richness of the microbial community can also affected its performance (Razaviarani & Buchanan, 2015). In AD process, the stages of acid production, acetic acid production, and methane production were completed by different functional ora . By studying the community structure of bacteria and archaea in anaerobic digestion, the internal relationship and interaction between functional ora and anaerobic digestion performance can be obtained (Abbassi-Guendouz et al., 2012;Rui et al., 2015). Maria et al.(2017) reported the microbial community structure of an AD slurry re ux system, showing that Firmicutes and Bacteroidetes were dominant, while the dominant groups of archaea were Methanosarcina (Acidophilic and Hydrogenotrophic Methanogens) and Methanobacteria (Hydrogenotrophic Methanogens). Li et al. (2018) used corn stalk as a substrate to carry out AD with slurry re ux and found that the system was enriched in Bacteroidetes and Firmicutes. Gulhane et al. (2017) reported that the recycling of sewage increased the number of hydrolyzing and fermenting microorganisms that consume volatile fatty acids (VFAs). Different AD substrates determine the species and abundance of bacteria in the system, which would also lead to differences in the activities of various enzymes (Rui et al., 2015). Microbial characteristics could not only provide an early warning of system stability but also revealed the AD mechanism of the system (Abbassi-Guendouz et al., 2012;Rui et al., 2015). Currently, there was a lack of research on the continuous AD of corn stalk, especially relating to the role of bacteria in restoring system function after a sudden increase in OLR.
Therefore, this study examined system stability and the in uence of the microbial community, using an AD system which incorporated slurry re ux. Corn stalk were used as the substrate, and various organic load rate (OLR) were set. This study aimed to provide a scienti c basis for the application of slurry re ux in biogas engineering technology.

2.1.Substrates and Inoculum
Corn stalks were obtained from the Beishan Scienti c Research Base of Shenyang Agricultural University. The inoculum was obtained from the anaerobic digestion pond of the Comprehensive Energy Demonstration Base of Shenyang Agricultural University. The physicochemical parameters of the corn stalks and inoculum were shown in Table 1.  (Fig. 1).The total volume of the reactor was 10 L(working volume 8 L), the temperature was set at (37 ± 1) °C, and total solid (TS) content was 10%, and the mixing ratio of inoculum and substrate was 1:3 (calculated from the total dry weight of the material). Re ux was carried out once every 3 d; after the material was ltered. Up to 75% leachate was taken from the re ux group (R2) to make up for the water and was mixed with corn stalk, while water was used instead of the non-re ux group (R1). The test was conducted using an intermittent stirring method; stirring was carried out for 20 min every 2 h, and the rotating speed was set at 30 r/min. The rst 10 days of the process comprised the start-up period (no data). Slurry re ux started on day 11, and was conducted for 81 days. According to the different OLRs, the test was divided into three phases: Phase (0-27 d after re ux) had an OLR of 2.0 gTS/(L·d), Phase (28-54 d after re ux) had an OLR of 3.0 gTS/(L·d), and Phase (55-81 d after re ux) had an OLR of 4.0 gTS/(L·d).

2.3.Analytical Techniques and Statistical Method
The pH values, total solid (TS) and volatile solid (VS) concentrations were measured according to the Standard Methods (APHA, 2012). The oxidation reduction potential (ORP) was measured using a Hach HQ40D multi-parameter water quality analyzer. A portable biogas analyzer was used for compositional analysis of the biogas. VFA concentrations were determined using gas chromatography on an FID detector 19091N-133 (30 m × 250 µm × 0.26 µm) column (Gu et al., 2014). Ammonia nitrogen was used for determination in the sodium reagent colorimetric method.
From the two groups of anaerobic digestion reactors with varying conditions, a total of six samples were selected for microbial community structure analysis and detection, these samples represented pre-re ux material (R1-1;R2-1), the end of Phase (R2-2), the end of Phase (R1-3;R2-3), and the end of Phase (R2-4). 'GACTGGAGTTCCTTGGCACCCGAGAA TTCCAGGACTACVSGGGTA TCTAA T-3'. PCR products with normal ampli ed fragments of bacteria and archaea above 400 bp were treated with 0.6 magnetic beads (Agencourt AMPure XP).
After treatment, the samples were sent to Shanghai Sangon Bioengineering Co., Ltd for sequencing. The detection types were bacteria and archaea, the platform was Miseq2 ×300 bp, and the library was the NCBI16S database. Operational taxonomic unit (OTU) clustering and species annotation were performed for more than 97% of the sequences, and Alpha diversity analysis and relative abundance analysis were also performed.

Results
3.1.In uence of slurry re ux on methane production rate and cumulative methane production The methane production rate was one of the important indicators for measuring the operating e ciency of the anaerobic digestion system. The methane production rate at each phase of the anaerobic digestion process was shown in Fig. 2(a), and the cumulative methane production at each phase was shown in Fig.   2(b).
In Phase , the methane production of R1 and R2 increased to a maximum and then decreased gradually, and the highest methane production rates were 234.78 mL/gTS and 250.73 mL/gTS, respectively. During Phase , the OLR was increased to 3 gTS/(L·d) and the methane production rate gradually increased. The maximum methane production rate reached 187.25 mL/gTS and 215.36 mL/gTS, respectively. In Phase , the OLR increased to 4 gTS/(L·d), and the average methane production rates in this phase were 100.41 mL/gTS and 171.21 mL/gTS, respectively, which decreased by 23.35% and 3.63% compared with that in Phase . It was worth noting that the methane production rate of R2 was better than that of R1 for the majority of the AD operation, and in R2 the average methane production rate was 45.8% higher than that of R1. The variation of methane production rate in the anaerobic digestion system was affected by both increased OLR and the slurry re ux. The methane production rate presented a downward trend with increasing OLR, and the implementation of slurry re ux could effectively alleviate this trend.
The cumulative methane production in Phase -, was 73.97 L, 84.88 L, and 86.76 L in R1 and 95.14 L, 115.12 L, and 147.92 L, in R2. The methane yield of R2 (363.30 L) was signi cantly higher than that of R1 (245.61 L), indicating that slurry re ux could promote methane production. During the three phases of operation in R2, the cumulative methane yield increased by 28.6%, 35.6%, and 70.5%, respectively, relative to R1. Slurry re ux enabled more complete degradation in the organic fermentation system, reduced the loss of microbes, and improved methane production. Such ndings are consistent with those of previous studies Lu & You, 2015). However,  and  found that slurry re ux inhibited biogas production and reduced methane production. After preliminary analysis, they suggested that this might be caused by an increase in slurry viscosity and the accumulation of salt ions, during long-term operation of the slurry re ux process, and that this required further investigation. In this study, the system was operated for 91 days, and no signi cant reduction in methane production was observed.
3.2.In uence of slurry re ux on system stability System stability is one of the key factors affecting the operation of the AD system. The variations in pH value, VFAs, ammonia nitrogen concentration, ORP, and other indexes (Fig. 3) indirectly re ected the stability of the AD system.
Prior to slurry re ux, VFA concentrations in R1 and R2 (Fig. 3b) were in a state of decline from continuous feeding, and the initial VFAs concentrations were 2310.99 mg/L and 2101.90 mg/L, respectively. With continuous feeding, the pH value of R1 (Fig. 3a) showed a low value of 6.39, while the pH value of R2 was relatively stable. On the 15th day of slurry re ux, the VFA concentration in R2 decreased to 142.79 mg/L, while in R1, it decreased to 194.73 mg/L on the 21st day. When the OLR was low, the effect of slurry re ux on the degradation of VFAs in the anaerobic digestion system was not obvious. In Phase , the concentration of VFAs was low in both groups of anaerobic digestion systems, At the beginning of Phase , although the increase in OLR caused a slight increase in the concentration of VFAs in R1, it soon became stable. In Phase , the OLR increased to 4 gTS/(L·d). Between 69-89 d, both experimental groups showed VFA concentration uctuations, and the VFA concentration of R1 increased signi cantly. It was worth noting that the concentration of propionate (282.4 mg/L) during this phase was higher than the concentration of acetate (162.4 mg/L). Previous research has shown (Ying et al., 2014) that the ratio of acetate to propionate can be used as a system performance index; in an AD system when the concentration of propionate exceeded that of acetate by a certain threshold, it heralded a failure in the anaerobic digestion system (Wagner et al., 2014). At the same time, with the increase in VFAs concentration in R1, the pH value of R1 decreased from 7.11-5.92, which was beyond the suitable range of most methanogens. Anaerobic digestion was inhibited, as previously mentioned in the methane production analysis. R2 was maintained at pH 7.1, and there was no evident VFA accumulation under high OLR (4 gTS/(L·d)); slurry re ux improved the utilization rate of VFAs and the system stability. This was because the slurry re ux increased the microbial quantity of hydrolysis and fermentation in the AD system, which could consume VFAs (Gulhane et al., 2017). Although increases in OLR of the system caused local uctuations in VFA concentrations, there was no accumulation of VFAs or inhibition of the AD system.
During the whole process of anaerobic digestion, the concentration of ammonia nitrogen (Fig. 3c) showed a decreasing trend, and was generally within 1500 mg/L; thus, it did not reach the inhibitory concentration and did not cause the accumulation of ammonia nitrogen. Although ammonia was an important index for measuring the nitrogen source in a system, previous studies have found that both high and low ammonia concentrations are not conducive to the growth and metabolism of microorganisms (Han et al., 2016;Yenigün & Demirel, 2013). An ammonia nitrogen concentration of 200 mg/L was signi cant for the microorganisms in the AD system, while concentrations that were too low do not meet the metabolic requirements of microorganisms. In the later phase of the anaerobic digestion process, the concentration of ammonia nitrogen was generally stable and uctuated around 200 mg/L, which was similar to the results of previous research .
During 9-45 d, the ORP of R1 (Fig. 3d) was at a stable stage (-405.71 ± 15 mV). The ORP in R2 was lower than this (-342 ± 5 mV). As the experiment went on, the ORP of R1 began to increase gradually. When the OLR was increased to 4 gTS/(L·d) in Phase , the ORP of R1 exceeded that of R2, which might be related to the increase in VFA concentration in R1. On the contrary, slurry re ux could improve the stability of the system so that the ORP of R2 was always within the appropriate range, and the average values of the three stages were − 347.9 mV, -336.9 mV, and − 331.1 mV, respectively. 3.3.Microbial community structure before and after slurry re ux

3.3.1.Analysis of microbial community diversity and richness
Microorganism methods, based on high-throughput sequencing, could return fairly complete DNA sequences and provide useful and tractable information about the microbial community, and have been widely used to analyze microbial activity during the anaerobic digestion process (Li et al., 2016b;Ying et al., 2014). The methane production and stability of the AD system depended largely on the activity of the microbial community structure. The diversity and richness of the microbial community affected the performance of the AD system (Razaviarani & Buchanan, 2015). Changes to operating conditions of the anaerobic digestion reactor also had signi cant effects on the microbial community structure (Si et al., 2016) in OTUs with a similarity level of 97%. A comparison of Alpha diversity, of anaerobic digestion bacteria in the no re ux and re ux groups, was shown in Table 2. By comparing the Shannon and Chao1 index of the groups without slurry re ux (R1-1, R1-3) with those of the slurry re ux group (R2-1, R2-3), it was clear that with increasing AD operation time, R1 and R2 both exhibited a reduction in bacterial community diversity compared to the initial phase, of 1.1% and 6.4%, respectively. The Chao1 of R1 decreased by 5.3%, and that of R2 increased by 7.1%. The diversity of the bacterial community decreased gradually, while the richness of the bacterial community increased continuously. In conclusion, slurry re ux promoted the evolution of bacterial community structure in the AD system, and increased the bacterial community richness, which was consistent with the results of previous studies (Gulhane et al., 2017;Li et al., 2018).
The alpha diversity of archaea in the anaerobic digestion system was shown in Table 3. The diversity of arctic communities in the non-re ux group (R1-1, R1-3) and the re ux group (R2-1, R2-3) increased by 39.0% and 19.6%, respectively. According to the Chao1 index, among archaea, the richness of R1 increased by 37.9%, while that of R2 decreased by 17.6%. It could be seen from the indexes in Table 3 that the richness and diversity of bacteria in the AD system were higher than those of archaea, which was mainly caused by the difference in genetic diversity between bacteria and archaea (Guo et al., 2014). The diversity and richness of bacterial and archaeal communities were correlated with the stability of the system. The disappearance of bacterial populations indicated the disappearance of bacterial strati cation and reduced evolution of new communities, which was a successful expression of the biochemical activity of the anaerobic system. However, further studies were needed to clarify the relationship between the stability of the anaerobic system and the composition of the microbial communities.

3.3.2.Bacterial community compositions
The relative abundance of bacterial communities at the phylum level was shown in Fig. 4.
As could be seen from Fig. 4, there were six dominant phyla in this study, namely Bacteroidetes, Firmicutes, Proteobacteria, Chloro exi, Errucomicrobia, and Synergistetes, which together accounted for 83.14-91.75% of the total phyla level. Firmicutes and Bacteroidetes were the dominant phyla in the two experimental groups, and their relative abundance accounted for 43.29-63.62% of the total level.
Bacteroidetes were important hydrolytic, acid-producing, and fermentation bacteria, with several functions in the AD process (Weber, 2015). As the experiment progressed, the non-re ux group (R1-1; R1-3) and re ux group (R2-1; R2-3) showed a decreasing trend in the relative abundance of Bacteroidetes, particularly in R2-3, while the relative abundance was higher (23.27%) in the re ux group (R2-3) was higher, which indicated that operating the AD system with slurry re ux resulted in better hydrolytic acidi cation (Regueiro et al., 2012).
In addition, compared with the non-re ux group (R1-1, R1-3), the relative abundance of Proteobacteria in the re ux group (R2-1, R2-3) was 17.46% and 7.5%. The relative abundance of Proteobacteria showed a downward trend, and was higher in the re ux group than in the non-re ux group (2.4%). The relative abundance of Chloro exi in the non-re ow group (R1-1, R1-3) and the re ow group (R2-1, R2-3) was 1.1%, 15.29%, and 16.05%, 29.52%, respectively. The relative abundance of Chloro exi continued to increase, which was promoted by the slurry re ux.
The community structure and relative abundance of bacteria in the re ux group (R2-1, R2-2, R2-3, and R2-4) were shown in Fig. 4. When OLR increased from 2 gTS/(L·d) to 4 gTS/(L·d), the relative abundance of Bacteroidetes rst decreased before increasing, which was opposite to the reaction exhibited by Proteobacteria. Firmicutes decreased from 23.35-14.57% during the anaerobic digestion process. With the increasing in OLRs, the peak value of Verrucomicrobia reached 25.84% at the end of process, and the relative abundance of Verrucomicrobia increased by 23.16% compared with the initial phase (R2-1). Chloro exi reached a maximum when OLR was 3 gTS/(L·d). The relative abundance of Chloro exi during the four phases was 16.05%, 23.36%, 29.52%, and 13.52%, respectively. The relative abundance of each microbial community changed signi cantly, indicating that the bacterial community structure was signi cantly affected by the OLR (Jing et al., 2019).
At the genus level (Fig. 5b), the dominant archaea included Methanothrix, Methanospirillum, Methanobacterium, Methanosphaerula, and Methanomassiliicoccus. Methanothrix belonged to the actotrophic methanogens, which could convert organic acids into methane. The relative abundances of Methanothrix in the non-re ux group (R1-1, R1-3) and the re ux group (R2-1, R2-3) were 58.24%, 15.34%, and 53.44%, 38.76%, respectively. During the AD process, the re ux group (R2-3) had a higher relative abundance of Methanothrix than the non-re ux group (R1-3), which might be due to the continuous e uent in the non-re ux group system and the lower VFA content.
Methanospirillum, Methanobacterium, Methanosphaerula, and Methanomassiliicoccus all belong to the hydrotrophic methanogens (Bassani et al., 2017;Girma et al., 2017;Ilaria et al., 2015;Lee et al., 2017). With increased OLR, the combined relative abundances of the four species in the non-re ux group (R1-1, R1-3) and re ux group (R2-1, R2-3) were 29.79%, 68.05%, and 17.17%, 24.14%, respectively. Hydrotrophic methanogens were dominant in the non-re ux group (R1-3), while methanogens in the re ux group (R2-3) were still mainly actotrophic methanogens. This was because the increase in OLR changed the main methanogens pathway in the AD system, and slurry re ux would slow down this trend. Additionally, analysis of the relative abundance of bacteria (Fig. 4) also con rmed this, as the relative abundance of Firmicutes, which contained many known SAOB, was higher in R1 than in R2. It exhibited a reciprocal relationship with hydrotrophic methanogens, which could convert acetate into H 2 and CO 2 to generate methane (Zhu et al., 2019).
The structure of archaea at genus levels, during the different phases, was shown in Fig. 5. The relative abundance of Methanothrix in the re ux group was 53.44%, 45.16%, 38.76%, and 41.75%, respectively. The main methane-producing pathway was always the acetophilic pathway, but the increase in OLR changed the relative abundance of Methanothrix (Jing et al., 2019). The second-most dominant archaea were Crenarchaeota. In the process of slurry re ux, the relative abundance of Crenarchaeota exhibited an increasing trend, reaching the highest value (36.48%) in R2-3. When the OLR reached 4 gTS/(d·L) (R2-4), the relative abundance began to decline rapidly to 16.2%. In R2, the relative abundance of hydrogentrophic methanogens increased with the increase in OLR throughout the AD process, reaching 40.68% at the end of the reaction (R2-4), which was consistent with previous studies Ros et al., 2017). Under high OLR conditions, hydrogentrophic methanogens occupied the dominant position in the whole anaerobic system.

Discussion
Slurry re ux improved the system performance of AD, reduce the discharge of slurry and increase the methane production rate. The utilization of VFAs was improved by the slurry re ux; pH and ORP were maintained in the suitable range, which showed signi cant increase in the system stability of AD. The concentration of ammonia nitrogen decreased gradually. Results from high-throughput sequencing technique for microbial analysis indicated slurry re ux increased the richness of bacterial and archaea community and decreased the diversity of community. The dominant bacteria did not change signi cantly, the increase of organic load changed the main methanogenic pathway in the AD system, slurry re ux could slow down this trend.
Abbreviations CSTR: Continuous stirred tank reactor; OUT: Operational taxonomic unit; VFAs: Volatile fatty acids; ORP: Oxidation-reduction potential; TS: Total solid; VS: Volatile solid; OLR: Organic loading rate Declarations Authors'contributions LZ performed the experiments and analyzed the raw data. YG conceived the study, designed the experiments and drafted the manuscript. JMZ , JXS and ZW participated in the design of the study. carried out the model analysis, and JYL and WK helped to perform the experiments. All authors read and approved the nal manuscript. Figure 1 Schematic diagram of the CSTR Figure 2 Methane production rate (a); cumulative gas production (b); Phase -Phase , respectively, had organic load rates of 2, 3, and 4 gTS/(L·d). Percent of archaeal community abundance at the genus level