The performance and archaeal community shifts in a modified anaerobic baffled reactor treating sweet potato starch wastewater at ambient temperatures

A conventional anaerobic baffled reactors (ABRs) treating high strength sweet potato starch wastewater at ambient temperatures resulted in acidification and bad performances. After modification, the acidification was remitted and COD removal efficiencies reached 92.73% at high temperatures and were maintained at 71.19% at low temperatures. Moreover, as much as 1.014 ± 0.056 L CH4/L/d were collected at Stage III. The q-PCR results revealed that the largest methanogen populations emerged at Stage III as well, which was 5.29 × 108 mcrA copies per milliliter sludge. A comparable shift in the archaeal community structure at different stages and acetoclastic methanogens Methanosaeta predominated the archaeal community in every compartment in Stages I (63.73%) and II (48.63%). Finally, the net energy gains analysis at mesophilic, thermophilic, and ambient temperature revealed that modified ABR at ambient temperature was not only economical but also profitable and could generated 3.68 KJ energy per gram COD removed.

However, when treating high strength SPSW using ABRs in south China, there are some limitations reported by other researchers. First, ABR's plug-flow structure caused a much higher load than average in the front compartments and incomplete degradation of large organic matter into volatile fatty acids (VFAs). An increase in VFAs concentration could lead to a simultaneous reduction in pH and exceed the optimal acid-base range (6.6-7.7) for microorganisms, such as methanogens, which could contribute to the decline in COD removal efficiency and methane production 12 . Second, a non-homogeneous influent flow rate caused by the simple baffled distribution structure and high particulate matter concentrations in SPSW could contribute to increased dead space and therefore, decrease the reactor's volume utilization 13,14 . Finally, most studies resorted to above-ambient temperatures to maintain good ABR performance and maximize CH 4 yield, overlooking the energy input to the process and hence, loss of net energy gain. A large amount of energy is usually needed to heat wastewater to mesophilic or thermophilic fermentation temperatures, particularly in subtropical and temperate zones. For instance, the wastewater usually heated to 35 °C, which could increase treatment costs by ~0.39 US Dollars per ton compared with 25 °C. The relatively high costs of anaerobic SPSW treatment has limited its application, while operated at low temperatures (10-15 °C), a fall in overall COD removal efficiency is usually observed in ABR performance, even when fed low strength wastewater (4000 mg/LCOD) 11 . SPSW (containing saccharides, proteins, and fats, etc.) degradation in an ABR is a synergistic and complex bioprocess that involves different anaerobic microbial groups. These functional microorganisms separate spatially along the course of the reactor and degrade various organic compounds in a concerted effort into CH 4 , CO 2 , and H 2 O through hydrolysis, acetogenesis, and methanogenesis 15 . Generally, hydrolytic and acidogenic bacteria and methanogens are the two major groups involved in anaerobic digestion, with methanogenesis being a rate-limiting step and requiring effective control to achieve good treatment performance 16 . Therefore, it is important to elucidate the composition and function of the methanogen communities and their shifts in response to environmental changes in ABRs, which can be used to optimize ABR operation. However, few studies have focused on the methanogen community, its shifts in response to environmental changes, and its relationship with ABR performance when run at ambient temperatures. Among all the culture-independent molecular techniques, real-time PCR and pyrosequencing are the most powerful methods that can provide significant insights into the composition and evolution of microbial community structure in the ABR system.
In this study, a conventional ABR was constructed to treat SPSW and its performance was investigated in the first period. Limitations mentioned above were observed during the conventional ABR operation. To solve these problems, the conventional ABR was then reconstructed in the second period. First, we rearranged the reaction compartments in a decline to save the problem of acidulation caused by a high hydraulic loading rate in the front ABR compartments. In doing this, the volume load in the front part of the ABR decreased properly. Furthermore, to avoid a channeling effect, we changed the baffle structure by perforating 1-2 cm diameter holes on the slanted edge. We then hung plastic membranes in each reaction compartment to increase the attachable area for bacterial and therefore, enhanced the reaction rate. After these improvements, the performance of the modified ABR treating high strength SPSW was investigated at ambient temperatures. Meanwhile we evaluated the benefit of a modified ABR operated at ambient temperatures in a subtropical zone compared with those operated at mesophilic or thermophilic fermentation temperatures, in the context of net energy gain. Additionally, quantitative methanogen community shifts and changes in archaeal community structure in relation to operation conditions at ambient temperatures was monitored by real-time PCR and high-throughput 16 S rRNA gene sequencing, respectively.

Results and Discussion
ABR Performance at ambient temperature. Performance of the conventional and modified ABRs during the entire study period are shown in Tables 1 and 2. The pH increased gradually as the SPSW advanced from compartment 1 to compartment 5 during all five stages in both ABRs. In the conventional ABR, average pH values (5.34 ± 0.75 and 5.69 ± 0.75, respectively) in the first and second compartments were significantly lower than that in the last three compartments (6.32 ± 0.38, 6.64 ± 0.33, and 6.81 ± 0.31 on average) at the p = 0.01 level (Table 1). Additionally, pH in all five compartments dropped continuously from Stages II to IV. This tendency was particularly obvious in the first two compartments, in which the pH decreased from 5.80 ± 0.41 and 6.03 ± 0.40, respectively, in Stage I to 4.62 ± 0.14 and 4.90 ± 0.1, respectively, in Stage IV. Similarly, the total COD removal efficiency fell by more than 20% in Stage IV compared with Stage II. The decline in COD removal efficiency might have been caused by decrease of pH in the first two compartments in the conventional ABR and therefore influence the utility of organic matters in SPSW. pH optima for hydrolytic and acidogenic bacteria is between 5.5 and 6.5 [17][18][19] and methanogenic microorganisms is around pH 7 [20][21][22] . Afterwards, excessive acidification led to the collapse of the conventional ABR.
Similar transformation rules of pH were observed in the improved ABR, in which the pH was significantly lower in the first two compartments compared with the last three (Table 2). However, after reconstruction, the COD removal efficiency did not decrease so rapidly in the first two compartments. The overall COD removal efficiencies in Stages III and IV were remained at 78.30 ± 6.01% and 71.19 ± 6.65%, respectively, which were much higher than the results reported by Alette A.M. et al. 23 . In their research, COD removal efficiencies dropped from 80% to 70% and then to 60% when the operation temperature decreased from 35 °C to 20 °C and then to 10 °C, however, what should be mentioned was that the HRT was 10 h. Moreover, the improved ABR did not collapse because of excessive acidulation at any point in the experiment. And the COD removal performance of the improved ABR was more stable than the conventional ABR with the standard deviation being 11 versus 15. Furthermore, COD removal in the improved ABR was less affected by temperature than in the conventional ABR. The correlation coefficient was 0.840, p < 0.01 for the conventional ABR and 0.419, p < 0.01 for the improved ABR. Because of the poor construction design and treatment performance, very little biogas was collected and detected in the conventional ABR (data not shown). However, quite a large quantity of biogas was collected from the improved ABR. As shown in Table 2, the methane production rate (MPR) first increased, then decreased over the course of the experiment and reached a maximum in Stage III. The max MPR was 1.014 ± 0.056 L CH 4 /L/d, much larger than the results reported by Yu, H. et al. 24 and R. Grover et al. 25 , which were 0.30 and 0.35 L CH 4 /L/d, when treating municipal wastewater and pulp and paper mill black liquors, respectively. Moreover, this observations were not in accordance with previous reports that longer HRT and higher temperature could promote methane production 26,27 . The drop of temperature and increase of HRT did not cause decline of COD removal of

Changes in total methanogenic archaea population. The methanogen populations of modified ABR
were determined by quantifying the gene encoding the alpha subunit of methyl-coenzyme M reductase (mcrA). As shown in Fig. 2, the average number of copies of the mcrA gene in the bioreactor increased from Stages I to III and decreased in Stage IV, which were 1.2 × 10 8 ± 2.1 × 10 7 , 2.12 × 10 8 ± 1.3 × 10 7 , 5.29 × 10 8 ± 1.3 × 10 7 , and 1.60 × 10 8 ± 1.4 × 10 8 copies per milliliter sludge, respectively. In Stage I, the methanogenic population in the first two compartments was smaller than that in the seed sludge. The mcrA gene were 1.99 × 10 7 ± 1.2 × 10 6 and 2.03 × 10 7 ± 1.8 × 10 6 copies/mL in compartment 1 and compartment 2, versus 8.25 × 10 7 ± 8.4 × 10 7 copies/mL in seed sludge. The decrease in the population might have been the result of disadvantageous pH conditions for methanogens, pH optima for which was reported to be around 7. In the last three compartments, the mcrA gene copy numbers increased by 36, 46, and 45%, respectively. Although the mcrA gene copy numbers were different in each compartment, the methane production rates were similar. This phenomenon suggested that the surviving microorganisms in the set-up period could not function efficiently. The methanogen numbers kept increasing in Stage II, with the third compartment being the largest, which was consistent with the methane production rate. In Stage III, the mcrA gene copies did not decrease as a result of the temperature drop. In contrast, methanogen abundance increased 2.8-, 4.6-, 0.9-, 1.5-, and 0.1-fold in each compartment compared with Stage II and reached their peak during the experimental period. In Stage IV, the mcrA gene copy numbers were still greater than that in the seed sludge, which meant that the refitting of the conventional ABR provided suitable conditions for methanogen growth, even at low temperatures. Analysis of the high-throughput sequencing profiles of the biomass samples taken from the improved ABR on days 1 (seed sludge), 43 (Stage I), 139 (Stage II), 178 (Stage III), and 275 (Stage IV) revealed a distinct shift in the archaeal community structure. The changes in the archaeal community in different treatments at the genus level is showed in Fig. 3. In the seed sludge, the archaeal genus Methanobacterium, hydrogenotrophic methanogens that typically dominate in livestock manure fermentation, was initially predominant 31,32 . After the improved ABR had been inoculated, their abundance decreased significantly from 46.68% to 4.94-15.40% on average on the following stages with the lowest abundance recorded in Stage III. The genus was then replaced as the major archaeal group by Methanosaeta in every compartment in Stages I and II, with an average abundance of 63.73 and 48.63%, respectively. Methanosaeta are acetoclastic methanogens and are the most dominant archaea genera in the anaerobic digestion of food waste 27,33 . On days 43 and 139 the predominant archaeal species in the five compartments along the ABR did not change, which suggested that a partial spatial separation of archaea along the ABR had not taken place. In Stages III and IV, the number of Methanosaeta dropped significantly in the front compartments. Take compartments 1 and 2 for instance, Methanosaeta abundance decreased by 84.59 and 91.53%, respectively, from Stages II to IV. Meanwhile, the genus Methanobrevibacter became the dominant methanogen in the first four compartments in Stage III and the first three compartments in Stage IV. A proliferation of Methanobrevibacter and a parallel decrease in Methanosaeta in Stage II to Stage IV might have resulted from a decline in pH in the  front compartments. Acidic conditions have a negative effect on Methanosaeta growth; thus, Methanobacteriales was more tolerant to acidulation compared to Methanosaeta 34 . An increase in pH occurred along the reactor in Stages III and IV, which was caused by a significant increase in the Methanosaeta populations, 83.04 and 83.92%, respectively. On day 178, the dominant archaea in each compartment started to differentiate, with Methanobrevibacter being the most abundant methanogen in the first four compartments and Methanosaeta in the fifth compartment. The situation was similar on day 275, when Methanobrevibacter was the dominant archaea in the first three compartments and Methanosaeta in the last two compartments. Performance improved in Stage III and no partial spatial separation of archaea took place with the changes in predominant methanogens in the initial compartments and final compartments. The partial spatial separation of archaea in the improved ABR, with hydrogenotrophic methanogens (Methanobrevibacter) in the front compartments and acetoclastic methanogens (Methanosaeta) in the final compartments, is beneficial to COD degradation and methane production.

Conclusion
The performance and archaeal community shifts in a modified anaerobic baffled reactor treating sweet potato starch wastewater at ambient temperatures were investigated in this study. The following conclusions can be drawn based on our results: • After modification, the acidification was remitted in the front compartments and the pH was maintained at 5.61 ± 0.17 in the first compartments compared with 4.62 ± 0.14 in the conventional ABR. • COD removal efficiencies in the improved ABR reach 92.73% at high temperatures (~30 °C) and remained at 71.19% at low temperatures (~10 °C), increasing by 12.37 and 17.67% compared with the conventional ABR. • As much as 1.014 ± 0.056 L CH 4 /L/d was collected from the improved ABR in Stage III, however, methane could not be collected from the conventional ABR. • The modified ABR provided suitable conditions for methanogen growth at both low and high temperatures.
The largest methanogen populations emerged at Stage III, which was 5.29 × 10 8 ± 1.3 × 10 7 mcrA copies per milliliter sludge on average. Methanogen quantities remained at high levels when temperature decreased. • A distinct shift in the archaeal community structure took place during the experimental period in the modified ABR. Acetoclastic methanogens, i.e., Methanosaeta, predominated the archaeal community in Stages I and II and the genera Methanosaeta and Methanobrevibacter predominates in different compartments in Stages III and IV. • Partial spatial separation of archaea along the ABR occurred in Stages III and IV, with the genus Methanobrevibacter becoming the dominant methanogen in the front compartments and Methanosaeta dominating subsequent compartments. • Net energy gains (NEG) revealed that operation of the modified ABR at ambient temperature was not only economical but also profitable and could generated 0.86,3.67,3.68 and 2.30 KJ energy per gram COD removed, respectively, from Stage I to Stage IV.

Methods
Conventional ABR structure. During the first period of the investigation, from June 2011 to December 2011, a pilot-scale conventional structure ABR was configured to treat SPWS at ambient temperatures, as shown in Fig. 4. The bioreactor with a working volume of 480 L (L × W × H = 150 cm × 40 cm × 80 cm) was divided into five equal compartments by baffles. At the top of each compartment, a biogas outlet was installed to collect and measure the biogas generated inside. The characteristics of the conventional ABR used in this experiment are listed in Table 3.
Improved ABR structure. During the second period of the experiment, the conventional ABR structure was improved to solve the abovementioned problems, as shown in Fig. 5. The improved laboratory scale ABR unit was made of transparent Plexiglas (100 cm long, 15 cm wide and 60 cm high) with an available capacity of 60 L. The bioreactor was then divided into five reaction compartments by vertical slanted edge baffles. Each reaction compartment was further divided into two parts, the down-and up-comer regions, which encouraged mixing within each compartment. To avoid acidulation caused by a much higher load than average in the first two compartments, we arranged the five reaction compartments in a decline, with the volume ratio of 12:10.8:9.7:8.7:7.9.
In doing this, the volume load of the front part of the ABR decreased properly. Furthermore, to avoid a channeling effect and decrease dead space, we perforated nine circular holes on slanted edge of baffles in each reaction compartment. The diameter of those holes was 2, 1.5, and 1 cm, respectively, with the largest holes on the top and the smallest on the bottom. Moreover, we also added porous double circle plastic ring carriers (see Fig. 5) in each reaction compartment to increase the attachable area for bacterial and therefore, enhanced the reaction rate. The main characteristics of the carriers were: diameter, 80 mm; pitch, 30 mm; surface area, 800 m 2 /m 3 ; and porosity 90%. The characteristics of the improved ABR used in this experiment are listed in Table 4.

Experimental plans and operating conditions. The experiment was carried out in Changsha, Hunan
Province from June 2011 to April 2013 at ambient temperature ranging from 5 to 30 °C in a subtropical hilly region. The whole experiment was divided into two periods according to the ABR used: the operation of conventional ABR (Period I, 2011.7-2011.12) and the operation of improved ABR (Period II, 2012.04-2013.03). The experimental plans and operating conditions for both periods are described in the following sections.
Conventional ABR operation. Characterization of the sweet potato wastewater and digested sludge: The sweet potato wastewater used in this research was obtained from a sweet potato wastewater treatment plant adjusting tank in a company extracting starch from sweet potato for glass noodle production in Changsha, China. The characteristics of the wastewater were as follows: COD (mg/L) = 8000-12000; BOD (mg/L) = 6000-9000; TN (mg/L) = 400-600; TP (mg/L) = 70-120; pH = 4.5-5.5. The SPSW was diluted and pH was adjusted to 11.5 with NaOH to neutralize acid produced during reactor operation. The ABR was inoculated with the anaerobic stabilized sludge from a pig farm digester at Baisha Town in Changsha, Hunan, China. The amount of sludge used for the ABR inoculation was 288 L (60% of the working volume). The characteristics of the inoculated sludge were: 21.3 ± 0.4 g/L total suspended solid (TSS), 15.2 ± 0.7 g/L volatile suspended solid (VSS), and pH = 7.8. Experimental design and conventional ABR operating conditions: The conventional ABR operation was divided into four phases with respect to the influent COD level, operating temperature, and COD removal performance: Phase I (Set-up phase, days 1-24); Phase II (Operation phase at high temperature, days 25-58); Phase III (Operation phase at moderate temperature, days 59-137); Phase IV (Operation phase at low temperature, days 138-172). More conventional ABR operating information is listed in Table 5. The reactor was continuously fed with sweet potato wastewater at room temperature from the feed tank.
Improved ABR operation. The origin and characteristics of SPSW and seed sludge used in the improved ABR were the same as those used in the conventional ABR. The SPSW was diluted and pH was adjusted to 7.5 with NaOH before used. Because the reactor was improved and in order to save NaOH dosage, the pH value of SPSW  Table 3. Characteristics of the conventional ABR. used in improved ABR was not adjusted to 11.5, the same value as conventional ABR. We inoculated 36 L (60% of the working volume) seed sludge into the improved ABR. Similar to the conventional ABR operation, the improved ABR was divided into four phases: Phase I (Set-up phase, days 1-76); Phase II (Operation phase at high temperature, days 77-139); Phase III (Operation phase at moderate temperature, days 140-205); Phase IV (Operation phase at low temperature, day 206-303). More improved ABR operating information is listed in Table 6. The reactor was continuously fed with sweet potato wastewater at room temperature from the feed tank.
Analytical methods. To investigate each chamber's performance in both conventional and improved ABRs, wastewater samples were taken with three duplicates at the influent and the upper part of each internal chamber, and COD was measured. Sludge was taken from sample ports with three duplicates at the bottom of every compartment for total DNA extraction, qPCR and archaea community analysis. The amount of biogas generated in every compartment was measured by wet gas flow meter and the biogas CH 4 concentrations were measured by gas chromatography.
DNA extraction, quantification, and 16S rRNA gene sequences. Total

Statistical analysis. Phylogenetic analysis.
Raw reads obtained through high-throughput sequencing process were further treated for the purpose of an ecological analysis of archaeal populations. Data quality control and analyses were mostly performed using the QIIME (v1.8.0) 37 . Single-end reads were quality filtered with Trimmomatic tool using the following options: TRAILING:20, MINLEN:200 and CROP:200, to remove trailing sequences below a phred quality score of 20 38 . Overlapping pair-end reads were connected with COPE (V1.2.3.3), followed by detection of Chimeric sequences by USEARCH 39,40 . Operational taxonomic units (OTUs) were picked de novo from quality-filtered reads using a 97% similarity cut off and assigned to a taxonomic lineage using QIIME (v1.8.0).
Improved ABR net energy gain analysis. To assess the benefit of the improved ABR operated at ambient temperatures, the net energy gain (NEG) was adopted and calculated based on data from this study and fermentative biogas production researches at mesophilic or thermophilic temperatures 41 . The approach for assessing net energy gain introduced by Perera et al. is extended here to replace energy produced by hydrogen to methane generated in reactors. Thus, the theoretical net energy gain, E N [kJ/g COD in feedstock] in this study is defined as the total energy produced equivalent to methane volume generated in reactors, E CH 4