Introduction
Food waste constitutes much of the organic fraction of municipal solid waste (OFMSW). OFMSW is comprised in large part by organic solubles that can be easily converted to volatile fatty acids (VFA). This makes food waste an ideal substrate for biogas production. However, excessive VFA conversion at an early stage of digestion may cause a drastic drop in pH, inhibiting methanogenesis. Two-stage anaerobic digestion (AD) systems separate acid fermentation and methanogenesis for the purpose of optimizing reactor conditions for the distinctly different microbes that carry out these functions. The first stage (acid fermentation) is maintained at typically low hydraulic residence times (HRT; 2–3 days) resulting in a washout of the acid-consuming organisms and low pH (5–6). The second stage (methanogenesis) is operated at HRT of 20–30 days and pH of 6–8, facilitating proliferation of the slow-growing methanogenic archaea. Application of two-stage AD systems for food waste has proven effective for resolving the pH inhibition issues of one-stage systemsReference Cho, Park and Chang 1 – Reference Shen, Yuan, Pang, Chen, Zhu, Zou, Liu, Ma, Yu and Li 6 .
Methanogenic archaea population as well as community dynamics have been characterized during startup of AD reactorsReference Angenent, Sung and Raskin 7 – Reference Ike, Inoue, Miyano, Liu, Sei, Soda and Kadoshin 10 , as well as steady-state two-stage systemsReference Klocke, Nettmann, Bergmann, Mundt, Souidi, Mumme and Linke 3 , Reference Shin, Han, Lim, Lee and Hwang 4 . The results of these studies indicate that the overall process performance is relatively stable compared to the dynamic changes in the microbial community and the acidogenic performance. Also, they point to differences in the composition of the methanogenic archaea community within different sized two-phase biogas reactors; however, previous research has not compared methanogenic archaeal community structure between the two- and one-stage systems. They conclude that the effect of these different methanogenic populations on the stability of the entire process is of particular interest for the operation of biogas reactors.
Based upon the work above, the effect of transitioning from one-stage to two-stage AD on population and diversity of methanogenic archaea remains unclear. In this study, the population and diversity of methanogenic archaea during conversion of a full-scale one-stage anaerobic digester to a two-stage anaerobic digester treating food waste was explored and compared to digester performance (biogas production and methane content).
Materials and Methods
Analytical methods
Total solids (TS) and volatile solids (VS) were measured according to standard methodsReference Eaton, Clesceri, Rice, Greenberg and Franson 11 . Chemical oxygen demand (COD) was measured with a Hach DR2700 instrument (Hach, Loveland, CO) using Hach digestion solution vials for COD (20–1500 mg l−1) following the manufacturer's protocol. The pH of the samples was measured using a VWR symphony pH meter (VWR, Radnor, PA). Biogas production was measured automatically by an AC250 gas meter (Elster, Raleigh, NC) and recorded at one cubic foot accuracy every 30 min by the data recording software (Mango). A methane concentration sensor (BlueSens, Herten, Germany) was installed in the gas line to allow continuous monitoring of the methane concentration (±0.01%) in the produced biogas.
Reactor operation
The anaerobic digester on the Clarkson University campus treats pre-consumer food waste from the campus cafeterias, and is part of a sustainable year-round cold-climate aeroponics greenhouse and energy recovery system for production of leafy vegetables and food waste reduction. The digester system is comprised of three 5-m3 insulated stainless steel tanks (Fig. 1). A fourth 5.7-m3 polyethylene tank holds the effluent of tanks 2 and 3 until use on Clarkson grounds as a fertilizer. Each reactor is hydraulically mixed using a Moyno 500 grinder pump (Moyno, Springfield, OH)Reference Grimberg, Hilderbrandt, Kinnunen and Rogers 12 . Mixing frequency for each tank was 20 min every 4 h. Approximately 8.5 m3 of anaerobic digester sludge (15.5 g COD l−1, TS 1.24% and VS 0.74%) from the Potsdam Sewage Treatment Plant (Potsdam, NY) was used to seed the food waste digester, and the system was operated as a single-phase digester for 1 year prior to this study.
Figure 1. The AD system used in this study consisted of three reaction vessels. Reactor 1 was employed as a one-stage AD system for Phase I of this study. During Phase II, reactor 1 was transitioned to operation as a fermentation reactor and reactors 2 and 3 were used for methanogenesis.
The composition and amount of food waste fed to the digester during the course of the study varied with waste food production in the cafeterias, and ranged from 18 to 90 kg d−1. Average and standard deviations of feed characteristics were 273±164 g COD l−1, VS of 18.58±8.76% and VS/TS of 95.23±1.64%. Food waste fed to the reactor consisted mostly of fruit and vegetable matter, but also pasta, bread and meat.
Digester performance and archaea community structure were monitored during conversion of the mesophilic anaerobic digester system from a one-stage (Phase I) to two-stage (Phase II) operation (Fig. 1). One-stage operation was with an average HRT of 72±16 days (7-day running average). In two-stage operation, reactor vessel 1 was operated at 20% capacity as an acid fermentation reactor (hydrolysis, acidogenesis and acetogenesis; average HRT=39±18 days). Methanogenesis reactor vessels 2 and 3 received effluent from the fermentation reactor and operated in parallel and at 90% capacity (HRT=318±176 days). HRT of this reactor configuration was greater than that reported for typical one- and two-stage AD systems treating food waste (10–30 days for one-stage mesophilic AD systems, 2–3 days for the fermentation step of a two-stage AD systems and 20–30 days for methanogenesis in two-stage systemsReference Metcalf, Eddy and Tchobanoglous 13 ). During the transition between Phase I and Phase II operation, there was a heating system failure, resulting in reactor temperatures dropping temporarily to 20 °C for 26 days.
DNA extraction and real-time quantitative polymerase chain reaction (qPCR)
DNA was extracted from all samples using the PowerSoil® DNA Isolation Kit (MoBio, Carlsbad, CA) following the manufacturer's protocol, except that 20 μl salmon testis gDNA (Sigma Aldrich, St. Louis, MO) was added to each bead tube prior to bead milling, which served as an exogenous extraction and amplification control. The expected concentration of salmon testis gDNA in each extract was 44.6 ng μl−1, assuming 100% recovery and no PCR inhibition. Actual recovery of salmon testis gDNA was measured for each sample using Sketa qPCR, described previouslyReference Rogers, Donnelly, Peed, Kelty, Mondal, Zhong and Shanks 14 . Real-time qPCR assays were performed with a Roche LightCycler 480 (Roche, Basel, Switzerland). Reaction mixtures (25 μl) contained 12.5 μl of 2×LightCycler 480 Probe Master (Roche), 1.25 μl forward and reverse primers, 5.0 μl of fluorogenic probe, 5.0 μl template DNA and 5.0 μl PCR-grade water (Roche). Extraction blanks were used to test for the presence of extraneous DNA contamination introduced during laboratory procedures. Extracted DNA was stored frozen at −20 °C until use (less than 3 weeks).
Methanogens were quantified by real-time qPCR targeting the methyl-coenzyme-M reductase (mcrA) gene fragment of the MCR I operonReference Nettmann, Bergmann, Mundt, Linke and Klocke 15 . MCR catalyses the reduction of methyl coenzyme-M (CH3-S-CoM) with coenzyme B (HS-CoB) to methane (CH4) and CoM-S-S-CoB under strictly anaerobic conditions, and is the key enzyme of methanogenesis. There are two forms of the MCR operon (MCRI and MCRII); MCRI is believed to be conserved in all methanogens whereas MCRII has only been shown to be present in members of the orders Methanobacteriales and Methanococcales Reference Luton, Wayne, Sharp and Riley 16 .
Real-time qPCR SYBR Green I assays were run on a Roche LightCycler® 480 (Roche). Reaction mixes included 5 μl template DNA, 12.5 μl Lightcycler® 480 SYBR Green I Master (10,000× concentration; Invitrogen, Carlsbad, CA), 1.25 μl ML forward and ML reverse primers described previouslyReference Nettmann, Bergmann, Mundt, Linke and Klocke 15 , Reference Freitag and Prosser 17 , 5.0 μl template DNA and 5 μl PCR-grade water (Roche). qPCR was carried out on 96-well reaction plates using a hot-start protocol. An initial denaturation step of 15 min at 95 °C was followed by 40 cycles of denaturation for 30 s at 95 °C, annealing for 45 s at 55 °C, extension for 45 s at 68 °C, fluorescence reading was taken at the end of 8 s at 79 °C step to allow dissociation of possible primer dimers and unspecific amplification products. The qPCR was finalized by an extension step at 68 °C for 10 min and the subsequent melt curve analysis for confirmation of amplicon specificity by verification of identical, clearly defined melting peaks. Melting curve analysis was obtained by increasing the temperature from 55 to 95 °C gradually with the fluorescence signal being measured every 0.5 °CReference Freitag and Prosser 17 .
Standard curves for qPCR were constructed from a 701 bp PCR amplicon spanning the mcrA region of genomic DNA of Methanosarcina barkeri (DSM 804) using the primers M. barkeri forward and reverse described previously by Freitag and ProsserReference Freitag and Prosser 17 . PCR products were processed using an UltraClean® PCR Clean-Up Kit (MoBio) and quantified using Quant-iT™ dsDNA HS Assay Kits (Invitrogen) following the manufacturers’ instructions. Verification of the mcrA fragment was completed by electrophoresis in 2.2% Agarose FlashGel™ DNA cassettes (Lonza, Basel, Switzerland) at 275 V for 5 min according to manufacturer's instructions. Standard curves for mcrA qPCR spanned from 10 DNA copies to 1×107 DNA copies and were run in triplicate on each 96-well qPCR reaction plate.
Denaturing gradient gel electrophoresis (DGGE)
DGGE analysis of the hypervariable region 3 (V3) in archaeal 16S rRNA was used to investigate the diversity of archaea in the anaerobic digesters during the course of study. Nested PCR was used to prepare the gene products for DGGE analysis. The first PCR amplification was with the primer set Arch349F/Arch806R previously described by Takai and HorikoshiReference Takai and Horikoshi 18 and included 10 min of initial denaturation at 95 °C followed by 35 cycles of denaturation at 95 °C for 25 s; annealing 58 °C 1 min, and a single final extension for 5 min at 72 °C. The 500 bp amplicons were used as template DNA for the second round of amplification with the primer set PARCH340f and PARCH519r targeting archaea V3 region as described by Overås et al.Reference Ovreås, Forney, Daae and Torsvik 19 A 40 bp G+C-rich sequence was included on the 5′ end of the forward primerReference Ovreås, Forney, Daae and Torsvik 19 . Both reaction mixes included 12.5 μl LightCycler® 480 Probes Master (Roche), 1.25 μl forward and reverse primers, 5.0 μl template DNA and 5 μl PCR-grade water (Roche).
The nested PCR products were electrophoresed in 2.2% Agarose FlashGel™ DNA cassettes (Lonza) at 275 V for 5 min to confirm the successful amplification of targeted fragment (200 bp). Successfully amplified samples were analyzed by DGGE. PCR products were loaded on 8% polyacrylamide denaturing gel with a denaturing gradient of 30–70% top to bottom (100% denaturing solution includes 7 M urea and 40% formamide). Electrophoresis was run for 14 h at 80 V in 1×TAE buffer using a D-code electrophoresis system (BioRad, Hercules, CA). The gel was stained for 30 min in 1× SYBR Gold staining solution (Invitrogen) to observe the bands by ultra-violet transillumination.
At least two different DGGE runs were carried out for all samples and for both loading orders of the samples on the gel, in order to estimate the reproducibility of the statistical analysis of the DGGE profiles generated with different loading schemes of samples.
DGGE fingerprint analysis
All DGGE fingerprint images were imported to GelCompar II software version 6.6 (Applied Maths, Austin, TX) for cluster analysis of fingerprint similarity. The fingerprints were normalized using the DGGE standard marker as an external reference. Cluster analysis was performed on the resulting similarity matrix using the unweighted pair group method with arithmetic means algorithm, resulting in dendrograms that graphically displayed the similarities among fingerprints.
The diversity of archaea in each phase was determined by calculating the Shannon index (H′). The Shannon–Weaver diversity indexReference Shannon and Weaver 20 was calculated as:
$$ H{\prime } - \sum\limits_{i = 1}^S {\,pi\,{\rm ln}\, (\,pi )} $$
where pi is the proportion of bands present in the samples and S is the total number of bands in each research phase. Shannon diversity index is a quantitative measure that reflects how many different types, in this case species, represented by separation of DGGE bands, there are in a dataset. It also takes into account how evenly the basic entities are distributed among the types. The value increases when the number of types increases and when the evenness increases.
Statistical analysis
The aim of the statistical analysis was to ascertain the correlations between the population of methanogens, as represented by mcrA gene copy number, and the methane yield [liters CH4 per kg VS (L-CH4 (kg-VS)−1) added to the digester]. The initial data analysis using Anderson–Darling test showed that the variables (methane yield and the methanogens concentration) do not follow a normal distribution, thus statistical analysis was performed using non-parametric tests. Spearman's rank correlation coefficients were used to determine whether a relationship between the methane yield and mcrA gene copy number existed. Pair-wise comparisons of mcrA between different reactors and the difference in methane yield during one- and two-stage operation were performed using the Mann–Whitney test. The 95% confidence intervals are constructed and the default value for level of significance was taken as 0.05 (α=0.05). All statistical tests were run in Minitab 15 (Minitab Inc, State College, Pennsylvania).
Results and Discussion
Effect of digester operation on the concentration of methanogens
In this study, the microbial community within a two-stage anaerobic food digester was quantified by measuring the presence of the functional gene, mcrA. The study consisted of two operational phases: a one-stage and a two-stage system. The characteristics of each phase are shown in Table 1, Figures 2 and 3. During the single phase operation mcrA concentrations, biogas production and VS loading rate qualitatively correlate well (Fig. 2) while during the two-stage operation biogas production and mcrA concentrations remain relatively constant and do not respond to temporal variation in the VS loading rate (Fig. 3). This is presumably due to the low loading rates and their relatively short temporal variations.
Figure 2. Performance of the one-stage AD reactor; (a) concentration of the mcrA gene; (b) biogas production; (c) VS loading rate.
Figure 3. Performance of the two-stage AD reactor; (a) concentration of the mcrA gene; (b) biogas production; (c) VS loading rate.
Table 1. Characteristics of the two research phases. T 1 represents the fermentation step; T 2 and T 3 represent the methane-producing step of a two-stage system.
The running average HRT during the first phase was 72±16 days (Fig. 4a). Due to the slow growth rate of methanogens, reactors operated at HRT suitable for acetogens (lower than 20 days) inhibit the accumulation of methanogens, and this is one of the limiting factors influencing the community composition in one-stage digestersReference Metcalf, Eddy and Tchobanoglous 13 . The concentration of mcrA during one-stage operation was 2.48×109 copies ml−1. The average methanogens concentration in one-stage mesophilic digesters reported by others ranged from 105–108 copies ml−1 Reference Cho, Lee, Kim and Hwang 9 , Reference Bialek, Kim, Lee, Collins, Mahony and O'Flaherty 21 . Greater than typical HRT almost certainly led to higher mcrA concentrations in the one-stage digester of this study as compared to that reported by others.
Figure 4. HRT (days) during (a) one-stage operation; (b) two-stage operation calculated as a 7-day running average.
During transition from one- to two-stage operation, a heating system malfunction for 26 days led to a drop in the average reactor temperatures to 20±2.09 °C over the course of 3 days. During this transition, the digester was operated as a two-stage system with no heating control. The average biogas flow rate decreased to 1766 liters d−1 with the average methane concentration of 52±5.5%, slightly lower than one-stage operation. Average normalized CH4 production decreased to 125±43 L-CH4 (kg-VS)−1 destroyed, less than one-third that at mesophilic temperatures.
Following transition, a two-stage mesophilic condition was established (37±1.9 °C). Average biogas flow rate increased to 3653 liters d−1 and the average methane concentration reached 61±6.4%. Normalized CH4 production was 446 L-CH4 (kg-VS)−1 added. HRT of 39 days in the fermentation reactor led to a drop in the pH to 6.71, inhibiting the growth of methanogens only slightly. Conversely, separation of acid fermentation from methanogenesis facilitated an increase in mcrA in the methanogenesis reactor vessels (Stage 2) to 1.76×1010 copies ml−1 (Fig. 3a), significantly greater than one-stage operation (P=0.0026). Biogas flow rate of 3653 liters d−1 (P=0.0505) and methane content 61% (P=0.0018) during two-stage operation were also significantly greater than during one-stage operation. Similar observations as those above were reported by othersReference Shen, Yuan, Pang, Chen, Zhu, Zou, Liu, Ma, Yu and Li 6 , Reference Lin, Zuo, Ji, Chen, Liu, Wang and Yang 22 , Reference Rincón, Borja, Martín and Martín 23 .
A fivefold increase in mcrA gene concentration during two-stage operation corresponded to the increase in methane production rates per kg COD destroyed (Table 1). Freitag and Prosser, as well as Lv et al.Reference Freitag and Prosser 17 , Reference Lv, Zhang and Yu 24 reported that mcrA gene abundance did not correlate to methane yield in their studies. Conversely, some researchers have demonstrated a positive and significant correlation between the rate of biogas production and methanogen abundance in one-stage digestersReference Traversi, Villa, Lorenzi, Degan and Gilli 25 , Reference Singh, Singh, Upadhyay, Joshi, Tripathi and Dubey 26 . Similar to our study, others have related that the performance of an anaerobic digester is positively influenced by both the relative abundance of microbial populations and the community compositionReference Lin, Zuo, Ji, Chen, Liu, Wang and Yang 22 , Reference Traversi, Villa, Lorenzi, Degan and Gilli 25 , Reference Lee, Behera, Kim and Park 27 , Reference Williams, Williams, Dinsdale, Guwy and Esteves 28 . The measurement of mcrA gene regulation by real-time reverse transcriptase qPCR would almost certainly provide a better measure on methanogen activity and may better correlate to methane production than functional gene presence and community diversity.
Shifts in microbial community diversity
Highly diverse ecosystems, such as activated sludge, sediments and soils, have DGGE banding patterns that are very complex to interpretReference Boon, Windt, Verstraete and Top 29 . To better understand the profile, computer-aided analysis methods are necessary to study the patterns.
The DGGE patterns obtained with archaea-specific primers did not yield many intense bands; however, a few dominant bands were present in some samples. The DGGE patterns of the archaeal community profile formed three distinct clusters by research phase. Clustering of DGGE patterns from the transition phase (between Phases I and II) was distinct from one-stage and two-stage operation, likely a result of the temporary decrease in temperature in the reactor vessels.
Based on the DGGE banding patterns, the Shannon diversity index H was calculated. During one-stage operation, H was 2.98, much less than that during the two-stage operation (H=7.29). These results are consistent with those published previously. For example, the Shannon index for methanogenic archaea in a mesophilic, high cell density hybrid anaerobic reactor was previously reported to be 7.76Reference Kundu, Sharma and Sreekrishnan 30 . Other findings report Shannon index ranging from 2.35 to 2.84 for one-stage mesophilic anaerobic reactorsReference Klocke, Nettmann, Bergmann, Mundt, Souidi, Mumme and Linke 3 , Reference Nettmann, Bergmann, Mundt, Linke and Klocke 15 , Reference Peng, Song, Wang, Yuan and Liu 31 . In a study by Lim et al.Reference Lim, Chen, Ho and Wang 32 , higher diversity in the two-stage methanogenic reactors than in the one-stage reactors led to 23% greater methane production. The main reason for greater diversity, and possibly increased methane production, of the two-stage system was greater solubilization (represented by the reduction of solids) and acidification in the two-stage digester. These researchers also suggested that greater populations of hydrogen-utilizing methanogens were present in the methanogenic reactor of the two-stage system, leading to greater hydrogen production supporting the growth of hydrogen-utilizing methanogens. Therefore, a two-stage design may help balance acetogenic and methanogenic microorganisms; should hydrogen gas levels rise within the one-stage system, the balance will be disrupted and the methane yields decreased. However, during this study, the hydrogen production was not measured.
The number and intensity of bands in the DGGE gels do not necessarily provide an accurate picture of the number and abundance of corresponding species within the microbial community. According to previous researchReference Boon, Windt, Verstraete and Top 29 , one organism may produce more than one DGGE band because of multiple, heterogeneous rRNA operons. Therefore, the banding pattern only reflects the most abundant rRNA types. The diversity index calculated from the DGGE banding patterns of amplified 16S rRNA sequences must be taken only as an indication of diversity, and not the absolute measure of diversity, in a microbial community.
This study compared the populations and diversity of methanogens involved in AD of food waste during one- and two-stage operations. Significantly higher methane yields as well as archaea diversity were observed during the two-stage operation. Furthermore, the concentration of functional mcrA gene was determined to be greater in the two-stage operation. The results of this study suggest that using a two-phase design for AD of food waste results in higher methane yields than in a one-stage design.
