Bioresource Technology xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Bacterial and methanogenic archaeal communities during the single-stage anaerobic digestion of high-strength food wastewater Hyun Min Jang a, Ji Hyun Kim a, Jeong Hyub Ha a, Jong Moon Park b,c,⇑ a

School of Environmental Science and Engineering, Pohang University of Science and Technology, San 31, Hyoja-dong, Pohang 790-784, South Korea Department of Chemical Engineering, Pohang University of Science and Technology, San 31, Hyoja-dong, Pohang 790-784, South Korea c Division of Advanced Nuclear Engineering, Pohang University of Science and Technology, San 31, Hyoja-dong, Pohang 790-784, South Korea b

h i g h l i g h t s

g r a p h i c a l a b s t r a c t

 Applicability of AD process was

evaluated for the treatment of highstrength FWW.  Microbial communities were examined by barcodedpyrosequencing and qPCR.  Bacterial communities were highly affected by the change of organic loading rate.  Methanogenic species shifted from aceticlastic to hydrogenotrophic methanogens.

a r t i c l e

i n f o

Article history: Available online xxxx Keywords: High-strength food wastewater Single-stage anaerobic digestion Barcoded-pyrosequencing Quantitative real-time PCR (qPCR) Multivariate statistical analysis

a b s t r a c t Single-stage anaerobic digestion (AD) was operated to treat high-strength food wastewater (FWW) derived from food waste recycling facilities at two different organic loading rates (OLRs) of 3.5 (Phase I) and 7 (Phase II) kg COD/m3 d. Changes in composition of microbial communities were investigated using quantitative real-time PCR (qPCR) and barcoded-pyrosequencing. At the high FWW loading rate, AD showed efficient performance (i.e., organic matter removal and methane production). Bacterial communities were represented by the phyla Bacteroidetes, Firmicutes, Synergistetes and Actinobacteria. During the entire digestion process, the relative abundance phylum Chloroflexi decreased significantly. The qPCR analysis demonstrated that the methanogenic communities shifted from aceticlastic (Methanosarcinales) to hydrogenotrophic methanogens (Methanobacteriales and Methanomicrobiales) with high increase in the proportion of syntrophic bacterial communities. Canonical correspondence analysis revealed a strong relationship between reactor performance and microbial community shifts. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction Food wastewater (FWW) is an organic-rich byproduct that is produced in large quantities during recycling of food waste (FW). Korea alone produces more than 9000 tons FWW/d (MOE, 2012). Before 2013, most FWW was discarded by ocean dumping, but this

⇑ Corresponding authors. Tel.: +82 54 279 2275; fax: +82 54 279 8659. E-mail address: [email protected] (J.M. Park).

practice was banned by the London Convention 97 protocol in January of 2013. Thus, treatment of FWW is very important for environmental protection and development of appropriate treatment technology has become an urgent task. Anaerobic digestion (AD) is widely used to treat organic wastes. Single-stage continuous mesophilic AD reactors are the most attractive for decentralized, medium or small-scale AD because they have high process stability and low cost, and because they do not require specialized operating skills. Meanwhile, the AD is performed by diverse microorganisms involved in serial

http://dx.doi.org/10.1016/j.biortech.2014.02.028 0960-8524/Ó 2014 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Jang, H.M., et al. Bacterial and methanogenic archaeal communities during the single-stage anaerobic digestion of highstrength food wastewater. Bioresour. Technol. (2014), http://dx.doi.org/10.1016/j.biortech.2014.02.028

2

H.M. Jang et al. / Bioresource Technology xxx (2014) xxx–xxx

biochemical reactions, i.e., hydrolysis, acidogenesis, acetogenesis and methanogenesis. Therefore, an improved understanding of microbial communities in AD would be useful to optimize reactor performance (i.e., organic matter removal and methane production). The microbial community structure in AD can be quantified rapidly and inexpensively using quantitative real-time polymerase chain reaction (qPCR) and pyrosequencing. qPCR enables real-time detection and quantification of specific sequences in a DNA sample, either as an absolute number of copies or as a relative amount. Pyrosequencing enables the sequencing of thousands to millions of molecules in parallel in a single run, thereby enabling inexpensive investigation of microbial community changes during biological processes (Kim et al., 2013; Lee et al., 2012b; Sundberg et al., 2013). Furthermore, with the help of specific barcode, highthroughput sequences of multiple samples are available in parallel to more fully examine various microbial species (Parameswaran et al., 2007). In this study, barcoded pyrosequencing was used to identify key phylotypes of bacteria, and a methanogen-specific qPCR technique (Jang et al., 2014) was used to quantify 16S rRNA gene copy numbers of total bacteria, archaea and four major methanogen orders in the single-stage AD of high-strength FWW. Multivariate statistical analysis (canonical correspondence analysis; CCA) and microbial community comparison (UniFrac hierarchical clustering and principal coordinated analysis; PCoA) were applied to investigate the relationship between reactor performance and microbial community structure. 2. Methods 2.1. Preparation of feedstock The feedstock was collected from a FW recycling facility in Pohang, Korea. About 50 tons/d of FWW has been generated in this facility. We proceeded pretreatment and storage directly after sampling from this facility as described in previous research (Jang et al., 2013). In addition, detailed physico-chemical characteristics of the FWW used in this study are presented in Table 1.

with working volume of 6 L and mesophilic condition (35 ± 0.2 °C) (Fig. 1). The reactor was seeded with mesophilic anaerobic sludge (Table 1) taken from a successfully-operated full-scale mesophilic anaerobic plant (only treated sewage sludge) in Daegu, Korea. The reactor was fed four times a day using a peristaltic pump (Cole-ParmerÒ) controlled by a timer and relay, and FWW was applied at two different HRTs (40-d, 20-d) at corresponding organic loading rates (OLRs) of 3.5 and 7 kg COD/m3 d sequentially over a period of 204 days. 2.3. Physico-chemical analysis As described in Standard Methods (APHA-AWWA-WEF, 1998), total solids (TS), volatile solids (VS), total chemical oxygen demand (TCOD), total alkalinity (TA), total nitrogen (TN) and total phosphorus (TP) were measured in samples from the reactor at intervals of 3 d. The concentration of soluble organic matter including soluble COD (SCOD), ammonia (NHþ 4 –N), soluble TN (STN), and soluble TP (STP) were measured after filtering through a 0.45-lm pore-size filter (Whatman, USA). The concentrations of nitrite ðNO 2 Þ, nitrate 3 ðNO 3 Þ, and orthophosphate (PO4 –P) were determined using an ion chromatograph (ICS-1000, DIONEX Co., USA). The pH and oxidation reduction potential (ORP) in each reactor were continuously measured and recorded using a pH meter (405-DPAS-SC-K85, METTLER TOLLEDO, Switzerland) and ORP meter (Pt-4805, METTLER TOLLEDO, Switzerland). After filtering step through a 0.22-lm pore-size filter (Whatman, USA), organic acids were quantified using a high performance liquid chromatograph (HPLC, Agilent Technology 1100 series, Agilent Inc., USA) equipped with an organic acid and alcohol analysis column (Aminex HPX-87H, BIORAD Inc., USA), a refractive index detector (RID), and a diode array detector (DAD). Biogas volume from the reactor was quantified using a water displacement method, and composition was detected using a gas chromatograph (Model 6890 N, Agilent Inc., USA) equipped with a pulsed discharged detector (PDD). The quantified organic acid values were converted theoretically to g COD/L by using the following conversion factors: lactic acid: 1.07, acetic acid: 1.07, propionic acid: 1.51, butyric acid: 1.82, succinic acid: 0.95. 2.4. Microbial community and population analysis

2.2. Reactor operation To evaluate reactor performance and microbial communities, a lab-scale single-stage anaerobic reactor was used; it was operated

2.4.1. Genomic DNA extraction and qPCR analysis Total genomic DNA from samples (Seed, Phases I and II) was extracted and stored as described previously (Jang et al., 2014). To

Table 1 Physico-chemical characteristics of Seed and Feedstock used in this study. Parameter

Seed (Anaerobic sludge)

Feedstock (Food wastewater)

pH TS (g/L) VS (g/L) TCOD (g/L) SCOD (g/L) TN (g N/L) NHþ 4 –N (g-NH3/L)  NO 2 (g NO2 -N/L)  NO3 (g NO 3 -N/L) Total organic acid (g COD/L) Lactic acid (g COD/L) Acetic acid (g COD/L) Propionic acid (g COD/L) Butyric acid (g COD/L) Succinic acid (g COD/L)

7.42 (0.25) 24.4 (0.21) 12.4 (0.31) 16.88 (0.12) 0.52 (0.03) 4.38 (0.45) 0.98 (0.04) – – –

4.31 (0.02) 118.49 (3.54) 106.52 (3.41) 139.58 (2.79) 90.46 (1.53) 1.94 (0.15) 0.57 (0.03) – – 69.47 (1.46)

– – – – –

40.74 (1.04) 14.35 (0.59) 3.32 (0.31) 9.94 (0.51) 1.12 (0.04)

Values are expressed as average (standard deviations). ‘–’: not detected.

Fig. 1. Schematic diagram of lab-scale single-stage AD process for high-strength FWW.

Please cite this article in press as: Jang, H.M., et al. Bacterial and methanogenic archaeal communities during the single-stage anaerobic digestion of highstrength food wastewater. Bioresour. Technol. (2014), http://dx.doi.org/10.1016/j.biortech.2014.02.028

H.M. Jang et al. / Bioresource Technology xxx (2014) xxx–xxx

quantify 16S rRNA gene copy numbers of total bacteria, archaea and four major methanogen orders (Methanobacteriales, MBT; Methanococcales, MCC; Methanomicrobiales, MMB; and Methanosarcinales, MSL), qPCR amplification and fluorescence detection were conducted using an Applied Biosystems 7300 real-time PCR system (Applied Biosystems, Forster City, USA) with group and order specific primer sets and representative strain reported in previous research (Jang et al., 2014). 2.4.2. Pyrosequencing analysis To investigate in-depth bacterial communities, hypervariable regions within bacterial 16s rRNA genes were amplified with universal primers (Table 2): Bac27F (50 -adaptor A-Barcode-AC-GAG TTT GAT CMT GGC TCA G-30 )/Bac541R (50 -adaptor B-Barcode-ACWTT ACC GCG GCT GCT GG-30 ) (Lee et al., 2014). PCR amplification was conducted using a FastStart High Fidelity PCR system (Roche, Branford, CT); the protocol was (1) initial denaturation at 94 °C for 4 min; (2) 35 cycles of 94 °C for 15 s, 55 °C for 45 s, and 72 °C for 1 min; (3) a final extension at 72 °C for 8 min. All PCR products were purified using a PCR purification kit (Solgent, Korea) and concentrations of the nucleic acid were measured using a fluorometer with Quant-iT™ PicoGreenÒ dsDNA Assay Kit (Invitrogen TM, California). After pooling of equal amount of PCR products from each sample, their sequencings were performed with 454-GS-FLX Titanium (Roche, Branford) using the massively parallel pyrosequencing protocol by a sequencing company (Macrogen, Korea). 2.4.3. Pyrosequencing data and multivariate statistical analysis The raw 16S rRNA gene sequences of bacteria obtained from pyrosequencing were initially sorted using the RDP Pyrosequencing Pipeline Initial Process (http://pyro.cme.msu.edu/) based on the barcode and filter criteria as follows: maximum number (N) of ambiguous bases = 1, minimum read quality score = 20 and minimum sequence length = 400 bp. Then adapters, barcodes and primers in all raw sequence were trimmed to reduce errors of clustering. Extraction of chimeric sequences was conducted using the MOTHUR program (www.mothur.org/) (Lee et al., 2014). The multiple clean sequences were aligned using the fast, secondary-structure aware INFERNAL aligner and were clustered into operational taxonomic units (OTUs) defined by 3% max distance (97% similarity) using complete linkage-clustering method provided in RDP pyrosequencing pipeline. The Shannon–Weaver index (Shannon and Weaver, 1963), Chao1 richness index (Chao and Bunge, 2002) and Evenness were calculated using clustered sequence data using the Shannon Index and Chao1 estimators of the RDP pyrosequencing pipeline. Rarefaction curves were also constructed with a 3% dissimilarity cut-off value using the RDP pyrosequencing pipeline. The taxonomic classification of clean sequences obtained from each sample was conducted using the RDP classifier 2.5 trained on 16S rRNA training set 9 (Wang et al., 2007) with an 80% confidence cut-off. The comparison of microbial communities was conducted using UniFrac analysis (http://unifrac.colorado.edu/) based on the phylogenetic information as described previously (Lee et al.,

Table 2 Adaptor and barcode sequences used in the pyrosequencing. Name

Sequence (50 –30 )

Adaptor sequences Adaptor A Adaptor B

CCTATCCCCTGTGTGCCTTGGCAGTCTCAG CCATCTCATCCCTGCGTGTCTCCGACTCAG

Bacterial barcode sequences Seed (0 d) Phase I (0–102 d) Phase II (102–204 d)

CATGCTC ATACGTACG AGACAGTACAG

3

2012a). Briefly, clean sequences were clustered into OTUs by using CD-HIT (www.bioinformatics.org/cd-hit/) and representative sequences from CD-HIT were aligned using NAST based on the Greengenes database. Then, a phylogenetic tree was constructed using PHYLIP software and the Kimura two-parameter model for the hierarchical clustering of bacterial communities in the unweighted UniFrac analysis. Also PCoA was performed to confirm the multiple communities identified by the UniFrac analysis. To evaluate the correlations, a multivariate CCA of the relative abundance of microbial communities, reactor parameters and reactor performance was performed using R software (http://cran. r-project.org/) with the vegan library. 3. Results and discussion 3.1. Reactor performance The reactor showed stable performance and methane production at 40-d and 20-d HRTs over 204 days (Fig. 2). In general, loading of high-strength feedstock on the biological reactor is a major cause of the reactor performance deterioration due to increase of environment stress, especially unfavorable pH condition (Salminen and Rintala, 2002). Although the feedstock used in this study had low pH (4.3), pH in the reactor was maintained between 6.75 and 7.33 (Fig. 2a), which is known to be a favorable pH range for organic matter degradation and methane production during single-stage AD. Also, sufficient total alkalinity which is regarded as buffering capacity in the AD was maintained in the range of 2.23 to 3.29 g CaCO3/L throughout digestion. When HRT was decreased from 40-d (Phase I) to 20-d (Phase II) (OLR increased from 3.5 to 7 kg COD/m3 d), solid (TS and VS) and total organic matter (TCOD) concentrations in the reactor increased slightly (Fig. 2b and c), but the efficiencies of solid and organic matter removal remained stable: 63–68% of TS, 77–80% of VS and 71–78% of TCOD were removed during the overall digestion. The FWW contained abundant organic matter, in both solid and soluble forms (SCOD) (Table 1). Of the SCOD, organic acid (converted to g COD/L) accounted for over 76%; over 58% was lactic acid, which is produced by partial lactic acid fermentation during the recycling process. Overall, significant SCOD removal efficiency (over 96%) was observed (Fig. 2c) and lactic acid was not detected in the reactor (Fig. 2d). During Phase I, most of the organic acids were consumed and only acetic acid (