Bioresource Technology 193 (2015) 53–61

Contents lists available at ScienceDirect

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

Anaerobic treatment of rice winery wastewater in an upflow filter packed with steel slag under different hydraulic loading conditions Yeadam Jo a, Jaai Kim a, Seokhwan Hwang b, Changsoo Lee a,⇑ a b

School of Urban and Environmental Engineering, Ulsan National Institute of Science and Technology (UNIST), 50 UNIST-gil, Eonyang-eup, Ulju-gun, Ulsan 689-798, Republic of Korea School of Environmental Science and Engineering, POSTECH, Pohang, Gyungbuk 790-784, Republic of Korea

h i g h l i g h t s  A UAnF packed with blast-furnace slag achieved effective anaerobic biomethanation of RWW.  The UAnF reactor retained stable performance at 10–2.2-day HRTs with 84–92% COD removal.  The maximum methane yield of 0.33 L/g CODremoved was achieved at 3.4-day HRT with >85% COD removal.  The substrate removal and methane production kinetics of the UAnF were successfully evaluated.  Methanosaeta strains likely formed the dominant methanogen group in the filter biofilm.

a r t i c l e

i n f o

Article history: Received 29 April 2015 Received in revised form 6 June 2015 Accepted 10 June 2015 Available online 17 June 2015 Keywords: Blast furnace slag Kinetic evaluation Microbial community structure Rice-washing drainage Upflow anaerobic filter

a b s t r a c t Rice-washing drainage (RWD), a strong organic wastewater, was anaerobically treated using an upflow filter filled with blast-furnace slag. The continuous performance of the reactor was examined at varying hydraulic retention times (HRTs). The reactor achieved 91.7% chemical oxygen demand removal (CODr) for a 10-day HRT (0.6 g COD/L d organic loading rate) and maintained fairly stable performance until the HRT was shortened to 2.2 days (CODr > 84%). Further decreases in HRT caused process deterioration (CODr < 50% and pH < 5.5 for a 0.7-day HRT). The methane production rate increased with decreasing HRT to reach the peak level for a 1.3-day HRT, whereas the yield was significantly greater for 3.4-day or longer HRTs. The substrate removal and methane production kinetics were successfully evaluated, and the generated kinetic models produced good performance predictions. The methanogenic activity of the reactor likely relies on the filter biofilm, with Methanosaeta being the main driver. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Rice is a staple food for approximately half of the world’s population and among the most widely grown crops worldwide. Rice is used not only for direct consumption but also for making a variety of processed foods, such as noodles, snacks, sweeteners, thickeners, and alcoholic beverages, owing to its high starch content. Rice should first be cleaned by repeated washing prior to use in food processing, and thus a large quantity of organic-rich rice-washing drainage (RWD) is produced from this step. Proper treatment of RWD is a costly operation that often involves highly energy-intensive facilities, such as filter presses or spray dryers, owing to its high organic load (Watanabe et al., 2009). Anaerobic digestion (AD) often provides a viable alternative to aerobic ⇑ Corresponding author. Tel.: +82 52 217 2822; fax: +82 52 217 2819. E-mail address: [email protected] (C. Lee). http://dx.doi.org/10.1016/j.biortech.2015.06.046 0960-8524/Ó 2015 Elsevier Ltd. All rights reserved.

treatment for handling high-strength organic wastes because the operating costs can be reduced by producing biogas (mainly methane), eliminating aeration, and generating less sludge. Conventional anaerobic processes for treating organic waste streams have mostly employed a continuously stirred tank reactor (CSTR) mode. Although simple and easy to operate, a CSTR configuration has some limitations related to the slow growth of anaerobic microorganisms, particularly methanogens, such as low reaction rate and long hydraulic retention time (HRT). Extensive efforts have therefore been directed towards retaining a higher biomass density by decoupling the HRT and solids retention time (SRT), leading to the development of various high-rate anaerobic reactors, such as the upflow anaerobic sludge blanket (UASB), expanded granular sludge bed (EGSB), anaerobic filter, anaerobic baffled reactor, and membrane bioreactor (de Lemos Chernicharo, 2007). Such ‘high-rate’ anaerobic reactors are generally able to handle a higher organic loading rate (OLR), i.e., a shorter HRT,

54

Y. Jo et al. / Bioresource Technology 193 (2015) 53–61

compared to a CSTR but are sensitive to clogging and fouling from the deposition of suspended particles (de Lemos Chernicharo, 2007). Therefore, careful consideration of the wastewater characteristics is required, particularly for a process treating a waste stream of high suspended solids content. An anaerobic filter is an attached-growth process that immobilizes microorganisms on the surface of the packing material to build up biofilms (de Lemos Chernicharo, 2007). Therefore, how to effectively capture and retain active biomass within the filter matrix is a critical factor that decides the performance and robustness of an anaerobic filter reactor. Various natural and man-made materials, for example, pebbles, shells, wooden blocks, rubber sheets, plastic rings, granulated activated carbon, and tire rubber, have been used as packing material for anaerobic filters that treat various wastewaters (Loupasaki and Diamadopoulos, 2013). Using an appropriate packing material is important for the performance of anaerobic filters as the physicochemical characteristics of the filter media, e.g., the porosity and specific surface area, have significant effects on the biomass attachment. Increasing attempts have recently been made to recycle industrial by-products or wastes as reactor packing material and thereby to make the process more economically and environmentally feasible (Loupasaki and Diamadopoulos, 2013). This study has focused on the potential of granulated blast-furnace slag (BFS), a by-product derived from iron ore processing for steel manufacture, as a filter medium for bioreactors. In addition to its long-standing use as a secondary cementing material, BFS has been used in a number of other applications, such as road construction, landfill cover, sewage trickling filter media, and contaminant sorbent (Proctor et al., 2000; Claveau-Mallet et al., 2013; Loupasaki and Diamadopoulos, 2013). BFS has a limited leachability of heavy metals and does not pose an environmental concern (Proctor et al., 2000). Additionally, its porous and bulky structure offers a large specific surface area to support cell attachment for biofilm growth (Akbarnejad et al., 2014). These features, together with its inexpensive price, show the potential of BFS as an attractive reactor packing material. Although BFS has often been used as a packing material for aerobic filter processes and constructed wetlands (Chaudhary et al., 2003; Korkusuz et al., 2005; Loupasaki and Diamadopoulos, 2013), its use in AD processes as a filter medium has been little studied. In this study, to address this shortfall, an upflow anaerobic filter (UAnF) packed with BFS grains was examined for continuous biomethanation of RWD from a local rice wine factory. The reactor performance was monitored at varying HRTs in terms of organic removal and methane production during a long-term operation for 380 days. For better comprehension of the reactor system, analyses of the process kinetics and the microbial community structure were also performed.

2. Methods

Table 1 Physicochemical characteristics of rice washing wastewater.

a b

Unit

Value

pH Total COD Soluble COD Total solids Volatile solids Total nitrogen Total phosphorus Carbon Hydrogen Oxygen Nitrogen Sulfur

g/L g/L g/L g/L mg/L mg/L % dwa % dw % dw % dw % dw

5.14 13–16 3.7–7.1 11.8–13.7 10.1–12.3 215–329 220–360 42.8 (0.52)b 8.68 (0.53) 35.2 (0.21) 2.62 (0.03) 0.22 (0.04)

dw, dry weight. Standard deviations are in parentheses.

2.2. Reactor setup and operation A CSTR with a 2-L working volume and a UAnF with a 1.4-L working volume were run in parallel to anaerobically treat RWD. Both reactors were made of glass and sealed with gas-tight fittings. RWD was fed to the reactors after being diluted to a COD concentration of 6.0 ± 0.2 g/L. The CSTR had a height-to-diameter ratio of approximately 1, and the UAnF was made in a column that had an internal diameter of 9 cm and a height of 60 cm. The UAnF was filled with BFS grains of 5–10 mm in size, which had been repeatedly washed with water, to a height of 36.5 cm, and a perforated steel plate was placed at the bottom to keep the packing media in position and to evenly distribute the influent flow. The BFS grains, slowly cooled with ambient air, were obtained from a local steel manufacturer and composed mostly (>97%) of nonmetallic CaO, SiO2, Al2O3, and MgO. The basic characteristics of the BFS grains used are shown in Table 2. The operational liquid height was set to 50 cm to have a working volume of 1.4 L. The effluent was recirculated to the reactor inlet at a flow rate of 10 mL/min by peristaltic pumping. The schematic diagram of the UAnF system is illustrated in Fig. 1. Both the CSTR and UnAF were inoculated with the anaerobic seed sludge at a volume ratio of 1:1 and operated at 35 ± 2 °C. The influent pH was adjusted to neutral (7.0 ± 0.3) with 3 N NaOH solution, and no further control was carried out in the reactors. The UnAF was initially run with intermittent feeding and recirculation during the start-up period of 50 days for the buildup of biomass in the filter zone and then subjected to step decreases in HRT from 10 to 7.0, 5.3, 3.4, 2.2, 1.3, and 0.7 days (Phases 1–7). For each HRT tested, the steady-state data were collected after at least three turnovers of the HRT. On the conclusion of the experiment at the shortest HRT of 0.7 days, the HRT was restored to 3.1 days (Phase 8) where the biomass samples for the

Table 2 Characteristics of blast furnace grains used as filter media.

2.1. Wastewater and inoculum source The RWD used as the reactor feed was obtained from a local Takju (Korean traditional rice wine) manufacturer producing approximately 50 m3 of alcohol and 15 m3 of RWD daily. The collected wastewater was stored at 4 °C before use to avoid decomposition. The basic characteristics of the RWD used in this study are shown in Table 1. Anaerobic seed sludge was collected from a local full-scale sewage treatment plant in Busan, Korea. Prior to inoculation, the sludge was sieved (with a 860-lm mesh) to remove coarse particles. The total suspended solids (SS) content of the seed sludge was 18.7 g/L, with 66% of it being volatile suspended solids (VSS).

Parameter

a b

Parameter

Unit

Value

Size Total surface area Bulk density Porosity CaO SiO2 Al2O3 MgO TiO2 MnO Fe2O3

mm m2/g g/cm3 % % dwb % dw % dw % dw % dw % dw % dw

5–10 2.37 (0.24)a 2.30 (0.29) 19.9 (9.35) 42.1 35.9 14.7 4.72 0.58 0.35 0.29

Standard deviations are in parentheses. dw, dry weight.

Y. Jo et al. / Bioresource Technology 193 (2015) 53–61

55

reported in this study have been deposited in the GenBank database: KP771679–771694. 2.4. Cluster analysis of DGGE profiles The bacterial and archaeal DGGE gel images were processed using TotalLab 1D software (TotalLab, Newcastle, UK), and a binary matrix was constructed from each gel by scoring the presence or absence of individual bands in all lanes as 1 or 0. Cluster analysis on the obtained matrices was carried out by the UPGMA (unweighted pair group method with arithmetic means) algorithm to measure the relationships between the analyzed microbial community structures based on the DGGE profiles. Distance matrix estimation and cluster dendrogram construction were performed using PAST 3.03 software. 2.5. Analytical methods

Fig. 1. Schematic diagram of the upflow anaerobic filter (UAnF) reactor.

microbial community analysis were collected. The experimental reactor was operated for a total of 380 days including the start-up period.

2.3. Molecular fingerprinting and sequencing Suspended and attached biomass samples were collected from five different positions in the UAnF for microbial community analysis. A mixed liquor sample was taken from the mid-depth of the liquid layer above the filter zone (S1), and BFS grains with biofilms (Fig. S1) were taken from the filter zone at four different heights: the top (S2), 2/3 height (S3), 1/3 height (S4), and the bottom (S5). Five BFS grains were randomly collected at each height, and the attached biomass was recovered by vigorous vortexing in 20 mL of distilled water (DW) for 5 min. A 1-mL aliquot was taken from each of the five samples and washed in a microtube by repeated pelleting and suspending in DW as previously described (Kim et al., 2013). A 200-lL portion of each resuspension was loaded in an automated nucleic acid extractor (Exiprogen, Bioneer, Daejeon, Korea), and the total DNA was prepared following the manufacturer’s instructions. The purified DNA was eluted in 200 lL of elution buffer and stored at 20 °C until use. Bacterial and archaeal 16S rRNA genes were amplified by touch-down polymerase chain reaction (PCR) using two domain-specific primer sets, BAC338F/805R and ARC787F/1059R, respectively, as previously described (Kim et al., 2013). The resulting PCR fragments (20 lL) were electrophoresed on 8% polyacrylamide gels for 16 h at 80 V in a D-code system (Bio-Rad, Hercules, CA). The denaturant gradients in the bacterial and archaeal DGGE gels were 20–60% and 35–65%, respectively, where 100% denaturant corresponds to 7 M urea and 40% (v/v) formamide. The DGGE gels were stained with SYBR Safe DNA gel stain (Molecular Probe, Eugene, OR) and visualized under blue light transillumination. Bands of interest were cut out of the gels and eluted in 40 lL of sterile DW. An aliquot of the elution (2 lL) was reamplified by PCR using the same primers as for the DGGE analysis but without the GC-rich sequence attached. The resulting PCR products were gel-purified and cloned into the pGEM-T Easy vector (Promega, Madison, WI). The insert sequences were analyzed using the T7 primer and compared against the GenBank and RDP databases. The RDP classifier was used for taxonomic classification at a bootstrap confidence threshold of 80%. All nucleotide sequences

COD was determined colorimetrically using an HS-COD-MR kit (Humas, Daejon, Korea), and solids were measured according to standard methods. Volatile fatty acids (VFAs, C2–C7) were measured using a 7820A gas chromatograph (Agilent, Palo Alto, CA) with a flame ionization detector and an Innowax column (Agilent). Biogas composition was analyzed using another 7820A gas chromatograph with a thermal conductivity detector and a ShinCarbon ST column (Restek, Bellefonte, PA). Samples filtered through a 0.45-lm-pore membrane filter were used for soluble COD and VFAs measurements. The C, H, O, N, and S contents (dry weight basis) of RWD were determined using an elemental analyzer (Flash 2000, Thermo Scientific, Delft, The Netherlands). The structural and compositional characteristics of the BFS were analyzed by mercury intrusion porosimetry using an AutoPore IV 9500 (Micromeritics, Norcross, GA) and by inductively coupled plasma-optical emission spectrometry using an Optima 5300 DV (Perkin Elmer, Waltham, MA), respectively. All analyses were replicated at least twice. 3. Results and discussion 3.1. Reactor performance The CSTR showed a gradual decrease in pH after the start of operation and failed to reach a steady state of stable performance at the design HRT of 10 days (Fig. S2). COD removal and methane production deteriorated rapidly after three turnovers of the HRT, followed by a buildup of VFAs along with a pH drop to below 5. The CSTR operation was halted after six turnovers of the reactor volume, and the treatment efficiency was 95%) when the HRT was 3.4 days or shorter, whereas the contribution was markedly lower at longer HRTs: 81%, 54%, and 47% at 5.3-, 7.0-, and 10-day HRTs, respectively. This suggests that acid fermentation was more facilitated at shorter HRTs, i.e., higher OLRs.

The biogas volume was corrected to standard temperature and pressure conditions (0 °C and 1 atm). The methane production rate (MPR, methane produced per unit reactor volume per day) increased step-wise up to 743 mL/L d, whereas the CODr decreased gradually with decreasing HRT from 10 to 1.3 days. However, the further reduction in HRT to 0.7 days caused a large drop in MPR (>20% decrease compared to that at the 1.3-day HRT), indicating a significant deterioration in process stability between the HRTs of 1.3 and 0.7 days. Interestingly, in contrast, the methane yield (YM, methane produced per unit mass of substrate consumed) remained high at 0.31–0.33 L/g COD until the HRT was shortened to 3.4 days, where the maximum YM was shown. Further decreases in HRT induced drastic drops in YM down to 0.16 L/g COD at the 0.7-day HRT (>50% decrease compared to the maximum), suggesting that the UAnF was more efficient in converting RWD organics into methane at 3.4-day or longer HRTs than at other shorter HRTs. The significant decrease in YM with the decrease in the HRT implies that methanogenesis was not the dominant electron sink in the reactor under the short HRT conditions tested (62.2-day HRT). Such changes might be due to alterations in community composition and/or biofilm structure of the reactor biomass, which can influence the mass transfer and energy flow. The excess sludge production (SP, excess sludge produced per unit mass of substrate fed) was between 0.04 and 0.09 g SS/g COD at all HRTs except for 0.7 days. These SP values are comparable to or lower than those reported in other studies for anaerobic filters treating high-strength organic wastewater (Hamdi and Ellouz, 1993; Cresson et al., 2007). A great increase in SP to 0.14 g SS/g COD occurred for the 0.7-day HRT, indicating that a larger detachment of the filter biomass was likely induced by the shortened HRT. A loss of methanogenic biofilm in an attached-growth reactor is generally related to high upflow velocity (i.e., high shear stress) or high OLR (i.e., fast growth and fast death) (Mahmoud et al., 2003; Michaud et al., 2005). The increase in SP in our reactor seems to be due more to the increased OLR than to the increased upflow velocity, given the effluent recirculation rate as high as 10 mL/min (i.e., 10 turnovers of the working volume/day). Overall, the BFS-packed UAnF was demonstrated to be suitable for anaerobic treatment of RWD, with the CODr range of 75.6– 91.7% for HRTs of 1.3–10 days. After collecting the steady-state data at the 0.7-day HRT, the reactor reverted to an HRT of 3.1 days (close to 3.4 days where the maximum YM with a high CODr of 85.5% was observed). After over 10 turnovers of re-stabilization, the reactor attained a comparable steady-state performance (CODr, 86.2%; MPR, 477 mL/L d; YM, 0.29 L/g COD) to that recorded for the 3.4-day HRT (CODr, 85.5%; MPR, 482 mL/L d; YM, 0.33 L/g COD). The re-stabilized reactor with the 3.1-day HRT was then analyzed for bacterial and archaeal community structures.

3.2. Kinetic analysis of the anaerobic filter The modified Stover–Kincannon model, among the most widely applied mathematical models for immobilized systems, was employed to describe the kinetics of organic removal and methane production (Yu et al., 1998). This model determines the substrate removal rate as a function of the OLR using the following equation:

U¼

Q U m ðQSi =VÞ ðSi  Se Þ ¼ V K B þ ðQSi =VÞ

ð1Þ

where U is the substrate removal rate (g COD/L d); Q is the inflow rate (L/d); V is the reactor working volume (L); Si is the influent substrate concentration (g COD/L); Se is the effluent substrate concentration (g COD/L); Um is the maximum substrate removal rate

Y. Jo et al. / Bioresource Technology 193 (2015) 53–61

57

constant (g COD/L d); and KB is the saturation value constant (g COD/L). Eq. (1) was fitted to the steady-state experimental data (Fig. 2; except for that from Phase 7 where process deterioration occurred) using SigmaPlot 10.0 (Systat Software Inc., San Jose, CA, USA). The fitted curve exhibited an excellent regression with the experimental data (Fig. 3), which confirms the validity of the fitted equation to describe the relationship between the OLR (=QSi/V) and the substrate removal rate. From the obtained equation (Eq. (2)), the kinetic constants Um and KB were determined to be 18.65 and 19.58 g COD/L d, respectively.

U¼

Q 18:65ðQSi =VÞ ðSi  Se Þ ¼ V 19:58 þ ðQSi =VÞ

ð2Þ

Eq. (2) can be easily rearranged into Eq. (3), which can be used to estimate the effluent substrate concentration at a given OLR for the UAnF.

Se ¼ Si 

18:65Si 19:58 þ QSi =V

ð3Þ

Similarly, the methane production kinetics can also be described as a function of OLR as for the substrate removal kinetics (Yu et al., 1998):

M¼

M m ðQSi =VÞ MB þ ðQSi =VÞ

ð4Þ

where M is the MPR (L/L d); Mm is the maximum MPR (L/L d); and MB is a constant (L/L d). As for the evaluation of the substrate removal kinetics, two kinetic constants, Mm and MB, were determined by fitting Eq. (4) to the experimental data (Fig. 4). The equation exhibited an excellent fit (R2 = 0.990), with the estimated Mm and MB values being 1.943 and 5.473 L/L d, respectively (Eq. (5)).

M¼

1:442ðQSi =VÞ 3:531 þ ðQSi =VÞ

ð5Þ

The curve fitting results suggest that Eq. (5) is highly adequate to approximate the response of MPR to the variations in OLR. This corresponds to the understanding that the substrate removal rate, which should have a direct relationship with MPR, is a function of OLR (Yu et al., 1998), as seen in Fig. 3. Overall, the kinetic relationships derived from the UAnF demonstrated good estimates of the process performance in terms of organic removal and methane production.

Fig. 3. Curve fitting of the substrate removal rate against the organic loading rate. The solid line represents the fitted curve, and the dashed lines the 95% confidence band.

Fig. 4. Curve fitting of the methane production rate against the organic loading rate. The solid line represents the fitted curve, and the dashed lines the 95% confidence band.

3.3. DGGE and phylogenetic affiliation For a comprehensive insight into the reactor microbial community structure, both suspended and attached microorganisms were analyzed by DGGE, followed by sequencing. The bacterial and archaeal DGGE profiles obtained from different height positions (S1–5) are presented in Fig. 5. A total of sixteen bands, eleven bacterial (RWB1 to 11) and five archaeal (RWA1 to 5), were excised from the gels and sequenced for phylogenetic analysis. Four bacterial sequences were closely related (P97% sequence similarity) to known species of the phyla Bacteroidetes (RWB1), Firmicutes (RWB5), and Proteobacteria (RWB4 and 11) frequently observed in AD environments (Table 3). RWB1 showed 100% sequence similarity to Chryseobacterium bovis, which was originally isolated from raw cow milk. C. bovis strains can grow anaerobically as well as aerobically and ferment glucose and other sugars to produce acids (Hantsis-Zacharov et al., 2008). A recent study identified Chryseobacterium as a dominant bacterial group in an anaerobic fluidized bed reactor fed with synthetic glucose medium for fermentative hydrogen production (Shida et al., 2012). The organism corresponding to RWB1 was therefore likely to be involved in acid fermentation of carbohydrates in the feed. Although the DGGE band intensity is not robustly quantitative, this is also supported by the markedly higher intensity of the band in the sample from the filter bottom near the RWD inlet (Fig. 5A). RWB4 was closely related to several Acinetobacter species. Although Acinetobacter is generally regarded as an obligate aerobic genus, its members have been observed to be active and abundant in different anaerobic or oxygen-limited environments, including AD systems (Supaphol et al., 2011; Baek et al., 2014; Resende et al., 2014). Their role in AD environments could be associated with the oxidation of organic matter or sulfides through anaerobic respiration (Supaphol et al., 2011; Baek et al., 2014). RWB5, affiliated with the genus Clostridium with a bootstrap value of 94%, is most closely related to Clostridium populeti. Clostridium species are common and abundant in AD processes for treating high-strength organic wastes, and C. populeti is a polysaccharolytic species capable of fermenting glucose and cellulose to produce H2, CO2, acetate, butyrate, and lactate (De Vos et al., 2009). The C. populeti-like population corresponding to RWB5 was thus likely to be responsible for the decomposition of sugars and polysaccharides in the reactor. RWB11 was closely related to several species of different genera of the family Rhodobacteraceae, whose species often occur in anaerobic fermentation environments (Cusick et al., 2010; Nelson et al., 2012; Baek et al., 2014) and are unclassifiable at the genus level (bootstrap

58

Y. Jo et al. / Bioresource Technology 193 (2015) 53–61

was classified in the genus Ruminococcus with a bootstrap value of 99%. Ruminococcus species, widely found in the rumen and intestines of mammals, are obligately anaerobic and grow by fermentation of carbohydrates (De Vos et al., 2009). R. flavefaciens and various other species are also capable of utilizing fibers, for example, cellulose and hemicellulose, and are thus thought to play a vital role in the digestion of cellulosic material in anaerobic processes. RWB7 and 10 were classified in the family Spirochaetaceae, whose relatives are often found in abundance in anaerobic

Table 3 Phylogenetic affiliation of bacterial 16S rRNA gene sequences from DGGE bands. Band

Closest species and taxon

Accession number

%Similarity

Classificationa

RWB1

Chryseobacterium bovis Chryseobacterium hominis Chryseobacterium arothri

NR044166

100

Chryseobacterium

NR042517

99.6

EF554408

99.6

RWB2

clone MADSa8 clone AEDQ1DH07 Solobacterium moorei

AB669229 CU922950 GU470893

99.8 99.6 78.7

Bacteria

RWB3

clone TSSUR002_02 clone F05.ab1 Clostridium swellfunianum

AB488224

93.7

Clostridia

EU136383 NR126179

93.3 91.2

Acinetobacter tandoii Acinetobacter haemolyticus Acinetobacter parvus

KM070561

99.4

JX945720

99.4

KC920997

97.7

Clostridium populeti Acetivibrio ethanolgignens

X71853

98.2

NR104783

97.5

RWB6

clone VHW_F_N1 clone T1_2_7 Ruminococcus flavefaciens

JQ085675 AB824503 JF970206

99.1 96.8 96.8

Ruminococcus

RWB7

Clone GB7 clone MADSa69 Spirochaetes bacterium SA-10

KJ679870 AB669259 AY695841

99.8 99.8 94.0

Spirochaetaceae

RWB8

clone QEDN3BD02 clone 07Nov1-02 Catabacter hongkongensis

CU925891 AB618372 JF514887

99.3 98.9 90.8

Clostridiales

RWB9

clone SGE310D clone B6_319 Spirochaetes bacterium SA-10

GU390008 HM228845 AY695841

98.3 97.6 95.7

Bacteria

RWB10

clone HAW-RM372-B-1209d-A3 clone QEDQ3CF03 Spirochaetes bacterium SA-10

FN563261

100.0

Spirochaetaceae

CU922923 AY695841

100.0 93.8

Paenirhodobacter enshiensis Rhodobacter vinaykumarii Roseicitreum antarcticum Paracoccus siganidrum

NR125604

99.8

AM600642

98.9

NR116571

98.6

KF817620

98.4

RWB4

RWB5

Fig. 5. Bacterial (A) and archaeal (B) DGGE fingerprints generated from the UAnF reactor. Samples for DNA recovery were collected at five different positions of the reactor: mid-depth of the liquid layer above the filter zone (S1), the top (S2), 2/3 height (S3), 1/3 height (S4), and the bottom (S5) of the slag bed.

value < 60%). The nearest species, Paenirhodobacter enshiensis, is a facultative fermenter recently isolated from the soil, and it shares high 16S rRNA gene sequence similarities of up to 97.1% with its Rhodobacter relatives. However, P. enshiensis lacks photosynthetic membrane structures and pigments and grows chemoheterotrophically, and Rhodobacter species are all photosynthetic (Wang et al., 2014). Accordingly, given the reactor operating conditions, RWB11 was supposedly derived from a fermentative Paenirhodobacterrelated bacterium. Interestingly, RWB4, 5, and 11 all show significantly higher intensities in the liquid sample than in the filter samples (Fig. 5A), indicating that the corresponding bacteria were more prominent in the planktonic community. The remaining seven bacterial sequences were not closely related to known species, and RWB3 was not even related to any uncultured clone in the databases. Although RWB3 was assigned to the class Clostridia, including Clostridium and other fermentative genera, the putative role of the corresponding population to this band is unclear. RWB2 and 9 were unclassifiable even at the phylum level, although they showed high sequence similarities to environmental clones from anaerobic processes. This implies that the bacterial diversity in AD environments is still largely unknown. RWB6 showed a meaningful sequence similarity of 96.8% (although below the cutoff limit) to Ruminococcus flavefaciens and

RWB11

Acinetobacter

Clostridium

Rhodobacteraceae

a The lowest rank assigned by the RDP Classifier at a bootstrap confidence threshold of 80%.

Y. Jo et al. / Bioresource Technology 193 (2015) 53–61

digesters (Lee et al., 2012; Shen et al., 2013; Li et al., 2015). Chemoorganotrophic Spirochaetaceae species utilize carbohydrates and/or amino acids for fermentative growth and produce mainly ethanol, acetate, H2, and CO2 (Krieg et al., 2011). The Spirochaetaceae-related microbe deduced from RWB 7 and 10 were therefore likely involved in degradation of polysaccharides and/or proteins in RWD. Of note here is that RWB10 appeared exclusively in the filter samples (lanes S2–5 in Fig. 5A), which means that the bacterium corresponding to this band could be active only when bound to the filter biofilm (i.e., attached growth). Given that Spirochaetes species can anaerobically reduce Fe(III) using organic material as an electron donor (Lentini et al., 2012), this observation suggests that the Spirochaetaceae population might have grown by anaerobic respiration with reducible metal ions from the BFS grains as the electron acceptors (Baek et al., 2014). RWB8 was classified in the order Clostridiales, whose members are fermentative and occur in abundance in anaerobic environments, and is closely related to several uncultured bacterial clones from AD processes for treating sewage sludge. Although its specific function is uncertain, the Clostridiales population corresponding to RWB8 supposedly participated in fermentation of organic matter in the reactor. The bacterial populations related to the bands commonly observed in all reactor samples with strong intensities (i.e., RWB2, 7, and 8) likely contributed to the hydrolysis and acidogenesis of organic compounds in the liquid zones (in suspended-growth form), as well as in the filter zones (in attached-growth form). The recovered archaeal sequences were all closely related to known methanogen species across three orders: hydrogenotrophic Methanomicrobiales (RWA1) and Methanobacteriales (RWA5) and aceticlastic Methanosarcinales (RWA2–4) (Table 4). This result is in agreement with the general understanding that archaea in AD environments are mostly methanogens (Yu et al., 2005). The archaeal DGGE fingerprints, although not robustly quantitative, suggest that the reactor likely relied on the slag bed biofilm for its methanogenic activity. The methanogen community of the attached biomass seems to be largely dominated by aceticlastic populations (RWA2–4), suggesting that the aceticlastic pathway was likely the major route of methanogenesis in the reactor. RWA2 and 3 were both assigned to the strictly aceticlastic genus Methanosaeta, whereas RWA4 was assigned to the genus Methanosarcina, which is metabolically more versatile (Boone and Castenholz, 2001). Methanosaeta is known to dominate over Methanosarcina in a stable AD system where the residual VFA concentration is low due to the differences in their biokinetic characteristics, represented by the lower maximum specific growth rate (lmax) and lower half-saturation coefficient (KS) on the common substrate, acetate (Ahring, 2003). The dominance of Methanosaeta-related bands therefore suggests that the reactor environment (CODr > 85% and residual VFA content < 400 mg/L in

Table 4 Phylogenetic affiliation of archaeal 16S rRNA gene sequences from DGGE bands. Band

Closest species and taxon

Accession number

% Similarity

Classificationa

RWA1 RWA2

Methanolinea tarda Methanosaeta concilii Methanosaeta concilii Methanosarcina siciliae Methanobacterium beijingense

NR028163 NR102903

98.9 99.6

Methanolinea Methanosaeta

NR102903

99.6

Methanosaeta

U89773

99.3

Methanosarcina

EU544027

98.5

Methanobacterium

RWA3 RWA4 RWA5

a The lowest rank assigned by the RDP Classifier at a bootstrap confidence threshold of 80%.

59

Phase 8; Fig. 2) was favorable for the development of a Methanosaeta-dominated community. Additionally, the filter likely provided supporting matrices for attached growth of Methanosaeta species, which otherwise would be washed out of the reactor due to their slow growth (the doubling time (Td) for Methanosaeta concilii is 4–7 days (Garcia et al., 2000)). Under attached-growth conditions, Methanosaeta also holds the advantage of its filamentous morphology for constructing the biofilm matrix (McHugh et al., 2003). This is in accord with the fact that the Methanosaeta-related bands were detected with dominant intensities in the filter samples but not clearly visible in the liquid sample. Methanosarcina-related RWA4, although in relatively low intensities, occurred in all biomass samples, including that from the liquid zone, which could be somewhat attributed to the faster growth rate of Methanosarcina (the Td for Methanosarcina siciliae is 7 h (Elberson and Sowers, 1997)). The two remaining bands, RWA1 and 5, were, alternately, closely related to Methanolinea and Methanobacterium species, respectively. Although the hydrogenotrophic pathway likely had a limited contribution to the methanogenesis in the reactor given the limited occurrence of RWA1 and 5, it could have played a critical role in the syntrophic fatty acid oxidation essential for effective stabilization of long-chain fatty acids in anaerobic environments (Ahring, 2003). The Methanolinea- and Methanobacterium-related methanogens corresponding to these bands were supposedly involved in the syntrophic relationship, as hydrogen scavengers, in the reactor (Fig. 2B). The occurrence of RWA1 and 5 exclusively in the filter samples (lanes S2–5 in Fig. 5B) may be, at least in part, related to the favorable conditions for forming a tight consortium of syntrophic VFA-oxidizing bacteria and hydrogenotrophic methanogens in the biofilm matrix, given that the close proximity between the syntrophic partners is essential to make the overall reaction feasible (de Bok et al., 2004). In such a case, the bacterial populations occurring in the filter zone only (e.g., RWB10) might possibly be the syntrophic partners for fatty acid oxidation (Fig. 5A). 3.4. Microbial community structure No apparent differences in bacterial and archaeal DGGE band patterns were observed among the filter samples at different heights, whereas the suspended biomass in the liquid zone showed DGGE profiles that were distinct from those analyzed from the filter biofilms (Fig. 5). This was further confirmed by the UPGMA cluster dendrograms constructed on the DGGE profiles (Fig. 6). In both the bacterial and archaeal dendrograms, all DGGE profiles from the filter samples were very closely clustered together (Sorensen similarity (SS) > 96%), with that obtained from the liquid sample being distantly located from the cluster. This indicates that there was little variation in either the bacterial or archaeal community structures across different levels of the slag bed and that the biomass in the liquid zone had an evidently different microbial community structure. This observation suggests that the difference in growth mode between the inside and outside of the filter (i.e., attached growth versus suspended growth) had a significant influence on the development of the microbial community structure, whereas the height position in the slag bed in the UAnF reactor did not. The archaeal community structures in all of the filter samples were identical (i.e., 100% similarity (SS) when compared based on the presence and absence of individual DGGE bands). This may be ascribed to the simple band patterns (i.e., low diversity), likely reflecting the less diversity of archaea compared to bacteria in AD systems due to the very limited substrate range of methanogens (Boone and Castenholz, 2001). The DGGE fingerprints show that the filter biomass had a higher diversity (i.e., more bands produced) in both bacterial and archaeal communities relative to the suspended biomass. This may be explained by the ability of

60

Y. Jo et al. / Bioresource Technology 193 (2015) 53–61

support of the Korea Ministry of Environment (MOE) through the ‘‘Waste-to-Energy Human Resource Development’’ Project. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.biortech.2015.06. 046. References

Fig. 6. Cluster dendrograms of bacterial (A) and archaeal (B) community structures analyzed from DGGE fingerprints. Samples for DNA recovery were collected at five different positions of the reactor: mid-depth of the liquid layer above the filter zone (S1), the top (S2), 2/3 height (S3), 1/3 height (S4), and the bottom (S5) of the slag bed.

biofilms to maintain a high microbial density within their matrices (i.e., extended biomass retention time) and thereby to allow slow-growing populations (e.g., Methanosaeta is suggested as the dominant methanogen group in our reactor) to remain in the system. A healthy and stable biofilm structure is therefore critical for the performance of high-rate AD processes based on attached-growth culture. The overall experimental results demonstrated the robustness of the BFS-packed UAnF reactor developed in this study to deal with RWD and also of the biofilm community structure involved in the process under varied operating conditions.

4. Conclusions An UAnF with BFS media was constructed and examined for the biomethanation of RWD under varying hydraulic loading conditions. The reactor retained steady treatment efficiency (CODr > 84%) until the HRT was reduced to 2.2 days. The YM remained high at 0.31–0.33 L/g COD at 3.4-day or longer HRTs, whereas it dropped greatly at shorter HRTs. The substrate removal and methane production kinetics were evaluated, and reliable kinetic equations were successfully constructed. Molecular analysis suggested that the methanogenic activity of the reactor seems to rely on the filter biofilm. Overall, the UAnF process proved effective for continuous RWD treatment and biomethanation. Acknowledgements This work was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (2014R1A1A1002329) and the Ministry of Education (2013R1A1A2062963). The authors are also grateful for the

Ahring, B.K., 2003. Biomethanation I, vol. 81. Springer, New York, USA. Akbarnejad, S., Houben, L.J.M., Molenaar, A.A.A., 2014. Application of aging methods to evaluate the long-term performance of road bases containing blast furnace slag materials. Road Mater. Pavement Des. 15, 488–506. Baek, G., Kim, J., Lee, C., 2014. Influence of ferric oxyhydroxide addition on biomethanation of waste activated sludge in a continuous reactor. Bioresour. Technol. 166, 596–601. Boone, D.R., Castenholz, R.W. (eds), 2001. Bergey’s manual of systematic bacteriology, vol. 1: the archaea and the deeply branching and phototrophic bacteria, 2nd ed. Springer, New York. Chaudhary, D., Vigneswaran, S., Ngo, H.-H., Shim, W., Moon, H., 2003. Biofilter in water and wastewater treatment. Korean J. Chem. Eng. 20, 1054–1065. Claveau-Mallet, D., Wallace, S., Comeau, Y., 2013. Removal of phosphorus, fluoride and metals from a gypsum mining leachate using steel slag filters. Water Res. 47, 1512–1520. Cresson, R., Escudié, R., Carrère, H., Delgenès, J.-P., Bernet, N., 2007. Influence of hydrodynamic conditions on the start-up of methanogenic inverse turbulent bed reactors. Water Res. 41, 603–612. Cusick, R.D., Kiely, P.D., Logan, B.E., 2010. A monetary comparison of energy recovered from microbial fuel cells and microbial electrolysis cells fed winery or domestic wastewaters. Int. J. Hydrogen Energy 35, 8855–8861. de Bok, F.A.M., Plugge, C.M., Stams, A.J.M., 2004. Interspecies electron transfer in methanogenic propionate degrading consortia. Water Res. 38, 1368–1375. de Lemos Chernicharo, C.A., 2007. Anaerobic Reactors. IWA Publishing, London, UK. De Vos, P., Garrity, G.M., Jones, D., Krieg, N.R., Ludwig, W., Rainey, F.A., Schleifer, K.H., Whitman, W.B. (eds.), 2009. Bergey’s manual of systematic bacteriology, vol. 3: the fermicutes, 2nd ed. Springer, New York. Elberson, M.A., Sowers, K.R., 1997. Isolation of an aceticlastic strain of Methanosarcina siciliae from marine canyon sediments and emendation of the species description for Methanosarcina siciliae. Int. J. Syst. Bacteriol. 47, 1258– 1261. Garcia, J.-L., Patel, B.K.C., Ollivier, B., 2000. Taxonomic, phylogenetic, and ecological diversity of methanogenic archaea. Anaerobe 6, 205–226. Hamdi, M., Ellouz, R., 1993. Treatment of detoxified olive mill wastewaters by anaerobic filter and aerobic fluidized bed processes. Environ. Technol. 14, 183– 188. Hantsis-Zacharov, E., Senderovich, Y., Halpern, M., 2008. Chryseobacterium bovis sp. nov., isolated from raw cow’s milk. Int. J. Syst. Evol. Microbiol. 58, 1024–1028. Kim, J., Lee, S., Lee, C., 2013. Comparative study of changes in reaction profile and microbial community structure in two anaerobic repeated-batch reactors started up with different seed sludges. Bioresour. Technol. 129, 495–505. Korkusuz, E.A., Bekliog˘lu, M., Demirer, G.N., 2005. Comparison of the treatment performances of blast furnace slag-based and gravel-based vertical flow wetlands operated identically for domestic wastewater treatment in Turkey. Ecol. Eng. 24, 185–198. Krieg, N.R., Staley, J.T., Brown, D.R., Hedlund, B.P., Paster, B.J., Ward, N., Ludwig, W., Whitman, W.B., 2011. Bergey’s manual of systematic bacteriology, vol. 4: the Bacteroidetes, Spirochaetes, Tenericutes (Mollicutes), Acidobacteria, Fibrobacteres, Fusobacteria, Dictyoglomi, Gemmatimonadetes, Lentisphaerae, Verrucomicrobia, Chlamydiae, and Planctomycetes, 2nd ed. Springer, New York. Lee, S.-H., Kang, H.-J., Lee, Y.H., Lee, T.J., Han, K., Choi, Y., Park, H.-D., 2012. Monitoring bacterial community structure and variability in time scale in fullscale anaerobic digesters. J. Environ. Monit. 14, 1893–1905. Lentini, C.J., Wankel, S.D., Hansel, C.M., 2012. Enriched iron(III)-reducing bacterial communities are shaped by carbon substrate and iron oxide mineralogy. Front. Microbiol. 3, 404. Li, Y.-F., Nelson, M., Chen, P.-H., Graf, J., Li, Y., Yu, Z., 2015. Comparison of the microbial communities in solid-state anaerobic digestion (SS-AD) reactors operated at mesophilic and thermophilic temperatures. Appl. Microbiol. Biotechnol. 99, 969–980. Loupasaki, E., Diamadopoulos, E., 2013. Attached growth systems for wastewater treatment in small and rural communities: a review. J. Chem. Technol. Biotechnol. 88, 190–204. Mahmoud, N., Zeeman, G., Gijzen, H., Lettinga, G., 2003. Solids removal in upflow anaerobic reactors, a review. Bioresour. Technol. 90, 1–9. McHugh, S., Carton, M., Mahony, T., O’Flaherty, V., 2003. Methanogenic population structure in a variety of anaerobic bioreactors. FEMS Microbiol. Lett. 219, 297– 304. Michaud, S., Bernet, N., Buffière, P., Delgenès, J.P., 2005. Use of the methane yield to indicate the metabolic behaviour of methanogenic biofilms. Process Biochem. 40, 2751–2755.

Y. Jo et al. / Bioresource Technology 193 (2015) 53–61 Nelson, M.C., Morrison, M., Schanbacher, F., Yu, Z., 2012. Shifts in microbial community structure of granular and liquid biomass in response to changes to infeed and digester design in anaerobic digesters receiving food-processing wastes. Bioresour. Technol. 107, 135–143. Proctor, D., Fehling, K., Shay, E., Wittenborn, J., Green, J., Avent, C., Bigham, R., Connolly, M., Lee, B., Shepker, T., 2000. Physical and chemical characteristics of blast furnace, basic oxygen furnace, and electric arc furnace steel industry slags. Environ. Sci. Technol. 34, 1576–1582. Resende, J.A., Silva, V.L., de Oliveira, T.L.R., de Oliveira Fortunato, S., da Costa Carneiro, J., Otenio, M.H., Diniz, C.G., 2014. Prevalence and persistence of potentially pathogenic and antibiotic resistant bacteria during anaerobic digestion treatment of cattle manure. Bioresour. Technol. 153, 284–291. Shen, P., Zhang, J., Zhang, J., Jiang, C., Tang, X., Li, J., Zhang, M., Wu, B., 2013. Changes in microbial community structure in two anaerobic systems to treat bagasse spraying wastewater with and without addition of molasses alcohol wastewater. Bioresour. Technol. 131, 333–340. Shida, G.M., Sader, L.T., Cavalcante de Amorim, E.L., Sakamoto, I.K., Maintinguer, S.I., Saavedra, N.K., Amâncio Varesche, M.B., Silva, E.L., 2012. Performance and

61

composition of bacterial communities in anaerobic fluidized bed reactors for hydrogen production: effects of organic loading rate and alkalinity. Int. J. Hydrogen Energy 37, 16925–16934. Supaphol, S., Jenkins, S.N., Intomo, P., Waite, I.S., O’Donnell, A.G., 2011. Microbial community dynamics in mesophilic anaerobic co-digestion of mixed waste. Bioresour. Technol. 102, 4021–4027. Wang, D., Liu, H., Zheng, S., Wang, G., 2014. Paenirhodobacter enshiensis gen. nov., sp. nov., a non-photosynthetic bacterium isolated from soil, and emended descriptions of the genera Rhodobacter and Haematobacter. Int. J. Syst. Evol. Microbiol. 64, 551–558. Watanabe, M., Takahashi, M., Sasano, K., Kashiwamura, T., Ozaki, Y., Tsuiki, T., Hidaka, H., Kanemoto, S., 2009. Bioethanol production from rice washing drainage and rice bran. J. Biosci. Bioeng. 108, 524–526. Yu, H., Wilson, F., Tay, J.-H., 1998. Kinetic analysis of an anaerobic filter treating soybean wastewater. Water Res. 32, 3341–3352. Yu, Y., Lee, C., Kim, J., Hwang, S., 2005. Group-specific primer and probe sets to detect methanogenic communities using quantitative real-time polymerase chain reaction. Biotechnol. Bioeng. 89, 670–679.