Reactor performance and microbial community dynamics during anaerobic co-digestion of municipal wastewater sludge with restaurant grease waste at steady state and overloading stages

Download Reactor performance and microbial community dynamics during anaerobic co-digestion of municipal wastewater sludge with restaurant grease waste at steady state and overloading stages

Post on 24-Feb-2017

216 views

Category:

Documents

2 download

TRANSCRIPT

Bioresource Technology 172 (2014) 232240Contents lists available at ScienceDirectBioresource Technologyjournal homepage: www.elsevier .com/locate /bior techReactor performance and microbial community dynamicsduring anaerobic co-digestion of municipal wastewater sludgewith restaurant grease waste at steady state and overloading stageshttp://dx.doi.org/10.1016/j.biortech.2014.09.0460960-8524/ 2014 Elsevier Ltd. All rights reserved. Corresponding author.E-mail address: razaviar@ualberta.ca (V. Razaviarani).Vahid Razaviarani , Ian D. BuchananDepartment of Civil and Environmental Engineering, University of Alberta, 9105-116 St, T6G 2W2 Edmonton, Alberta, Canadah i g h l i g h t s Linkage between reactor performance and bacterial and archaeal community dynamics. GTW co-digestion with MWS at steady state and overloading conditions. Two 10 L bench-scale reactors operated at mesophilic and 20 days SRT. Pyrosequencing determined the sequence abundance of bacterial and archaeal communities. The CCA revealed the linkage between microbial dynamics and environmental variables.a r t i c l e i n f oArticle history:Received 28 June 2014Received in revised form 9 September 2014Accepted 10 September 2014Available online 21 September 2014Keywords:Mesophilic anaerobic co-digestionMunicipal wastewater sludgeGrease wasteMicrobial dynamicsReactor performancea b s t r a c tLinkage between reactor performance and microbial community dynamics was investigated duringmesophilic anaerobic co-digestion of restaurant grease waste (GTW) with municipal wastewater sludge(MWS) using 10 L completely mixed reactors and a 20 day SRT. Test reactors received a mixture of GTWand MWS while control reactors received only MWS. Addition of GTW to the test reactors enhanced thebiogas production and methane yield by up to 65% and 120%, respectively. Pyrosequencing revealed thatMethanosaeta andMethanomicrobium were the dominant acetoclastic and hydrogenotrophic methanogengenera, respectively, during stable reactor operation. The number of Methanosarcina and Methanomicro-bium sequences increased and that ofMethanosaeta declined when the proportion of GTW in the feed wasincreased to cause an overload condition. Under this overload condition, the pH, alkalinity and methaneproduction decreased and VFA concentrations increased dramatically. Candidatus cloacamonas, affiliatedwithin phylum Spirochaetes, were the dominant bacterial genus at all reactor loadings. 2014 Elsevier Ltd. All rights reserved.1. IntroductionDisposal and sustainable management of grease trap waste(GTW) have been a challenge for years due to a variety ofoperational issues and municipal disposal limitations. GTW is alipid-rich organic material collected from the waste streams of res-taurant and food service establishments. Direct disposal of thiswaste into the environment is no longer permitted by most munic-ipalities (Long et al., 2012). Anaerobic digestion as a robust alterna-tive technology is widely applied to stabilize municipalwastewater sludge (MWS) and many organic wastes economicallyand effectively. However, due to some intrinsic limitations of usingthis technology for the treatment of MWS, various pretreatmentsare required to improve its efficiency. GTW anaerobic co-digestion(ACD) with MWS has become a valuable alternative to improvenutrient balance in mixed substrates and enhance buffer capacity,biogas production and reactor performance (Zhu et al., 2011).Despite all the reported benefits of ACD systems, previousstudies have indicated that the performance and stability of suchsystems are dependent on reactor design as well as manyoperational and physico-chemical parameters such as substratecharacteristics, organic loading rates, temperature, and pH amongothers (Razaviarani et al., 2013a,b; Zhu et al., 2011). Anaerobicdigestion, as a syntrophic biological process, also is reliant onmicroorganisms activities via the four major stages of hydrolysis,acidogenesis, acetogenesis and methanogenesis. A deeperunderstanding and resolution of the linkage between microbialcommunity dynamics and process stability can provide invaluableinformation to predict the reactor performance. Yet, the microbialhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.biortech.2014.09.046&domain=pdfhttp://dx.doi.org/10.1016/j.biortech.2014.09.046mailto:razaviar@ualberta.cahttp://dx.doi.org/10.1016/j.biortech.2014.09.046http://www.sciencedirect.com/science/journal/09608524http://www.elsevier.com/locate/biortechV. Razaviarani, I.D. Buchanan / Bioresource Technology 172 (2014) 232240 233dynamics and their interactions still remained uncertain primarilydue to the complexity of the microbial activities within the inter-related biological reactions in these systems (Supaphol et al.,2011). Several studies were conducted over the last decade toinvestigate the microbial population structure in anaerobic diges-tion of lipid-rich waste (Palatsi et al., 2010; Pereira et al., 2002)with a focus mostly on the LCFA inhibition effects. Nevertheless,to the authors knowledge, microbial studies linked to reactor per-formance of the anaerobic co-digestion of GTW with MWS havenot yet been conducted.Among the available microbial fingerprinting techniques, dena-turing gradient gel electrophoresis (DGGE) and clone library arethe most popular methods used to evaluate the microbial popula-tions (Lee et al., 2010). However, using the DGGE method for inves-tigation of complex microbial populations is intricate due todrawbacks which include the identification of limited bands andco-migration of sequences. Also, the cloning technique and its dataanalysis are laborious and uneconomical. The development of 454Pyrosequencing, as a new generation of sequencing techniques,facilitates the investigation of microbial community dynamics invarious environments by identifying a larger number of sequencesmore quickly (Guo et al., 2014).The objective of this study was to investigate the reactor perfor-mance linked the microbial community dynamics of mesophilicACD of GTW and MWS at (1) steady-state conditions conductedat two different runs with different organic loadings and (2) duringreactor overload conditions. For this purpose, physico-chemicalanalysis along with the Pyrosequencing microbial technique wasperformed and reactors stabilities and performance were moni-tored accordingly with the associated microbial populationdynamics.2. Methods2.1. Inoculum and substratesMunicipal wastewater sludge (MWS) consisting of a 4:1 (v/v)mixture of primary sludge (PS) and thickened waste activatedsludge (WAS), was collected from a wastewater treatment plant(WWTP) in Edmonton, Alberta, Canada. Primary sludge is mixedwith thickened waste activated sludge in a 4:1 volumetric ratioat this WWTP before it is pumped to the on-site anaerobic digest-ers. The MWS used in the study was collected from a pipeline thatconveys this 4:1 mixture to the on-site digesters. GTW wasreceived from a local waste collection company in Edmonton,Alberta. Samples were stored at 4 C until their utilization. TheTable 1Characteristics of substrates and inoculum.Parameter Mean value standard deviation of 4 sampMWS1b MWS2cCOD (g/L) 75.3 1.2 40.9 TS (g/L) 63.9 1.0 22.9 VS (g/L) 35.7 0.9 18.3 TSS (g/L) 61.5 1.3 19.2 VSS (g/L) 34.8 1.4 15.6 VFA (mg/L) 2983 39 1640 TKNe (mgN/L) 804 794TANe (mgN/L) 346 332Alkalinitya (mg/L) 1850 15 2000 pH 5.8 0.1 5.7 0a Alkalinity represented as mg/L CaCO3.b Municipal wastewater sludge used in Stage 1.c Municipal wastewater sludge used in Stage 2.d Non-measured.e 1 sample (no sample standard deviation).GTW from restaurants and food services typically has some set-table solids which are not degraded during the anaerobic digestionand collect at the bottom of the reactor. Pre-treatment should bepracticed at WWTPs to remove these solids and water which canbe treated more efficiently in other process units. Therefore, beforeadding the GTW to the MWS, it was brought to room temperatureand then the top layer of grease, fats and oils (FOG) was separatedfrom the settable solids and water layers. This top layer of FOG wasthen blended with the MWS in the desired proportion. Digestedeffluent from a full-scale mesophilic anaerobic reactor at the sameEdmonton WWTP was used as the inoculum (biomass) for thestart-up of the reactors. The characteristics of substrates andinoculum are shown in Table 1.2.2. Reactor operation and loading protocolThe experiment was conducted at two separate stages, asshown in Table 2, with respect to the collection of MWS at differenttimes. During each stage, two identical 10 L (8 L working volume)reactors were mixed by magnetic stirrers and operated at37 0.5 C and a 20-day solids retention time (SRT). The reactorswere sustained at desired temperature using heating tape wrappedaround the reactors and the temperatures were monitored andcontrolled by Type K thermocouples and digital temperature con-trollers. An insulating jacket was also applied around each reactorto minimize heat loss.Each reactor was initially filled with 8 L of inoculum and thenthe reactors headspace was purged with nitrogen gas. Each day,0.4 L of digested material was withdrawn from each reactor andreplaced with the same volume of substrate to provide a 20 daySRT. Reactor 1 served as a control (C-1) and received only MWS,while reactor 2 was operated as the test digester (T-1) and wasfed a mixture of MWS and GTW based on a percentage of the con-trol reactor COD loading. Initially, for the first run of the experi-ment, C-1 and T-1 received an equal amount of MWS1 toestablish their baseline performance for a period of 30 days (1.5SRT). When the equivalence of the reactors performance wasestablished, the COD loading of T-1 reactor was increased withthe addition of a known volume of GTW to the MWS1 to obtainthe desired COD loading of 150% relative to the control reactor(C-1). This operating mode was continued for 3 SRT (60 days) toreach the steady-state conditions and then for another 10 day per-iod during which daily sampling was conducted. Similarly, for thesecond stage, the control (C-2) and test (T-2) reactors were fedwith same volume of MWS2 for 30 days to establish the baselineperformance. Then the test reactor (T-2) was fed with a mixtureles InoculumeGTW0.9 2697.8 52.5 22.90.8 776.8 16.7 22.30.5 776.2 13.8 12.20.8 47.7 1.0 20.90.5 47.6 0.9 11.328 nmd 8nm 2140nm 75518 nm nm.1 5.0 0.1 7.3Table 2Reactor organic loading rates (OLR) and their durations.Stage Reactor Substrate Duration OLR Relative loading(d) (g COD/L d) (%)1 C-1 MWS1 100 3.77 100T-1 MWS1 30 3.77 100T-1 GTW +MWS1 70 5.57 1502 C-2 MWS2 121 2.05 100T-2 MWS2 30 2.05 100T-2 GTW +MWS2 70 3.84 190T-20 GTW +MWS2 21 8.00 400234 V. Razaviarani, I.D. Buchanan / Bioresource Technology 172 (2014) 232240of GTW and MWS2 to reach 190% of the control reactor (C-2) CODloading. Steady state conditions were achieved after 3 SRTs whencoefficients of variation of effluent COD, VSS and methane dailymeasurements were less than 5%. Thereafter, for a period of10 days, samples were collected for the subsequent analyses. Inorder to investigate the reactors performance and microbial com-munity response to reactor overloading, GTW addition to T-2 wasincreased to achieve a COD loading approximately fourfold that ofC-2. This operating mode was continued for 3 weeks in the T-2reactor which hereafter is termed the overload condition. ReactorT-2 is indicated as T-20 when operated under this overload condi-tion. Reactors loading rates during the experiment are shown inTable 2.2.3. Physico-chemical analysisThe biogas flow rate from each reactor was quantified by a dig-ital gas flowmeter and logged to a lab computer. Agilent 7890A gaschromatograph (GC) was used to measure the CH4 content in thebiogas. The GC was equipped with an Agilent GS-Q column and aflame ionization detector (FID). The CO2 percentage in biogas wasintermittently measured using a Fyrite gas analyzer according tothe method specified by the manufacturer (Bacharach Inc.).Chemical oxygen demands (tCOD and sCOD) of substrates andeffluents were measured with the closed reflux (5220C) methodusing HACH COD reactor and GENESYS20 spectrophotometer. Totalsolids (TS), volatile solids (VS), total suspended solids (TSS) andvolatile suspended solids (VSS) were quantified using methods2540G and 2540D, respectively. Total alkalinity (TA), partial alka-linity (PA) and pH were determined using Thermix stirrer 120Sand ACCUMET AB15 Plus pH meter. The titration end points forPA and TA were pH 5.75 and 4.3, respectively according to Stan-dard Method 2320B. All the above analyses were performedaccording to the standard methods in triplicate (APHA, 2005). Forthe VFA measurement, the samples were centrifuged at 5000g for10 min and then the supernatants were filtered sequentiallythrough 0.45 lm and 0.2 lm nylon syringe filters. Individual vola-tile fatty acids (acetate, propionate, iso-butyrate, n-butyrate, iso-valerate and n-valerate) in the substrates and digested effluentswere analyzed by a Varian 430 gas chromatograph (GC) equippedwith a Stabilwax-DA capillary column and a flame ionizationdetector (FID). Total ammonia and TKN were measured by the Bio-chemical Analytical Service Laboratory staff in the University ofAlberta according to the standard methods.2.4. Microbial community analysis2.4.1. Sludge sampling and DNA extractionThe effluent sludge samples were collected from the reactorsduring the last 10 days of steady-state operation at each COD load-ing. Total genomic DNA was extracted from approximately 500 lLof well-homogenized sample using Fast DNA Spin kit for soil(Biomedical, USA) according to the manufacturers instructions.The effluent sludge samples were initially centrifuged at 10,000gfor 20 min, and the supernatant was decanted carefully to obtainthe settled biomass for DNA extraction. A NanoDrop 2000C spec-trophotometer was used to determine the concentrations, qualityand integrity of the extracted DNA. Extracted DNA samples werestored at 20 C until submitted to the microbiology lab for thepyrosequencing analysis.2.4.2. Pyrosequencing analysisSamples were collected from each reactor on the sixth day ofthe 10-day steady state sampling period. The 16S rRNA genes wereamplified using bar-coded universal bacterial and archaeal primersfor each sample. The primer sequences are as follows; bacterialuniversal (27F: AGR GTT TGA TCM TGG CTC AG, 519r: GTN TTACNG CGG CKG CTG) and archaeal universal (349F: GYG CAS CAGKCG MGA AW, 806r: GGA CTA CVS GGG TAT CTA AT). The PCRreactions were conducted in a single step 30 cycle PCR using Hot-StarTaq Plus Master Mix Kit (Qiagen, Valencia, CA) under followingconditions: initial denaturation at 94 C for 3 min, followed by 28cycles of 94 C for 30 s; annealing at 53 C for 40 s and elongationat 72 C for 1 min; after which a final elongation step at 72 C for5 min was performed. All the amplified 16S rRNA from differentsamples was mixed in equal concentrations and purified usingAgencourt Ampure beads (Agencourt Bioscience Corporation, MA,USA). All samples were subjected to pyrosequencing using Roche454 FLX Titanium instruments and reagents according to the man-ufactures guidelines. The nucleotide sequence reads were sortedout using a proprietary analysis pipeline. Initially, the barcodesequences shorter than 200 bp, non-16S rRNA sequences andsequences with homo-polymer runs exceeding 6 bp were removedfrom barcode sorted sequences. Then, sequences were de-noisedand chimera sequences of selected reads were also removed. Oper-ational taxonomic units (OTUs) of each sequence were identifiedafter removal of singleton sequences, clustering at 3% divergence(97% similarity). Final OTUs were taxonomically classified usingBLAST and database derived from GreenGenes, RDPll and NCBl(DeSantis et al., 2006) and compiled into each taxonomic level intoboth count and percentage datasets. The count dataset con-tains the actual number of sequences, while the percentage datasetis defined as the ratio of number of assigned sequence reads ofspecific taxon divided by the number of total sequence reads.2.5. Statistical analysisThe correlation between microbial communities (bacteria andarchaea) and reactors performance and stability parameters weredetermined by the canonical correspondence analysis (CCA) usingthe XLSTAT software version 2014.3. Results and discussion3.1. Process stability and reactor performanceThe reactors operated in two stages were carried out in differentOLRs (Table 2) and the obtained experimental results are shown inTable 3. The methane yield per unit VS applied increased withincreasing the GTW COD addition to the test reactors T-1 and T-2relative to the control reactors C-1 and C-2, respectively until areduction in methane yield was observed in T-2 at the 8.0 gCOD/L d loading (the overloaded condition). The cause of this reductionwas investigated by monitoring the critical parameters includingpH, VFA and alkalinity. As shown in Table 3, during the experiment,the pH remained practically constant with values ranging between7.25 and 7.33 in the reactors. Yet at the overloaded condition, areduction in the pH value was observed. In the same way, theTable 3Reactor stability and performance parameters at different loadings.Parameters Mean value standard deviation of 5 samplesStage 1 Stage 2Reactor C-1 T-1 C-2 T-2 T-20Loading (g COD/L d) 3.77 5.57 2.05 3.84 8.00pH 7.30 0.04 7.28 0.01 7.33 0.01 7.25 0.03 6.20 0.10VFA (mg/L) 39.5 1.1 26.7 1.1 54.0 2.5 9.5 0.6 1200.0 71.0Acetic ac. (mg/L) 21.7 1.0 15.2 1.0 33.8 1.2 8.5 0.3 677.0 38.0Propionic ac. (mg/L) 7.0 0.2 3.1 0.1 4.4 0.5 1.0 0.1 324.0 1.0Butyric ac. (mg/L) 22.2 1.1 10.4 0.6 6.4 0.2 nd1 90.4 5.1Valeric ac. (mg/L) 4.5 0.4 1.9 0.3 9.5 0.3 nd 108.6 4.9Alkalinity (mg/L) 4060 48 3930 46 4180 40 3780 29 2590 39COD (g/L) 43.5 1.0 58.0 0.8 24.0 0.5 30.0 0.7 110.0 1.3VS (g/L) 17.5 0.4 17.0 0.4 9.7 0.3 10.0 0.3 35.0 0.5Biogas production (L/d) 8.2 0.1 13.5 0.2 4.5 0.1 10.0 0.2 5.2 0.1Methane yield (LCH4/gVSadded) 0.34 0.01 0.50 0.01 0.33 0.01 0.61 0.01 0.12 0.081 Not detected.V. Razaviarani, I.D. Buchanan / Bioresource Technology 172 (2014) 232240 235VFA remained low during the experiment until a sudden increasewas observed during the overloaded condition. The pH reductionand VFA accumulation was associated with a marked reductionin biogas production and methane yield as well as decreasedCOD and VS removal.The VFA/alkalinity ratio is a reliable indicator of process stabil-ity with the process being stable when this ratio is less than 0.30.4 (Rincn et al., 2008). As can be easily calculated from Table 3,this ratio was well below the limit range during the experimentuntil the overloaded condition in reactor T-2. At this loading, a sub-stantial increase in the VFA/alkalinity ratio to the value of 0.6 wasobserved. Furthermore, acetate and propionate were the major fer-mentation products consisting of 56% and 27% of the total VFA,respectively at this loading. However, low concentrations of ace-tate in reactors T-1 and T-2 coincided with increased biogas pro-duction and methane yield compared to the control reactorsprior to the T-2 overloaded condition. This could be possibly as aresult of good acclimation of microbial populations, particularlyacetoclastic methanogens, to the addition of GTW to the reactors.GTW addition at the 150% and 190% COD relative loadings to T-1and T-2 resulted in 65% and 120% increases in biogas productionrelative to C-1 and C-2, respectively. The relative loadings alsoresulted in 47% and 85% increases in the methane yield relativeto the C-1 and C-2 reactors, respectively. Yet, as shown in Table 3,when the relative loading was increased to 400% in T-2, its biogasproduction and methane yield decreased by 48% and 80%, respec-tively compared to that observed during the previous 190% relativeloading. It is noted that under this overloaded condition the non-acetate VFAs also increased considerably compared to the otherloading levels which may suggest a reduction in the syntrophicacetogenic populations.3.2. Methanogenic community structure of the ACDThe methanogenic community is generally less diverse thanthat of the bacterial community and can occupy a limited ecologi-cal niche in anaerobic reactors. The type and origin of inoculumand substrate along with the environmental conditions can haveconsiderable effects on the archaeal distribution (Leclerc et al.,2004). As shown in Table 4, the majority of archaeal 16S rRNA genesequences are assigned to the classes Methanomicrobia and Met-hanobacteria. The sequence distributions at the family and orderlevels are shown in Fig. 1. Typically, methanogens are categorizedas either acetoclastic or hydrogenotrophic according to the sub-strate they utilize. Acetoclastic methanogens utilize acetate to pro-duce methane while hydrogenotrophic methanogens consume CO2and H2 to produce methane.Start-up is a critical step in the operation of an anaerobicdigester and is highly dependent on the microbial source, the sizeof the inoculum and the initial operation mode (Griffin et al.,1998). The archaeal populations in the inoculum used as seedthroughout the experiment were found to be predominantly ofthe Methanosarcinales order. Genus Methanosaeta sequences werepresent in the greatest relative abundance (2619 sequences, 78%of total archaeal sequences), followed by Methanomicrobium (262,8%), Methanosarcina (190, 6%), Methanospirillum (168, 5%) andMethanobacterium (105, 3%). Therefore, 84% of the 3344 sequencesobtained from the inoculum were of acetoclastic methanogens ofgenera Methanosaeta and Methanosarcina. McMahon et al. (2004)observed that anaerobic digesters with high levels of archaea ofwhich Methanosaeta was the dominant acetoclastic methanogenstarted up well, whereas those having low levels of archaea inwhich the dominant methanogens were Methanosarcina andMethanobacteria experienced problems at start-up. The spectrumof sequence abundance in the inoculum used in the current studyindicates that the seed was collected from a stable anaerobic diges-ter where the sequences belonging to acetoclastic methanogenswere over 5-fold more abundant than those of hydrogenotrophicmethanogens.As shown in Fig. 1, regardless of the different batches of MWS(Table 1) fed into the control reactors (C-1) and (C-2) during Stages1 and 2, the proportion of acetoclastic (Methanosarcinales)sequences was approximately the same as that of hydrogenotroph-ic (Methanomicrobiales and Methanobacteriales) methanogens inthese reactors. The proportion of acetoclastic methanogensincreased when COD loading was increased to the test reactors(T-1 and T-2). Acetoclastic methanogens have been found to pre-dominate under stable operating conditions where the molecularhydrogen concentration is low (Kim et al., 2014). Methanosaetaare the dominant acetoclastic methanogen at low acetate concen-trations under stable operating conditions. This is in agreementwith the parameter values listed in Table 3 for the T-1 and T-2operating conditions.During Stage 1, the 150% relative COD loading to reactor T-1resulted in only a 3% decrease in the total number of archaealsequences compared to the C-1 reactor (see Table 4). Althoughthe archaeal community in reactor C-1 was distributed equallybetween hydrogenotrophic and acetoclastic methanogens, 67% ofthe archaeal sequences in T-1 consisted of acetoclastic methano-gens, predominantly of the familyMethanosaetaceae (Fig. 1). There-fore, the 47% increase in methane yield in T-1 compared to the C-1(see Table 3) was accompanied by a 35% increase in the relativeabundance of acetoclastic methanogens and a 30% decrease in ace-tate concentration. Typically, over 70% of methane production isTable 4Abundance of archaeal phylogenetic groups in reactors.Stage 1 Stage 2Reactor C-1 T-1 C-2 T-2 T-2% loading 100 150 100 190 400Genus (similarity %) Number (percent) of sequences in relevant genus1MethanomicrobiumH (97) 2210 (42.0) 1245 (24.4) 958 (41.1) 647 (18.0) 944 (28.3)1MethanospirillumH (99) 111 (2.1) 77 (1.5) 33 (1.4) 154 (4.3) 177 (5.3)1MethanoculleusH (99) 184 (3.5) 0 (0) 75 (3.2) 104 (2.9) 33 (1.0)1MethanosarcinaA/H (99) 247 (4.7) 388 (7.6) 86 (3.7) 104 (2.9) 737 (22.1)1MethanosaetaA (99) 2310 (43.9) 3041 (59.6) 1114 (47.8) 2461 (68.5) 1161 (34.8)2MethanobacteriumH (97) 79 (1.5) 128 (2.5) 33 (1.4) 65 (1.8) 73 (2.2)2MethanosphaeraH (99) 111 (2.1) 66 (1.3) 0 (0) 54 (1.5) 87 (2.6)2MethanobrevibacterH(99) 11 (0.2) 148 (2.9) 33 (1.4) 0 (0) 80 (2.4)Other 0 (0) 10 (0.2) 0 (0) 4 (0.1) 43 (1.3)Totals 5263 (100) 5103 (100) 2332 (100) 3593 (100) 3335 (100)Classes: 1Methanomicrobia and 2Methanobacteria.H: hydrogenotrophic methanogens; A: acetoclastic methanogens.0102030405060708090100ControlTest1ControlTest2OverloadingStage 1 Stage 2Relative abundance (%)MethanobacteriaceaeMethanospirillaceaeMethanomicrobiaceaeMethanosaetaceaeMethanosarcinaceaeMethanobacterialesMethanomicrobialesMethanosarcinalesFig. 1. Distribution of family level archaeal community categorized in brackets at order level.236 V. Razaviarani, I.D. Buchanan / Bioresource Technology 172 (2014) 232240carried out by this group of archaea under stable reactor conditions(Kundu et al., 2014). The genus Methanosaeta has a lower half sat-uration coefficient than does the genus Methanosarcina (Conklinet al., 2006; McMahon et al., 2004; among others). Therefore,Methanosaeta is dominant in the presence of a low acetate concen-tration. Hori et al. (2006) reported that the genus Methanosaetagrows more rapidly than Methanosarcina at acetate concentrationslower than 1 mM while the genus Methanosarcina are the predom-inant acetoclastic methanogens at acetate concentrations higherthan 1 mM. As shown in Table 3, the acetate concentration in theT-1 reactor was well below the 1 mM threshold where the genusMethanosaeta would predominate over the genus Methanosarcina.During Stage 2, 3592 archaeal sequences were obtained forreactor T-1 during its 190% relative COD loading, whereas the con-trol C-2 reactor yielded 2330 archaeal sequences. This represents a54% increase in T-2 archaeal sequences relative to C-2.Approximately 76% of the archaeal sequences from T-2 belongedto acetoclastic methanogens with relative abundances of 69% and3% for Methanosaeta and Methanosarcina, respectively. This is con-sistent with Methanosaeta outcompeting Methanosarcina in thepresence of a low acetate concentration (see Table 3). The increasein the relative abundance of acetoclastic methanogens wasaccompanied by 120% and 85% increases in biogas productionand methane yield in the T-2 reactor relative to C-2, respectively(see Table 3). The predominance of this genus also represents a sta-ble reactor as indicated in previous studies (Ariesyady et al., 2007;Raskin et al., 1994).The distribution and dynamics of methanogens in the anaerobicco-digestion of lipid-rich materials under mesophilic conditionshave not been documented. Martn-Gonzlez et al. (2011) investi-gated the thermophilic anaerobic co-digestion of FOG with theorganic fraction of municipal solid wastes and observed the genusMethanosarcina to be the predominant acetoclastic methanogens intheir samples. It should be noted that they did not detect the genusMethanosaeta during their investigation. In contrast, the genusMethanosaeta represented the largest number of acetoclastic pop-ulations in the present study, whereas the relative abundance ofMethanosarcina sequences remained low during all steady statesampling periods and were only observed to increase in numberduring the overloading period (Fig. 1).As shown in Fig. 1, the hydrogenotrophic pathway was primar-ily associated with the family Methanomicrobiaceae within theorder Methanomicrobiales at all loadings. Members of the orderMethanobacteriales were present throughout the study in allV. Razaviarani, I.D. Buchanan / Bioresource Technology 172 (2014) 232240 237reactors, but their numbers of sequences accounted for only 1.43% of the total methanogen population. The increase of GTW load-ing at the 400% relative COD loading (overload condition) resultedin changes in both the acetoclastic and hydrogenotrophic commu-nities as shown in Fig. 1. The total number of archaeal sequencesobtained during the overload condition (reactor T-20) decreasedby 7% compared to those obtained from reactor T-2 (the same reac-tor) during the 190% relative COD loading. Thus, the total numberof archaeal sequences changed little despite signs of instability inthe reactor such as low pH, high VFA concentrations and reducedbiogas production (Table 3), among others. The major changewas in the distribution of the archaeal community. As shown inFig. 1, the relative abundance of hydrogenotrophic methanogensincreased considerably compared to that at the previous 190% rel-ative COD loading. The relative sequence abundance Methanomicr-obium within the family Methanomicrobiaceae and the orderMethanomicrobiales increased from 18.0% to 28.3%. This indicatesthat the community response to the GTW overloading that causeda reduction in the reactor performance was an increase in theactivity of hydrogenotrophic methanogens due presumably to anincrease in the H2 partial pressure. Padmasiri et al. (2007) alsoobserved similar behavior of hydrogenotrophic methanogens dur-ing decreased reactor performance. Kim et al. (2014) reportedMet-hanoculleus bourgensis of the order Methanomicrobiales to becomethe predominant methanogen as the reactor used in their studyapproached unstable conditions. High H2 partial pressure in ananaerobic digester can hamper the syntrophic relationshipbetween its microbial communities (Kundu et al., 2014). This cancause the dominance of hydrogenotrophic populations such asMethanobacteriales andMethanomicrobiales over acetoclastic meth-anogens. Although the hydrogenotrophic communities did notbecome predominant at the 400% relative COD loading in termsof sequence abundance, the proportion of hydrogenotrophic meth-anogens did increase to approximately 42% of the total archaealsequences. If Methanosarcina which can produce methane usingeither H2/CO2, acetate or methyl are included.The Methanosarcina sequences increased to 22% of the totalnumber of archaeal sequences during the overloading conditionfrom 2.9% at the previous 190% relative COD loading (Table 4).In contrast, the relative abundance of Methanosaeta sequencesdecreased by approximately 50% under the overloadingconditions. This can be attributed primarily to their growthkinetics as discussed previously, which allows Methanosarcinato outcompete Methanosaeta in the presence of high acetateconcentrations.3.3. Bacterial community structure of the ACDThe bacterial 16S rRNA gene sequence at different taxonomiclevels and loading percentages are summarized in Fig. 2 andTable 5. The total sequence abundance of bacteria in the inoculumsample was 9139. This is almost three times the total sequenceabundance of the archaeal population in the inoculum sample.The genus Candidatus cloacamonas was the dominant bacterialcommunity in the inoculum with 7951 of the total 9140 bacterialsequence reads and remained dominant in the reactors at all load-ing rates (Table 5). C. cloacamonas ssp. are syntrophic fermentationbacteria found in a number of anaerobic digesters and wererecently categorized under the phylum Spirochaetes (Pelletieret al., 2008). Spirochaetes are gram-negative bacteria with a distinc-tive spiral shape and are able to ferment carbohydrates and aminoacids into mainly acetate, H2 and CO2 in anaerobic digesters (Leeet al., 2013).During the two 100% MWS loadings stages, as shown in Fig. 2,the bacterial distributions remained almost constant in the controlreactors regardless of the different batches of sludge used.However, compared to the inoculum, the total number of bacterialsequences was reduced by 10% and 41% in control reactors C-1 andC-2, respectively (Table 5). These changes could be attributed toseveral factors including alteration and availability of substrates.In these reactors the bacterial sequences were predominantly affil-iated with the genus C. cloacamonas within the phylum Spirochae-tes with the relative abundance of 82.0% and 82.2% in C-1 and C-2,respectively. The number of sequences in the phylum Actinobacte-ria in inoculumwas very low, but was slightly increased in the con-trol reactors possibly due to the variation of MWS and feedingmode in the bench-scale reactors.The total number of bacterial sequences was reduced by 33% inreactor T-1 at the 150% relative COD loading, compared to that ofthe C-1 reactor, while the distribution of the bacterial communitychanged very little, as shown in Table 5. Sequences from four phylawere identified in the reactors as shown in Table 5. The dominantsequence identified in reactor T-1 was associated with the phylumSpirochaetes (80.4%), followed by the phyla Proteobacteria (12.1%),Actinobacteria (1.5%) and Chloroflexi (2.1%). While the genus C. clo-acamonas remained the dominant community in reactor T-1, thesequence abundance of the genus cEsherichia within the phylumProteobacteria increased by 12% compared to that in the C-1 reactor(see Table 5). In addition, although the genera Anaerolinea andCaldilinea within the phylum Chloroflexi were minor populationsin the control reactor, C-1, with the addition of GTW, the sequenceabundance of phylum Chloroflexi decreased by 68% compared tothe C-1 reactor and the affiliated genus Caldilinea was not measur-able (see Table 5).During Stage 2, the 190% relative COD loading to the T-2 reactorresulted in little change to the total sequence abundance of bacte-ria compared to the C-2 reactor (Table 5). Despite the addition ofGTW to reactor T-2, the relative distribution of bacterial strainsremained similar to that in the control reactor (Table 5) with thegreatest changes in relative sequence abundance being withinthe genera Esherichia (+54%) and Caldilinea (100%), which carryout acidogenesis and fermentation, respectively. As shown inFig. 2 and Table 5, the genus C. cloacamonas remained the prevail-ing bacterial community with a relative sequence abundance of80.3% in the T-2 reactor at the 190% relative COD loading.At the 400% relative COD loading, the total sequence abundanceof bacterial populations decreased by approximately 18% in theT-20 reactor compared to the control reactor (C-2) and to reactorT-2 at the 190% relative COD loading (Table 5). The larger changesin relative sequence abundances shown in Table 5 at the highestrelative COD loading were in the genera Esherichia (+20%) andAnaerolinea (57%), compared to those in reactor T-2. The genusCaldilinea remained undetectable in reactor T-20. As shown inFig. 2, the genus C. cloacamonas within the phylum Spirochaetesconstituted the dominant percentage of the total bacterialsequences under all loading conditions. The abundance and distri-bution of Spirochaetes in anaerobic digesters have been rarelyinvestigated because of the inadequate knowledge of this phylum.The relative sequence abundance of phylum Proteobacteria, as thesecond largest population found in this study at all loadings,increased by 71% in reactor T-20 under the overloaded conditionscompared to that in C-2. Chen et al. (2008) reported that evenlow concentrations of long chain fatty acids (LCFA) from the degra-dation of lipid-rich materials could be detrimental to anaerobes,particularly for the gram-positive bacteria. The phylum Actinobac-teriawas the only gram-positive bacterial community found in thisstudy. However, the relative abundance of genera in this phylumwas low in all reactors. The affiliated genus Demequina completelydisappeared after the GTW was added even during the safe load-ings of the T-1 and T-2 reactors. However, the response of thegenus Dermatophilus was not as clear (Table 5). LCFA inhibitionwas not investigated in this study, however because GTW is0102030405060708090100Control1Test1Control2Test2OverloadingStage 1 Stage2Relative abumdance (%)AnaerolinealesEnterobacterialesDesulfarculalesMicrococcalesCandidatus cloacamonasCaldilinealesOtherAnaerolinea - Proteobacteria - ProteobacteriaActinobacteriaSpirochaetesCaldilinesFig. 2. Distribution of order level bacterial community categorized in brackets to class level.Table 5Abundance of phylogenetic groups of bacteria in reactors.Stage 1 Stage 2Reactor C-1 T-1 C-2 T-2 T-20% loading 100 150 100 190 400Genus (similarity %) Number (percent) of sequences in relevant genus1SpaerochaetaU (95) 33 (0.4) 22 (0.4) 15 (0.3) 14 (0.3) 0 (0)1C. cloacamonasS (99) 6744 (82.0) 4395 (80.0) 3979 (82.2) 3881 (80.3) 3148 (78.9)2DemequinaF (99) 16 (0.2) 0 (0) 15 (0.3) 0 (0) 0 (0)2DermatophilusH (92) 164 (2.0) 82 (1.5) 58 (1.2) 77 (1.6) 56 (1.4)3DesulfarculusA (93) 181 (2.2) 110 (2.0) 82 (1.7) 72 (1.5) 104 (2.6)3EscherichiaA (97) 592 (7.2) 665 (12.1) 392 (8.1) 604 (12.5) 567 (14.2)4AnaerolineaF (95) 189 (2.3) 115 (2.1) 116 (2.4) 111 (2.3) 40 (1.0)4CaldilineaF (94) 173 (2.1) 0 (0) 111 (2.3) 0 (0) 0 (0)Other 132 (1.6) 104 (1.9) 73 (1.5) 72 (1.5) 76 (1.9)Totals 8224 (100) 5494 (100) 4841 (100) 4833 (100) 3990 (100)Phyla: 1Spirochaetes, 2Actinobacteria, 3Proteobacteria and 4Chloroflexi.U: unknown putative function, S: syntrophic, F: fermentative, H: hydrolytic, A: acidogenic.238 V. Razaviarani, I.D. Buchanan / Bioresource Technology 172 (2014) 232240converted to LCFA before the breakdown to VFA compounds, thedecrease in the phylum Actinobacteria after the addition of GTWcould be attributed to the existence of such intermediate lipid-deg-radation products. It should be noted that, as Chen et al. (2008)indicated, because methanogens cell wall resembles that ofgram-positive bacteria they are more vulnerable to the LCFA con-centration than is the bacterial community.3.4. Correlation between environmental parameters and microbialdynamicsThe correlation between the microbial (bacterial and archaeal)community and the reactors performance and stability parameterswas investigated by performing a Canonical Correspondence Anal-ysis (CCA). The CCA evidenced significant correlations (p < 0.0001)between the microbial communities (major genera of bacterial andarchaeal communities), and the environmental variables pH, alka-linity and VFA concentration. As shown in Fig. 3, both ordinationaxes of the CCA triplot combine to explain 96.7% of bacterial andarchaeal community variations, indicating that these environmen-tal variables were major factors shaping the microbial communitydynamics. The significance analysis of the environmental variablesrevealed that VFA accounted for much of the difference in bothbacterial and archaeal community distributions arising from theaddition of GTW to the feed. The VFA variable was negatively asso-ciated with the pH and alkalinity variables, as would be expected.The archaeal and bacterial genera shown in Fig. 3 representapproximately 90% and 95% of total sequence abundance ofarchaea and bacteria listed in Tables 4 and 5, respectively. Thegenus C. cloacamonas, represented as B4 in Fig. 3, is located nearthe origin of the plot. This location indicates that the number ofC. cloacamonas sequences had little response to changes in theenvironmental variables during the experiment. This is possiblydue to its syntrophic characteristics. The location of Escherichia(B2) indicates that this genus was more abundant in test digestersthan in the controls and was somewhat more abundant at higherthan average VFA concentration and lower than average pH andalkalinity. In fact the abundance of Escherichia sequences had thelowest coefficient of variation of any bacterial genus and werepresent in greatest abundance at low to moderate VFA concentra-tion. Therefore, the variation in Escherichia sequence abundancemay be due to random error or to an environmental variable notconsidered during the analysis. Additionally, Braak andVerdonschot (1995) indicate that inferences drawn from a triplotare not always accurate because the plot represents multi-dimen-sional relationships in only two dimensions.The sequence abundance of Anaerolinea (B1) was greater at lowVFA concentrations and higher pH and alkalinity. The genusC-1T-1C-2T-2T-2'A1A2A3B1B2B3B4pHVFAAlkalinity-0.6-0.4-0.200.20.40.6-1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2F2 (38.27 %)F1 (58.43 %)Fig. 3. Canonical correspondence analysis (CCA) triplot showing the relationship between the relative sequence abundance of the major bacterial and archaeal communitiesand the environmental variables pH, alkalinity and VFA concentration in the different stages of GTW COD loadings (d). The environmental variables are shown as vectors inthe plot. The open triangle symbols (4) represent the genera of bacterial populations that include B1: Anaerolinea; B2: Esherichia; B3: Dermatophilus; B4: C. cloacamonas. Theopen circle symbols (s) represent the genera of archaeal populations that include A1: Methanomicrobium; A2: Methanosarcina; A3: Methanosaeta.V. Razaviarani, I.D. Buchanan / Bioresource Technology 172 (2014) 232240 239Dermatophilus (B3) represented at most 3% of the total sequencesunder any loading condition and showed very little response toany of the environmental factors considered (Tables 5).In terms of the reactor loading conditions (sites), Fig. 3 showsthat the environmental conditions in reactors T-1 and T-2 weresimilar, as were those in C-1 and C-2. However, the environmentalconditions in reactor T-20 were characterized by higher than aver-age VFA and lower than average pH and alkalinity. The number ofgenus Escherichia (B2) sequences was highest during the moderateGTW loadings (T-1 and T-2), but this genus reached its greatest rel-ative sequence abundance during the overloading condition (T-20).As shown in Fig. 3, it is apparent that the archaeal genera pointsare not as closely clustered around the origin as are those of thebacterial community, indicating a generally greater response tochanges in environmental factors than was the case for the bacte-rial genera. This was expected because methanogenic communitiesare generally more sensitive to changes in environmental condi-tions, including pH, alkalinity and VFA concentration.With respect to the archaeal community, Methanomicrobium(A1) and Methanosaeta (A3) were the dominant methanogensthroughout the study, with the genus Methanosaeta being particu-larly abundant in reactors T-1 and T-2. As shown in Fig. 3, thesequence abundance of Methanosaeta was greatest under averagevalues of the environmental factors considered. This suggests thatthese acetoclasticmethanogenswere important populations duringstable operation and increased in sequence diversity with the addi-tion of GTW at the 150% and 190% relative COD loadings. The num-bers of sequences of genus Methanosarcina (A2) were greatest athigh VFA concentration during the overload condition in T-2. Thisbehavior could be attributed to the higher growth rates of Methan-osarcina at high acetate concentrations and their ability to producemethane via either the acetoclastic or hydrogenotrophic pathway.The 50% reduction in Methanosaeta sequence abundance duringthe overload condition in reactor T-20 compared to that in the samereactor (T-2) at the 190% relative COD loading was accompaniedwith an 80% reduction in methane yield during the overloadingcondition compared to that at the 190% relative COD loading inT-2. Although the sequence abundance of Methanosarcina andMethanomicrobium increased considerably during overloading,these populations could not compensate for the loss of Methanosa-eta sequences in methane production within the 21 days of reactoroperation at the 400% relative COD loading. The triplot shown inFig. 3 indicates that the genus Methanomicrobium (A1) was mostabundant under average environmental conditions and thatsequence numbers were similar under all loading conditions.Recently, a similar archaeal composition was reported duringthe mesophilic anaerobic co-digestion of organic wastes (Ikeet al., 2010; Supaphol et al., 2011). This suggests that co-digestionof mixed organic wastes can promote a superior diversity of nutri-ents which can result in a broader diversity of microbial popula-tions and greater reactor stability and performance. It should benoted that, understanding the role of syntrophic bacteria such asC. cloacamonas within the phylum Spirochaetes and their interac-tions with methanogens is key to the understanding of reactor per-formance and should be investigated further, particularly in theanaerobic co-digestion of mixed-substrates.4. ConclusionsMesophilic co-digestion of GTWwith MWS under stable operat-ing conditions led to enhanced biogas production and methaneyield with acetoclastic methanogens (Methanosaeta) being thedominant population in terms of genus diversity. Increasing theproportion of GTW in digester feed to 300% COD loading relativeto the MWS in the feed resulted in a decline in pH, alkalinity, bio-gas production and methane yield and an increase in VFA concen-tration. At this loading the absolute and relative sequenceabundance of acetoclastic methanogens was reduced while thesequence abundance of hydrogenotrophic methanogens increased.The genus C. cloacamonas of the phylum Spirochaetes remaineddominant at all digester loadings throughout the study. The CCAtriplot indicated that VFA concentration accounted for much ofthe major shifts in microbial sequence abundance.240 V. Razaviarani, I.D. Buchanan / Bioresource Technology 172 (2014) 232240AcknowledgementsFunding for this project was provided by a grant from EPCORWater Services Inc. and a matching CRD grant (CRDPJ 401343 10) from the Natural Sciences and Engineering Research Councilof Canada.Appendix A. Supplementary dataSupplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.biortech.2014.09.046.ReferencesAPHA, 2005. Standard Methods for The Examination of Water and Wastewater,21st ed. American Public Health Association Washangton DC, USA.Ariesyady, H.D., Ito, T., Okabe, S., 2007. Functional bacterial and archaealcommunity structures of major trophic groups in a full-scale anaerobic sludgedigester. Water Res. 41, 15541568.Chen, Y., Cheng, J.J., Creamer, K.S., 2008. Inhibition of anaerobic digestion process: Areview. Bioresour. Technol. 99, 40444064.Conklin, A., Stensel, H.D., Ferguson, J., 2006. Growth kinetics and competitionbetween Methanosarcina and Methanosaeta in mesophilic anaerobic digestion.Water Environ. Res. 78, 486496.DeSantis, T.Z., Hugenholtz, P., Larsen, N., Rojas, M., Brodie, E.L., Keller, K., Huber, T.,Dalevi, D., Hu, P., Andersen, G.L., 2006. Greengenes, a chimera-checked 16SrRNA gene database and workbench compatible with ARB. Appl. Environ.Microbiol. 72, 50695072.Griffin, M.E., McMahon, K.D., Mackie, R.I., Raskin, L., 1998. Methanogenic populationdynamics during start-up of anaerobic digesters treating municipal solid wasteand biosolids. Biotechnol. Bioeng. 57, 342355.Guo, X., Wang, C., Sun, F., Zhu, W., Wu, W., 2014. A comparison of microbialcharacteristics between the thermophilic and mesophilic anaerobic digestersexposed to elevated food waste loadings. Bioresour. Technol. 152, 420428.Hori, T., Haruta, S., Ueno, Y., Ishii, M., Igarashi, Y., 2006. Dynamic transition of amethanogenic population in response to the concentration of volatile fatty acidsin a thermophilic anaerobic digester. Appl. Environ. Microbiol. 72, 16231630.Ike, M., Inoue, D., Miyano, T., Liu, T.T., Sei, K., Soda, S., Kadoshin, S., 2010. Microbialpopulation dynamics during startup of a full-scale anaerobic digester treatingindustrial food waste in Kyoto eco-energy project. Bioresour. Technol. 101,39523957.Kim, S., Bae, J., Choi, O., Ju, D., Lee, J., Sung, H., Park, S., Sang, B.-I., Um, Y., 2014. Apilot scale two-stage anaerobic digester treating food waste leachate (FWL):performance and microbial structure analysis using pyrosequencing. ProcessBiochem. 49, 301308.Kundu, K., Bergmann, I., Klocke, M., Sharma, S., Sreekrishnan, T.R., 2014. Impact ofabrupt temperature increase on the performance of an anaerobic hybridbioreactor and its intrinsic microbial community. Bioresour. Technol. 168, 7279.Leclerc, M., Delgnes, J.-P., Godon, J.-J., 2004. Diversity of the archaeal community in44 anaerobic digesters as determined by single strand conformationpolymorphism analysis and 16S rDNA sequencing. Environ. Microbe 6, 809819.Lee, C., Kim, J., Shin, S.G., OFlaherty, V., Hwang, S., 2010. Quantitative and qualitativetransitions of methanogen community structure during the batch anaerobicdigestion of cheese-processing wastewater. Appl. Microbe Biotechnol. 87,19631973.Lee, S.-H., Park, J.-H., Kang, H.-J., Lee, Y.H., Lee, T.J., Park, H.-D., 2013. Distributionand abundance of Spirochaetes in full-scale anaerobic digesters. Bioresour.Technol. 145, 2532.Long, J.H., Aziz, T.N., Reyes III, F.L.d.l., Ducoste, J.J., 2012. Anaerobic co-digestion offat, oil, and grease (FOG): a review of gas production and process limitations.Process Saf. Environ. 90, 231245.Martn-Gonzlez, L., Castro, R., Pereira, M.A., Alves, M.M., Font, X., Vicent, T., 2011.Thermophilic co-digestion of organic fraction of municipal solid wastes withFOG wastes from a sewage treatment plant: reactor performance and microbialcommunity monitoring. Bioresour. Technol. 102, 47344741.McMahon, K.D., Zheng, D., Stams, A.J., Mackie, R.I., Raskin, L., 2004. Microbialpopulation dynamics during start-up and overload conditions of anaerobicdigesters treating municipal solid waste and sewage sludge. Biotechnol. Bioeng.87, 823834.Padmasiri, S.I., Zhang, J., Fitch, M., Norddahl, B., Morgenroth, E., Raskin, L., 2007.Methanogenic population dynamics and performance of an anaerobicmembrane bioreactor (AnMBR) treating swine manure under high shearconditions. Water Res. 41, 134144.Palatsi, J., Illa, J., Prenafeta-Bold, F.X., Laureni, M., Fernandez, B., Angelidaki, I.,Flotats, X., 2010. Long-chain fatty acids inhibition and adaptation process inanaerobic thermophilic digestion: batch tests, microbial community structureand mathematical modelling. Bioresour. Technol. 101, 22432251.Pelletier, E., Kreimeyer, A., Bocs, S., Rouy, Z., Gyapay, G., Chouari, R., Rivire, D.,Ganesan, A., Daegelen, P., Sghir, A., 2008. Candidatus Cloacamonasacidaminovorans: genome sequence reconstruction provides a first glimpseof a new bacterial division. J. Bacteriol. 190, 25722579.Pereira, M.A., Roest, K., Stams, A.J., Mota, M., Alves, M., Akkermans, A.D., 2002.Molecular monitoring of microbial diversity in expanded granular sludge bed(EGSB) reactors treating oleic acid. FEMS Microbiol. Ecol. 41, 95103.Raskin, L., Poulsen, L.K., Noguera, D.R., Rittmann, B.E., Stahl, D.A., 1994.Quantification of methanogenic groups in anaerobic biological reactors byoligonucleotide probe hybridization. Appl. Environ. Microbe 60, 12411248.Razaviarani, V., Buchanan, I.D., Malik, S., Katalambula, H., 2013a. Pilot-scaleanaerobic co-digestion of municipal wastewater sludge with restaurantgrease trap waste. J. Environ. Manage. 123, 2633.Razaviarani, V., Buchanan, I.D., Malik, S., Katalambula, H., 2013b. Pilot scaleanaerobic co-digestion of municipal wastewater sludge with biodiesel wasteglycerin. Bioresour. Technol. 133, 206212.Rincn, B., Borja, R., Gonzlez, J.M., Portillo, M.C., Siz-Jimnez, C., 2008. Influence oforganic loading rate and hydraulic retention time on the performance, stabilityand microbial communities of one-stage anaerobic digestion of two-phase olivemill solid residue. Biochem. Eng. J. 40, 253261.Supaphol, S., Jenkins, S.N., Intomo, P., Waite, I.S., ODonnell, A.G., 2011. Microbialcommunity dynamics in mesophilic anaerobic co-digestion of mixed waste.Bioresour. Technol. 102, 40214027.Braak, C.J.E., M Verdonschot, P.E., 1995. Canonical correspondence analysis andrelated multivariate methods in aquatic ecology. Aquat. Sci. 57, 255289.Zhu, Z., Hsueh, M.K., He, Q., 2011. Enhancing biomethanation of municipal wastesludge with grease trap waste as a co-substrate. Renew. Energy 36, 18021807.http://dx.doi.org/10.1016/j.biortech.2014.09.046http://dx.doi.org/10.1016/j.biortech.2014.09.046http://refhub.elsevier.com/S0960-8524(14)01297-8/h0005http://refhub.elsevier.com/S0960-8524(14)01297-8/h0005http://refhub.elsevier.com/S0960-8524(14)01297-8/h0010http://refhub.elsevier.com/S0960-8524(14)01297-8/h0010http://refhub.elsevier.com/S0960-8524(14)01297-8/h0010http://refhub.elsevier.com/S0960-8524(14)01297-8/h9000http://refhub.elsevier.com/S0960-8524(14)01297-8/h9000http://refhub.elsevier.com/S0960-8524(14)01297-8/h0020http://refhub.elsevier.com/S0960-8524(14)01297-8/h0020http://refhub.elsevier.com/S0960-8524(14)01297-8/h0020http://refhub.elsevier.com/S0960-8524(14)01297-8/h0030http://refhub.elsevier.com/S0960-8524(14)01297-8/h0030http://refhub.elsevier.com/S0960-8524(14)01297-8/h0030http://refhub.elsevier.com/S0960-8524(14)01297-8/h0030http://refhub.elsevier.com/S0960-8524(14)01297-8/h0040http://refhub.elsevier.com/S0960-8524(14)01297-8/h0040http://refhub.elsevier.com/S0960-8524(14)01297-8/h0040http://refhub.elsevier.com/S0960-8524(14)01297-8/h0045http://refhub.elsevier.com/S0960-8524(14)01297-8/h0045http://refhub.elsevier.com/S0960-8524(14)01297-8/h0045http://refhub.elsevier.com/S0960-8524(14)01297-8/h0055http://refhub.elsevier.com/S0960-8524(14)01297-8/h0055http://refhub.elsevier.com/S0960-8524(14)01297-8/h0055http://refhub.elsevier.com/S0960-8524(14)01297-8/h0065http://refhub.elsevier.com/S0960-8524(14)01297-8/h0065http://refhub.elsevier.com/S0960-8524(14)01297-8/h0065http://refhub.elsevier.com/S0960-8524(14)01297-8/h0065http://refhub.elsevier.com/S0960-8524(14)01297-8/h0070http://refhub.elsevier.com/S0960-8524(14)01297-8/h0070http://refhub.elsevier.com/S0960-8524(14)01297-8/h0070http://refhub.elsevier.com/S0960-8524(14)01297-8/h0070http://refhub.elsevier.com/S0960-8524(14)01297-8/h9005http://refhub.elsevier.com/S0960-8524(14)01297-8/h9005http://refhub.elsevier.com/S0960-8524(14)01297-8/h9005http://refhub.elsevier.com/S0960-8524(14)01297-8/h9005http://refhub.elsevier.com/S0960-8524(14)01297-8/h0080http://refhub.elsevier.com/S0960-8524(14)01297-8/h0080http://refhub.elsevier.com/S0960-8524(14)01297-8/h0080http://refhub.elsevier.com/S0960-8524(14)01297-8/h0080http://refhub.elsevier.com/S0960-8524(14)01297-8/h0085http://refhub.elsevier.com/S0960-8524(14)01297-8/h0085http://refhub.elsevier.com/S0960-8524(14)01297-8/h0085http://refhub.elsevier.com/S0960-8524(14)01297-8/h0085http://refhub.elsevier.com/S0960-8524(14)01297-8/h0090http://refhub.elsevier.com/S0960-8524(14)01297-8/h0090http://refhub.elsevier.com/S0960-8524(14)01297-8/h0090http://refhub.elsevier.com/S0960-8524(14)01297-8/h0100http://refhub.elsevier.com/S0960-8524(14)01297-8/h0100http://refhub.elsevier.com/S0960-8524(14)01297-8/h0100http://refhub.elsevier.com/S0960-8524(14)01297-8/h0105http://refhub.elsevier.com/S0960-8524(14)01297-8/h0105http://refhub.elsevier.com/S0960-8524(14)01297-8/h0105http://refhub.elsevier.com/S0960-8524(14)01297-8/h0105http://refhub.elsevier.com/S0960-8524(14)01297-8/h0110http://refhub.elsevier.com/S0960-8524(14)01297-8/h0110http://refhub.elsevier.com/S0960-8524(14)01297-8/h0110http://refhub.elsevier.com/S0960-8524(14)01297-8/h0110http://refhub.elsevier.com/S0960-8524(14)01297-8/h0115http://refhub.elsevier.com/S0960-8524(14)01297-8/h0115http://refhub.elsevier.com/S0960-8524(14)01297-8/h0115http://refhub.elsevier.com/S0960-8524(14)01297-8/h0115http://refhub.elsevier.com/S0960-8524(14)01297-8/h0120http://refhub.elsevier.com/S0960-8524(14)01297-8/h0120http://refhub.elsevier.com/S0960-8524(14)01297-8/h0120http://refhub.elsevier.com/S0960-8524(14)01297-8/h0120http://refhub.elsevier.com/S0960-8524(14)01297-8/h0125http://refhub.elsevier.com/S0960-8524(14)01297-8/h0125http://refhub.elsevier.com/S0960-8524(14)01297-8/h0125http://refhub.elsevier.com/S0960-8524(14)01297-8/h0125http://refhub.elsevier.com/S0960-8524(14)01297-8/h0130http://refhub.elsevier.com/S0960-8524(14)01297-8/h0130http://refhub.elsevier.com/S0960-8524(14)01297-8/h0130http://refhub.elsevier.com/S0960-8524(14)01297-8/h0135http://refhub.elsevier.com/S0960-8524(14)01297-8/h0135http://refhub.elsevier.com/S0960-8524(14)01297-8/h0135http://refhub.elsevier.com/S0960-8524(14)01297-8/h0140http://refhub.elsevier.com/S0960-8524(14)01297-8/h0140http://refhub.elsevier.com/S0960-8524(14)01297-8/h0140http://refhub.elsevier.com/S0960-8524(14)01297-8/h0145http://refhub.elsevier.com/S0960-8524(14)01297-8/h0145http://refhub.elsevier.com/S0960-8524(14)01297-8/h0145http://refhub.elsevier.com/S0960-8524(14)01297-8/h0150http://refhub.elsevier.com/S0960-8524(14)01297-8/h0150http://refhub.elsevier.com/S0960-8524(14)01297-8/h0150http://refhub.elsevier.com/S0960-8524(14)01297-8/h0150http://refhub.elsevier.com/S0960-8524(14)01297-8/h0160http://refhub.elsevier.com/S0960-8524(14)01297-8/h0160http://refhub.elsevier.com/S0960-8524(14)01297-8/h0160http://refhub.elsevier.com/S0960-8524(14)01297-8/h0170http://refhub.elsevier.com/S0960-8524(14)01297-8/h0170http://refhub.elsevier.com/S0960-8524(14)01297-8/h0175http://refhub.elsevier.com/S0960-8524(14)01297-8/h0175Reactor performance and microbial community dynamics during anaerobic co-digestion of municipal wastewater sludge with restaurant grease waste at steady state and overloading stages1 Introduction2 Methods2.1 Inoculum and substrates2.2 Reactor operation and loading protocol2.3 Physico-chemical analysis2.4 Microbial community analysis2.4.1 Sludge sampling and DNA extraction2.4.2 Pyrosequencing analysis2.5 Statistical analysis3 Results and discussion3.1 Process stability and reactor performance3.2 Methanogenic community structure of the ACD3.3 Bacterial community structure of the ACD3.4 Correlation between environmental parameters and microbial dynamics4 ConclusionsAcknowledgementsAppendix A Supplementary dataReferences

Recommended

View more >