Oa

LD

a

ARR2AA

KBKLRS

1

eob2bepictfttio

CT

h0

Journal of Biotechnology 192 (2014) 179–186

Contents lists available at ScienceDirect

Journal of Biotechnology

j ourna l ho me page: www.elsev ier .com/ locate / jb io tec

utlining microbial community dynamics during temperature dropnd subsequent recovery period in anaerobic co-digestion systems

eticia Regueiro ∗, Marta Carballa, Juan M. Lemaepartment of Chemical Engineering, Institute of Technology, University of Santiago de Compostela, 15782 Santiago de Compostela, Spain

r t i c l e i n f o

rticle history:eceived 19 August 2014eceived in revised form6 September 2014ccepted 7 October 2014vailable online 16 October 2014

eywords:acteroidetes

a b s t r a c t

To improve the stability of anaerobic reactors, more knowledge is required about how the different com-munities react against operating perturbations and which specific ones respond better. The objectiveof this work was to monitor the changes in microbial community structure of an anaerobic digesterduring a temperature drop by applying different complementary molecular techniques. Temperaturedecrease led to an increase of Bacteroidales order, Porphyromonadaceae family and Bacteroides genus anda decrease in Syntrophomonas and Clostridium genera. Once the temperature was restored, the reactorrecovered the steady state performance without requiring any modification in operational conditions orin the microbiome. During the recovery period, Sedimentibacter genus and Porphyromonadaceae family

ineticsow temperature performanceecovery periodyntrophomonadaceae

played an important role in the degradation of the accumulated volatile fatty acids. The hydrogenotrophicmethanogens appeared to be the keystone archaeal population at low temperatures as well as in therecovery period. This study stands out that the understanding of microbial community dynamics dur-ing temperature drop could be utilized to develop strategies for the mitigation of temperature changeconsequences and speed up the recovery of stable reactor performance.

© 2014 Elsevier B.V. All rights reserved.

. Introduction

Anaerobic digesters are often exposed to operational and/ornvironmental perturbations, such as temperature drops, organicverloadings, or the entry of toxic compounds, leading to insta-ility due to a stressed microbial community (Kleyböcker et al.,012). Reactor microbiome responds differently to distinct distur-ances, resulting in an imbalance in the trophic network (Leitãot al., 2006). This usually causes an accumulation of intermediateroducts, such as volatile fatty acids (VFAs), which may in turn

nhibit microbial numbers, eventually reducing the process effi-iency. Allison and Martiny (2008) have indicated the necessityo understand the relationship between the microbiome and theirunction in the anaerobic process, especially in deteriorated sys-ems in order to identify those populations that better respond

o each individual operational stressor with the ultimate goal ofmproving the stability of anaerobic reactors and the control basedn microbial management.

∗ Corresponding author at: Instituto de Investigacións Tecnolóxicas (IIT)/Constantino Candeira s/n, 15782 Santiago de Compostela, Spain.el.: +34 881 816016; fax: +34 881 816702.

E-mail address: leticia.regueiro.abelleira@gmail.com (L. Regueiro).

ttp://dx.doi.org/10.1016/j.jbiotec.2014.10.007168-1656/© 2014 Elsevier B.V. All rights reserved.

One common perturbation occurring in anaerobic reactors istemperature variations (Poh and Chong, 2009). Many studies haveinvestigated microbiomes in anaerobic reactors working at dif-ferent temperatures and they point out that community profilesvaried in each temperature range. Lee et al. (2012) detected alarge presence of Firmicutes in thermophilic range compared tomesophilic one, and the opposite trend for Bacteroidetes phylum.Leven et al. (2007) showed that Chloroflexi and Bacteroidetes rep-resented the major phyla at mesophilic temperatures, whilst thephylum Thermotogae was the dominant group in the thermophilicrange. In relation to the archaeal community, Methanomicrobialesdominated at 15 ◦C, and Methanosaeta and Methanomicrobialeswere the most abundant at 37 ◦C (Bialek et al., 2012). Summariz-ing, these studies confirm the temperature as a parameter affectingthe structure of the microbial communities, even more than theincrease in the organic loading rate (OLR) (Guo et al., 2014), butthey lack transient membership of the microbial community dur-ing the days of the temperature perturbation. Moreover, to the bestof our knowledge, there have been no studies published, whichmonitored the microbial community dynamics during the recov-

ery period after restoration of temperature. All this understandingcould be utilized to develop strategies for the mitigation of tem-perature change consequences and speed up the recovery of stablereactor performance based on microbial knowledge.

180 L. Regueiro et al. / Journal of Biotech

Table 1Physico-chemical characteristics of pig manure (PM), molasses residues (MR) andfish waste (FW).

PMa MR FW

TS (g kg−1) 50 ± 15 835 304VS (g kg−1) 40 ± 10 707 282TKN-N (g N kg−1) 3.1 ± 1.0 57 19NH4

+-N (g N kg−1) 2.8 ± 0.7 15 0.7CODtotal (g O2 kg−1) 62 ± 20 723 567TA (g CaCO3 kg−1) 10.5 ± 3.0 N.D. N.D.PA (g CaCO3 kg−1) 4.5 ± 2.0 N.D. N.D.Lipids (g kg−1) N.D. N.D. 35

TS: total solids; VS: volatile solids; TKN-N: total Kjeldahl nitrogen; NH4+-N: ammo-

nium; CODtotal: total chemical oxygen demand; TA: total alkalinity; PA: partiala

t

id(itmc(c

2

2

hw(mcst(dd

swis

2

2

2aTfddatGBmst

lkalinity; N.D: not detected.a Standard deviations are only shown for pig manure since different batches of

his substrate were necessary during the whole experiment (over 400 days).

Therefore, the objective of this work was to monitor the changesn microbial community structure of an anaerobic reactor not onlyuring a temperature drop from mesophilic (37 ◦C) to psychrophilic17 ◦C) range but also during the recovery period after restor-ng temperature to 37 ◦C. The experimental design consisted ofwo types of perturbation (gradual and abrupt) and three comple-

entary molecular techniques were applied to follow microbialommunity dynamics: Denaturing Gradient Electrophoresis GelDGGE), Fluorescence In Situ Hybridization (FISH) and a 16s rRNAharacterization with Illumina MiSeq platform sequencing.

. Materials and methods

.1. Substrates and inoculum

Pig manure (PM) was obtained from a fattening hog farm (3000eads) located in Santiago de Compostela (Spain). Fish waste (FW)as collected in a canning industry located on the Galician coast

Spain) and consisted of processing remains of albacore. Beetolasses residues (MR) were obtained from a sugar processing

ompany in A Coruna (Spain) and consisted of by-products of theugar-extraction process. Fish waste was grounded prior to use andhe substrates were stored at 4 ◦C. All residues were characterizedTable 1) in terms of total and volatile solids content, total Kjel-ahl nitrogen and ammonium concentration, total chemical oxygenemand (COD), total and partial alkalinity, and lipid content.

A mixture (50:50, v/v) of two anaerobic sludges, one from aewage sludge anaerobic digester and the second from a breweryastewater anaerobic reactor, was used as inoculum. The initial

n-reactor inoculum concentration was 15 g of volatile suspendedolids (VSS) per liter.

.2. Anaerobic digester operation

.2.1. Experimental setupOne continuously stirred tank reactor (160 rpm, Heidolph RZR

041), with a working volume of approximately 10 L, was oper-ted in two consecutive experimental runs (Experiments 1 and 2).he reactor was fed semi-continuously (once a day draw-off andeeding) with a mixture of PM, FW, and MR (60–20–20, COD basis)uring both experiments. This mixture was prepared every week,iluted with tap water according to the applied OLR, and storedt 4 ◦C. Temperature, pH, stirring speed, biogas production (Rit-er milligascounters, Dr. Ing. Ritter Apparatebau GmbH, Bochum,ermany) and biogas composition (AwiFLEX model from AWITE

ioenergie GmbH) were monitored on-line. Samples of reactorixed liquor were taken twice a week for VFAs, total COD, total

uspended solids (TSS), VSS, alkalinity and ammonium determina-ions. Biomass samples were taken weekly for molecular analysis,

nology 192 (2014) 179–186

except during the perturbation period, when a more frequentsampling was conducted (every 2–3 days).

2.2.2. Operational strategyExperiment 1 was divided in three periods: start-up (0–30 days),

gradual increase of OLR from 0.5 to 2 g COD L−1 d−1 and steady stateperformance at the latter OLR (31–150 days), and the perturbationperiod (151–195). The same periods were applied in Experiment 2,but the perturbation period lasted from day 151 to 161 and a fourthperiod, which we refer to as the recovery period (162–212 days),was also included. In addition, the perturbation differed betweenboth experimental runs. In Experiment 1, a gradual temperaturedecrease was conducted by lowering the temperature from 37 to17 ◦C at a rate of 2 ◦C per day and keeping the latter temperatureconstant during the last 35 operating days (around 2 hydraulicretention times). The temperature was modified by adjusting thetemperature set point of the surrounding water bath. In Experi-ment 2, an abrupt perturbation (the temperature dropped from37 to 17 ◦C (room temperature) in 24 h) was performed by turn-ing off the water bath. After working at 17 ◦C for 10 days, thewater bath was again connected at day 161 (the temperature ofthe reactor rose to 37 ◦C in one day) and maintained until theend of the Experiment 2. Moreover, the effect of using a differ-ent PM batch was also evaluated in Experiment 2, since typicallythe manure has different characteristics depending on the seasonalperiod.

2.3. Analytical methods

VFA (acetic, propionic, i-butyric, n-butiric, i-valeric and n-valeric) were analyzed by gas chromatography (HP, 5890A)equipped with a Flame Ionization Detector (HP, 7673A). COD, solids,TKN, ammonium, alkalinity and lipids were determined accordingto standard methods (APHA, 1998).

2.4. Molecular techniques

A selection of biomass samples were analyzed with DGGE plussequencing, FISH and also 16s rRNA characterization with the Illu-mina MiSeq platform with the main focus of determining themicrobial community dynamics during the perturbation and recov-ery periods. In this way, 21 samples from Experiment 1 (days 0, 15,48, 66, 85, 95, 104, 124, 135, 145, 150, 152, 154, 157, 159, 161, 164,171, 178, 185 and 194) and 14 samples from Experiment 2 (days0, 49, 98, 150, 152, 155, 157, 161, 164, 168, 173, 181, 197 and 211)were selected.

2.4.1. Denaturing gradient gel electrophoresis and sequencingDNA extraction, PCR, DGGE and sequencing were performed

according to Regueiro et al. (2012), based on the primers U968-fand L1401-r for Bacteria and the primers A109 (T)-f and 515-r forArchaea. Cluster analysis was conducted using Bionumerics soft-ware v.6.1 (Applied Maths, Sint-Martens-Latem, Belgium).

2.4.2. Fluorescent in situ hybridizationProbe sequences and formamide concentrations were applied

according to probeBase following the protocol explained byRegueiro et al. (2012). The probes used were: Eub338mix (Bacteria),CFX1223 (Chloroflexi), CFB562 (Bacteroidetes), LGC354 (Firmicutes),Arc915 (Archaea), Ms821 (Methanosarcina), Mx825 (Methanosaeta)

and MB1174 (Methanobacteriales). The abundance of each popula-tion was qualitatively evaluated by DAIME program (Daims et al.,2006). At least six photos were taken per 20 �L of fixed sample.

L. Regueiro et al. / Journal of Biotechnology 192 (2014) 179–186 181

Fig. 1. Volumetric biogas production and organic loading rate applied during Experiment 1 (A) and Experiment 2 (B) and volatile fatty acids concentrations in Experiment 1( turbe

2

5geffafiEcQa1iDCwa

upqt2twtu2apm

C) and Experiment 2 (D) during the last operational days, when the reactor was dis

.4.3. 16s rRNA characterization using Illumina MiSeq platformFor Illumina MiSeq, the extracted DNA was PCR amplified using

15-forward primers and 806-reverse Golay barcoded primers tar-eting the V4 region of the 16S rRNA gene according to Gilbertt al. (2010). The PCR cycle conditions consisted of 94 ◦C for 3 minor initial denaturation; 25 cycles for denaturation at 94 ◦C for 45 sollowed by annealing at 50 ◦C for 60 s, elongation at 72 ◦C for 90 snd final elongation for 10 min at 72 ◦C. PCR products were con-rmed by gel electrophoresis and then cleaned using the MagBind-Z pure cleaning kit (Omega Bio-tek Inc., Norcross, GA, USA). DNAoncentration of the amplicons was quantified using Invitrogenuant-It Pico Green DNA quantification kit (Life Technologies, USA)nd samples were subsequently pooled at equimolar ratios (around00 ng of each sample). Libraries were sent in for pair-end sequenc-

ng (2 × 250 bp) on the Illumina MiSeq platform (Illumina, Saniego, CA, USA) at the Cornell University Biotechnology Resourceenter (Ithaca, NY, USA). Only bacterial population was followedith Illumina, since the primers were not appropriate to study the

rchaeal community.Computational analysis of the sequencing reads was performed

sing the Quantitative Insights into Microbial Ecology (QIIME v1.7)latform (Caporaso et al., 2010). After joining the paired-end reads,uality filtering (using the default values in QIIME with the excep-ion that the minimum acceptable Phred quality score was set to5) and demultiplexing was performed. Closed-reference opera-ional taxonomic unit (OTU) picking with the default uclust methodas used to group sequences into OTUs at 97% identity. Represen-

ative sequences selected for each OTU were assigned taxonomysing the Greengenes reference database (Werner et al., 2011), May

013 version. OTU assignment resulted in 3362 OTUs total witht least one read per OTU. Principal coordinate analysis (PCoA)lot was generated based on weighted Unifrac pairwise distanceatrices.

d.

3. Results and discussion

3.1. Anaerobic reactor performance

The first two periods were practically identical inboth experimental runs. During the start-up with anOLR of 0.5–0.7 g COD L−1 d−1, the biogas production was≈0.22–0.35 L L−1 d−1 (data not shown) corresponding to amethanization efficiency of 50%, while it increased up to≈0.85–0.95 L L−1 d−1 when the OLR increased to 2 g COD L−1 d−1

(Fig. 1), with a methanization efficiency ≈65%. Throughout thesteady-state performance at an OLR of ≈2 g COD L−1 d−1 andprior to both perturbation periods, the ammonium, TSS and VSSconcentrations were around 2 g L−1, 6 g L−1, and 5 g L−1, respec-tively. In addition, VFAs levels were negligible, the pH rangedbetween 7.5–7.9 and the alkalinity was around 6–7 g CaCO3 L−1.This indicates that the digester performed under stable conditions.

When temperature gradually decreased in Experiment 1(Fig. 1A), a drop in the biogas production occurred along with agradual accumulation of VFAs (Fig. 1C). In only 10 days (day 160),the biogas production decreased by ≈50% (Fig. 1A), but once thetemperature remained constant at 17 ◦C, the biogas productionstabilized to ≈0.3 L L−1 d−1, corresponding to a COD removal effi-ciency of around 18–20%. During this perturbation period, aceticand propionic acids accumulated in the reactor, reaching peak con-centrations of 2.5 and 2 g L−1, respectively, at day 170 (Fig. 1C),but then they decreased and stabilized at approximately 1 g L−1.The concentrations of butyric and valeric acids remained below0.2 and 0.3 g L−1, respectively, over the entire perturbation time.

Despite the high VFAs levels, no “souring” behavior was observedin the digester, since the pH values stayed constant and in the neu-tral range (7.5–7.8), probably due to the high alkalinity levels ofthe reactor (around 7–8 g CaCO3 L−1) provided by the manure. The

182 L. Regueiro et al. / Journal of Biotechnology 192 (2014) 179–186

F C (daya

aid

wdpTidtth

ittt

3

ht(ocacrap

pi3h(sth(aatCts

ig. 2. COD balances in steady-state conditions at 37 ◦C (days 132–150) and at 17 ◦

nd methanogenic activity at 17 ◦C and 37 ◦C.

mmonium concentration decreased to 1.5 g L−1 and TSS and VSSncreased to 7 and 6 g L−1, respectively, probably due to a loweregradation of proteins and organic material at low temperature.

In Experiment 2, the abrupt temperature drop provoked a step-ise decrease in the biogas production from 0.98 to 0.45 L L−1 d−1

uring 10 days (Fig. 1B) with the concomitant accumulation of pro-ionic and acetic acids up to 2.5 and 2 g L−1, respectively (Fig. 1D).hese maximum values were similar to those obtained in Exper-ment 1 (Fig. 1C), so the type of perturbation (gradual or abrupt)id not influence the system response at macroscopic level inerms of volatile fatty acids accumulation, as well as biogas produc-ion, although valeric and butyric acid concentrations were slightlyigher (Fig. 1D).

At day 161, when the biogas production had dropped 50% (asn Experiment 1), the temperature was returned to 37 ◦C initiatinghe recovery period, and less than 5 days were necessary to recoverhe biogas production rate (around 1 L L−1 d−1). In contrast, moreime was needed to degrade the accumulated VFAs (Fig. 1D).

.2. Influence of temperature on reactor kinetics

To evaluate the effect of temperature on theydrolytic–acidogenic and methanogenic activities in the reactor,he COD balances during the steady-state at high (37 ◦C) and low17 ◦C) temperature were calculated (Fig. 2). The biological activityf the hydrolytic–acidogenic population at each temperature wasalculated in terms of particulate hydrolytic–acidogenic activitynd soluble hydrolytic–acidogenic activity considering in bothases the COD degraded (particulate and soluble, respectively) pereactor’s volume (10 L) and per day (17 days). The methanogenicctivity was determined as the methane generated (in COD basis)er liter reactor and per day.

It can be observed that temperature strongly affected thearticulate hydrolytic–acidogenic activity, since almost no activ-

ty was observed at 17 ◦C (0.05 g COD L−1 d−1). Meanwhile, at7 ◦C, it was around 0.61 g COD L−1 d−1 (Fig. 2). The solubleydrolytic–acidogenic activity varied from 0.71 (37 ◦C) to 0.2817 ◦C) g COD L−1 d−1 (Fig. 2), so microbial communities degradingoluble matter were less affected than those degrading par-iculate COD at 17 ◦C. Moreover, both particulate and solubleydrolytic–acidogenic activities were similar at 37 ◦C. Alvarez et al.2008) showed that when the anaerobic reactor performance faces

temperature decrease, the hydrolysis rates of both particulatend soluble organic matter decrease, but our results indicate that

he particulate one was much more affected, since the particulateOD was negligible hydrolyzed by anaerobic bacterial communi-ies at 17 ◦C. Moreover, Komemoto et al. (2009) indicated that theolubilization rate treating food waste was accelerated under both

s 172–190) and calculated particulate and soluble hydrolytic–acidogenic activities

mesophilic and thermophilic conditions; but it was particularlyhigh under mesophilic ones, indicating a clear relationship betweensolubilization and operating temperature.

The methanogenic activity was also lower at 17 ◦C, around0.35 g COD L−1 d−1, than at 37 ◦C (1.34 g COD L−1 d−1) (Fig. 2). This4-fold decrease in the methanogenic activity was higher thanthe decline in the biogas production rate (around 50%), but itcan be explained by the fact that methane content in the bio-gas lowered as well, from 60% to 45%. McHugh et al. (2004) alsoobserved a decrease in the specific methanogenic activity between2 to 4 fold (depending on the synthetic substrate treated (acetate,ethanol, etc.)) working at 22 ◦C respect to the activity at 37 ◦C.In addition, the methanogenic activity was similar to the sum ofboth hydrolytic–acidogenic activities, regardless the temperature,explaining the steady-state performance. Residual VFA levels at17 ◦C are the consequence of the VFAs accumulated during theperturbation period.

3.3. Microbiome evolution due to acidification by temperaturedecrease

During the stable performance period prior to temperature drop(days 100–150), the microbial analysis of the bacterial and archaealcommunity structures following the DGGE patterns revealed a highlevel of consistency, and only small differences were found duringthe gradual increase of the OLR (days 30–100), such as the dis-appearance of one Chloroflexi band-related when the OLR reached1.2 g COD L−1 d−1 (Figure S1) in Experiment 1. Presence of Chlo-roflexi was also followed by FISH, confirming its activity at low OLR(data not shown). In the archaeal community, results during non-perturbed periods detected higher levels of Methanosaeta species(data not shown).

The bacterial community prior to temperature drop (Fig. 3,day 150), was mainly represented by Firmicutes (65–75%) andBacteroidetes (15–25%) phyla, being Syntrophomonadaceae andClostridiaceae the most abundant families (Fig. 4) and Syn-trophomonas and Clostridium the main genera (Fig. 5), in bothexperimental runs. However clear differences were observed inthe dominant families between Experiment 1 and 2 at day 150.This was also achieved in the DGGE tracks (Fig. 6), since althoughqualitatively the populations identified were similar, excludingthe non-presence of Firmicutes related-band in Experiment 2,their bands profile varied. The archaeal community was dom-inated by Methanosarcinales (DGGE, data not shown), mainly

Methanosaetaceae and Methanosarcinaceae communities (FISH,Table 2), although the Methanosarcinales (sum of Methanosaetaceaeand Methanosarcinaceae) presence was higher (65%) in Experiment1 than in the second one (35%).

L. Regueiro et al. / Journal of Biotechnology 192 (2014) 179–186 183

Fig. 3. Relative abundance of the most prevalent bacterial phyla in Experiments 1 and 2. From the total phylum, only those that exceeded 1% in abundance have been takeninto account for the analysis.

F and 2i

d(r(t

TSo

ig. 4. Relative abundance of the most prevalent bacterial families in Experiments 1nto account for the analysis.

During gradual temperature decrease (Experiment 1), the abun-ance of Bacteroidetes phylum increased in detriment of FirmicutesFig. 3 and Table 2). The reduction of Firmicutes was mainly

elated with the decrease in two families, Syntrophomonadaceae8–9%) and Clostridiaceae (9–10%) (Fig. 4), and specifically inwo genera Clostridium and Syntrophomonas (Fig. 5), whereas

able 2emi-quantitative percentages of Bacteria–Archaea (B–A), Methanosaeta and Methanosarcine, at different operational days during Experiment 1 (E1) and Experiment 2 (E2) determ

Sampling time Ratio B–A (%) Methanosaeta (%)

E1 E2 E1 E2 E1 E2

150 150 50–50 60–40 45 30

159 152 60–40 70–30 30 5

171 157 70–30 70–30 10 2

178 168 80–20 70–30 20 20

194 197 80–20 70–30 30 25

. From the total family, only those that exceeded 1% in abundance have been taken

Bacteroidetes increase was associated to Bacteroidales order andBacteroidaceae family (Fig. 4). Talbot et al. (2008) showed a decreasein Syntrophomonadaceae in a poor performing anaerobic digester,

suggesting an inhibitory effect by high acetate concentrationson this population. Similarly, Regueiro et al. (2012) detectedhigher abundance of Syntrophomonas in good performing industrial

na in the archaeal population, and Bacteroidetes and �Proteobacteria in the bacterialined by FISH.

Methanosarcina (%) Bacteroidetes (%) �Proteobacteria (%)

E1 E2 E1 E2 E1 E2

20 5 5 2 2 540 15 15 8 5 1260 25 20 15 10 1040 10 25 20 15 1220 10 25 15 18 5

184 L. Regueiro et al. / Journal of Biotechnology 192 (2014) 179–186

F and 2i

rptrdcumCtcbc(ttt(r

ig. 5. Relative abundance of the most prevalent bacterial genus in Experiments 1

nto account for the analysis.

eactors, indicating that a high percentage of this family is a wellerformance indicator. Cirne et al. (2007) observed a decrease inhe genus Clostridium (included in Clostridiaceae family) when theate of hydrolysis became a limiting factor in the overall degra-ation. This is consistent with the results of our study, where alear decrease in the hydrolytic activity, mainly over the partic-late COD (Fig. 2), occurred. Therefore, Clostridiaceae are familyembers that seem to be responsible of degrading the particulate

OD of solid residues. The higher presence of Bacteroidetes at loweremperature suggests a major importance of this phylum on theonversion of organic material at 17 ◦C, as previously mentionedy Leven et al. (2007). A little increase in the Proteobacteria per-entage was also detected during gradual temperature decreaseFig. 3) and FISH confirmed that the abundance of Gammapro-eobacteria species was higher (from about 2% to 15–18% of the

otal active Bacteria) by the end of this experimental run. In con-rast, this variation could not be clearly appreciated at family levelFig. 4), although a band closely matched to Gammaproteobacte-ia class and one Pseudomonas genus related-band appeared at

Fig. 6. DGGE of Bacteria during Experiment 1 (A) and Ex

. From the total genus, only those that exceeded 1% in abundance have been taken

low temperatures in DGGE gel (Fig. 6A). Therefore, the presenceof Pseudomonas seems to be also important to lead with low tem-peratures (Duran et al., 2006). Ruminococcaceae band was weakerwith time (Fig. 6A), but it did not disappear completely duringthe gradual perturbation. This variation was not detected in thesequencing results from the Illumina platform (Fig. 4), probablyexplained by the fact that Ruminococcaceae family comprises a greatnumber of different OTUs, whereas the DGGE band means one spe-cific microorganism. In other words, the influence of the weakerRuminococcaceae specie in the total family percentage was almostnegligible (Fig. 4).

FISH analyses revealed that after the gradual perturbation thetotal archaeal abundance decreased, and the ratio Bacteria–Archaeachanged from 50–50 at day 150 to 80–20 at day 194 in Experiment1 (Table 2). A significant decrease in the methanogenic population

(from 34% to 2–5%) was also observed by Liu et al. (2002) with theincrease in VFAs concentration.

Archaeal community data showed a distinct shift fromMethanosaeta dominance to Methanosarcina dominance in

periment 2 (B) with specific populations marked.

iotech

rpsistcsat(s2mabodabeti5tna

i(tBabtbiidco

3

obftaMe(ptrt

dtibav

c

L. Regueiro et al. / Journal of B

esponse to gradual temperature drop (Table 2) in the acetoclasticathway. After 10 days working at 17 ◦C (day 171), Methanosarcinahowed its maximum percentage (60%) coinciding with the max-mum VFAs concentration (Fig. 1 C). Methanosarcina is able tohift its methanogenic pathway according to acetic acid level dueo their metabolism, making them more flexible against severalhanges (Qu et al., 2009). In psychrophilic anaerobic digestionystems, the hydrogenotrophic pathway is more favorable than thecetoclastic one (Lettinga, 1999). The reason is supported by thehermodynamics of the different conversions at low temperatureKotsyurbenko, 2005) since the hydrogenotrophic methanogenictep requires less energy than the acetoclastic one (Lettinga et al.,001). De Vrieze et al. (2012) also found that Methanosarcina com-unities are able to tolerate several environmental stresses, such

s high acetate. Members of Methanobacteriales only representedetween 2 and 3% of the total Archaea population during the wholeperation time (data not shown), although it increased slightlyuring the operation at 17 ◦C (3–5%), indicating that this order had

minimum role regardless the operating temperature. One DGGEand related with Methanomicrobiales population appeared at thend of the Experiment 1 (data not shown), specifically relatedo Methanolinea, but this could not be elucidated by FISH. Yet, its important to note that using the FISH technique, only around0–60% of the total archaeal population was deciphered. Probablyhe remaining fraction consisted of Methanomicrobiales commu-ities, well-known to play important roles in low-temperaturenaerobic reactors (Collins et al., 2006).

Similarly to Experiment 1, the abrupt temperature drop (Exper-ment 2, Fig. 3, day 157) provoked a decrease in Firmicutes phylumSyntrophomonadaceae and Clostridiaceae families (Fig. 4), and Syn-rophomonas and Clostridium genera (Fig. 5)) and an increase inacteroidetes one (Fig. 3, day 157). Gammaproteobacteria percentagelso increased from 5 to 12% between days 150 and 160 (Table 2),ut in contrast to Experiment 1, Pseudomonadaceae seems noto vary after abrupt temperature decrease (Fig. 6A) and generaelonging to Tissierellaceae family, such as Gallicola, experienced an

mportant increase in the abrupt temperature drop (Fig. 5). Follow-ng the archaeal community, Methanosarcina species increased inetriment of Methanosaeta (Table 2); however, the maximum per-entages (25%) were lower than in experiment 1 (60%), pointingut that gradual drop was more severe than the abrupt change.

.4. Gradual vs. abrupt temperature decrease

In general terms, the same trend was observed in both typesf perturbation (gradual and abrupt), regardless of the manureatch used. Therefore the microbial lineages responsible to con-ront a temperature drop, independently of the substrate fed orhe shock length, belonged to Bacteroidales order and Pseudomon-daceae family, in bacterial domain, and to Methanosarcina andethanomicrobiales in the archaeal one. However, some differ-

nces could be observed between both experiments. Tissierellaceaemainly Gallicola (Fig. 5)) seems to be important during abrupt tem-erature drop, but its presence did not varied significantly duringhe gradual perturbation (Fig. 4). These differences are probablyelated to the manure used rather than to the temperature varia-ion.

PCoA analysis allowed separating between the two variablesriving the dynamics of microbial community structure (Figure S2):emperature and the manure batch. The great influence of substraten the microbial communities present in anaerobic digesters haseen already reported (Regueiro et al., 2014; Zhang et al., 2014),

nd here this fact was quite obvious, since it explained microbialariability between both experiments.

Moreover, samples from 170 on are clustered with the samplesollected when the reactor was working at OLR of 1.2 g COD L−1 d−1

nology 192 (2014) 179–186 185

(Figure S3). This microscopic result is consistent with the macro-scopic performance, since the biogas production rate at 17 ◦C (days170–191), working at OLR of 2 COD L−1 d−1, was similar to thatachieved at 37 ◦C when the reactor was working at lower OLR,around 1–1.2 g COD L−1 d−1 (data not shown). Regueiro et al. (2012)have been previously found a correlation between microbial com-munity structure and microbial degradation rates, which has beenconfirmed with these results.

3.5. Recovery period implied collaboration of different phyla

During the recovery period after temperature shock, bacterialcommunity did not recover the previous structure prior to abrupttemperature drop during the first days working at 37 ◦C again(Figs. 3–5B). However, the DGGE gel shows similar pattern on day211, once the VFAs concentration decreased (Fig. 5B). It indicates, atleast, a partial recovery to the previous microbial state. The changesinitiated during temperature perturbation were accentuated dur-ing the first days of the recovery period. Thereby, Bacteroidetesphylum (Table 2) and Porphyromonadaceae family (Fig. 4) continuedto grow.

Clostridium, Lachnospiraceae, Spirochaetes (Fig. 6B) and Sedi-mentibacter (Fig. 5) appeared or increased its presence and theyremained until VFAs levels decreased, even when the temperaturehad already been several days at 37 ◦C. Once the VFAs concen-tration was practically negligible, these populations disappearedaccording to DGGE data (Fig. 6B). Spirochaetes role in anaerobicdigestion is not clear yet (Delbes et al., 2000; Lee et al., 2013),but some recent results suggest that they could be related withthe metabolism of acetate (Lee et al., 2013). Our findings cor-roborated this function, since this phylum appeared around day180, and from this day on, the concentration of acetate startedto decrease (Fig. 1D). The appearance of the spore-forming grampositive bacteria Clostridium and Lachnospiraceae at high VFA con-centrations was previously reported (Town et al., 2014). Withinthe archaeal community, the percentage explained by FISH wasquite low. Apparently other archaeal species, like Methanomi-crobiales, had a strong effect in the methane production duringthis recovery period. But it seems that once the VFAs were con-sumed, Methanosaeta percentage increased. Similar to the bacterialdomain, the archaeal community is intended to recover the initialsteady state prior to perturbation period.

Therefore, once the mesophilic range was re-established, thereactor returned to a similar steady state performance withoutrequiring any modification in operational conditions (lower OLR,stop feeding, etc.) or in the microbiome (reinoculation, bioaugmen-tation, etc.). Once more the latter indicates the resilient characterof the anaerobic microbiome to resist and surpass short pertur-bation periods (Werner et al., 2011), since it rebounds followingtemperature shock.

4. Conclusions

Regardless the type of perturbation (gradual or abrupt) andthe variability in the substrate used, temperature shock leads toan increase in Bacteroidetes phylum (Bacteroidales order and Bac-teroides genus in gradual perturbation and Porphyromonadaceaefamily in abrupt one) and Gammaproteobacteria class and a cleardecrease in Syntrophomonas and Clostridium genera. Moreover,the methanogenic hydrogenotrophic pathway replaces the ace-togenic one at low temperatures. Once the temperature was

restored, Clostridium, Sedimentibacter and Bacteroidetes (mainlyPorphyromonadaceae) played an important role in the degrada-tion of the accumulated intermediates (mainly VFAs). Finally, thecombination of different microbial identification and quantification

1 iotech

ttptt

A

CaCfph

A

f2

R

A

A

A

B

C

C

C

D

D

D

D

86 L. Regueiro et al. / Journal of B

echniques allowed to double check which communities would behe most affected by temperature decrease and in the recoveryeriod. This knowledge could be utilized to develop strategies forhe mitigation of temperature change consequences and speed uphe recovery of stable reactor performance.

cknowledgements

This research was supported by theMinistry of Economy andompetitiveness through COMDIGEST (CTM2010-17196) projectnd the Ramón y Cajal contract (RYC-2012–10397) to Dr. Martaarballa and by the Xunta de Galicia through GRC program co-

unded by FEDER (GRC 2013-032), and MicroDAN (EM2012/087)rojects. I would also like to acknowledge to Largus Angenent foris contribution to this research and for his helpful suggestions.

ppendix A. Supplementary data

Supplementary data associated with this article can beound, in the online version, at http://dx.doi.org/10.1016/j.jbiotec.014.10.007.

eferences

llison, S.D., Martiny, J.B., 2008. Colloquium paper: resistance, resilience, and redun-dancy in microbial communities. Proc. Natl. Acad. Sci. U.S.A. 105, 11512–11519.

lvarez, J.A., Armstrong, E., Gomez, M., Soto, M., 2008. Anaerobic treatment of low-strength municipal wastewater by a two-stage pilot plant under psychrophilicconditions. Bioresour. Technol. 99 (15), 7051–7062.

PHA, 1998. Standard Methods for the Examination of Water and Waste Water, 20thed. American Public Health Association, Washington, DC.

ialek, K., Kumar, A., Mahony, T., Lens, P.N., O’Flaherty, V., 2012. Microbial commu-nity structure and dynamics in anaerobic fluidized-bed and granular sludge-bedreactors: influence of operational temperature and reactor configuration. Micro-bial Biotechnol. 5 (6), 738–752.

aporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello,E.K., Fierer, N., Pena, A.G., Goodrich, J.K., Gordon, J.I., et al., 2010. QIIME allowsanalysis of high-throughput community sequencing data. Nat. Methods 7,335–336.

irne, D.G., Lehtomäki, A., Björnsson, L., Blackall, L.L., 2007. Hydrolysis and microbialcommunity analyses in two-stage anaerobic digestion of energy crops. J. Appl.Microbiol. 103 (3), 516–527.

ollins, G., Enright, A.M., Scully, C., Mahony, T., O’Flaherty, V., 2006. Applicationand biomolecular monitoring of psychrophilic anaerobic digestion. Water Sci.Technol. 54, 41–47.

aims, H., Lücker, S., Wagner, M., 2006. Daime, a novel image analysis program formicrobial ecology and biofilm research. Environ. Microbiol. 8, 200–213.

e Vrieze, J., Hennebel, T., Boon, N., Verstraete, W., 2012. Methanosarcina: the redis-covered methanogen for heavy duty biomethanation. Bioresour. Technol. 112,1–9.

elbes, C., Moletta, R., Godon, J.J., 2000. Monitoring of activity dynamics of an anaer-obic digester bacterial community using 16S rRNA polymerase chain reaction– single-strand conformation polymorphism analysis. Environ. Microbiol. 2 (5),

506–515.

uran, M., Tepe, N., Yurtsever, D., Punzi, V., Bruno, C., Mehta, R., 2006. Bioaugmentinganaerobic digestion of biosolids with selected strains of Bacillus, Pseudomonas,and Actinomycetes species for increased methanogenesis and odor control. Appl.Microbiol. Biotechnol. 73, 960–966.

nology 192 (2014) 179–186

Gilbert, J.A., Meyer, F., Jansson, J., Gordon, J., Pace, N., Tiedje, J., Ley, R., Fierer, N., Field,D., Kyrpides, N., et al., 2010. The Earth Microbiome Project: meeting report ofthe “1st” EMP Meeting on Sample Selection and Acquisition at Argonne NationalLaboratory October 6th 2010. Stand. Genomic Sci. 3, 249–253.

Guo, X., Wang, C., Sun, F., Zhu, W., Wu, W., 2014. A comparison of microbial charac-teristics between the thermophilic and mesophilic anaerobic digesters exposedto elevated food waste loadings. Bioresour. Technol. 152, 420–428.

Kleyböcker, A., Liebrich, M., Verstraete, W., Kraume, M., Würdemann, H., 2012. Earlywarning indicators for process failure due to organic overloading by rapeseedoil in one-stage continuously stirred tank reactor, sewage sludge and wastedigesters. Bioresour. Technol. 123, 534–541.

Komemoto, K., Lim, Y.G., Nagao, N., Onoue, Y., Niwa, C., Toda, T., 2009. Effect oftemperature on VFA’s and biogas production in anaerobic solubilization of foodwaste. Waste Manag. 29 (12), 2950–2955.

Kotsyurbenko, O.R., 2005. Trophic interactions in the methanogenic microbial com-munity of low-temperature terrestrial ecosystems. FEMS Microbiol. Ecol. 53,3–13.

Lee, S.H., Kang, H.J., Lee, Y.H., Lee, T.J., Han, K., Choi, Y., Park, H.D., 2012. Moni-toring bacterial community structure and variability in time scale in full-scaleanaerobic digesters. J. Environ. Monit. 14 (7), 1893–1905.

Lee, S.H., Park, J.H., Kang, H.J., Lee, Y.H., Lee, T.J., Park, H.D., 2013. Distribution andabundance of Spirochaetes in full-scale anaerobic digesters. Bioresour. Technol.145, 25–32.

Leitão, R.C., van Haandel, A.C., Zeeman, G., Lettinga, G., 2006. The effects ofoperational and environmental variations on anaerobic wastewater treatmentsystems: a review. Bioresour. Technol. 97, 1105–1118.

Lettinga, G., 1999. Psychrophilic anaerobic treatment of low strength wastewaters.Water Sci. Technol. 39, 203–210.

Lettinga, G., Rebac, S., Zeeman, G., 2001. Challenge of psychrophilic anaerobicwastewater treatment. Trends Biotechnol. 19, 363–370.

Leven, L., Eriksson, A.R.B., Schnürer, A., 2007. Effect of process temperature on bacte-rial and archaeal communities in two methanogenic bioreactors treating organichousehold waste. FEMS Microbiol. Ecol. 59, 683–693.

Liu, W.T., Chan, O.C., Fang, H.H., 2002. Microbial community dynamics during start-up of acidogenic anaerobic reactors. Water Res. 36 (13), 3203–3210.

McHugh, S., Carton, M., Collins, G., O‘Flaherty, V., 2004. Reactor performanceand microbial community dynamics during anaerobic biological treatment ofwastewaters at 16–37 ◦C. FEMS Microbiol. Ecol. 48 (3), 369–378.

Poh, P.E., Chong, M.F., 2009. Development of anaerobic digestion methods for palmoil mill effluent (POME) treatment. Bioresour. Technol. 100, 1–9.

Qu, X., Vavilin, V.A., Mazéas, L., Lemunier, M., Duquennoi, C., He, P.J., Bouchez,T., 2009. Anaerobic biodegradation of cellulosic material: batch experimentsand modelling based on isotopic data and focusing on aceticlastic and non-aceticlastic methanogenesis. Waste Manag. 29 (6), 1828–1837.

Regueiro, L., Veiga, P., Figueroa, M., Alonso-Gutierrez, J., Stams, A.J., Lema, J.M.,Carballa, M., 2012. Relationship between microbial activity and microbial com-munity structure in six full-scale anaerobic digesters. Microbiol. Res. 167 (10),581–589.

Regueiro, L., Veiga, P., Figueroa, M., Lema, J.M., Carballa, M., 2014. Influence of tran-sitional states on the microbial ecology of anaerobic digesters treating solidwastes. Appl. Microbiol. Biotechnol. 98, 2015–2027.

Talbot, G., Topp, E., Palin, M.F., Masse, D.I., 2008. Evaluation of molecular meth-ods used for establishing the interactions and functions of microorganisms inanaerobic bioreactors. Water Res. 42 (3), 513–537.

Town, J.R., Links, M.G., Fonstad, T.A., Dumonceaux, T.J., 2014. Molecular char-acterization of anaerobic digester microbial communities identifies microor-ganisms that correlate to reactor performance. Bioresour. Technol. 151,249–257.

Werner, J.J., Knights, D., Garcia, M.L., Scalfone, N.B., Smith, S., Yarasheskim, K., Cum-mings, T.A., Beers, A.R., Knight, R., Angenent, L.T., 2011. Bacterial community

structures are unique and resilient in full-scale bioenergy systems. Proc. Natl.Acad. Sci. U.S.A. 108 (10), 4158–4163.

Zhang, W., Werner, J.J., Matthew, A.T., Angenent, L.T., 2014. Substrates type drivesvariation in reactor microbiomes of anaerobic digesters. Bioresour. Technol. 151(1), 397–440.