ww.sciencedirect.com

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 7 0 9 8e7 1 0 8

Available online at w

ScienceDirect

journal homepage: www.elsevier .com/locate /watres

Drivers of microbial community composition inmesophilic and thermophilic temperature-phasedanaerobic digestion pre-treatment reactors

Hasina M. Pervin a, Paul G. Dennis a,b, Hui J. Lim a, Gene W. Tyson a,b,Damien J. Batstone a, Philip L. Bond a,*aAdvanced Water Management Centre, The University of Queensland, Brisbane, Queensland 4072, AustraliabAustralian Centre for Ecogenomics, School of Chemistry and Molecular Biosciences, The University of Queensland,

Brisbane, Queensland 4072, Australia

a r t i c l e i n f o

Article history:

Received 8 April 2013

Received in revised form

2 July 2013

Accepted 4 July 2013

Available online 23 October 2013

Keywords:

Temperature-phased anaerobic

digestion

Molecular microbial ecology

Thermophilic pre-treatment

Hydrolysisefermentation

Abbreviations: AD, Anaerobic Digestion;Hydraulic Retention Time; MP, Mesophilic Pmorphism; TP, Thermophilic Pre-treatment;RFLP, Terminal Restriction Fragment Length* Corresponding author. Advanced Water Ma

Research Rd., St. Lucia, Queensland 4072, AuE-mail address: phil.bond@awmc.uq.edu.

0043-1354/$ e see front matter ª 2013 Elsevhttp://dx.doi.org/10.1016/j.watres.2013.07.053

a b s t r a c t

Temperature-phased anaerobic digestion (TPAD) is an emerging technology that facilitates

improved performance and pathogen destruction in anaerobic sewage sludge digestion by

optimising conditions for 1) hydrolytic and acidogenic organisms in a first-stage/pre-

treatment reactor and then 2) methogenic populations in a second stage reactor. Pre-

treatment reactors are typically operated at 55e65 �C and as such select for thermophilic

bacterial communities. However, details of key microbial populations in hydrolytic com-

munities and links to functionality are very limited. In this study, experimental thermophilic

pre-treatment (TP) andcontrolmesophilic pre-treatment (MP) reactorswereoperatedasfirst-

stages of TPAD systems treating activated sludge for 340 days. The TP system was operated

sequentially at 50, 60 and 65 �C,while theMP rectorwasheld at 35 �C for the entireperiod. The

compositionofmicrobial communities associatedwith theMPandTPpre-treatment reactors

was characterised weekly using terminal-restriction fragment length polymorphism (T-

RFLP) supported by clone library sequencing of 16S rRNA gene amplicons. The outcomes of

this approachwere confirmedusing 454 pyrosequencing of gene amplicons andfluorescence

in-situ hybridisation (FISH). TP associated bacterial communities were dominated by pop-

ulations affiliated to the Firmicutes, Thermotogae, Proteobacteria and Chloroflexi. In particular

there was a progression from Thermotogae to Lutispora and Coprothermobacter and diversity

decreased as temperature and hydrolysis performance increased. While change in the

composition of TP associated bacterial communities was attributable to temperature, that of

MP associated bacterial communities was related to the composition of the incoming feed.

This study determined processes driving the dynamics of keymicrobial populations that are

correlated with an enhanced hydrolytic functionality of the TPAD system.

ª 2013 Elsevier Ltd. All rights reserved.

BLAST, Basic Local Alignment Search Tool; FISH, Fluorescent In situ Hybridization; HRT,re-treatment; OTU, Operational Taxonomic Unit; RFLP, Restriction Fragment Length Poly-TPAD, Temperature-Phased Anaerobic Digestion; T-RF, Terminal Restriction Fragment; T-Polymorphism.nagement Centre (AWMC), The University of Queensland, Level 6 Gehrmann Building (60),stralia. Tel.: þ61 (0)7 3446 3226; fax: þ61 (0)7 3365 4726.au (P.L. Bond).

ier Ltd. All rights reserved.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 7 0 9 8e7 1 0 8 7099

1. Introduction

observed to be both stable (Weiss et al., 2008) and variable

Anaerobic digestion (AD) is an economically and environ-

mental attractive microbially driven process that can be used

to generatemethane from biosolids. AD comprises hydrolysis,

fermentation, acetogenesis and methanogenesis (Weiland,

2010). Bacteria generally dominate the first three stages;

however, archaea drive methanogenesis and this latter stage

is generally well characterised (Amani et al., 2010). In com-

parison, detailed knowledge of the populations responsible for

hydrolysis and fermentation is lacking. Hydrolysis is a key

rateelimiting step as the products are rapidly consumed by

methanogens (Li and Noike, 1992; Rittmann, 2008). Greater

insight into the key microorganisms involved, their dynamics

and details of their activities can provide opportunities for

manipulation of populations and functions for enhancing

process performance, reactor efficiency and troubleshooting

management of AD.

Temperature-phased anaerobic digestion (TPAD) is an

emerging process, in which enhanced hydrolysis, facilitated

by a pre-treatment reactor operated at 50e70 �C, can be

combined with enhanced stability of the overall process (Paul

et al., 2012). A TPAD system may typically comprise of a

thermophilic pre-treatment reactor with a short (2e4 days)

hydraulic retention time (HRT) followed by the main reactor

operating at mesophilic temperatures with a 10e15 day HRT

(Ge et al., 2011). The functionality of the pre-treatment stage is

mainly associated with hydrolysis and acidogenesis, which

are mediated by bacteria. The second stage function is pri-

marily acetogenesis and methanogenesis, and this will

comprise a mix of bacteria and archaea. TPAD is particularly

applicable to AD of activated sludge, as it allows increased

performance regarding hydrolysis of organic solids and

methane production at a moderate energy input, as well as

full pathogen removal (Paul et al., 2012).

There have been limited molecular-based studies of mi-

crobial communities in AD systems. These studies largely

focus on the methanogenic communities (Kobayashi et al.,

2009) although those that include analyses of bacterial com-

munities reveal mostly novel phylotypes (Chouari et al., 2005).

For example, in a survey of publically available 16S rRNA gene

amplicon sequences associated with anaerobic digesters,

Nelson et al. (2011) found that ca. 60% of bacteria could not be

assigned to any established genus (Nelson et al., 2011). Mi-

crobial communities of seven mesophilic full scale digesters

have been examined and, bacteria of the Chloroflexi, Betapro-

teobacteria, Bacteridetes and Synergistetes were observed as core

groups (Riviere et al., 2009). However, details of the bacterial

populations within those phyla and the ecological role of

those are yet to be determined. The compositions of AD-

associated bacterial communities are seen to be influenced

by reactor parameters. For example, pH fluctuations have

been observed to correlate with changes in bacterial com-

munitieswithin a thermophilic AD over time (Hori et al., 2006),

and similarly, pH changes in a mesophilic multistage AD

system were implicated in bacterial community differences

between phases of the reactor (Supaphol et al., 2011). In AD

systems run with the same operational conditions over time,

however, the composition of bacterial communities has been

(Fernandez et al., 1999; Pycke et al., 2011). A recent study

compared bacterial communities of seven full scale anaerobic

reactors having a temperature range of 35e52 �C (Lee et al.,

2012). It was found that temperature was a major factor

affectingmicrobial community compositions, and some of the

phylotypes could be weakly linked with some process per-

formance parameters (Lee et al., 2012). In a recent study of a

TPAD pre-treatment stage treating primary sludge, over 90%

of the sequences were from previously undetected bacteria

(Pervin et al., 2013). This study identified specific thermophilic

populations, but could not link performance to community

changes. Additionally, the treated waste was primary sludge,

and although feed communities were not determined, the

variable nature of the feed was implied to influence the pre-

treatment bacterial communities. It is evident that studies of

these bacterial communities are limited and knowledge of the

ecology and how that may be related to the system operation

is just beginning to develop. In particular, the pre-treatment

stage in a temperature-phased anaerobic process is

appealing from a microbial perspective, since it is novel in

temperature and performance, and studies comparing ther-

mophilic and mesophilic systems would be revealing.

Consequently, there are limitations with previous studies

with respect to determining hydrolytic communities, aswell as

with identification of dynamics of key microbial populations.

Additionally, insight is required as towhether functionality is a

general community characteristic, or whether it can be linked

to key populations. It is very difficult to address this fully, but

examination of TPAD operating at different temperatures, and

on a feed more consistent than primary sludge, such as acti-

vated sludge, offers the opportunity to associate microorgan-

isms with reactor performance. In this study we attempt to

address this gap through multiple molecular approaches

including high analytical frequency through T-RFLP, objective

identification through 16S rRNA gene clone-libraries and

pyrotag sequencing, as well as PCR-independent analysis

through fluorescence in-situ hybridisation. The overall objec-

tivewas to identify community populations thatmay be linked

to increased temperature and performance.

2. Materials and methods

2.1. Reactor operation and performance

Two TPAD systems, each consisting of two stages: 1) a 0.6 L

pre-treatment reactor with a hydraulic retention time (HRT) of

two days and 2) a 4 L main reactor (methanogenic) with a 14

day HRT, were operated for 340 days. In one system, the pre-

treatment was performed under mesophilic conditions at

35 �C (MP), while the other was operated under thermophilic

conditions at 50 �C for 186 days, 60 �C for 100 days and 65 �C for

60 days (TP). In both systems the second-stage reactor was

operated at 35 �C. Details of the reactor operations are

described in Ge et al. (2011). The reactors were inoculatedwith

sludge from another lab-scale mesophilic TPAD system, this

provided a relevant and common starting point for both re-

actors. The TPADs were fed 0.3 L per day of activated sludge

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 7 0 9 8e7 1 0 87100

collected from a wastewater treatment plant in Gold Coast

city, Australia. The activated sludge feed was collected

monthly and stored at below 4 �C. The reactors were mixed

usingmagnetic stirrer bars. The performance of each reactor was

characterised by measuring quantities of total solids (TS),

volatile solids (VS), volatile fatty acid (VFA), chemical oxygen

demand (COD), total Kjeldahl nitrogen (TKN), ammo-

niumenitrogen (NH4þ �N) and methane gas as described in

Ge et al. (2011).

2.2. Characterisation of microbial communities

2.2.1. Sample collection and DNA extractionEffluent sludge samples were collected at regular intervals

from both pre-treatment reactors during the operation of the

TPAD systems (Figure S1), these samples were immediately

frozen at �20 �C for DNA extraction. Additional sludge sam-

ples were collected from the feed reservoir, which was

replenished daily with waste activated sludge (WAS), and are

denoted as ‘Feed’ in this study. Genomic DNA was extracted

using the Fast DNA SPIN Kit for soil according to the manu-

facturers instructions (Q-Bio gene, Australia).

2.2.2. Microbial community fingerprinting by T-RFLPBacterial community structure was assessed using terminal

restriction fragment length polymorphism fingerprinting (T-

RFLP; Supplementary information). Briefly, PCR was per-

formed using the 63F (50-AGTTTGATCCTGGCTCAG-30) and

1389R (50-ACGGGCGGTGTRC-30) primers (Marchesi et al., 1998)

followed by purification and restriction enzyme digestion in

duplicates usingMSPI andHaeIII (Fermentas Inc., Hanover,MD,

USA). Terminal restriction fragments (T-RFs) were analysed

using an automated DNA sequencer (ABI Prism TM 3730). T-RF

sizes and their corresponding areas were measured using

GeneMarker� (version 1.75; SoftGenetics, LLC., State College,

PA, USA). Data were standardised by expressing each T-RF

peak area as a percentage of the total population for that

profile and normalised using a constant percentage threshold

method as previously described (Sait et al., 2003).

The likely origin of T-RF peaks was identified by producing

in-silico profiles of cloned genes (see Section 2.2.3) using the T-

RFLPMAP software (NERC environmental bioinformatics

centre, UK). Although the T-RFs identities are putative, the

confidence of the suggested identifications was increased by

using two enzyme digestions and coinciding T-RFLP abun-

dance patterns of the two digestions during the reactor

operation (Figure S3). The TAP t-RFLP tool (Marsh et al., 2000)

of the Ribosomal Database Project was used to verify the

affiliation of peaks and to infer matches when those were not

available from the clone library database. Both the MSPI and

HaeIII digest results were used for identifications, and the T-

RFLP profiles of the different digestions were in good agree-

ment (Fig. S3). However, as MSPI generated the more detailed

T-RFLP profiles, further analysis and results presented in this

study are based on those.

2.2.3. 16S rRNA gene amplicon cloning and sequencingBacterial and archaeal 16S rRNA genes of the communities

were determined by PCR, cloning and sequencing at 100 days

operation for MP and TP, and also at 240 days operation for TP.

Amplicons were generated by PCR, as described previously

(Bond et al., 1995), using the primer pairs 27F (50eTTTGATCCT

GGCTCAGe30) and 1492R (50eGGTTACCTTGTACGACTTe30) forbacteria, and Arc8F (50eTCCGGTTGATCCTGCCe30) and

Arc927R (50e CCCGCCAATTCCTTTAAGTTTCe30) for archaea

(Singh et al., 2006). Clone librarieswere constructed based on a

total of 187 positive clones (Supplementary information). The

full length clone sequences were phylogenetically analysed

using representative bacterial sequences (ARB database,

greengenes.lbl.gov) and evolutionary distance analysis in

the ARB software package (Ludwig et al., 2004). The topology of

the phylogenetic tree was used to aid probe design (Section

2.2.5).

2.2.4. 16S rRNA gene amplicon pyrosequencingBacterial and archaeal 16S rRNA gene amplicons were exam-

ined by pyrosequencing from MP and TP reactor samples at

days 240 and 324, this was when TP was operated at 60 �C and

65 �C respectively. Gene amplicons were generated by PCR with

primers 926F and 1392R (Engelbrektson et al., 2010) that were

modified on the 50 end to contain the 454 FLX Titanium Lib L

adapters B and A, respectively. The reverse primers also

contained a 5-6 base barcode sequence positioned between

the primer sequence and the adapter. A unique barcode was

used for each sample. Following amplification (see Supple-

mentary Information) purified and normalised amplicons

were submitted to Macrogen (South Korea) for 454 pyrose-

quencing (Supplementary information). Sequence analysis

was performed as described previously (Dennis et al., 2013)

using QIIME (Caporaso et al., 2010) and UCHIME (Edgar et al.,

2011).

2.2.5. Fluorescence in situ hybridization (FISH)Probes for FISH for each of the three major OTUs in the clone

library were designed as previously described using ARB

(Hugenholtz et al., 2002), and then checked for specificity

using BLAST (http://www.ncbi.nlm.nih.gov/BLAST/) and

probeCheck (Loy et al., 2008). Probes were synthesised and

labelled at the 50 end with either Cy3, Cy5 or FITC (Gene-

Works, Australia). FISH was performed using previously re-

ported phylogenetic group specific probes and newly

designed species specific probes (Table S1) on reactor sam-

ples fixed in either 4% paraformaldehyde (Gram negatives) or

ethanol (Gram positives) according to established procedures

(Amann et al., 1995). To assess probe specificities FISH was

performed on fixed reactor samples with those probes at

different stringencies by increasing the formamide concen-

tration in the hybridization buffer in 5% intervals and

decreasing the NaCl concentration in the respective wash

buffers accordingly. The most stringent conditions that gave

the brightest signal with the presumed target cells were

considered as optimal for subsequent quantification. Micro-

bial cells hybridising to the fluorescently labelled probes were

observed with a Zeiss Axioplan LSM510 confocal laser-

scanning microscope (CLSM) using standard excitation and

emission wavelengths. FISH images were used for quanti-

fying the population of target microbe using the DAIME

version 1.3.1 software (Daims et al., 2006). The images were

optimised by adjusting the pixel level to reduce background

noise and to achieve a good contrast of the target organism

Table 1 e Reactor performance parameters as determined by Ge et al. (2011).

Parameter 35 �Ca 50 �Cb 60 �Cb 65 �Cb

Hydrolysis speed khyd (d�1) 0.2 � 0.1 0.12 � 0.06 0.5 � 0.1 0.7 � 0.2

Hydrolysis extent fd 0.4 � 0.1 0.4 � 0.1 0.41 � 0.04 0.51 � 0.04

a Measured in MP system.b Measured in TP system.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 7 0 9 8e7 1 0 8 7101

over the general population. At least 25 images were used to

obtain an accurate biovolume fraction of the target microbe.

All biovolume fractions obtained have a minimum congru-

ency level of 90% to ensure that the quantification values

obtained were reliable.

2.3. Statistical analyses

Turnover in the composition of T-RFs between samples was

visualised using Principle Component Analysis (PCA). Data

were Hellinger transformed prior to analysis. The effect of

treatment parameters on the composition of microbial com-

munities was determined using permutational multivariate

analysis of variance (PERMANOVA). Statistical analyses were

conducted using R statistical package (R Development Core

Team, 2004).

3. Results

3.1. Reactor operation and performance

A detailed report and comparison of the performance of the

TPADs is presented in Ge et al. (2011). Briefly, both TPAD sys-

tems successfully produced methane throughout the 340 day

operation and the thermophilic pre-treatment (TP) reactor

demonstrated significantly enhanced AD performance.

Increased volatile solids destruction, solubilisation and pro-

duction of volatile fatty acids was evident in the TP reactor in

Table 2 e Putative identification of the most abundant T-RFs inphylogenetic analysis of cloned sequences. TP 50, TP 60, TP 65microorganisms during the periods of the TP reactor operationreactor operation (at 35 �C throughout the experiment).

T-Rf length asper MspI (bp)

Putative identification determinedfrom clone sequence affiliation

56 Comamonas sp.

102 Uncultured Betaproteobacterium

103 Uncultured Gammaproteobacterium

112 Novosphingobium mathurense

236 Uncultured Thermotogae sp.

256 Coprothermobacter sp.a

271 Lutispora thermophila

450 Pseudomonas meridian

451 Zoogloea resiniphila

454 Thauera sp.

480 Clostridium cellulovorans

482 Uncultured Chloroflexi

489 Clostridium sp.

a Obtained from 100% matching fullelength representative sequence to

comparison to the mesophilic pre-treatment (MP) reactor (Ge

et al., 2011). The solubilisation of organic solids during the

operationof theTP reactor increased from15%at50 �Cto27%at

60 �C and 65 �C. In comparison, solubilisation of organic solids

in the MP reactor was only 7% (based on the chemical oxygen

demandbalance).Additionally, theammoniaenitrogenrelease

in the TP reactor was constantly higher than that in the MP

reactor (Ge et al., 2011). Model based analysis determined key

kinetic parameters related to the extent (fd) and speed (khyd) of

organic solid degradation (Table 1). Both the speed and amount

of hydrolysis increased with increasing temperature in the TP

reactor, but these parameters did not differ between the MP

reactor at 35 �C and the TP reactor at 50 �C.

3.2. Identification of key populations

Sequencing of 16S rRNA amplicons from the clone library

revealed a range of organisms that were associated with the

pre-treatment reactors (Table 2 and Fig. 1). These sequences

were subjected to in-silico restriction enzyme digests which

facilitated their association with peaks from the T-RFLP anal-

ysis (Fig. 2). The composition of microbial communities asso-

ciated with the MP and TP reactors differed significantly

(PERMANOVA, P < 0.001), with the TP reactor (50e65 �C) beingassociated with members of the Thermotogae, Lutispora and

Coprothermobacter (Fig. 3). The TP reactor communities were

dominated by these key populations, particularly at the higher

temperatures of 60 �C and 65 �C (Fig. 2). In comparison, the MP

reactor (35 �C) communities were not dominated by particular

the pre-treatment reactors as determined from theand MP 35 indicate average abundance of particularat 50 �C, 60 �C, and 65 �C respectively, and during the MP

Average abundance (%) as per MSPI

TP 50 TP 60 TP 65 MP 35

0.28 0.02 0.06 1.06

1.40 5.79 2.60 2.21

3.58 1.65 2.33 5.21

3.67 1.63 1.35 4.20

12.18 28.58 3.77 0.73

0.15 3.92 10.45 0.35

3.94 17.23 21.46 0.68

5.15 2.64 3.42 2.14

0.41 3.33 2.02 3.67

4.37 1.85 1.81 6.58

3.30 0.33 0.81 1.79

4.91 1.91 2.79 2.01

0.00 0.03 0.02 3.57

respective pyrotag sequence analysis.

Fig. 1 eMicrobial community analysis of TP samples by FISH, T-RFLP, and 16S rRNA gene pyrotag sequencing, including; (A)

organisms identified by the THERMO846 probe as Thermotogae sp. at day 240 (60 �C) (magenta) against other bacteria in blue

(EUB338), (B) organisms identified by the LUTI1250 probe as Lutispora thermophile at day 294 (65 �C) (magenta) against other

bacteria in blue; (C)Methanosarcina thermophila as identified by the SARC1645 (orange) with other Archaea (ARC915) in green

and bacteria in blue; and (D) comparison of community analyses by pyrotag sequencing (Pyro 240, Pyro 325) and T-RFLP at

days 240 (60 �C) and 325 (65 �C). Abundant populations detected are highlighted by arrows. (For interpretation of the

references to colour in this figure legend, the reader is referred to the web version of this article.)

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 7 0 9 8e7 1 0 87102

bacterial populations, but were associated with representa-

tives ofGammaproteobacteria,Clostridium and Zoogloea.Archaeal

populations related to the Methanosarcinaceae represented

approximately 10% of the total population in the TP reactor at

60 �C (Figure S4). This findingwas in agreementwith the results

of FISH analyses using the probe SARC1645 which specifically

targetsMethanosarcina thermophila (Fig. 1). Only lownumbers of

archaea were detected by FISH (<1%) in the MP reactor.

3.3. The effect of temperature of bacterial communitycomposition

Within the TP reactor, 36% of variation in microbial commu-

nity composition was attributable to reactor temperature

(PERMANOVA, P < 0.001) with: 1) Thermotogae being more

abundant at 50 �C and 60 �C, 2) Lutispora thermophila being

more abundant at 60 �C and 65 �C, and 3) Coprothermobacter

being more abundant at 65 �C (Fig. 3). To confirm these find-

ings two oligonucleotide FISH probes, THERMO846 and

LUTI1250, were designed to target Thermotogae and Lutispora

thermophila, respectively. Thermotogae populations and Lutis-

pora thermophila were detected by FISH at 13.3 � 4.7% and

3.8 � 1% of the bacteria in TP (60 �C) at day 240 (Fig. 1A and

Figure S5). At the highest temperature (65 �C, day 294) Lutispora

thermophila was more abundant at 8.4 � 2.8% and Thermotogae

sp. had decreased to 2.6 � 1.6% (Fig. 1B and Figure S5). These

trends were in agreement with the results obtained using T-

RFLP and pyrosequencing at 60 and 65 �C; however, the overall

values were slightly lower by FISH. This discrepancy may be

explained by the FISH probe not capturing the full diversity of

the populations detected by sequencing and T-RFLP. For

example, 43% of the clone sequences affiliated with Clostridia

did not completely match the LGC354 mix probes used for

detection of the Firmicutes.

Fig. 2 e Abundance patterns of bacteria in (A) TP and (B) MP as revealed by T-RFLP with MspI enzyme; T-RFs having

abundance <5% of the total community were pooled into a group called ‘others’. Different T-RFs are indicated by the

colours and the taxonomic affiliation of those with a corresponding taxonomic affiliation from in-silico restriction enzyme

digests of 16S rRNA gene amplicon sequences (Table 2) is shown. The temperature in the thermophilic reactor was 50 �C on

days 35e186, 60 �C on days 187e287, and 65 �C on days 288e340. The temperature in the mesophilic reactor was 35 �Cthroughout the experiment. (For interpretation of the references to colour in this figure legend, the reader is referred to the

web version of this article.)

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 7 0 9 8e7 1 0 8 7103

3.4. Changes in bacterial community composition overtime

The largest changes in community composition in the TP

reactor were associated with changes in operating

temperature (Figs. 2 and 3). However, approximately, 6% and

19% of variation in bacterial community composition was

attributable to changes that occurred over time within a

temperature level in the TP and MP pre-treatment reactors,

respectively (PERMANOVA, P < 0.05; Fig. 3). The changes that

Fig. 3 e PCA ordination representing variation in the composition of bacterial communities detected using T-RFLP

fingerprinting of 16S rRNA gene amplicons. Samples are represented by circles; the size of which represents the number of

weeks since the start of the experiment and the colour represents the temperature of the reactor at any given time. Samples

from the same reactor are joined by arrows that indicate the progression of time. The crosses represent T-RFs, of which the

most discriminating are labelled. T-RF labelled in bold font represent those with a corresponding taxonomic affiliation from

in-silico restriction enzyme digests of 16S rRNA gene amplicon sequences (Table 2). The red text highlights the time periods

of samples that are discussed in the main text. (For interpretation of the references to colour in this figure legend, the reader

is referred to the web version of this article.)

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 7 0 9 8e7 1 0 87104

occurred in the TP pre-treatment reactor during operation at

50 �C, 60 �C and 65 �C were related to populations that were

influenced by the temperature changes. For example, at 50 �Cthe community differed over time due to a gradual increase in

the abundance of the Thermotogae population (T-RF 236),

whereas at 65 �C the community differed over time due to the

same population gradually decreasing (Fig. 3).

Changes that occurred in the MP pre-treatment reactor,

however, appear to have been related to changes in the

composition of the microbial communities associated with

the feed (Fig. 3). Feed-associated microbial communities

differed significantly over time, with feed samples becoming

more enrichedwith three populations (T-RFs 396, 450 and 489)

as the experiment progressed. These populations included

Pseudomonas meridiana and a representative of the Clostridium

(PERMANOVA, P < 0.001; Fig. 4). The largest change in the

composition of the MP reactor associated bacteria community

was apparent after 208 days andwas related to an enrichment

of the P. meridiana and Clostridium populations (T-RFs 450 and

489) also observed to vary in the feed. Another obvious change

in the composition of the MP reactor associated bacteria

community was apparent after 255 days (Figs. 3 and 4). The

change was related to an enrichment of the P. meridiana pop-

ulation which was also more abundant in feed from days

250e280 relative to previous feed-stocks (Fig. 4). Feed com-

munity composition is further implicated as a driver of MP

pre-treatment reactor associated microbial community

composition by a PERMANOVA analysis, which demonstrated

that the primary axis scores from the PCA of the feed associ-

atedmicrobial communities (Fig. 4) were significant predictors

of variation in the composition of the MP pre-treatment

reactor associated communities (P ¼ 0.016).

4. Discussion

4.1. Dynamic nature of microbial communities andcorrelation to reactor conditions

We were interested to detect differences in the communities

of the two pre-treatment reactors and correlate those to the

operating conditions and performance of the reactors. The

importance of temperature for methanogenic archaeal spe-

cies composition has been shown previously (Karakashev

et al., 2005), however, the influence of temperature driving

bacterial species composition is yet to be established in AD.

The communities in both TP and MP were dynamic as

revealed by T-RFLP, FISH and pyrosequencing. The thermo-

philic system was clearly dynamic in response to operating

temperature. The largest variations in the TP reactor bacterial

communities coincided with the changes in operation tem-

perature (Fig. 3). Consequently this study successfully

Fig. 4 e PCA ordination representing variation in the composition of feed-associated bacterial communities. Samples are

represented by circles. The size of the circles is proportional to the number of weeks since the start of the experiment.

Arrows indicate the progression of time. The crosses represent T-RFs of which the most discriminating are labelled. T-RF

labelled in bold font represents those with a corresponding taxonomic affiliation from in-silico restriction enzyme digests of

16S rRNA gene amplicon sequences (Table 2). The first feed sample was characterised 100 days into the experiment after

that each new batch of feed was subjected to T-RFLP fingerprinting. The temperature in the thermophilic reactor was 50 �Con days 35e186, 60 �C on days 187e287, and 65 �C on days 288e340. The temperature in the mesophilic reactor was 35 �Cthroughout the experiment.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 7 0 9 8e7 1 0 8 7105

correlated microbial community changes, particularly those

of the bacteria, to operating conditions of increased temper-

ature in anaerobic digestion.

The TP reactor communities were dominated by particular

bacteria at the different stages of temperature operation. This

is likely to be related to the different temperature optimums of

the bacteria. For example, at 65 �C Thermotogae sp. was out

competed in the TP reactor by Lutispora thermophila and other

thermophiles; possibly this temperature being above the or-

ganisms optimal range. Another possibility was the Thermo-

togae sp. abundance was not favoured by the high

concentrations of ammonium or organic acids, since the total

VFA concentration increased three foldwhen the temperature

increased from 60 to 65 �C in the TP reactor (Ge et al., 2011).

Thus, reactor conditions influenced the presence and domi-

nance of specific populations resulting in the community dy-

namics at various temperatures.

The Feed community was seen to significantly affect the

bacterial community structure in the MP reactor. Correla-

tion of variation of Feed communities with that in MP

reactor communities was evident by PCA ordination (Fig. 3).

Likely, contributing to the MP communities were: survival of

facultative anaerobic Feed populations due to the similar

temperature conditions of the MP reactor and WAS, changes

in the batches of Feed community composition and dead

cell DNA from the Feed. Contributions of the Feed commu-

nities were not so evident in the TP reactor communities

and this could be explained by the selection pressure of

temperature in TP having a stronger influence on microbial

community composition in comparison with the feed

material.

Some variation in the TP reactor bacterial communities

were detected within periods of constant operating tempera-

ture. Variation in bacterial communities has been detected

during the stable operation of AD reactors (Fernandez et al.,

1999; Pycke et al., 2011). However, this variation in TP could

be attributed to adjustment of population abundance

following reactor temperature changes and consequential

performance changes, such as changing VFA levels, occurring

through the periods of constant temperature operation.

4.2. Microbial community composition and reactorperformance

The principal parameters for performance of anaerobic

digestion are stated in Section 3.1. However, it is important

that hydrolysis in TP at 50 �C was similar to that in MP, and

only increased once temperature increased to 60 �C, and the

key dominating populations shifted frommixed communities

and Thermotogae to Lutispora thermophila and Coprothermobacter

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 7 0 9 8e7 1 0 87106

as the temperature increased to 65 �C. This indicates that

while at 50 �C the community was strongly directed by tem-

perature (to Thermotogae), it was only once temperature

increased and caused a population shift particularly to Lutis-

pora thermophila and Coprothermobacter that performance in

terms of hydrolysis increased substantially. While the com-

munity changes could be related to increased hydrolysis,

these changes also coincided with other changes, including

increases in ammonia and organic acid concentrations (Ge

et al., 2011). It should be noted also that these increases in

digestion performance were related to increased protein hy-

drolysis, a particularly important feature for WAS digestion.

In the TP reactor methane production was significantly

higher than in the MP reactor. This coincided with the pres-

ence ofMethanosarcina thermophilawhich was abundant in TP.

In contrast, FISH (results not shown) and pyrosequencing

indicated that archaea were in very low abundance in MP.

4.3. Possible functions of key organisms in thepre-treatment reactors

We assume the abundant organisms in the pre-treatment

reactors were playing important roles in the AD perfor-

mance. While detection of particular 16S rRNA genes is not

proof of phenotype activity, it is possible to suggest potential

functions of the abundant microorganisms in the pre-

treatment reactors. Thermotogae has been previously detec-

ted in thermophilic anaerobic digesters (Chen et al., 2004;

Leven et al., 2007), and in mesophilic AD (Nesbo et al., 2006),

however, on this occasion they were not detected in the MP

reactor. Members of the Thermotogae are thermophilic anaer-

obes that excrete hydrolytic enzymes to catalyse a wide range

of polysaccharides to acetate, carbon dioxide and hydrogen as

the main fermentation products (Huber and Hanning, 2007).

Thermotogae are also implicated with interspecies hydrogen

transfer (Johnson et al., 2006), and the presence of Meth-

anosarcina thermophila, implicates a possible syntrophy

contributing to the methane production in TP.

Lutispora thermophila were originally isolated from an

anaerobic bioreactor operating at 55 �C (Shiratori et al., 2008),

and is a fermentater that strictly utilises amino acids for

growth. Coprothermobacter, like Lutispora are within the Clos-

trida subphylum of the Firmicutes. Also, similar to Lutispora,

Coprothermobacter are detected in thermophilic andmesophilic

anaerobic digestionand theyhavepreference for fermentation

of protein and amino acids as opposed to carbohydrate

fermentation (Etchebehere et al., 1998). This function of these

organisms coincides with the abundance of these sequences

detected with the higher levels of NH4þ and VFA in the TP

reactor at higher temperatures. Additionally, it highlights the

importance of protein degradation and fermentation for the

performance of AD systems.Coprothermobacter also produceH2

and improved growth of these is detected in the presence of H2

utilising methanogens (Sasaki et al., 2011).

Methanosarcina thermophila, the primary archaeon in the TP

reactor, are detected in previous studies of thermophilic

anaerobic digestion systems (Kobayashi et al., 2008; Leven

et al., 2007). In general, Methanosarcina are thought respon-

sible for methane production in anaerobic digestion systems

when acetate concentrations are high (Jetten et al., 1990;

McMahon et al., 2001). However, Methanosarcina thermophila

is capable of H2/CO2 conversion tomethane (Mladenovska and

Ahring, 2000), and consequently the nature of this methano-

genesis is of interest in terms of the metabolic pathway uti-

lised. Additionally, the growth rates of organisms in the TP

reactor (HRT of 2 days) would be faster than typically expected

for methanogens. However, high growth rates, such as a 12 h

doubling time on acetate, are reported for various Meth-

anosarcina spp. (Mladenovska and Ahring, 2000).

5. Conclusion

The mesophilic pre-treatment reactor bacterial communities

were heavily influenced by the feed, while the thermophilic

reactor was less diverse, and had dominant populations of

Thermotogae sp., Lutispora thermophila, and Coprothermobacter,

shifting progressively from the first to the last as temperature

was increased from 50 �C to 65 �C. Functionality was higher at

60 �C and 65 �C, when the process wasmore dominated by the

latter two organisms, indicating that while temperature can

direct community, there will be optimums related to the

emergence of key populations that we suggest are implicated

for the enhanced hydrolytic ability. A particularly important

outcome was the consistency in outputs from the multiple

methods applied, with key populations being quantified

consistently by FISH, T-RFLP (full 16S rRNA gene sequence)

and 454 pyrosequencing (partial 16S rRNA gene sequence).

Acknowledgements

We thank the Queensland Government and Environmental

Biotechnology Cooperative Research Centre (EBCRC),

Australia for supporting this work as a sub-project of “Small-

medium scale organic solids stabilization”. The authors

gratefully acknowledge the contributions of Dr. Frances

Slater, Dr. Huoqing Ge, and Dr. Paul Jensen.

Appendix A. Supplementary data

Supplementary data related to this article can be found at

http://dx.doi.org/10.1016/j.watres.2013.07.053.

r e f e r e n c e s

Amani, T., Nosrati, M., Sreekrishnan, T.R., 2010. Anaerobicdigestion from the viewpoint of microbiological, chemical,and operational aspects e a review. Environ. Rev. 18, 255e278.

Amann, R.I., Ludwig, W., Schleifer, K.H., 1995. Phylogeneticidentification and in-situ detection of individual microbial-cells without cultivation. Microbiol. Rev. 59 (1), 143e169.

Bond, P.L., Hugenholtz, P., Keller, J., Blackall, L.L., 1995. Bacterialcommunity structures of phosphate-removing and non-phosphate-removing activated sludges from sequencing batchreactors. Appl. Environ. Microbiol. 61 (5), 1910e1916.

Caporaso, 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., Huttley, G.A., Kelley, S.T.,

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 7 0 9 8e7 1 0 8 7107

Knights, D., Koenig, J.E., Ley, R.E., Lozupone, C.A.,McDonald, D., Muegge, B.D., Pirrung, M., Reeder, J.,Sevinsky, J.R., Tumbaugh, P.J., Walters, W.A., Widmann, J.,Yatsunenko, T., Zaneveld, J., Knight, R., 2010. QIIME allowsanalysis of high-throughput community sequencing data. Nat.Methods 7 (5), 335e336.

Chen, C.L., Macarie, H., Ramirez, I., Olmos, A., Ong, S.L.,Monroy, O., Liu, W.T., 2004. Microbial community structure ina thermophilic anaerobic hybrid reactor degradingterephthalate. Microbiology-Sgm 150, 3429e3440.

Chouari, R., Le Paslier, D., Daegelen, P., Ginestet, P.,Weissenbach, J., Sghir, A., 2005. Novel predominant archaealand bacterial groups revealed by molecular analysis of ananaerobic sludge digester. Environ. Microbiol. 7 (8), 1104e1115.

Daims, H., Lucker, S., Wagner, M., 2006. daime, a novel imageanalysis program for microbial ecology and biofilm research.Environ. Microbiol. 8 (2), 200e213.

Dennis, P.G., Guo, K., Imelfort, M., Jensen, P., Tyson, G.W.,Rabaey, K., 2013. Spatial uniformity of microbial diversity in acontinuous bioelectrochemical system. Bioresour. Technol.129, 599e605.

Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., Knight, R., 2011.UCHIME improves sensitivity and speed of chimera detection.Bioinformatics 27 (16), 2194e2200.

Engelbrektson, A., Kunin, V., Wrighton, K.C., Zvenigorodsky, N.,Chen, F., Ochman, H., Hugenholtz, P., 2010. Experimentalfactors affecting PCR-based estimates of microbial speciesrichness and evenness. ISME J. 4 (5), 642e647.

Etchebehere, C., Pavan, M.E., Zorzopulos, J., Soubes, M., Muxi, L.,1998. Coprothermobacter platensis sp. nov., a new anaerobicproteolytic thermophilic bacterium isolated from an anaerobicmesophilic sludge. Int. J. Syst. Bacteriol. 48, 1297e1304.

Fernandez, A., Huang, S.Y., Seston, S., Xing, J., Hickey, R.,Criddle, C., Tiedje, J., 1999. How stable is stable? Functionversus community composition. Appl. Environ. Microbiol. 65(8), 3697e3704.

Ge, H., Jensen, P.D., Batstone, D.J., 2011. Temperature phasedanaerobic digestion increases apparent hydrolysis rate forwaste activated sludge. Water Res. 45 (4), 1597e1606.

Hori, T., Haruta, S., Ueno, Y., Ishii, M., Igarashi, Y., 2006. Dynamictransition of a methanogenic population in response to theconcentration of volatile fatty acids in a thermophilicanaerobic digester. Appl. Environ. Microbiol. 72 (2), 1623e1630.

Huber, R., Hanning, M., 2007. The Prokaryotes. Springer NewYork, New York, NY, pp. 899e922.

Hugenholtz, P., Tyson, G.W., Blackall, L.L., 2002. Design andevaluation of 16S rRNA-targeted oligonucleotide probes forfluorescence in situ hybridization. Methods Mol. Biol. 179,29e42.

Jetten, M.S.M., Stams, A.J.M., Zehnder, A.J.B., 1990. Acetatethreshold values and acetate activating enzymes inmethanogenic bacteria. Fems Microbiol. Ecol. 73 (4), 339e344.

Johnson, M.R., Conners, S.B., Montero, C.I., Chou, C.J.,Shockley, K.R., Kelly, R.M., 2006. The Thermotoga maritimaphenotype is impacted by syntrophic interaction withMethanococcus jannaschii in hyperthermophilic coculture. Appl.Environ. Microbiol. 72 (1), 811e818.

Karakashev, D., Batstone, D.J., Angelidaki, I., 2005. Influence ofenvironmental conditions on methanogenic compositions inanaerobic biogas reactors. Appl. Environ. Microbiol. 71 (1),331e338.

Kobayashi, T., Li, Y.Y., Harada, H., 2008. Analysis of microbialcommunity structure and diversity in the thermophilicanaerobic digestion of waste activated sludge. Water Sci.Technol. 57 (8), 1199e1205.

Kobayashi, T., Yasuda, D., Li, Y.-Y., Kubota, K., Harada, H., Yu, H.-Q., 2009. Characterization of start-up performance andarchaeal community shifts during anaerobic self-degradation

of waste-activated sludge. Bioresour. Technol. 100 (21),4981e4988.

Lee, S.-H., Kang, H.-J., Lee, Y.H., Lee, T.J., Han, K., Choi, Y.,Park, H.-D., 2012. Monitoring bacterial community structureand variability in time scale in full-scale anaerobic digesters. J.Environ. Monit. 14 (7), 1893e1905.

Leven, L., Eriksson, A.R.B., Schnurer, A., 2007. Effect of processtemperature on bacterial and archaeal communities in twomethanogenic bioreactors treating organic household waste.Fems Microbiol. Ecol. 59 (3), 683e693.

Li, Y.Y., Noike, T., 1992. Upgrading of anaerobic-digestion ofwaste activated-sludge by thermal pretreatment. Water Sci.Technol. 26 (3e4), 857e866.

Loy, A., Arnold, R., Tischler, P., Rattei, T., Wagner, M., Horn, M.,2008. probeCheck e a central resource for evaluatingoligonucleotide probe coverage and specificity. Environ.Microbiol. 10 (10), 2894e2898.

Ludwig, W., Strunk, O., Westram, R., Richter, L., Meier, H.,Yadhukumar, Buchner A., Lai, T., Steppi, S., Jobb, G.,Forster, W., Brettske, I., Gerber, S., Ginhart, A.W., Gross, O.,Grumann, S., Hermann, S., Jost, R., Konig, A., Liss, T.,Lussmann, R., May, M., Nonhoff, B., Reichel, B., Strehlow, R.,Stamatakis, A., Stuckmann, N., Vilbig, A., Lenke, M., Ludwig, T.,Bode, A., Schleifer, K.H., 2004. ARB: a software environment forsequence data. Nucleic Acids Res. 32 (4), 1363e1371.

Marchesi, J.R., Sato, T., Weightman, A.J., Martin, T.A., Fry, J.C.,Hiom, S.J., Wade, W.G., 1998. Design and evaluation of usefulbacterium-specific PCR primers that amplify genes coding forbacterial 16S rRNA. Appl. Environ. Microbiol. 64 (2), 795e799.

Marsh, T.L., Saxman, P., Cole, J., Tiedje, J., 2000. Terminalrestriction fragment length polymorphism analysis program,a web-based research tool for microbial community analysis.Appl. Environ. Microbiol. 66 (8), 3616e3620.

McMahon, K.D., Stroot, P.G., Mackie, R.I., Raskin, L., 2001.Anaerobic codigestion of municipal solid waste and biosolidsunder various mixing conditions e II: microbial populationdynamics. Water Res. 35 (7), 1817e1827.

Mladenovska, Z., Ahring, B.K., 2000. Growth kinetics ofthermophilic Methanosarcina spp. isolated from full-scalebiogas plants treating animal manures. Fems Microbiol. Ecol.31 (3), 225e229.

Nelson, M.C., Morrison, M., Yu, Z.T., 2011. A meta-analysis of themicrobial diversity observed in anaerobic digesters. Bioresour.Technol. 102 (4), 3730e3739.

Nesbo, C.L., Dlutek, M., Zhwcybayeva, O., Doolittle, W.F., 2006.Evidence for existence of “mesotogas”, members of the orderthermotogales adapted to low-temperature environments.Appl. Environ. Microbiol. 72 (7), 5061e5068.

Paul, E., Carrere, H., Batstone, D.J., 2012. In: Paul, E., Liu, Y. (Eds.),Biological Sludge Minimization and Biomaterials/BioenergyRecovery Technologies. John Wiley & Sons, Inc., New Jersey,pp. 373e404.

Pervin, H.M., Batstone, D.J., Bond, P.L., 2013. Previouslyunclassified bacteria dominate during thermophilic andmesophilic anaerobic pre-treatment of primary sludge. Syst.Appl. Microbiol. 36 (4), 281e290.

Pycke, B.F.G., Etchebehere, C., Van de Caveye, P., Negroni, A.,Verstraete, W., Boon, N., 2011. A time-course analysis of fourfull-scale anaerobic digesters in relation to the dynamics ofchange of their microbial communities. Water Sci. Technol. 63(4), 769e775.

Rittmann, B.E., 2008. Opportunities for renewable bioenergy usingmicroorganisms. Biotechnol. Bioeng. 100 (2), 203e212.

Riviere, D., Desvignes, V., Pelletier, E., Chaussonnerie, S.,Guermazi, S., Weissenbach, J., Li, T., Camacho, P., Sghir, A.,2009. Towards the definition of a core of microorganismsinvolved in anaerobic digestion of sludge. ISME J. 3 (6),700e714.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 7 0 9 8e7 1 0 87108

Sait, L., Galic, M., Strugnell, R.A., Janssen, P.H., 2003. Secretoryantibodies do not affect the composition of the bacterialmicrobiota in the terminal ileum of 10-week-old mice. Appl.Environ. Microbiol. 69 (4), 2100e2109.

Sasaki, K., Morita, M., Sasaki, D., Nagaoka, J., Matsumoto, N.,Ohmura, N., Shinozaki, H., 2011. Syntrophic degradation ofproteinaceous materials by the thermophilic strainsCoprothermobacter proteolyticus and Methanothermobacterthermautotrophicus. J. Biosci. Bioeng. 112 (5), 469e472.

Shiratori, H., Ohiwa, H., Ikeno, H., Ayame, S., Kataoka, N., Miya, A.,Beppu, T., Ueda, K., 2008. Lutispora thermophila gen. nov., spnov., a thermophilic, spore-forming bacterium isolated from athermophilic methanogenic bioreactor digesting municipalsolid wastes. Int. J. Syst. Evol. Microbiol. 58, 964e969.

Singh, B.K., Nazaries, L., Munro, S., Anderson, I.C., Campbell, C.D.,2006. Use of multiplex terminal restriction fragment lengthpolymorphism for rapid and simultaneous analysis ofdifferent components of the soil microbial community. Appl.Environ. Microbiol. 72 (11), 7278e7285.

Supaphol, S., Jenkins, S.N., Intomo, P., Waite, I.S., O’Donnell, A.G.,2011. Microbial community dynamics in mesophilic anaerobicco-digestion of mixed waste. Bioresour. Technol. 102 (5),4021e4027.

Weiland, P., 2010. Biogas production: current state andperspectives. Appl. Microbiol. Biotechnol. 85 (4), 849e860.

Weiss, A., Jerome, V., Freitag, R., Mayer, H.K., 2008. Diversity ofthe resident microbiota in a thermophilic municipal biogasplant. Appl. Microbiol. Biotechnol. 81 (1), 163e173.