Culture-independent, molecular techniques were applied to the characterization of microbial communities of an anaerobic granular sludge obtained from a full-scale digester. Procedures were optimised for total DNA recovery and polymerase chain reaction (PCR) amplification of 16S rDNA using archaea- and eubacteria-specific oligonucleotide primers. Cloned PCR products were subsequently screened by amplified rDNA restriction analysis to identify operational taxonomic units (OTUs). Inserts from clones representing each OTU were sequenced and phylogenetic trees were prepared. In addition, the microbial communities were characterised using terminal restriction fragment length polymorphism (T-RFLP). The specific methanogenic activity of the biomass, against various substrates, was also ascertained. Two anaerobic bioreactors were seeded with granular and non-granular (i.e. crushed) aliquots of the characterised sludge, respectively, and used to investigate the treatment of a volatile fatty acid (VFA)-based synthetic wastewater, at a loading rate of 5 kg COD m−3 day−1 at low ambient temperatures (18°C). DNA was isolated from sludge samples during the test period and shifts in archaeal and eubacterial population structures were elucidated. The start-up period was successful with methane yields and COD removal efficiencies of 60–75% and 65–85%, respectively. Specific methanogenic activities of reactor biomass, obtained at the conclusion of the trial, indicated the development of psychrotolerant biomass during the 90-day experiment. Furthermore, the efficacy of T-RFLP as a molecular tool for use in the surveyance of engineered ecosystems was confirmed.
Anaerobic digestion of wastewater offers several advantages over more conventional processes, including a reduction in energy needed for aeration and the production of methane, a readily usable fuel. Currently, full-scale applications of anaerobic treatment are almost exclusively managed at temperatures exceeding 18°C , i.e. mesophilic (30–37°C) and thermophilic (55–65°C) operation. However, under moderate climate conditions many wastewaters, including domestic and industrial effluents, are discharged at a considerably lower temperature than the optima of biological wastewater treatment processes such as nitrification, denitrification and mesophilic methanogenesis . Moreover, the degradation of organic matter with subsequent methane production occurs at low temperatures in most terrestrial ecosystems of boreal and northern climate zones . For example, methanogenesis has been described in tundra soil, pond sediments and deep-lake ecosystems . In addition, psychrophilic wastewater treatment technologies offer an attractive potential alternative to established systems, since the maintenance of mesophilic anaerobic digestion facilities is expensive . To this end, significant advances have recently been made in anaerobic reactor designs, which may be applicable to low-temperature anaerobic digestion.
Psychrophilic anaerobic treatment has been reported to occur at lower rates and with less stability than under mesophilic conditions [6,7]. The principal difficulty is the decreased level of mixing and fluidisation of reactor biomass, attributable to the reduced rate of biogas production during psychrophilic methanogenesis. This drawback has been overcome with a considerable degree of success by the introduction of the expanded granular sludge bed (EGSB) reactor . Although growing, the number of reports of successful low-temperature digestion [1,2,5,9] is modest and this research field is still in its infancy. Much scope exists for the application of further bioengineering innovations, which will allow for the implementation of psychrophilic anaerobic digestion on a more global basis for the treatment of a wide variety of wastewaters.
Despite the economic importance and widespread harnessing of anaerobic digestion in engineered systems and the development of de novo reactor types, coupled with an increased depth of understanding with regard to the complex microbial processes that occur in anaerobic digestion , relatively little knowledge has been acquired with regard to the structure and function of the microbial populations within the process. Understanding the dynamics of such communities would be difficult enough if each population could be isolated from the whole, cultivated and characterised. Our inability, however, to cultivate greater than 90% of the members from many communities, makes the task considerably more daunting . However, in recent years, nucleic acid-based, phylogenetic approaches have proven useful to describe the microbial ecology of such consortia, particularly when supported by physiological, microscopic and biochemical data. Molecular methods targeting the small subunit rRNA-encoding genes are currently used to investigate microbial communities in anaerobic digesters and polymerase chain reaction (PCR)-based molecular typing methods rapidly provide an image of the community structure of an ecosystem (e.g. [12,13]). One of these rRNA gene-based approaches is terminal restriction fragment length polymorphism (T-RFLP) analysis of 16S rRNA genes, which allows for the rapid identification of ribotypes from a variety of samples, including soils and sludges of environmental origin . Due to the sensitivity and high throughput of this method it is an ideal technique for comparative community analyses .
The goal of the present study was to investigate the microbial assemblages within an anaerobic, mesophilic sludge, used to treat a synthetic wastewater under psychrophilic conditions. Molecular techniques, 16S rRNA gene sequencing and T-RFLP analysis, were used to characterise the microbial community structure of the sludge and the population dynamics throughout the starting phase of reactor operation, respectively. Process data and physiological profiling analysis were used in conjunction with T-RFLP, which is confirmed here as a useful and effective tool for frequent biomonitoring. We examined the feasibility of anaerobic digestion of high-strength industrial wastewaters at low temperatures and recorded a successful start-up period for EGSB lab-scale reactors at 18°C, suggesting the prevalence of psychrotolerant organisms.
Materials and methods
Source of biomass
A granular, mesophilic, anaerobic sludge, obtained from a full-scale, 8000 m3 UASB bioreactor, operating at 37°C, at Archer Daniels Midland (ADM, Co. Cork, Ireland), treating citric acid production wastewater, was used for the present study.
Design and operation of psychrophilic anaerobic bioreactors
Two 3.5-l glass laboratory-scale EGSB reactors (R1 and R2) were used in the present study, as described by Colleran and Pender . A total mass of 70 g volatile suspended solids (VSS) of seed sludge were inoculated to each bioreactor. In order to evaluate the significance of granular aggregates for successful low-temperature digestion, R1 and R2 were seeded with crushed (non-granular) and granular sludge, respectively. The anaerobic reactors were fed a synthetic wastewater (pH 7.5±0.2) consisting of ethanol, butyrate, propionate and acetate, in the chemical oxygen demand (COD) ratio of 1:1:1:1, to a total of 10 g COD l−1. The synthetic influent was buffered with NaHCO3 and fortified, as described by Shelton and Tiedje , with macro- (10 ml l−1) and micro- (1 ml l−1) nutrients. Both reactors were maintained at 18±1°C and at a 2-day hydraulic retention time (HRT) for the entire test run of 90 days. Effluent was recycled to give a liquid upflow velocity of 5 m h−1 and samples of reactor effluent and biogas were routinely taken for volatile fatty acids (VFAs), COD, and CH4 determination as described previously .
Determination of specific methanogenic activity
The specific methanogenic activity (SMA) of the seed sludge and reactor sludge samples on day 90 was ascertained at 37, 30, 22 and 15°C. The tests were employed with acetate, butyrate, propionate, ethanol and H2CO2 as substrates as described previously . The sludge samples were washed and a final biomass concentration of 2–5 g VSS l−1 was added to anaerobic activity test medium, to give a final volume of 10 ml in 20 ml serum vials for liquid substrates and in 60 ml hypovials for gaseous substrates. Vials without any added substrate, or with the addition of N2CO2 in the case of the 60 ml vials, served as controls.
Extraction and PCR amplification of 16S rDNA
An extraction protocol was first optimised for the recovery of total genomic DNA from sludge granules. A number of methods were examined for isolation of nucleic acids, the merits of which were assessed by determination of the cell lysis efficiency, i.e. the fewest unlysed cells while maintaining a high yield of non-degraded, high molecular mass DNA. Microorganisms in sludge samples before and after DNA extraction were visualised microscopically according to the method of Bitton and co-workers , with the exception that Sybr-Gold (Molecular Probes, USA) was used as a stain instead of acridine orange. A sample (0.1 g) of crushed sludge was suspended in 500 μl of filter-sterilised water. An aliquot (100 μl) of this was added to 10 ml of filter-sterilised water. An 8 ml volume of the diluted sample was filtered onto black Isopore® (Whatman) membrane filters and 200 μl of Sybr-Gold (10× stock solution in TE buffer, pH 8) was added to the remaining 2 ml of diluted sample and left for 5 min. The remainder of the sample was then drawn onto the filter that was mounted onto a glass slide and a minimum amount of mineral oil was added under a coverslip. The samples were viewed using a Nikon Optiphot-2UV microscope fitted with a 100 W mercury bulb, a B-2A excitation filter for blue light, a 100× planar objective lens and 10× eyepieces. Fluorescent microscopy of sludge granules was carried out as described previously by Pender .
Aggregates were initially disassociated by grinding in a mortar and pestle, or by sonication (UltraSonik, Yucaipa, CA, USA) for 30 s, prior to microbial cell lysis using a chemical lysis approach, as described by Zhou et al. . Alternatively, aggregates were disrupted and cells were lysed by bead beating combined with chemical lysis. Different combinations of the above were tested and it was found that gently crushing the sludge granules with a pestle and mortar, before passing through a Soil DNA Kit (MoBio Laboratories) gave the highest yield of good-quality DNA with little shearing, and provided overall the best microbial cell lysis. Archaeal and eubacterial 16S rRNA genes were amplified with forward primer 21F (5′-TTCCGGTTGATCCYGCCGGA-3′)  and reverse primer 958R (5′-YCCGGCGTTGAMTCCAATT-3′) , and forward primer 27F (5′-GAGTTTGATCCTGGCTCAG-3′)  and reverse primer 1392R (5′-ACGGGCGGTGTGTRC-3′) , respectively. Reaction mixtures were prepared in a laminar air-flow biological cabinet (Nuaire, Plymouth, UK) and PCR was performed with 10 ng of DNA as the template, the primer set (0.25 μM l−1 of each primer), MgCl2 (0.125 mmol), 5 μl 10× NH4 buffer and 1 U of Taq DNA polymerase (Bioline). The cycle profiles used were denaturation at 95°C for 1.5 min, annealing at 55°C (archaeal) or 52°C (eubacterial) for 1.5 min and extension at 72°C for 1.5 min; the number of cycles was 30. PCR products were resolved by gel electrophoresis on 1% 1× TAE agarose gels, containing ethidium bromide (1 μg ml−1) and visualised by UV excitation. Primers were removed and the amplified products were concentrated, using a PCR Prep Purification kit (Promega) according to the manufacturer's protocol.
Cloning of 16S rDNA, amplified rDNA restriction analysis (ARDRA), sequencing and phylogenetic analysis of seed sludge biomass
PCR amplicons were ligated into the plasmid vector pCR® 2.1-TOPO (Invitrogen) and used to transform TOPO 10 (Invitrogen) competent Escherichia coli cells by following the manufacturer's instructions. Clone libraries were generated, by growing 96 (A1 to H12) archaeal and eubacterial clones, originating from the R1 and R2 seed sludge in respective micro-well plates (NUNC), containing 200 μl LB broth and 50 μg kanamycin ml−1 per well. Amplification products, generated using the vector-specific primers M13F (5′-GTTTTCCCAGTCACGAC-3′) and M13R (5′-CAGGAAACAGCTATGAC-3′), were obtained from clones and digested using the tetrameric restriction endonuclease HaeIII (Promega) at 37°C overnight. Resultant DNA fragments were separated electrophoretically in 3.5% 1× TAE high-resolution agarose gels, containing ethidium bromide as before. Operational taxonomic units (OTUs)  were identified, based on restriction cleavage patterns and clones representing the OTUs selected for sequencing. An alkaline-lysis miniprep kit (Qiagen) was used, as per the manufacturer's instructions, to prepare plasmid DNA from overnight cultures of positive transformants, and sequencing was achieved using vector-specific primers on a Licor gel sequencer (MWG Biotech, Milton Keynes, UK). Sequence data were compared with the nucleotide database using BLAST (Basic Local Alignment Search Tool ). Sequences from this study were manually aligned with sequences retrieved from the Ribosomal Database Project (RDP ), and the CHIMERA_CHECK software maintained at the RDP was used to identify potential chimeric sequences. Finally, phylogenetic trees were calculated with the Paup*4.0b8 phylogenetic inference package , using the Kimura-2 parameter correction . The partial 16S rRNA gene sequences determined in this study were deposited in the GenBank database under accession numbers AY161236 to AY161261 with the generic name of SSADM_E (Seed Sludge Archer Daniels Midland Eubacteria) for eubacterial clones and SSADM_A for archaeons.
T-RFLP analysis of community structure
T-RFLP was used, in parallel with clone library analysis, to examine the microbial community structure of the sample biomass in seed sludge, and in sludge samples recovered from the bioreactors on day 41 of operation and at the end of the test period (day 90). DNA was isolated from the biomass as described above. PCR was performed, as above, but using archaeal (21F and 958R) and eubacterial (27F and 1392R) primer sets, where forward (21F and 27F) and reverse (958R and 1392R) primers were labelled at the 5′ end with the phosphoramidite dyes 6-FAM and HEX, respectively. All PCR products were used directly for endonuclease restriction and separate digestions were processed, in the manufacturer's recommended buffers, with the 4-bp cutting enzymes HhaI and AluI (separately) at 37°C for 6 h. Terminal restriction fragments (TRFs) were sized using an automated ABI Prism genetic analyzer by Oswel, Southampton, UK. Genescan® 3.1 software was used to quantify the electropherogram output and sample data consisted of the size (base pairs), peak height and peak area for each TRF peak in sample profiles. The abundance of individual TRFs in a given sample was calculated, based on the total peak area of the TRF pattern under investigation. Those TRFs, however, with a peak area not greater than 2.5% of the total area of the sample were excluded from further analysis. TRF sizes were then rounded to the nearest bp, and 1-bp bins constructed, as per Clement et al. . More than 300 archaeal and 700 eubacterial in silico T-RFLP simulations were carried out using the primer combinations and endonuclease restriction sites, and sequences from both the RDP 16S rRNA database and those retrieved from the seed sludge clone libraries. Predicted TRF lengths were then used for comparison with actual TRFs obtained from sludge samples.
Psychrophilic reactor performance
A rapid start-up of the laboratory-scale reactors was achieved with less than 2500 mg COD l−1 present in the effluents of both reactors after 60 days of operation. Efficient and stable COD removal (generally between 70 and 75%), at the applied loading rate of 5 kg COD m−3 day−1, was maintained for the remaining 30 days of the start-up experiment and the two reactors were very similar in terms of performance during the 90 days of operation. The principal components of the effluents from both reactors were the VFA propionate and acetate (Fig. 1). Methane yields from both R1 and R2 typically constituted throughout the test period between 60 and 75% of total biogas produced (data not shown). The operational pH in the digesters was monitored during the trial and the extremities of this range were 6.8 and 7.7 (data not shown). There was no evidence of granulation of the biomass in R1, which was seeded with a crushed inoculum and the biomass consisted of a non-granular floculant sludge with no significant increase in the particle size distribution pattern to that determined for the seed sludge after the 90-day trial (data not shown).
Microbial population structure of the seed sludge as determined by 16S rRNA clone library and ARDRA analysis
The DNA extraction protocols carried out in this study provided cell lysis efficiencies of >99% according to microscopic analysis, which revealed fewer than 1% of all cells observed in seed sludge samples, prior to extractions, remained unlysed post-extraction (data not shown). By ARDRA, 13 different archaeal and 18 eubacterial OTUs were identified from the respective 96-clone libraries, and representative clones from each were chosen for sequencing. Chimeric clones, composed of 16S rRNA genes amplified from different organisms, can arise during PCR amplification of mixed DNA populations , and chimeras found here, which included five from the eubacterial study, were removed from further analysis. BLAST search results and phylogenetic reconstruction (Fig. 2) revealed that the majority (eight of 13) of the archaeal OTUs represented euryarchaeotal clones closely related to the methanogens, and, in particular, the order Methanosarcinales. Five OTUs, accounting for 56% of the total library, however, were found to represent close relatives of the Crenarchaeota. This indicates that, although greater apparent archaeal diversity was present within the euryarchaeotal segment of the community, most archaeons in the seed sludge were assigned to the terrestrial cluster of the Crenarchaeota. The function of Crenarchaeota, with respect to anaerobic digestion, is unclear as high levels of these organisms have not previously been reported in such systems. Further study is required to investigate the role, if any, of these organisms in anaerobic digestion.
The bacterial community, as revealed by this inventory (Fig. 3), included organisms closely related to the Syntrophus group of the δ-Proteobacteria (SSADM_ED8), the Flexibacter–Cytophaga–Bacteroides group (SSADM_EF11) and the β-Proteobacteria (SSADM_EG9), which composed 24%, 12% and 7% of the eubacterial clone library, respectively. The remaining 57% of the clones were represented by the Gram-positive organisms, including clones related to the Clostridia (SSADM_ED5, SSAMD_EE9). The primary functions of the major constituents of the eubacteria in anaerobic digester ecosystems are hydrolysis, fermentation and, as in this case, the syntrophic metabolism of organic acids, ketones and alcohols.
Diverisity of seed sludge and microbial population dynamics in the laboratory-scale reactors as determined by T-RFLP analysis
Archaeal TRF patterns derived from both endonucleases used included TRF peaks, which, when compared to predicted peaks from in silico experiments, suggested the presence of both euryarchaeotal and crenarchaeotal members. More specifically, T-RFLP revealed the dominance of Methanosarcina within the methanogenic population and the significant proportion of Crenarchaeota-like organisms within the sludge, thus supporting the conclusions of the clone libraries and sequence analysis. To begin with, the archaeal TRF profile of the seed biomass, generated using HhaI as restriction endonuclease and the universal (reverse) primer, revealed the dominance of peaks at 579 and 585 bp representing members of the Methanomicrobiales and marine Crenarchaeota, respectively (data not shown). However, the use of the forward primer for the same sample provided a more in-depth view of the structure of this community, with peaks representing the halophilic archaea, Methanosarcinales and Methanobacteria (Fig. 4). Secondly, archaeal T-RFLPs from the seed sludge, which were digested with AluI and generated using the reverse primer, described the presence of a peak at 171 bp, representing SSADM_AC4 and SSADM_AD3 from the clone library in this study, and larger peaks at 439 and 631 bp for which no matches were found in in silico experiments (data not shown). Again, results from experiments using the forward primer offered more clarity with respect to the composition of the community under investigation and suggested the presence of members of the Methanosarcinales and a peak at 92 bp, for which no match was found in the in silico experiments (data not shown).
T-RFLP profiles of the archaeal community within the seed sludge and the reactor biomass, at days 41 and 90, for the R1 and R2 experiments, using the reverse primer suggest that a stable archaeal community composition was maintained for the duration of both reactor trials. A dominant peak at 439 bp, for example, in the AluI-digested profiles was clear for both digesters (data not shown). The profiles for the R1 (Fig. 4(i)) and R2 (Fig. 4(ii)) experiments, using the forward primer and HhaI, illustrate the persistence of a peak throughout the trial period at 198 bp representing the Methanosarcinales population. Peaks at 328 and 330 bp represent organisms closely related to Methanosarcina vacuolata and Methanobacterium palustre, respectively (Fig. 4). Despite, differences in the levels of discrimination between forward and reverse primer-based T-RFLP, there was good agreement in the general patterns of community structure observed for both the archaeal and bacterial populations of the reactors.
Patterns obtained from eubacterial samples similarly supported the evidence as per ARDRA and clone library analysis, demonstrating the bacterial diversity of the biomass. Reverse primer-generated patterns for the seed biomass, which were restricted with HhaI, suggest a dominant Proteobacteria segment in the community (300 bp), with smaller peaks at 287 and 302 bp, representing green non-sulphur bacteria, and Fibrobacter and Gram-positive bacteria, respectively. The corresponding forward primer profile (Fig. 5), however, discriminates the composition of the proteobacterial community, revealing the presence of a ribotype similar to Helicobacter pylori at 98 bp, Iosphaera and Wollinea at 376 bp and also records a peak at 562 bp representing a ribotype similar to Arhodomonas. The eubacterial structure as described by the reverse primer profile, and using AluI as digestion enzyme, revealed the presence of members of the Flexibacter–Cytophaga–Bacteroides group by a peak at 73 bp and of clones from this study (SSADM_ED5, SSADM_EE3, SSADM_EE5 and SSADM_EG12) by a peak at 130 bp (data not shown). When the profile was obtained using the forward primer TRF, organisms related to Bacillus sp. were identified at the 73 bp peak and peaks were recorded at 175, 189 and 248 bp, representing no match, and organisms related to Haloanaerobium and Paracoccus, respectively (data not shown).
TRF profiles of the bacterial assemblages throughout the experiments for both bioreactors using the reverse primer show the persistent dominance of the Proteobacteria at 300 bp. This suggests an inherent stability within the bacterial populations of the two digesters. The TRF profiles, however, of bacterial communities within the seed sludge and the digester biomass, at days 41 and 90, for R1 (Fig. 5(i)) and R2 (Fig. 5(ii)), using the forward primer and HhaI, showed very different dynamics within the digester ecosystems. The peak at 98 bp in the seed sludge, representing members of the Proteobacteria, is not dominant in the R1 biomass by the trial conclusion. This peak almost disappeared by day 41 in the R2 biomass but re-emerges as a peak of interest by day 90. A peak at about 1125 bp, which could not be identified by in silico restrictions, is evident in the biomass of both bioreactors (Fig. 5). Overall, a more diverse bacterial population is represented by the profiles from the R1 biomass, which also appears to exhibit the least compositional stability.
It should be noted that the electronic simulations, from which predicted TRF peak lengths were obtained, were not exhaustive. Some peaks, therefore, were found in TRF profiles for which no matches were available through in silico experiments. Furthermore, some TRFs could represent microorganisms not yet added to the database, and were therefore used only to identify microbial groups rather than species. Despite this, T-RFLP does allow biomonitoring of the fate of these peaks during the sampling and experimental period, maintaining the integrity of the technology as a suitable biomonitoring tool.
SMA profiles of seed sludge and reactor biomass sampled on day 90
The mesophilic seed sludge displayed relatively high SMA values for ethanol, acetate and H2CO2 when measured at 37°C (Table 1). The activity with butyrate was low and long lag phases were observed before the onset of substrate conversion to methane. No propionate degrading activity was detectable. Considerably lower activities were observed under psychrophilic conditions, i.e. 15°C, than at 37°C (Table 1).
|Substrate/Test||Seed (day 0)||R1 (day 90)||R2 (day 90)|
|Substrate/Test||Seed (day 0)||R1 (day 90)||R2 (day 90)|
aLag 290 h.
bLag 78 h.
dLag 780 h.
eLag 590 h.
fLag 214 h.
gLag 215 h.
hLag 830 h.
iLag 430 h.
jLag 200 h.
The SMA activities of sludge biomass from reactor 1 at 37°C for ethanol, butyrate, acetate, propionate and H2CO2 increased by a factor of 5, 7, 13, 81 and 5, respectively, when measured after 90 days of operation (Table 1). A substantial increase was also apparent for activity with butyrate. The activity with propionate, however, remained low after 90 days of operation, despite the presence of propionate as a wastewater constituent. Stable degradation of acetate was achieved, thereby indicating greater reactor stability, as 70% of all methane formed during anaerobic digestion originates from acetate . Although lesser activities were observed at 30, 22 and 15°C, greater methanogenic activity than that of the seed sludge was evident with all of the above substrates. A similar SMA profile was established for R2 biomass (Table 1), with high rates of substrate conversion to biogas. This was represented by extremely high activities when the tests were carried out at 37°C. Low propionate activities were again observed.
The level of performance achieved by the two reactors investigated here was somewhat less than that reported previously for mesophilic reactors treating VFA wastewaters  and also of one set of psychrophilic systems seeded with a different sludge , but is comparable with other reported data for psychrophilic systems and affirms the potential of this approach . The methane yield of the biogas compares favourably with previous reports of both mesophilic  and psychrophilic  anaerobic digesters. One of the major concerns with regard to sub-ambient anaerobic reactors is the low biogas production rate . High methane production not only reflects increased methanogenic activity, but also promotes mixing and fluidisation of reactor biomass, thus inferring greater COD removal, due to increased sludge–substrate contact.
Propionate was the principal VFA present in the effluents of R1 and R2 throughout the trial. Rebac  reported that propionate oxidation is most sensitive in a psychrophilic anaerobic environment and may thus be the rate-limiting step for reactor operation under low-temperature conditions. The poor performance of the reactor biomass in terms of propionate degradation is reflected in the extremely low to non-existent metabolic activity of the reactor biomass with this substrate (Table 1). Furthermore, the high acetate levels observed in the reactor effluents may contribute to inhibition of the development of a propionate-degrading biomass, and the long lag phases observed before the onset of propionate degradation in batch tests, as previous studies have demonstrated that acetate has a strong inhibitory effect on propionate degradation in anaerobic bioreactors . Although the low levels of propionate degrading activity in the seed sludge were not desirable for the successful start-up of the digesters, it is clear that no population capable of occupying this niche had emerged after the 90-day trial. Longer experiments will determine whether the development of such a population is possible or whether high propionate degrading activity will be the key prerequisite for seeding low-temperature anaerobic digesters. Rebac  suggested that butyrate degradation was the most stable with respect to temperature fluctuations and it was seen here that the increase in the butyrate activity was the most pronounced of all the VFA substrates (seven-fold), and that butyrate degrading capacity could be developed during low-temperature reactor operation.
Crushing of the sludge granules allows for the penetration of the substrate to the core of the aggregates  and we suggest here that this may account for the greater acetoclastic methanogenic activities recorded for R1 biomass. The lack of granulation observed in the reactor seeded with crushed anaerobic sludge may be the result of unfavourable substrate conditions, e.g. the absence of carbohydrates in the influent . Alternatively, the applied loading rates and applied liquid upflow velocity may have been unsuitable for granulation at low temperatures, although granulation was observed under identical substrate, upflow velocity and loading rate conditions under mesophilic conditions . It is also unlikely to be a particular psychrophilic issue as granulation has been observed in low-temperature anaerobic reactors [G. Collins, unpublished data]. The exact liquid upflow velocity regime and loading rates required during start-up for the induction of granulation at low temperatures remain unclear and are an important area for further research before full-scale psychrophilic applications are contemplated.
The microbial populations involved in the metabolism of the substrates studied in batch tests remained mesophilic (temperature optima of 37°C) after the 90-day start-up period (Table 1). All SMA profiles, except those for H2CO2 and ethanol (in the case of R2) collated at 15°C for the reactor sludges were higher than those recorded at 37°C for the seed sludge, thus revealing a satisfactory development of the methanogenic activity of the community. We posit that the euryarchaeotal or methanogenic segment of the microbial community, therefore, may have become psychrotolerant and thus displayed elevated activities at 15°C. Chin and Conrad  demonstrated that H2CO2 was consumed mainly by methanogenesis at 30°C but by homoacetogenesis at 15°C, and comparisons of the relative amounts of accumulated intermediates, in that study, indicated that H2 consumption by methanogenesis was more sensitive to low temperatures than homoacetogenic hydrogen consumption. This may explain the absence of increased methanogenesis at 15°C of H2CO2. However, no evidence of homoacteogenic activity (acetate accumulation) was observed in batch activity tests (data not shown), and no significant development of new populations was detected by T-RFLP analysis during the 90-day trial. The relatively poor hydrogenophilic activity suggests that elevated hydrogen partial pressures may also have contributed to the poor conversion of propionate in the laboratory-scale reactors .
The high levels of acetate observed in the reactor effluents and the poor granulation of R2 biomass are also consistent with the observed dominance of Methanosarcina sp. in reactor biomass, as determined by both T-RFLP and clone library analysis, and although this organism has been shown to be present in granules and also to be capable of ‘spontaneous granulation’, it is generally accepted that the dominance of the filamentous acetoclastic Methanosaeta sp. is required for practically useful granules. Methanosaeta sp. are rod-shaped organisms that grow as filaments and are regarded as being important in the onset of granulation and the maintenance of stable granules during system perturbations [41,42]. In situ hybridisation studies carried out by Merkel et al.  on sludge from a mesophilic digester fed lactate, propionate, butyrate and acetate revealed that members of the genus Methanosaeta made up more than 90% of the archaeal community in the reactor. Other researchers have also identified Methanosaeta as the predominant acetoclastic methanogen in anaerobic reactors when using microscopic  and MPN methodologies . This organism is selected over the faster-growing Methanosarcina through its greater substrate affinity for acetate and is selected for in well-functioning anaerobic digesters with low levels of acetate . Consequently, the choice of inoculum and the high levels of acetate recorded in the reactors during this trial may well have inhibited granulation.
While the cell envelopes of some archaea are very fragile and easy to lyse in water, other members are almost unbreakable by conventional methods . Some have a particularly thick and rigid outer layer named the methanochondroitin , which shows a high resistance towards chemical lysing agents such as sodium dodecyl sulphate (SDS) or Triton X-100, and to physical disruption . In order to retrieve representative DNA yields for community structure analyses, therefore, a robust and efficient cell lysis procedure must be employed and the method used in this study fulfilled these criteria and allowed the generation of community profiles from the reactors during start-up. Although some researchers have reported a bias towards Methanosaeta-like species using cloning-based techniques , this was not observed in this case. In this study, TRF patterns obtained using the universal reverse primers 958R (Archaea) and 1392R (Bacteria) provided a ‘macro-view’ of the community structure within samples, where the conserved nature of the target region resulted in a summary of community structure only discriminated to the level of group, e.g. Methanobacteriales, based on the peak profiles. Conversely, T-RFLP profiles derived from fluorescently-labelled forward primers 21F (archaeal) and 27F (eubacterial) offered a more discriminatory and detailed description of population structure, down to genus level in some cases, as a consequence of the length heterogeneities at the 5′ end of the gene, within the V1, V2 and V3 regions . Both types of primer, however, are complementary and, when used together, provided a comprehensive snapshot of the status of microbial communities within environmental samples. The results described here, illustrate the use of these primers for T-RFLP, to provide high resolution and sensitivity for the detection of OTUs from a complex community, notwithstanding the potential technical problems of the method [14,52], including potential PCR bias [53,54]. The T-RFLP results demonstrated the relatively well replicated nature of the archaeal population structures and dynamics of both reactor systems and suggest that a reproducible and stable response to the environmental parameters occurred in both reactors. This is in agreement with other studies suggesting that stable, replicated archaeal populations develop in replicated reactor experiments . A peak at 630 bp is present in the archaeal TRF pattern for the seed biomass, obtained using AluI and the reverse primer, but is almost absent in R1 at day 90. The apparent decrease in the relative abundance of organisms represented by this peak may be due to the non-granular nature of the R1 biomass, which suggests the importance of granular sludge for the retention of important community members within such engineered systems. In summary, however, the archaeal communities showed no major shift in general community structure and T-RFLP data obtained suggest that the overall archaeal diversity of the sludge did not change throughout the trial. This supports the evidence of a psychrotolerant population within the biomass, as also suggested by the SMA measurements.
The apparent diversity within the bacterial community, based on the number of TRF peaks and ARDRA-derived OTUs, was apparently greater than that among the Archaea in the seed sludge and the population structures observed here are in agreement with the findings of previous authors who studied mesophilic sludges [56,57]. However, the limitations of the techniques used here should be noted since more than one representative TRF may be located under a T-RFLP peak and the ARDRA approach may underestimate the diversity of microbial populations in environmental samples . Furthermore, the bacterial populations in the reactors were more dynamic than the archaeal communities. It has been suggested, however, that an extremely dynamic community sustains a functionally stable ecosystem  and that prevalent members and diversity within the microbial communities of such ecosystems can change dramatically even over short periods.
The T-RFLP and clone library analyses provided correlating data on the archaeal and bacterial population structure of the sludge samples. The population structure recorded for the reactors also corresponded well with the SMA profiles and the reactor performance, for example, the elevated acetate concentrations observed in the reactor effluent and the poor granulation of R2 biomass was consistent with the observed dominance of Methanosarcina sp. in the reactor biomass, as determined by both T-RFLP and clone library analysis. In general, the polyphasic experimental approach employed here provided evidence that links between low-temperature metabolic activity profiles, microbial population structure and reactor performance may be established. However, the value of the approach will be more comprehensively tested during long-term reactor trials and by monitoring the performance and population dynamics of reactor biomass in response to changing environmental conditions.
In this study, T-RFLP was successfully used as a molecular approach for biomonitoring of the microbial communities in anaerobic digesters and presents a sensitive, high-throughput approach. Molecular techniques are now an invaluable tool for elucidating members of complex microbial communities, and when used in conjunction with process engineering and physiological measurements they will allow a more extensive knowledge and understanding of the physiology and biochemistry of the microbial populations involved in psychrophilic anaerobic digestion. In particular, long-term low-temperature reactor trials will be required to develop a comprehensive understanding of the microbial interactions which occur in these systems and also the ecological factors which will determine reactor performance, such as, whether new, psychrophilic, microbial populations would emerge over time.
The receipt of financial support from the Higher Education Authority (HEA) Programme for Research in Third Level Institutes (PRTLI)-Cycle II, through the Environmental Change Institute, NUI, Galway, and an Enterprise Ireland Research Scholarship to G.C. is gratefully acknowledged.