Anaerobic digestion of olive mill wastewaters is generally performed in anaerobic contact bioreactors where the removal of toxic phenols is often unsatisfactory. In the present work we show that a granular activated carbon packed-bed biofilm reactor can be successfully used to achieve effective and reproducible wastewater decontamination even at high organic loads. A comparison of 16S rRNA gene sequences of the inoculum and of biomass samples from different districts of the reactor revealed enrichment of specific microbial populations, probably minor members of the inoculum and/or of the olive mill wastewaters. They mainly consisted of the members of Proteobacteria, Flexibacter-Cytophaga-Bacteroides, and sulphate-reducing bacteria. The dominant sequence among Archaea (70% of clones) was closely related to Methanobacterium formicicum.
Olive mill wastewater
Granular activated carbon
16S rRNA gene
Olive mill wastewater (OMW) is the effluent from olive oil production. OMWs exhibit a high phytotoxicity and antibacterial potential because of their high chemical oxygen demand (COD) and content of phenolic compounds [1–3], and therefore have to be treated before discharge [1,2]. The most promising OMW treatment technology proposed thus far is anaerobic microbial digestion because of its ability to combine OMW bioremediation performances with the generation of CH4[1,4,5]. However, this process, generally performed in conventional dispersed growth bioreactors, is unable to completely remove toxic phenols [6–8], which currently hinders its large-scale application. Biological and chemical–physical pre-treatments for the removal of phenolic compounds from OMWs [6,7,9–12] or aerobic biological post-treatments directed to degrade phenols persisting in effluents of conventional digesters [13,14] have been proposed. In most cases, however, the proposed pretreatments were not effective [7,12], and all attempts to develop an integrated anaerobic–aerobic treatment resulted in processes that were difficult to manage under open mode of operation .
The possibility of increasing the biodegradation of OMW phenols using immobilized biomass anaerobic digesters packed with granular activated carbon (GAC) or “Manville” silica beads has recently been demonstrated [15, Bertin et al. unpublished data]. The GAC packed-bed reactor showed higher yields in COD and phenolic compound removal when compared to the silica bead reactor, and exhibited a better tolerance towards high organic loads than the dispersed growth anaerobic digester developed by Beccari and coworkers using the same microbial inoculum [7,12,13]. On the basis of these promising preliminary results, we decided to further investigate the GAC packed-bed OMW digester by studying its stability, biodegradation efficiency and CH4 productivity under a large range of high OMW loads. For a more complete assessment of the operation of the GAC-biofilm reactor, we also elucidated the structure and spatial distribution of microbial community within the reactor using a combination of terminal restriction fragment length polymorphism (T-RFLP), sequencing and phylogenetic analyses of 16S rRNA genes.
Despite numerous studies on anaerobic digestion of OMWs [1–4,13], this is the first report on the use of a GAC packed-bed biofilm reactor for OMWs disposal, and on the structure of microbial community within an OMW anaerobic digester evaluated with molecular techniques.
2 Materials and methods
Chemicals used in the analysis of COD, total phenolic compounds, SO4= and total protein and in the preparation of the samples for scanning electron microscopy were obtained from Sigma–Aldrich, Milan, Italy and from Carlo Erba, Milan, Italy. Granular activated carbon (GAC, as cylinders, 3 mm in diameter and 10 mm in length) was supplied by Chemviron Carbon (Feluy, Belgium). The solvents used for HPLC and ion chromatography were purchased from Baker Italia (Milan, Italy).
2.2 Inoculum and OMWs used in the reactor
The anaerobic consortium used as inoculum for the packed-bed loop reactor was the one recently developed by Beccari and coworkers [7,12]. Two industrial OMWs, i.e. OMW1 and OMW2, containing about 20 and 30 g/l of COD and 1.5 and 2.0 g/l of total phenolic compounds, respectively, were used. Two amended OMWs, i.e., AOMW1 and AOMW2, were prepared from OMW1 and OMW2, respectively, by diluting each OMW with an equal volume of tap water, and by amending them with Ca(OH)2 to increase their pH to 6.5, urea (0.45 g/l) and 1 N NaOH (to adjust their pH to 7.8 ± 0.2). The two AOMWs were placed in 4 l glass jars, where they were vigorously mixed with a magnetic stirrer and purged with 0.22 μm filter-sterilized (Millipore, MO, USA) O2-free N2 at room temperature for 3 h before use. AOMW1 and AOMW2 exhibited a COD of about 10 and 15 g/l, respectively. They contained about 650 and 800 mg/l of total phenols, 60 and 100 mg/l of SO4=, 13 and 20 mg/l of nitrate and 14 and 21 mg/l of chloride ions, respectively.
2.3 Bioreactor, its inoculation, running conditions and sampling
We used a hermetically closed glass column bioreactor (diameter: 80 mm; height: 450 mm) equipped with a recycle line and thermostated at 35 °C (through a water circulation system) (Fig. 1). The AOMW inlet was at the bottom, whereas the treated wastewaters plus the biogas produced in the reactor were jointly collected in a closed reservoir hydraulically connected to a 4 l “Mariotte” bottle through an outlet line on the top of the reactor. A redox and a pH probe (97–78 SC model and 81–04 model, respectively, ATI Orion, Boston, MA, USA) were placed at the top of the bioreactor. After its sterilization, the reactor was filled with 1.19 kg (dry weight) of GAC pre-sterilized in autoclave (110 °C, 30 min). The internal volume of the empty reactor system was 2.4 l, whereas its volume after GAC addition was 1.032 l. The packed reactor was purged with 0.22 μm filter-sterilized O2-free N2 for 1 h and then filled with a deoxygenated suspension of the microbial inoculum [20.1 mg (on dry-weight basis) per liter, prepared by resuspending the inoculum at 10% (v/v) in AOMW2 in a closed bottle purged with filter-sterilized O2-free N2]. The reactor medium was then recycled (upflow) at 23.1 ml/min for two weeks. To sustain biofilm formation, the reactor medium was then completely replaced with fresh deoxygenated AOMW2 that was recycled at 23.1 ml/min for two more weeks. Then, the reactor was forced to operate in continuous mode, by feeding it with either AOMW1 or AOMW2 at defined and increasing dilution rates (D, expressed as the ratio between AOMW influent flow rate and the reactor reaction volume). The recycle rate was increased proportionally with the dilution rate to achieve a reactor recycle ratio (defined as the ratio of the returned flow rate to the influent flow rate) of 77 identical for all experiments. Steady state conditions were attained when COD and phenolic compound concentrations (measured daily) in the effluent remained constant for at least a week. Steady state was typically achieved within two 2 weeks after a new dilution rate was set. Samples (6 ml) of the medium were taken daily through a sampling port placed along the recycle line (Fig. 1); they were passed through 0.22 μm cellulose-nitrate filters (Millipore, MO, USA) and then analyzed for COD, total phenolic compounds, volatile fatty acids (VFAs) and SO4= as detailed below. An aliquot of each sample was also analyzed by HPLC for OMW aromatic compounds and potential biodegradation aromatic metabolites. Biogas was quantified using the “Mariotte” bottle system (Fig. 1), while CH4 content was determined by gas chromatography of biogas samples collected at the reactor headspace. The amount of biomass immobilized in the reactor was determined at the end of the study. The reactor was opened, and triplicate samples (3 g) of GAC carrier were collected at 5, 18 and 36 cm (height, from the bottom) of the reactor packed-bed and subjected to analysis of their protein content using a modified Lowry method . Other samples (of about 5 g) of biofilm-covered GAC were collected from the same regions of the reactor, washed several times with a sterile physiological solution and prepared for scanning electron microscopy (SEM) . A third set of GAC samples (of about 20 g) collected from the same places in the reactor were washed and subjected to DNA extraction as described below.
Scheme of the anaerobic GAC packed-bed reactor developed in the study.
2.4 Extraction of total genomic DNA, purification, and PCR amplification
An extraction protocol was optimized for the recovery of total genomic DNA from samples of biofilm, reactor mobile phase, and from inoculum, reactor influent and effluent. Several methods were examined for the isolation of nucleic acids in order to obtain high yield of non-degraded high molecular mass DNA. Microorganisms were initially released from the biofilm samples by bead beating, by grinding in a mortar and pestle, or by sonication for 30 s, prior to microbial cell lysis using a chemical approach described by Zhou et al. . We found that bead beating for 2 h at 30 °C in TE buffer (10 mM Tris/HCl, pH 8.0, 1 mM EDTA) on rotary shaker (200 rpm), before passing through a Genomic DNA purification kit (Fermentas UAB, Lithuania) gave the highest yield of non-degraded high molecular mass DNA. Good yields of DNA were obtained also when using the DNeasy tissue kit (Qiagen, Italy). The amount and quality of nucleic acids were checked with electrophoresis.
Archaeal 16S rRNA genes were amplified with a primer set 3F-FAM and 1389R , and bacterial 16S rRNA genes with forward primer w017-FAM and reverse primer w002 . PCR was performed in a total volume of 50 μl containing 10–50 ng of community DNA template, 1 μM of each primer, 0.2 mM of (each) deoxynucleotide triphosphate, 5 μl of 10 ×Taq buffer, 10 μl of 5 ×Taq master and 1 U of Taq DNA polymerase (Eppendorf, AG, Germany). Amplification was performed in a thermal cycler (GeneAmp 9700; Applied Biosystems, Italia) with an initial denaturation step (95 °C for 5 min) followed by 30 cycles of denaturation (94 °C for 30 s), annealing (60 °C for 30 s with Bacteria primer set, 55 °C for 30 s with Archaea primer set), and extension (68 °C for 1 min and 30 s), and a single final extension step (68 °C for 7 min). Amplified DNA was verified by electrophoresis in 1% agarose gel in 1 × TAE buffer, and the amplifications product was purified with Wizard SV Gel and PCR Clean-Up System (Promega, Italia) according to the manufacturer's instruction to remove unincorporated nucleotides and labelled primers.
2.5 T-RFLP analysis
Fluorescently labeled PCR products (100 ng) were digested with 10 U of restriction enzyme (Invitrogen, Italia) at 37 °C for at least 4 h. T-RFLP profiles were generated using the restriction enzyme Rsa I. Additional profiles were generated using the restriction enzyme Hha I in order to confirm results obtained with Rsa I, and to assist in the effort to assign tentative phylogenetic affiliations to T-RFs. Aliquots (2 μl) were mixed with 19.5 μl of deionized formamide and 0.5 μl of ROX-labeled GS500 internal size standard (Applied Biosystems). Each sample was denaturated for 5 min at 95 °C and immediately chilled on ice before capillary electrophoresis on ABI Prism 310 Genetic Analyzer (Applied Biosystems) operating in a GeneScan mode with filter set D. Genescan 3.1 software was used to quantify the electropherogram output by setting the peak height threshold of 50 fluorescent units. The relative abundances of T-RFs in a given sample were calculated based on the peak height of the individual T-RF in relation to the total peak height of all T-RFs detected. Samples were run one more time if the cumulative peak height was below 9500 fluorescent units. Replicate T-RF profiles gave reproducible fingerprints.
2.6 DNA sequencing and phylogenetic analysis
Partial clone libraries of 16S rRNA genes were generated from community samples. Unlabeled PCR products, purified as described above, were cloned using the pGEM-T easy vector system (Promega) and Escherichia coli JM109 according to the manufacturer's instructions. From each library randomly selected clones were screened for positive inserts and by T-RF analysis after digestion with the endonucleases Rsa I and Hha I. Thirty clones from the bacterial libraries, representing 10 different T-RFs, and ten clones from archaeal libraries were subjected to cycle sequencing using the M13 primers and the BigDye terminator cycle sequencing ready reaction kit (Applied Biosystems). The DNA sequences were bi-directionally resolved on an ABI Prism 310 in a sequencing mode. Nucleotide sequences of about 500 bp were then assembled, checked for potential chimeric sequences using the CHIMERA-CHECK software, and compared with the sequences in the Ribosomal Database Project (RDP) database to identify the closest relatives. The phylogenetic analysis was carried out according to the maximum likelihood method and neighbor-joining topology using the appropriate tools of the RDP program package. Bootstrapping using 1000 replicates was performed to test reliability of the branches of the trees.
2.7 Nucleotide sequence accession numbers
The 16S rRNA gene sequences obtained in this study are available from the EMBL nucleotide sequence database under Accession No. genbank:AJ608921– genbank:AJ608930.
2.8 Analytical methods
HPLC analysis of AOMW aromatic compounds and of their biodegradation aromatic metabolites was performed with a Beckman Coulter chromatograph equipped with an UV–Vis diode array detector and an Ultrasphere ODS column as described previously . The concentration of total phenolic compounds of AOMWs was determined spectrophotometrically according to the Folin–Ciocalteu method  using 4-hydroxybenzoic acid as the standard. COD was determined following the APHA, AWWA, WPCF procedure  by titrating the residual oxidative agent (K2Cr2O7) with a 0.25 N solution of (NH4)2Fe(SO4)2× 6H2O. Volatile fatty acids were determined by gas chromatography according to Andreozzi et al. . Biogas was analyzed for CH4, CO2, N2 and O2 by gas chromatography and SO4= with an IC system as described by Fava et al. . COD (g/l) corresponding to CH4 (l) produced in the reactor was calculated considering that generally the biodegradation of 2.76 g of COD produces (at 35 °C) about 1 l of CH4. COD (g/l) consumed in the microbial reduction of sulphate occurring in AOMWs pumped into the reactor was calculated considering that 63.98 g of COD are required to reduce 1 mol of SO4= into S =.
3 Results and discussion
The possibility of improving the biodegradation of toxic phenolic compounds of two OMWs by performing their anaerobic digestion in a GAC packed-bed loop reactor has been recently demonstrated [15, Bertin et al., unpublished]. In this work, an identically configured biofilm reactor was developed and used to investigate the reproducibility, stability and the main microbial features of this innovative OMW anaerobic digestion technology.
3.1 Performance of the GAC-biofilm digester
The performance of the GAC-biofilm reactor developed in this study was preliminary investigated in seven sequential 3-week experiments (experiment nos. 1–7, Table 1) run at different and increased OMW organic loads (calculated by multiplying COD or phenolic compound content of the AOMW by the dilution rate at which the reactor operated). In general, the pollutant removal (expressed as COD depletion yields, which were calculated by dividing the amount of pollutant removed in the reactor under steady state conditions by the amount of pollutant occurring in the reactor influent) increased with the organic load, while phenol removal slightly decreased (Table 1). Notably, CH4 production (expressed as l of CH4 produced per g of COD removed) increased sharply from experiment no. 2 to no. 3 achieving values close to 0.2, and only slightly decreased during the successive experiments at higher organic loads (Table 1). The sharp increase in CH4 production observed during experiment no. 3 might be due to the completion of biofilm maturation, which usually takes place in the first two months of reactor operation, and which is characterized by high COD consumption . Once this process is ended, the supplied COD becomes completely available for CH4 production. No HPLC-detectable aromatic metabolites accumulated in the reactor throughout the seven experiments. On the contrary, a large array of VFAs occurred in the effluents, accounting for 30–60% of the effluent COD. Generally, acetate was more abundant than propionic acid that in turn was more abundant than all the other detected VFAs (i.e., iso-butyric acid, butyric acid and valeric acid). SO4= occurred in AOMW2 and AOMW1 at 101.1 ± 25.90 and 62.1 ± 4.91 mg/l, respectively; in general, only less than 5% of such SO4= amounts were detected in the effluents of the reactor under steady state conditions. Considering that 63.98 g of COD are required to microbiologically reduce 1 mol of SO4= to S =, it can be estimated that about 60 and 40 mg/l of COD of AOMW2 and AOMW1, respectively, (i.e., about 1% of the depleted COD), were sequestered by SO4= -reducing bacteria to methanogenesis. In all experiments, significant differences in pH and redox potential were observed between influents (7.8 ± 0.2 and −262 ± 12 mV, respectively) and effluents (5.2 ± 0.2 and −280 ± 23 mV, respectively) when steady state conditions were attained.
Methane production (l CH4 produced/g COD depleted)
10140 ± 326
905 ± 81.5
4.21 ± 0.14
0.38 ± 0.03
10100 ± 682
625 ± 140
6.99 ± 0.47
0.43 ± 0.10
15170 ± 186
574 ± 18.1
15.7 ± 0.19
0.60 ± 0.02
16030 ± 2120
777 ± 280
22.2 ± 2.93
1.08 ± 0.39
14470 ± 290
786 ± 150
30.1 ± 0.61
1.63 ± 0.32
14920 ± 1090
632 ± 110
41.3 ± 3.01
1.75 ± 0.29
16050 ± 1820
651 ± 190
55.6 ± 6.29
2.26 ± 0.66
15280 ± 1450
815 ± 130
33.0 ± 3.13
1.76 ± 0.29
↵a Data corresponding to the average (±SD) of single measurements carried out on at least five samples collected separately from the reactor influent during the third week of treatment (experiments nos. 1–7) or from the third to the eighth week of treatment (experiment no. 8), when steady state conditions were attained.
↵b Data obtained by dividing the difference between pollutant concentrations in the influent and the effluent of the reactor by the concentration values at which the pollutants occurred in the reactor influent. COD and phenol concentration in the effluents were determined through single measurements performed on at least five separate samples collected during the third week (experiment nos. 1–7) and from the third to the eighth week of treatment (experiment no. 8).
Taken together, these observations indicate that the GAC-biofilm reactor was characterized by a good versatility and tolerance towards high and variable OMW organic loads. Furthermore, the finding that OMW decontamination yields were very similar to those achieved with the previously developed GAC-biofilm digester at comparable organic loads (experiment nos. 1–3, Table 1; ) indicates that the GAC-biofilm technology is also characterized by relevant reproducibility.
To investigate the stability of the GAC-system, it was operated at a relatively high organic load for a 2-month period (experiment no. 8; Table 1). Fig. 2 shows the evolution of COD and phenolic compound concentration in the influent and effluent of the reactor throughout the whole experiment. In Fig. 2A, COD values consisting of the sum of the non-metabolized COD, COD corresponding to the detected CH4, and COD theoretically consumed in the reduction of depleted AOMW SO4= (that was removed by more than 95% also in this experiment) are also reported. A good correspondence between the COD introduced into the reactor and that leaving the process as COD, CH4, or “reduced SO4=” was observed only after the 20th day of the experiment. During the second month of the experiment, pollutant depletion yields and CH4 productivities were similar to those obtained in the 3-week experiment no. 5 (carried out at comparable AOMW2 COD and phenolic compound organic loads) (Table 1). As observed in the previous seven experiments, several VFAs were detected in the effluent of the reactor under steady state conditions: acetate, detected at 965.3 mg COD/l, was more abundant than propionic acid (787.0 mg COD/l), butyric acid (377.3 mg COD/l), iso-butyric acid (113.4 mg COD/l), and valeric acid (71.8 mg COD/l). No aromatic metabolites accumulated in the reactor throughout experiment no. 8, during which changes in pH and redox potential comparable to those reported for the previous experiments were recorded. These findings indicate that GAC-bioreactor is also characterized by a remarkable stability.
Profiles of COD (A) and phenolic compound concentration (B) in the influent (●) and in the effluent (△) of the GAC packed-bed reactor continuously fed at D= 2.07 (days−1) with AOMW2 (experiment no. 8, Table 1) throughout the 56 days of the treatment. COD values corresponding to the sum of the non-metabolized COD, COD converted into CH4 and that depleted through the AOMW SO4= -microbial reduction are also provided in the graph A (- - -). Data were obtained through single measurements performed on samples of the reactor influent and effluent collected at the times indicated in the graphs.
The volumetric productivity in terms of pollutant removal and CH4 production (expressed as removed pollutant/produced CH4 per day per reaction volume) exhibited by the reactor during experiment no. 8 is shown in Table 2, including the main features and performances of some of the other bench-scale OMW anaerobic digesters described in the literature [4,12–13,26–28]. Interestingly, the volumetric productivity exhibited by the GAC-bioreactor was significantly higher (by about 100% and 300% in terms of removal of COD and phenolic compounds, respectively, and by about 70% in terms of CH4 production) than those obtained with other up-flow packed-bed biofilm OMW digesters already described in the literature [4,26–28] and that obtained with the improved contact digester with the same microbial inoculum developed by Beccari et al. [12,13] (Table 2).
Comparative evaluation of the performances of the bench-scale continuous anaerobic OMW digesters described so far in the literature (excluding UASB systems)
Reaction volume (l)
COD load (g/l day)
Phenol load (g/l day)
COD removal yield
Phenol removal yield
Methane production [l CH4 produced/g COD depleted]
Volumetric productivity in COD removal (g/l day)
Volumetric productivity in phenols removal (g/l day)
Volumetric productivity in methane generation (l CH4/l day)
Wood chips packed-bed reactor
Folded polyethylene net packed reactor
Dispersed growth contact reactor
GAC packed-bed reactor
Data presented were calculated considering the treatment conditions under which each of the quoted bioreactor systems displayed the best performances. n.d.: not determined.
*MPUF: Macro-reticulated polyurethane foam.
3.2 Microbial biomass in the reactor
The properties of the GAC reactor were further investigated by determining total biomass and the structure of microbial communities within different reactor compartments at the end of the study, which is after 9 months of operation.
The amount of total immobilized biomass occurring at 5, 18 and 36 cm height of the reactor packed-bed was 38.3 ± 0.82, 37.5 ± 2.12 and 36.3 ± 1.22 (mg of dried biomass/g of dried support), respectively. Considering the average of these biomass values (37.34 mg/g) and the total amount of dried support used for packing the reactor (1190 g), the total immobilized biomass available in the GAC reactor was 44.5 g (dry weight basis). SEM observations indicate that biofilm developed on GAC cylinders was generally composed of rod-shaped bacterial cells, randomly distributed on the GAC surface (data not shown).
3.3 Structure and spatial distribution of microbial community in the reactor
The structure of the microbial community of the biofilm and of those occurring at different regions of the reactor was investigated with T-RFLP analysis. This technique generally provides a rapid and reproducible way to determine spatial shifts in the microbial community of complex ecosystems. We used this analysis to identify dominant 16S rRNA genes within a community and, by means of specific primers, to target either bacterial or archaeal DNA . However, similarly as in other techniques for studying the structure and dynamics of microbial communities, the resulting diversity might not be exactly representative of the real community composition: DNA extraction may introduce biases [20,24,30], amplification may select for some templates and bias the relative frequencies of genes in PCR products [31–33]. Osborn et al.  made similar observations during T-RFLP analysis of DNA samples from either PCB-polluted or pristine soil. These authors demonstrated that T-RFLP analysis is a powerful tool in microbial ecology and, once standardized, is a highly reproducible and robust technique for the rapid analysis of microbial community structure. We therefore assessed the reliability and robustness of our analysis in two ways; by evaluating the reproducibility of replicate T-RF profiles (using the same DNA templates and replicate DNA extractions performed with different methods) and by examining the effect of template concentration and of the number of PCR cycles on each profile. Replicate profiles from the same DNA sample were almost identical, suggesting that T-RF profiles were reliable fingerprints of the microbial communities present in the reactor. No additional T-RFs were detected in duplicate analyses upon dilution of the initial amount of template DNA. Furthermore, similar results were obtained by changing the DNA extraction procedure or the number of PCR cycles (in the range 30 to 35, data not shown). These results indicated that amplification biases cannot be completely excluded, however, in this study, they were limited.
Initial characterization of the microbial communities in the GAC reactor relied on T-RFLP fingerprints of 16S rRNA genes from biofilm samples collected at 5, 18 and 36 cm height of the reactor packed-bed and from samples of the reactor mobile-phase, influent, effluent and inoculum. Fluorescent amplifications were obtained from all samples when universal bacterial 16S-rDNA primers were used. T-RFLP analysis of duplicate samples confirmed that the obtained profiles were reproducible. The analyses of T-RF patterns produced by Rsa I or Hha I digestion were combined to achieve more accurate characterization of microbial communities. Rsa I digestion of amplicons generated with universal primers for Bacteria generated 17 peaks (Table 3), while Hha I digestions resulted in 10 different peaks (data not shown). Differences in profiles, as well as changes in absolute numbers of discernible peaks, could be seen among the various samples taken through the reactor. In profiles obtained with Rsa I digestion of the mobile phase samples, the major T-RF (peak height) was present at 98 bp (Table 3). In the T-RFLP fingerprints of the biofilm communities, a major peak was present at 386 bp in samples collected at 5 cm height of the reactor packed-bed, while a 280 bp fragment was predominant in samples from the middle (biofilm 18 cm) and the upper part (biofilm 36 cm) of the reactor (Table 3). Differences in major peaks were also observed in the corresponding T-RFs obtained with Hha I digestion (data not shown). As shown in Table 3, more taxa were detected in the mobile phase than in the biofilm samples; only a single T-RF, which was present along the reactor packed-bed, was found also in the influent, whereas almost all T-RFs were detected in the reactor effluent. PCR performed with Archaea°C for 1 min and 30 s), and a single final extension step (68 °C for 7 min). Amplified DNA was verified by electrophoresis in 1% agarose gel in 1 × TAE buffer, and the amplifications product was purified with Wizard SV Gel and PCR Clean-Up System (Promega, Italia) according to the manufacturer's instruction to remove unincorporated nucleotides and labelled primers.
Schematic representation of T-RFs obtained after Rsa I digestion of 16S rRNA genes amplified from DNA of samples of different districts of the GAC reactor with primer specific for Bacteria
T-RF length (bp)
The dominant (height) T-RF peak in each profile is indicated in boldface. Individual clones having a corresponding peak in the T-RF profiles are indicated below. The numbers indicate the relative abundance of individual T-RF. These values were calculated based on the peak height of individual T-RF in relation to the total peak height of all T-RFs detected in the respective community fingerprint pattern. The peak heights were automatically quantified by GeneScan software (PE Applied Biosystems), performing the analysis with a peak height threshold of 50 fluorescent units.
Schematic representation of T-RFs obtained after Rsa I digestion of 16S rRNA genes amplified from DNA of samples of different districts of the GAC reactor with primer specific for Archaea
T-RF length (bp)
The dominant (height) T-RF peak in each profile is indicated in boldface. The numbers indicate the relative abundance of individual T-RF. These values were calculated based on the peak height of individual T-RF in relation to the total peak height of all T-RFs detected in the respective community fingerprint pattern. The peak heights were automatically quantified by GeneScan software (PE Applied Biosystems), performing the analysis with a peak height threshold of 50 fluorescent units.
Several eubacteria, along with few highly abundant Archaea taxa, thus colonized different regions of the reactor (Tables 3 and 4). Bacterial and archaeal T-RFs profiles changed markedly along the reactor packed-bed (Tables 3 and 4) and this, according to previous findings , might be ascribed to the high heterogeneity that typically characterizes the composition of biofilms generated on porous carriers in packed-bed column reactors. Different distribution of bacterial and archaeal taxa among fixed- and mobile-phase may be ascribed to some mass transfer limitations that may have hindered the availability of substrates to biofilm composing cells, thus adversely affecting the growth and/or the persistence of some members of the biofilm community in the stationary phase [24,34]. Marked differences in the Bacteria fingerprints were present among the inoculum and various samples taken from the reactor (Table 3). Furthermore, none of the Archaeal T-RFs detected in the reactor apparently derived from the inoculum, which did not harbor any detectable taxa belonging to this domain (Table 4). These findings suggest that several members of the starter bacterial community were lost during the study and that many others, probably minor and undetectable members of the inoculum and/or of the employed AOMWs, were enriched in the reactor throughout the 9 months of operation. Similar evidence was reported by Sakano et al. , who analyzed the distribution of total, ammonia-oxidizing and denitrifying bacteria in packed-bed biofilm reactors developed for potable water recovery. Also other authors [36,37] suggest that inocula play a minor role on the development of sub-dominant species and the establishment of the final microbial community in anaerobic digesters treating wastewaters rich in simple and complex organic matter compared to operational conditions (i.e., temperature, pH or reactor configuration).
3.4 Analysis of clone libraries
To investigate bacterial diversity in detail and to identify the prominent bands in the T-RF patterns, partial clonal libraries of 16S rRNA genes were constructed from the inoculum, the biofilm, and the influent and effluent samples. T-RF screening of the clone libraries indicated that sequences representing abundant T-RFs of the community patterns were recovered in clone libraries (Table 3), although some components of the T-RF profiles were not recovered in our screening of the clone libraries. Conversely, some clones, which were present at low frequencies in the clone libraries, did not have a corresponding peak in the T-RF profiles (e.g. clone 31). Sequencing and BLAST search of 30 individual bacterial clones revealed ten different sequences grouped in five taxonomic groups (Fig. 3). Comparative analysis of these sequences with the RDP database showed significant similarities with 16S rRNA gene sequences of clones isolated from anaerobic digesters (Accession Nos. genbank:AF129860, genbank:U81680, genbank:U81676, genbank:U81706 and genbank:U81730 ), from an anaerobic consortium transforming trichlorobenzene (acc. number AJ009471) and from rumen bacterial communities (Accession No. genbank:AF001716); Sab values (similarity coefficient for query and matching sequences) were between 0.89 and 0.97. Analysis of bacterial 16S rRNA gene sequences from clones representing major peaks (peak height) from Rsa I-generated T-RF patterns indicated the presence of the members of the following taxa: Synergistes (clone B12, T-RF of 98 bp), Flexibacter-Cytophaga-Bacteroides group (clones B24, B25, T-RF of 280 bp; clone B27, T-RF of 434 bp) and γ-Proteobacteria (clone B1, T-RF of 386 bp). In addition, clones B2 (Syntrophus), B32 (Clostridium), B23 (Bacteroides) and B3 (Synergistes), could be matched with small T-RFs with the size of 106, 274, 282 and 414 bp, respectively, while clone B31 (Clostridium) did not generate a RsaI T-RF. Clone distribution was 40% Bacteroides group, 25% δ-Proteobacteria, 10% γ-Proteobacteria, 15% Anaerobaculum thermoterrenum group and 10%low G+C gram-positive group. A similar bacterial community composition was reported to occur in afluidized-bed reactor fed with wine distillation waste ; notably, in the present study the occurrence of SO4=-reducing bacteria was also documented,probably a result of the occurrence of SOSO4= in the AOMWs treated in the reactor. The dominant sequence of archaeal community (70% of the clones) was 100% similar to the 16S rRNA gene of Methanobacterium formicicum (T-RF of 80 bp), which is a hydrogenophylic methanogen. Similar structures of microbial communities were reported for different anaerobic reactors by Griffin et al.  and Leclerc et al. .
Phylogenetic tree of 16S rRNA genes recovered from clone libraries. The cloned sequences are indicated in boldface and the GeneBank accession numbers of sequences are in bracket. The distance matrix and phylogenetic tree were calculated by maximum likelihood method and neighbor-joining algorithm, respectively. The scale bar is in fixed nucleotide substitutions per sequence position. The numbers above the internal segments are the percentages of bootstrap replicates, which supported the maximum likelihood tree.
The presence of Archaea populations within the GAC reactor was limited with respect to that of Bacteria (Table 4). The same was found for some acidogenic reactors [37,38]. In addition, the detected Archaea were represented by a sole dominant species, i.e., Methanobacterium formicicum. This archaeal methanogenic strain was also dominant and persistent in other anaerobic digesters [20,27,39,40]. Methanosaeta and Methanosarcina species, which are acetoclastic methanogens, often found at high concentration in anaerobic digesters [20,30,36], occurred poorly in the GAC reactor, which is consistent with finding high concentrations of acetate in the reactor effluents. Limited occurrence of acetoclastic methanogens in the GAC reactor might be ascribed to the fact that its microbial community was investigated after a long phase of steady state operations, i.e. when the relative occurrence of these species, which typically predominate during the reactor start up stages [20,36], became low. In addition, the acidic pH environments, typically occurring in the reactor, might have been detrimental for these Archaea species that are often referred to as highly pH-sensitive methanogens [36,37,41]. The detected Archaea populations were preferentially distributed within the medium and upper regions of the GAC-packed bed (Table 4). This might be due to the fact that the continuous feeding of fresh AOMW operated through the bottom region of the reactor may have caused a marked local acidification due to the activity of acidogenic bacteria (which strongly populated that region) (Table 3), along with a high local availability of toxic phenolic compounds.
In conclusion, the anaerobic GAC biofilm digester preliminary described in our previous study (, Bertin et al., unpublished data) and better characterized in this one, enables an effective, reproducible and stable OMW-digesting process. It is tolerant to high OMW organic loads and capable of biodegradation and methanogenic performances higher than alternative bench-scale biofilters and dispersed growth digesters described so far [4,7,11–13,26–28]. The GAC biofilm digester developed herein thus presents a promising new technology for industrial disposal and valorization of OMWs. However, data on microbial populations occurring within the GAC bioreactor suggest that its methanogenic potential might be further improved by establishing operational conditions enabling extensive colonization of the reactor by methanogenic bacteria. This is the first report in which the performances of an innovative OMW anaerobic digester are assessed by evaluating the main chemical end physical parameters of the technology in combination with the structure of its microbial community. In addition, the results of this study highlight the importance of using microbial community structure analysis in combination with main chemical and physical parameters in the assessment of new biotechnological processes specifically designed for the disposal and valorization of agro-industrial wastewaters that are difficult to manage.
The Authors thank Prof. M. Majone and Prof. M. Beccari (Department of Chemistry, University of Rome “La Sapienza”, Italy) for their suggestions and help, R. Agnone for her participation in the research described in this paper, and the Frantoio Sant'Agata d'Oneglia (Imperia, Italy), for providing the OMWs employed in the study. The project was funded by the Italian MIUR (COFIN/PRIN 2000) and the Inter-University University Consortium “The Chemistry for the Environment”, Venezia, Italy.
(1998) Comparative effect of different aerobic pretreatments on the kinetics and macroenergetic parameters of anaerobic digestion of olive mill wastewater in continuous mode. Bioprocess Eng. 18, 127–134.
(2000) Enhancement of anaerobic treatability of olive oil mill effluents by addition of Ca(OH)2 and bentonite without intermediate solid/liquid separation. In: Proceedings of the 1st World Congress of the “International Water Association”, Paris, July 3–7.
(2001) An aerobic fixed-phase biofilm reactor system for the degradation of the low-molecular weight aromatic compounds occurring in the effluents of anaerobic digestors treating olive mill wastewaters. J. Biotechnol. 87, 161–177.
(2003) Use of packed-bed biofilm reactors in the anaerobic digestion of olive mill wastewater. In: Proceedings of the Second Bioremediation Conference, pp. 523–526, Chania, Crete, Greece, June 30–July 4.
(2003) Microbial reductive dechlorination of pre-existing PCBs and spiked 2,3,4,5,6-pentachlorobiphenyl in anaerobic slurries of a contaminated sediment of Venice Lagoon (Italy). FEMS Microbiol. Ecol. 44, 309–318.
(2002) Spatial distribution of total. Ammonia-oxidizing, and denitrifying bacteria in biological wastewater treatment reactors for bioregenerative life support. Appl. Environ. Microbiol. 68, 2285–2293.
(2001) Changes in product formation and bacterial community by dilution rate on carbohydrate fermentation by methanogenic microflora in continuous flow stirred tank reactor. Appl. Microbiol. Biotechnol. 57, 65–73.