Samples from three different positions of the biofiltration system were taken to evaluate the spatial distribution of the microbial population. Figure 3 shows a picture of the biofiltration system and the positions where the samples were collected. The samples at the top of the reactor "a" correspond to the inlet of the biogas stream mixed with different air fluxes, samples "b" and "c" correspond to the middle part of the reactor and sample "d" was located at the bottom of the biofilter (outlet of the biogas stream).

Additionally, the changes in the bacterial community were also determined by taking samples during long-term operation of the biofiltration system (Samples from 1a to 9a). The analysis was performed by a DDGE system using 16S rRNA as a bacteria-specific target for PCR amplification. Figure 4 shows an example of denaturing gradient gel DDGE (15% to 60%) from samples of different times of cultivation compared with the initial bacterial community. In summary, around 13 bands for the total bacterial community were systematically detected over long-term operation of the biofiltration system. (Samples 1a to 3a correspond to days 5, 10 and 20th of operation. Sample 4a was obtained at day 45th of operation, when the inlet load was increased to 3000 ppmv. Samples from 5a to 8a were obtained in days 90, 110 130 and 150 of operation, respectively.) In view of the total bacterial community, the bands remained constant until variations in intensity appeared. In lane 4a, lower intensity bands revealed a weakening pattern, which suggest a decrease in certain types of bacteria, when the H2S concentration increased from 1500 ppm to 3000 ppm. Both the decrease in removal efficiency and the decrease in the microbial population could be explained by the toxicity of the extremely high H2S concentration. This factor was assumed to be responsible for the disappearance of some of the microbial species. Increased intensity

in some bands in the gel (boxes A and B) demonstrated intensifications of specific band patterns. These data suggested the eventual dominance of H2S-oxidizing bacteria (SOB) and bacteria able to consume VFA.

To compare bacterial community between different samples and to determine possible changes in composition, the presence or absence of a band in a DDGE gel was analyzed using a binary system. A 0 value was assigned when the band was absent (i. e., different band is considered a different microorganism) and 1 when the band in two or more samples was present (i. e., same microorganism) at similar positions in the gel. Jaccard’s index and the Sorensen-Dice index could then be calculated. Table 2 shows the matrix constructed using the DDGE gel containing different band patterns obtained at different times of operation (lanes 1a to 9a) and at different lengths along the biofilter (9a, 9b, 9c and 9d). Nineteen different bands (arbitrarily named A to S) were found in the samples analyzed by gradient DDGE.

Once the number of bands that were similar or different between the two samples was determined, the similarity of the different samples was determined by calculating the Jaccard and Sorensen-Dice indexes. Two different aspects were analyzed: the similarity of the samples during the time of cultivation (lanes 1a to 9a) and the similarity at a different position in the reactor (lane 9a compared to 9b, 9c and 9d).

Regarding time of cultivation, the bands A, B and E to J constantly appeared in the microbial community, suggesting little change in the microbial populations during the operation of the biofilter. The most similar microbial communities were found in lanes 6a and 7a, with a Jaccard index of 0.875 and a Sorensen-Dice index of 0.933, which corresponded to the steady state of the biofiltration system at an average H2S inlet concentration of 1500 ppmv and a removal efficiency of 95%. In contrast, the least similarity was found between lanes 4a and 5a, 6a and 7a (Jaccar’s indexes of 0.438, 0.47 and 0.56, respectively). In lane 4a, the microbial community sample was exposed to an increased H2S concentration of 3000 ppmv. These data differed from those found by Maestre et al., 2010. These authors reported a wide phylogenetic diversity and showed that the initial populations became more specific, being the SOB the dominant community.

The similarity between the microbial communities along the biofilter was also calculated. For this purpose, the bands in lane 9a were compared with the bands in lanes 9b, 9c and 9d. Significant differences in the microbial population were observed at different lengths along
the biofilter. The highest divergence was found between lanes 9d and 9b, with a Jaccard index of 0.333 and a Sorensen-Dice index of 0.5. These data could be partially explained by the H2S concentration gradient: a higher concentration at the inlet of the biofilter and a lower concentration at the outlet (sample 9d). The accumulation of metabolic products could also explain the divergence. The highest similarity was found in samples of lanes 9b and 9c, which corresponded to the middle of the reactor, where apparently the environmental conditions were more homogeneous (Jaccard index of 0.833 and Sorensen-Dice index of 0.909). These data are in agreement with the results obtained by Maestre et al., 2010 and Omri et al., 2011 about the divergence in microbial populations along the reactor.

Sequence analysis of DNA extracted from single bands representing specific species were then used as an approach for further community characterization. Sequence analyses of bands (Table 3) revealed the predominant bacteria in the biofiltration system. The structure of the bacterial community sequenced was associated with microbial activity in the system as a function of the pollutant eliminated in the biofiltration system.

Table 3. Sequence analysis and species identification of the major (7) DGGE bands for the biofilter samples.

Band sequencing results showed that the dominant members of SOB consisted of Bosea thiooxidans and Thiobacillus sp (Table 3). Das et al., 1996 reported that Bosea thiooxidans was a new gram-negative bacterium isolated from agricultural soil and capable of oxidizing reduced inorganic sulfur compounds. Data showed that this microorganism was strictly aerobic. Experiments conducted to evaluate thiosulfate oxidation showed that the growth yield varied with the concentration of this compound; the greatest growth was observed at a
concentration of 5 g/L. Under these conditions, conversion of thiosulfate to sulfate was stoichiometric, and the pH of the medium decreased from 8.0 to 6.6. The distance matrix phylogenetic tree based on the level of difference between Bosea thiooxidans and 19 reference strains of the alpha subclass of the Proteobacteria indicated that strain BI-42 belonged to a new lineage located between the methylotrophs, the genus Beijerinckia, and the Rhodopseudomonas palustris group. No close relationship was found between the strain and other sulfur-oxidizing bacteria, such as Thiobacillus acidophilus and Acidiphilium species (Das et al., 1996). Microorganisms utilize sulfur compounds for the biosynthesis of cellular material or transform these compounds as part of a respiratory energy-generating process. Most of the known sulfur-oxidizing bacteria belong to the genera Thiobacillus, Thiothrix, Beggiatoa, Thiomicrospira, Achromatium, Desulfovibrio, Desulfomonas, Desulfococcus, and Desulfuromonas. Furthermore, members of the genus Thiobacillus have been studied extensively to increase understanding of the coupling of oxidation of reduced inorganic sulfur compounds to energy biosynthesis and assimilation of carbon dioxide.

The presence of Arcobacter butzleri, belonging to the Phylum Proteobacteria (e — Proteobacteria), could be associated with VFA degradation (these compounds were introduced into the biofilter through the biogas stream). This microorganism is able to grow under both aerobic and anaerobic conditions over a wide temperature range (15-42 °C). However, optimal growth occurs under microaerobic conditions (3-10% O2). Arcobacter

butzleri was also recently found as a member of the microbial population in a microbial fuel cell (MFC) used to produce electricity from synthetic domestic wastewater that contained a mixture of VFAs as electron donors (Freguia et al., 2010). Similar results were obtained by Nien et al., 2011, where an Arcobacter butzleri strain, ED-1, was also determined to be part of the microbial community of a MFC fed with acetate. Although aerobic species were predominant because the metabolic activity determined (sulfate as the main product), the DGGE showed that profile some facultative anaerobes were as part of the microbial population, which could be related to the trophic properties of the community, and the different substrates in the biogas stream (H2S and VFAs).

Some of the species found in the present study agreed with those previously reported in the literature for biofiltration systems used in the removal of reduced sulfur compounds. For example, Ding et al., 2006 studied a packed compost biofilter for the treatment of a mixture of H2S and methanol using 16S rRNA sequencing analysis. The authors established that the microbial community was composed of strains of Thiobacillus, Sulfobacillus, and Alicyclobacillus hesperidensis. In a biofilter packed with compost, activated carbon and sludge used for the removal of H2S, Chung, 2007 determined a microbial population composed of Pseudomonas citronellolis, P. fluorescens, P. putida, S. capitis, Bacillus subtilis and Paracoccus denitrificans. In a recently published work (Omri et al., 2011), it was reported that most bacteria in the operation samples were of the genera Pseudomonas sp., Moraxellacea, Acinetobacter and Exiguobacterium, which belong to the phyla Pseudomonadaceae, gamma — Proteobacteria and Firmicutes.

A neighbor-joining tree (Fig 5.) of partial 16S rRNA sequences (approximately 750 bp) was constructed in MEGA4 (Tamura et al., 2007) by considering sequences obtained and comparing them with others in the data bank.

4. Conclusion

The feasibility of CH4 was demonstrated. High VS removal, the increased methane yield, and the natural pH control during the stable period of the ADS was obtained by codigestion of VFW and MR, due to an adequate ratio of nutrients and the availability of proteins for new cell synthesis. However, the increasing MR concentration in the ADS increased the H2S concentration in the gas stream. The elimination of H2S and VFAs by a biofiltration system was successfully determined, reaching high removal efficiencies of both compounds (95% and 99%, respectively). This approach could allow the potential use of the biogas maintaining the methane (CH4) content throughout the process. The microbial population characterization of the bioifltration system showed that dominant members of SOB were Bosea thiooxidans and Thiobacillus sp. Some facultative anaerobes were also determined in the system, which could be explained by the composition of the biogas stream and the conditions at different length of the reactor.

5. Acknowledgment