International Biodeterioration & Biodegradation 63 (2009) 747–751 Contents lists available at ScienceDirect International Biodeterioration & Biodegradation journal homepage: www.elsevier.com/locate/ibiod Bioremediation of gasoline-contaminated groundwater in a pilot-scale packed-bed anaerobic reactor Dalva A. Souza a, *, Fabio A. Chinalia b, **, Eugenio Foresti a, Marcelo Zaiat a a Departamento de Hidráulica e Saneamento, Escola de Engenharia de São Carlos (EESC), Universidade de São Paulo (USP), Av. Trabalhador São-Carlense, 400, 13550-590 São Carlos, SP, Brazil b School of Applied Sciences, Cranfield University, Building 40, Cranfield, Bedfordshire MK43 0AL, UK a r t i c l e i n f o a b s t r a c t Article history: Received 28 January 2009 Received in revised form 28 May 2009 Accepted 28 May 2009 Available online 28 June 2009 This work reports on the anaerobic treatment of gasoline-contaminated groundwater in a pilot-scale horizontal-flow anaerobic immobilized biomass reactor inoculated with a methanogenic consortium. BTEX removal rates varied from 59 to 80%, with a COD removal efficiency of 95% during the 70 days of in situ trial. BTEX removal was presumably carried out by microbial syntrophic interactions, and at the observed concentrations, the interactions among the aromatic compounds may have enhanced overall biodegradation rates by allowing microbial growth instead of co-inhibiting biodegradation. There is enough evidence to support the conclusion that the pilot-scale reactor responded similarly to the lab-scale experiments previously reported for this design. Ó 2009 Elsevier Ltd. All rights reserved. Keywords: Anaerobic biodegradation BTEX Groundwater Pilot-scale Methanogenic consortium 1. Introduction Petroleum hydrocarbons are widespread environmental contaminants and hotspots pollution, as a result of spills, leaks and improper disposal are common source of public concern and health and safety risk assessments (Miège et al., 2003; Martinez et al., 2004; Ratola et al., 2006). The treatment of BTEX-containing groundwater has been studied by many authors using a variety of bench-scale bioreactor designs, but very few studies are available reporting on in situ treatment performances and feasible applications. For instance, several attempts have been made to characterize and optimize the operating variables applied to horizontal-flow anaerobic immobilized biomass reactors (HAIB reactors) for treating BTEX-contaminated wastewater in the Laboratory of Biological Processes at EESC/ USP, Brazil (de Nardi et al., 2005; Cattony et al., 2005; Gusmão et al., 2007) and, this work, is a practical test for the feasible application of these developments. Problems associated with scale-up from laboratorial experiments are major challenges facing the implementation of novel biological bioremediation approaches (Nakhla, 2003). Therefore, in this sense, this experiment is also a contribution and an * Corresponding author. Tel.: þ55 16 34130553. ** Corresponding author. Tel.: þ44 1234 750111x2794. E-mail addresses: dalsoza@gmail.com (D.A. Souza), chinalia@cranfield.ac.uk (F.A. Chinalia). 0964-8305/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.ibiod.2009.05.007 attempt to apply lab-scale developments into a full-scale technology to treat BTEX-contaminated groundwater. Bioremediation technologies offer the possibility of destroying various contaminants using enhanced natural biological activity. Comparatively to other non-biological methods these technologies show some advantages regarding handling, management and, very often, they reveal relatively low costs. Therefore, the main practical concern when developing such a technology is in relation to their robustness for easy transporting and on site implementation (Mollea et al., 2005; Antizar-Ladislao et al., 2006; Wu et al., 2008). After a short uncontained gasoline spillage event, a portion of the subsoil and groundwater located in the vicinity of a gas station in the city of Araraquara, state of São Paulo, Brazil, showed significant concentrations of BTEX. This event required immediate attention and, since the establishment had already been fined by the state’s environmental agency, the company liaised with the university and allowed an in situ test to be carried out. The contaminated soil was excavated, removed and treated elsewhere. The contaminated groundwater, generated during the process of soil removal, was pumped into a reservoir (first tank of Fig. 1), and, as solids were settling, the remaining liquid was treated in a HAIB reactor. The settled solids were later re-incorporated into the removed soils which were then destined to other specific treatment. The environmental impact resulting from BTEX contamination by Brazilian fossil fuels is greatly aggravated by the high ethanol 748 D.A. Souza et al. / International Biodeterioration & Biodegradation 63 (2009) 747–751 Fig. 1. Schematic representation of the pilot-scale HAIB reactor at the Araraquara gas station, Brazil. content in the gasoline (22–25%). For instance, ethanol is reported to enhance hydrocarbon concentrations in the water-phase and decrease BTEX biodegradation in situ. Therefore, spills of ethanolrich gasoline can spread BTEX contamination into a larger area within a shorter period of time. Conventional aerobic treatments may potentially increase air pollution, once such compounds are easily volatilized. On the other hand, the presence of ethanol and also significant concentrations of BTEX in solution are considered to be an ideal condition for a biological treatment based on a process carried out by methanogenic consortium. This type of biological treatment not only avoids the problem of air contamination by BTEX volatilization but the alternative sources of carbons such as ethanol are commonly observed to enhance anaerobic process by stimulating growth of adapted microbial biomass. Therefore, the aim of this work was to install and monitor the performance of a HAIB reactor treating BTEX-containing groundwater, as assessed by COD and BTEX removal, and to characterize the anaerobic microbial community of bacteria associated to the hydrocarbondegrading methanogenic consortium. 2. Material and methods Contaminated groundwater by ethanol-rich gasoline was collected at Araraquara gas station (Sao Paulo state, Brazil) and it was pumped into a storage tank from which the separated liquid phase was treated within a HAIB reactor (53 l, 3 m length and 15 cm diameter, Fig. 1). Local contaminated soil was excavated, removed and treated elsewhere. The whole design for the treatment of contaminated groundwater is shown in Fig. 1. The apparatus was set-up behind the gas station entrance and next to the working plant where soil had been excavated. Before transportation for in situ placement, HAIB reactor was constructed and metabolic activated in laboratorial conditions (the establishment of a methanogenic consortium) according to specifications given by Zaiat et al. (2000). Cubic polyurethane foam particles (5 mm in size and 23 kg/m3 apparent density) were used as physical support matrix for biomass immobilization. Before placement within the horizontal bioreactor, microbial biomass was immobilized into the polyurethane matrices as described by Zaiat et al. (1994). Biomass collected from an Up-flow Anaerobic Sludge Blanket reactor (UASB reactor) treating poultry slaughterhouse wastewater was used as inoculum. Briefly, cubic polyurethane matrices were left in contact with blended anaerobic biofilm (blended UASB reactor’s biomass) for several hours before being partially drained and introduced into the horizontal bioreactor. The HAIB reactor was then fed with prescreened (2 mm) domestic sewage during 8 days at a hydraulic detention time of 10 h in order to foment and sustain an active methanogenic consortium prior to the transportation and placement in the petrol station for the provision of contaminated groundwater treatment. After 8 days in the laboratory and with a methane production rate of about 8.12 mol/d, the HAIB reactor was then transported for in situ bioremediation. The contaminated groundwater was fed at a flow rate of 1–2.5 l/h, resulting in hydraulic retention times of 8.5–19.3 h during a period of 70 days. The performance of the HAIB reactor was evaluated by monitoring the COD, pH, volatile acids content, alkalinity (Standard Methods for the Examination of Water and Wastewater, 1998) and BTEX concentrations (by gas chromatography, HP 6890 FID, according to Moraes et al., 2000) analyzing and comparing influent and effluent samples. Samples of the biofilm for molecular analysis were collected at the influent and effluent opening ports. Sampling for molecular analyses were carried out before the addition of contaminated groundwater (in the end of 8 days period for methanogenic activation) and at the end of the trial (70 days). The biofilm was processed to assess the bacterial diversity as described by Gusmão et al. (2007). Briefly, polyurethane matrices were crushed using phosphate buffer, pestle/mortar and the suspended detached biomass was collected in 15 ml centrifuge tubes. The suspension was centrifuged for 1 min at 13,000 rpm and, applying the same centrifugation protocol, the pellet was washed twice in phosphate buffer before DNA extraction. Total DNA was extracted as described by Gusmão et al. (2007). Briefly, the extraction was carried out using 1:1:1:0.5 v/v/v/g of phenol:chloroform:phosphate buffer:glass beads to 0.5 g of microbial biomass (pellet), respectively. The suspension was vortexed in 15 ml centrifuge tubes for 1 min and the water-phase (1 ml of phosphate buffer), which was separated after centrifugation, was washed twice in phenol and twice in chloroform using intermittent steps of centrifugation. Finally, the process resulted in 0.1 ml final solution containing total microbial DNA. Polymerase chain reaction (PCR) and diversity assessments by denaturing gradient gel electrophoresis (DGGE) were performed using bacterial primers and conditions described by Muyzer et al. (1993). It was applied a denaturing gradient of 40–60% (40% of acrylamide/bis; solution 50  TAE; 40 or 60% of formamide and urea) and electrophoresis was carried out at a constant temperature of 65  C, 75 V for 16 h. Gels were stained with ethidium bromide. DGGE-profiling documentation was accomplished in an Eagle Eye TM III (Stratagene) under excitation of 254 nm UV, using Eagle Sight software. DGGE-bands were extracted using disposable blades and PCR re-amplified and cloned into a plasmid pCR 2.1 TOPO-TA easy vector systems according to the manufacturer instructions (Invitrogen). Positive clones were sequenced in ABI 377 DNA Sequencer (Perkin Elmer) using M13 primers (forward and reverse, separately). The resultant nucleotide sequences were assembled, checked for potential chimerical sequences and compared with the electronic database (NCBI and RDP) for the identification of the closest matches. Sequences were aligned using MEGA 4 software and phylogenetic tree was constructed using neighbor-joining Kimura 2-parameter. A Raup–Crick probability-based similarity test was used to statistically compare DGGE-profiling obtained between treatments. The test was made using a matrix of presence/absence of bands and PAST software v 1.9 (Hammer et al., 2001). D.A. Souza et al. / International Biodeterioration & Biodegradation 63 (2009) 747–751 Table 1 Mean values of operating variables assessed during the reactor’s 70-day trial. Parameters Average values (standard deviation) Total COD (mg l1) Filtered COD (mg l1) Alkalinity (mg CaCO3 l1) Volatile fatty acids (mg CH3COOH l1) pH Ethanol concentration Removal efficiency % Influent Effluent 577 552 9.0 17.6 6.6 247 36 (20) 20 (19) 65.2 (20.7) 22.1 (7.3) 6.6 (0.2) Not detected (69) (64) (9.5) (4.5) (0.2) (38) 94 (3) 96 (3) 3. Results and discussion Table 1 lists the average and standard deviation for COD, alkalinity, volatile fatty acids pH and ethanol concentrations during the 70 days of groundwater treatment. During the trial, the temperature was not controlled and varied from 17 to 34  C, with a mean value of 26  C. COD removal was high, with low variations. Significant groundwater COD values were due mainly to the contribution of ethanol which is commonly added to the Brazilian gasoline at a ration of 22–25% final concentration. Fig. 2 shows the variation of benzene, toluene, xylenes (m-p-o-xylene) and ethylbenzene concentrations in the influent and effluent of the HAIB reactor treating contaminated groundwater. This figure shows that the BTEX concentrations in the influent have decreased during the treatment (70 days). This was primarily an effect of dilution caused by the continuous pumping of contaminated groundwater showing gradual decreases in BTEX concentrations. As the source of BTEX contamination had been removed from the site before the biological treatment set-up (soil and leaking tank); BTEX concentrations 749 in the contaminated groundwater naturally diminished within time and continuous extraction (pumping into the storage tank). Secondly, it is not possible to rule out some volatilization during the process of pumping and storage. The initial stages of BTEX removal shown in Fig. 2 are believed to be associated with adsorption phenomena into the polyurethane foam matrices and the organic biomass within the bioreactor. According to calculations proposed by Cattony et al. (2005), saturation of BTEX adsorption may have occurred in this case after the first 5 days of operation. BTEX removal rates observed during the remaining days were therefore considered to be the potential resultant of biological degradation. On average, the efficiency of BTEX removal varied from 59 to 80% and the compounds with comparatively increasing removal rates were xylenes, benzene, toluene and ethylbenzene, in this order. It was not observed an apparent inhibition effect or negative interaction between individual BTEX during the treatment. It has been reported that biodegradation of individual aromatics may follow Monod kinetics at initial benzene and toluene concentrations below 100 mg/l, but these compounds can have an inhibitory effect to the biodegradation of other aromatics when above this concentration (Hamed et al., 2003). In this work, however, BTEX biodegradation were not inhibited and the results suggest an opposite trend. It is possible that the BTEX-ethanol mix enhanced biodegradation rates possibly by allowing microbial growth and the accumulation of active and diversified methanogenic-BTEX degrading biomass within the reactor. The production of methane was not assessed during the treatment. Methane production was only estimated during the period of activation (8 days) and, at the end of this period, it showed a rate of about 8.12 mol/d. On the other hand, an indication of methanogenesis can be made by comparing the decline in alkalinity between influent and effluent. While acetotrophic methanogenic archaea Benzene Toluene 20 Influent 20 Influent effluent 16 effluent 12 mg/L mg/L 16 8 4 12 8 4 0 0 10 20 30 40 50 60 70 10 20 30 Days 40 50 60 Xylenes 20 Ethylbenzene Influent Influent 20 Effluent Effluent 16 16 12 12 mg/L mg/L 70 Days 8 4 8 4 0 0 10 20 30 40 Days 50 60 70 10 20 30 40 50 60 70 Days Fig. 2. Comparison of hydrocarbon degradation rates of benzene, toluene, xylenes and ethylbenzene in the influent and effluent of the HAIB reactor during the period of groundwater treatment (70 days). % of Alkalinity and COD removal 750 D.A. Souza et al. / International Biodeterioration & Biodegradation 63 (2009) 747–751 100 90 80 70 60 50 40 30 20 10 0 0 10 20 30 40 50 60 70 days Fig. 3. Variations in alkalinity (-) and COD removal (C) during the period of the trial. A generate alkalinity by producing methane and carbon dioxide (alkalinity), hydrogenotrophic archaea consume carbon dioxide and hydrogen for the production of methane. The overall balance within an anaerobic methanogenic reactor consistently shows a significant decline in alkalinity (Isık and Sponza, 2005). Fig. 3 shows that the variation of alkalinity in the effluent was consistent with COD removal and considerably lower than the values found in the influent during the period of the treatment (Table 1). Stoichiometrics calculations of the hydrogenotrophic reaction suggest a 1:1 ratio between bicarbonate consumption (HCO3) and methane production (CH4). Therefore, as a rough approximation, the reduction of alkalinity, in methanogenic conditions, may be associated with the consumption of carbon dioxide and production of methane at a neutron pH. However, it is only an indication of a portion of the total potential for methane production within the system. It should be pointed out that pH in these systems are a function not only of bicarbonate concentration but also of VFA production, but Table 1 shows that variation of pH values were not significant during the period of trial even with the decrease in alkalinity. B Fig. 4. This figure shows 16S rRNA phylogenetic tree (B) of the DNA bands extracted from the DGGE-profiling of the HAIB reactor biofilm (A). Lanes 1 and 2 of the DGGE-profiling correspond to the sampling points closest to the influent entrance and effluent discharge from the horizontal reactor before the addition of contaminated groundwater, respectively. Lanes 3 and 4 show the respective profile at the end of the trial. The totals of 12 bands were numbered as LPB100 to LPB111, from the top to bottom of DGGE-profiling, and they are thus discriminated in the phylogenetic tree. The neighbor-joining tree was constructed using MEGA 4 software with Kimura 2-parameter algorithm. Bacterial specimens of type category were downloaded from the Ribosomal Database Project (full name in the tree) and Thermovibrio rubber and Thermovibrio ammoniificans were used as out-group. D.A. Souza et al. / International Biodeterioration & Biodegradation 63 (2009) 747–751 Fig. 4 shows bacterial diversity estimation (DGGE-profiling) and bacterial species’ composition in the anaerobic consortium (phylogenetic tree). Comparing the samples collected before treatment (lanes 1 and 2) and at the end of the trial (lanes 3 and 4) DGGE-profiling (Fig. 4) showed a reduction in the number of bands at the end of the treatment. There are also differences in the number of bands between lanes 1 and 2. At this point the horizontal reactor was being feed pre-screened domestic sewage for a short period of 8 days. Fluctuation on the bacterial diversity as a resultant of such a short period of adaptation can be expected. Factors such as substrate diffusion into the horizontal system and bacterial colonization of the horizontal packed-bed may be contributing in the process. With the addition of BTEX-contaminated groundwater for 70 days, however, most of the remaining bacterial species present in the reactor displayed a similar profile at both ends (Fig. 4). Furthermore, the DGGE-profiling suggested a significant adaptation of the bacterial community to the BTEX-contaminated groundwater. This hypothesis was tested using a Monte-Carlo-based statistical approach (Raup–Crick index) used to calculate similarities between samples (Hammer et al., 2001). This probability-based similarity index, when used to compare any two samples, generates a value between 0 and 1. At 90% interval of confidence, significant similarity/ dissimilarity is given by values 0.90 or 0.10, respectively. Using this interval, the Raup–Crick test indicated that DGGE lanes 3 and 4 are significantly different from lane 2. The probability similarity index value between lane 1 and lanes 3 and 4 were not higher than 0.21. Therefore, this result is a strong indication that shifts in the DGGEprofiles is a consequence of bacterial community adaptation to BTEXcontaminated groundwater. On the other hand, random fluctuation on the colonization of bioreactor’s polyurethane matrix may have been responsible for the specific presence of DGGE band LPB105 in lanes 1, 3 and 4 and its noted absence in lane 2. Despite of the short DNA sequence analyzed in this work, phylogenetic analysis indicated that band LPB104 and LPB105 matched strains closely related to Spirochaetes and Bacteroidetes-like species commonly present in environments exhibiting anaerobic biodegradation of aromatics (NCBI accession numbers AJ009450 and AB234438, respectively). This, in fact, is also a characteristic shared by all the isolates of the Proteobacteria and Firmicutes groups identified in this work (Fig. 4). Biodegradation of aromatics in anaerobic conditions differs significantly according to the redox-potential. In a methanogenic consortium, for instance, mineralization of aromatics may not be achievable by a single species alone. Therefore, at very low potential redox (methanogenic environment); some bacterial species may biodegrade hydrocarbons using a syntrophic interaction with methanogenic archaea. Several proteobacteria and Bacteroideteslike species, such as the ones observed in this work, are believed to be capable of such interaction (Ficker et al., 1999). 4. Conclusions At present, only limited data are available on the in situ microbial degradation of mixtures of aromatic hydrocarbons, and future research is needed to investigate the effects of biodegradation of multiple substrate contaminants. Nevertheless, this report concludes that, in the present concentrations, a mixture of multiple aromatics was significantly mitigated at rates similar to those achieved in previously reported lab-scale experiments. The BTEX removal rates varied from 59 to 80%, with a COD removal efficiency of 95% during the 70 days of in situ trial. The HAIB reactor was effective not only in removing the BTEX contamination but also in sustaining an anaerobic consortium, which probably mineralized the hydrocarbons through syntrophic interactions within the 751 microbial community, as suggested by the characterization of the bacterial diversity. Acknowledgements This work was supported by FAPESP and CNPQ (Brazilian research-funding agencies). The authors are indebted to Nickol do Brasil and Ecomark-Consultoria for their partnership, without which this experiment could not have been conducted. References Antizar-Ladislao, B., Lopez-Real, J.M., Beck, A.J., 2006. Bioremediation of polycyclic aromatic hydrocarbons (PAH) in an aged coal–tar-contaminated soil using different in-vessel composting approaches. Journal of Hazardous Materials 137, 1583–1588. Cattony, E.B.M., Chinalia, F.A., Ribeiro, R., Zaiat, M., Foresti, E., Varesche, M.B.A., 2005. Ethanol and toluene removal in a horizontal-flow anaerobic immobilized biomass reactor in the presence of sulfate. Biotechnology and Bioengineering 91, 244–253. de Nardi, I.R., Ribeiro, R., Zaiat, M., Foresti, E., 2005. Anaerobic packed-bed reactor for bioremediation of gasoline-contaminated aquifers. Process Biochemistry 40, 587–592. Ficker, M., Krastel, K., Orlicky, S., Edwards, E., 1999. Molecular characterization of a toluene-degrading methanogenic consortium. Applied and Environmental Microbiology 65, 5576–5585. Gusmão, V.R., Chinalia, F.A., Sakamoto, I.K., Varesche, M.B.A., 2007. Performance of a reactor containing denitrifying immobilized biomass in removing ethanol and aromatic hydrocarbons (BTEX) in a short operating period. Journal of Hazardous Materials B139, 301–309. Hamed, T., Bayraktar, E., Mehmetoglu, T., Mehmetoglu, U., 2003. Substrate interactions during the biodegradation of benzene, toluene and phenol mixtures. Process Biochemistry 39, 27–35. Hammer, Ø., Harper, D.A.T., Ryan, P.D., 2001. PAST: paleontological statistics software package for education and data analysis. Palaeontologia Electronica 4, 4–9. Isık, M., Sponza, D.T., 2005. Effects of alkalinity and co-substrate on the performance of an upflow anaerobic sludge blanket (UASB) reactor through decolorization of Congo Red azo dye. Bioresource Technology 96, 633–643. Martinez, E., Gros, M., Lacorte, S., Barceló, D., 2004. A simplified procedures for analysis os polycyclic aromatic hydrocarbons in water, sediments and mussels. Journal of Chromatography A 104 (7), 181–188. Miège, C., Dugay, J., Hennion, M.C., 2003. Optimization, validation and comparison of various extraction techniques for the trace determination of polycyclic aromatic hydrocarbons in sewage sludges by liquid chromatography coupled to diodearray and fluorescence detection. Journal of Chromatography A 995, 87–97. Mollea, C., Bosco, F., Ruggeri, R., 2005. Fungal biodegradation of naphthalene: microcosms studies. Chemosphere 60, 636–643. Moraes, E.M., Adorno, M.A.T., Zaiat, M., Foresti, E., 2000. Determination of volatile acids by gas chromatography in effluent from anaerobic reactors treating liquid and solid wastes (Determinação de Ácidos Voláteis por Cromatografia Gasosa em Efluentes de Reatores Anaeróbios Tratando Resı́duos Lı́quidos e Sólidos). In: VI Oficina e Seminário Latino-Americano de Digestão Anaeróbia, 2000, Recife PE. Anais. Recife - PE Editora Universitária - UFPE 2, 235–238. Muyzer, G., Wall, E.C., Uitterlinden, A.A.G., 1993. Profiling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-amplified genes coding for 16S rRNA. Applied and Environmental Microbiology 59, 695–700. Nakhla, G., 2003. Biokinetic modeling of in situ bioremediation of BTX compoundsimpact of process variables and scale-up implications. Water Research 37, 1296–1307. Ratola, N., Lacorte, S., Alves, A., Barceló, D., 2006. Analysis of polycyclic aromatic hydrocarbons in pine needles by gas chromatography–mass spectrometry. Comparison of different extraction and clean-up procedures. Journal of Chromatography A 1114, 198–204. Standard Methods for the Examination of Water and Wastewater, twentieth ed., 1998 American Public Health Association/American Water Works Association/ Water Environment Federation, Washington, DC, USA. Wu, Y., Luo, Y., Zou, D., Ni, J., Liu, W., Teng, Y., Li, Z., 2008. Bioremediation of polycyclic aromatic hydrocarbons contaminated soil with Monilinia sp.: degradation and microbial community analysis. Biodegradation 19, 247–257. Zaiat, M., Cabral, A.K.A., Foresti, E., 1994. Horizontal-flow anaerobic immobilized biomass reactor for treating wastewater: conception and preliminary evaluation of performance (Reator Anaeróbio Horizontal de Leito Fixo Para Tratamento de Águas Residuárias: Concepção e Avaliação Preliminar de Desempenho). Revista Brasileira de Engenharia Quı́mica 11, 33–42. Zaiat, M., Passig, F.H., Foresti, E., 2000. A mathematical model and criteria for designing horizontal-flow anaerobic immobilized biomass (HAIB) reactors. Bioresource Technology 71, 235–243.