This is the accepted version of the journal article: Fernandez Palacios, Eva; Lafuente Sancho, Francisco Javier; Mora Garrido, Mabel; [et al.]. «Exploring the performance limits of a sulfidogenic UASB during the long-term use of crude glycerol as electron donor». Science of the Total Environment, Vol. 688 (October 2019), p. 1184-1192. DOI 10.1016/j.scitotenv.2019.06.371 This version is available at https://ddd.uab.cat/record/266947 under the terms of the license 1 Exploring the performance limits of a sulfidogenic UASB during the 2 long-term use of crude glycerol as electron donor 3 E. Fernández-Palacios, J. Lafuente, M. Mora, D. Gabriel* GENOCOV Research Group, Department of Chemical, Biological and Environmental Engineering, Escola d’enginyeria, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain (*corresponding author: David.Gabriel@uab.cat) 4 5 ABSTRACT 6 SOx contained in flue gases and S-rich liquid effluents can be valorized to recover 7 elemental sulfur in a two-stage bioscrubbing process. The reduction of sulfate to sulfide 8 is the most crucial stage to be optimized. In this study, the long-term performance of an 9 up-flow anaerobic sludge blanket (UASB) reactor using crude glycerol as electron donor 10 was assessed. The UASB was operated for 400 days with different sulfate and organic 11 loading rates (SLR and OLR, respectively) and a COD/S-SO42- ratio ranging from 3.8 g 12 O2 g-1 S to 5.4 g O2 g-1 S. After inoculation with methanogenic, granular biomass, the 13 competition between sulfate-reducing and methanogenic microorganisms determined to 14 what extent dissolved sulfide and methane were produced. After the complete washout of 15 methanogens, which was revealed by next-generation sequencing analysis, the highest S- 16 EC was reached in the system. The highest average sulfate elimination capacity (S- 17 EC=4.3 kg S m-3d-1) was obtained at a COD/S-SO42- ratio of 5.4 g O2 g-1 S and an OLR 18 of 24.4 kg O2 m-3d-1 with a sulfate removal efficiency of 94%. The conversion of influent 19 COD to methane decreased from 12% to 2.5% as the SLR increased while a large fraction 20 of acetate (35% of the initial COD) was accumulated. Our data indicate that crude 21 glycerol can promote sulfidogenesis. However, the disappearance of methanogens in the 22 long-term due to the outcompetition by sulfate reducing bacteria, lead to such large 23 accumulation of acetate. 24 . 25 26 1 27 Keywords: crude glycerol, UASB, SRB, carbon sink, bioscrubber, SO2 valorization 28 29 1. INTRODUCTION 30 Combustion of sulfur-containing fuels, such as coal, natural gas, peat, wood and oil, 31 results in SO2 formation mainly generated in the energetic and industrial sectors (Klimont 32 et al., 2013). These emissions are usually treated through physical-chemical processes 33 that are expensive and generate additional effluents requiring further processing and 34 energy inputs (Srivastava and Jozewicz., 2001; Philip and Deshusses., 2003). As an 35 example, aqueous slurries with high sulfite and sulfate content are generated from wet 36 flue gas desulfurization (FGD) with sodium hydroxide. The development of 37 environmentally friendly alternatives to valorize not only SO2 from FGD but also S-rich 38 liquid effluents is clearly needed. The two-stage bioscrubber concept described in Figure 39 S1 (Supplementary Information section) is a potential alternative process to recover 40 elemental sulfur from such gaseous effluents (Fernández et al., 2017). The process 41 consists of a first scrubbing stage for SOx absorption in water at slightly alkaline pH, 42 followed by two-stage biological process to obtain elemental sulfur. The biological 43 process converts, firstly, sulfate to total dissolved sulfide (TDS) using an organic waste 44 as C source and electron donor and, secondly, TDS to elemental sulfur through a partial 45 oxidation performed under oxygen limiting conditions. Partial TDS oxidation can be also 46 performed through autotrophic denitrification. 47 Sulfate to TDS reduction has been studied under a range of substrates such as sewage or 48 methanol under a range of operating conditions including thermophilic processes (Qian 49 et al., 2015; Jiang et al., 2013; Weijma et al., 2000). However, the sulfate reduction stage 50 is still the one that requires further economic and technical improvements (Chen et al., 51 2014). Sulfate reduction, which is catalyzed by sulfate-reducing bacteria (SRB) 2 52 (Liamleam and Annachhatre, 2007), can be carried out with a large assortment of organic 53 wastes and under different operating conditions. Recently, crude glycerol has been 54 proposed as a competitive substrate to reduce high loads of sulfate to TDS in batch tests 55 as well as during the start-up of an up-flow anaerobic sludge blanket reactor (UASB) 56 (Mora et al., 2018). Crude glycerol is a waste organic effluent produced in the biodiesel 57 industry with an exceptional COD concentration (≈800 g COD L-1) that does not require 58 any additional treatment before its use as carbon source. In most of the recent research, 59 crude glycerol has been used as a suitable substrate for biogas production in anaerobic 60 systems (Nakazawa et al., 2015) or as a co-substrate in anaerobic digestion to increase 61 biogas production (Nghiem et al., 2014; Athanasoulia et al., 2014). Despite different 62 approaches to reduce sulfate from S-rich streams have been investigated using pure 63 glycerol (Santos et al., 2017) the potential of crude glycerol has been poorly explored in 64 sulfidogenic reactors. . In fact, to the best of our knowledge, assessment of the long-term 65 performance of a UASB using crude glycerol as electron donor for sulfate reduction has 66 not been addressed before. 67 One of the main problems related to the start-up of a reactor for sulfate reduction with 68 organic matter is the competition between SRB and methanogens. Since the inoculum is 69 usually obtained from full-scale anaerobic digesters targeting methane production, the 70 enrichment of sulfate reducing bacteria (SRB) becomes a decisive threat between sulfate 71 reduction and methane production. SRB and methanogens competition for the common 72 intermediates in the anaerobic degradation process has been widely reported, which 73 results in a variable performance of the reactor. Then, the origin of the inoculum becomes 74 critical as it contains diverse microbial populations leading to differences in initial activity 75 and substrate adaptation (De Vrieze et al., 2015). Some variables that have been studied 76 to assess this competition are COD to SO42- ratio (COD/S ratio), TOC/S ratio, organic 3 77 loading rates (OLR), sulfate loading rates (SLR) and the type of electron donor used to 78 reduce sulfate (Pol et al., 1998). Most of them have been studied using different electron 79 donors, such as glucose (O’Reilly and Colleran, 2006), lactate (Zhou et al., 2014), ethanol 80 (Hu et al., 2015) and Volatile Fatty Acid (VFA) mixtures (acetate, propionate and 81 butyrate) (Omil et al., 1996; Omil et al., 1998; Lens et al., 1998) but, there are no reports 82 on the long-term operation using a substrate with a significant fraction of slowly 83 hydrolysable carbon source such as that contained in crude glycerol. Despite the 84 competition of SRB over methanogens has been widely described, the use of crude 85 glycerol as carbon source implies the production of metabolites through its fermentation 86 that may lead to microbial diversity changes that have not been yet explored. It remains 87 uncertain if such competition may be beneficial or not to process performance. 88 Another important parameter in the start-up and long-term operation of UASB reactors 89 for sulfate reduction is biomass granulation. Granular biomass provides a strong structure 90 and good settling properties that contribute to high biomass retention, and stands up 91 against possible shock and high loading rates (Liu and Tay., 2004). As demonstrated by 92 De Vrieze et al. (2015), selecting an inoculum according to your objective is crucial for 93 a robust operation. In the current study, granular sludge from methanogenic anaerobic 94 digesters is used as inoculum in UASB bioreactors for sulfate reduction (Mora et al., 95 2018) considering that no granular SRB reactors are currently operated in the field. Long- 96 term operation may lead to microbial diversity changes that could affect UASB 97 performance during the long-term operation of such sulfidogenic reactors. 98 To better understand the limits and applicability of sulfate reduction, this study aimed at 99 assessing 1) the limits of the process in terms of sulfate reducing capacities and 2) the 100 long-term performance of a UASB for the treatment of synthetic sulfate-rich effluents to 101 produce TDS using crude glycerol as electron donor. This study not only provides new 4 102 information regarding S-rich streams valorization but also assesses the use of crude 103 glycerol specifically for sulfate reduction through the analysis of C sinks to the main 104 bioprocess occurring in the system. 105 106 2. MATERIALS AND METHODS 107 2.1. Experimental setup 108 A jacketed glass-made UASB reactor of 2.5 L, with a granular sludge volume of 1L, was 109 used in this study. A detailed scheme of the UASB is presented in Figure 1. During the 110 operation, inlet pH ranged between 8.4 and 8.6 and temperature was controlled at 35ºC 111 by a thermostatic bath connected to the water jacket of the reactor (Figure 1). The 112 composition of the mineral medium was (g L-1): K2HPO4 (3), NH4Cl (0.2) dissolved in 113 tap water to add macro- and micronutrients and adjusted to pH=8.8-9.0 with NaOH (2 114 M). Because of the difficulty to control pH in a plug-flow type bioreactor, the pH in the 115 UASB was not controlled. However, the buffering capacity of the mineral medium 116 allowed maintaining the outlet pH above 7 along the whole operation of the reactor (data 117 not shown). Mineral medium was pumped at a flow rate of 0.5 L h-1, once mixed with the 118 organic influent, from the bottom to the top of the UASB (up-flow velocity of 0.25 m h- 1 119 ). The flow rate of the organic influent was set at 30 mL h-1. Hence, crude glycerol was 120 diluted to adjust the inlet COD concentration. The hydraulic residence time (HRT), 121 calculated as that corresponding to the reaction volume only (sludge blanket), was 2h. 122 Biogas produced in the UASB was collected in a 5 L Tedlar bag (FlexFoil, SKC Inc.) to 123 monitor its composition and flow rate. Inlet and outlet flows were also sampled every 124 two/three days to analyze COD, S compounds (sulfate, thiosulfate and TDS) and VFA. 125 2.2 Operating conditions and short-term experiments 5 126 Granular sludge obtained from an anaerobic digester treating wastewater in a pulp and 127 paper industry was used to inoculate the UASB reactor to reach an initial Volatile 128 Suspended Solids (VSS) concentration of 28 g VSS L-1. As shown in Table 1, the reactor 129 was operated during 400 days at different sulfate inlet concentrations. Inlet sulfate 130 concentrations ranging from 235±17 mg S-SO42- L-1 to 859±30 mg S-SO42- L-1 were fed 131 by adding sodium sulfate to the mineral medium. Different SLR and OLR were tested 132 during the long-term operation of the UASB in order to assess the sulfate reducing 133 capacity of the system 134 The operation was divided into 6 different periods according to the initial sulfate inlet 135 concentrations and the COD/S ratio tested (Table 1). Period I focused on the UASB start- 136 up to enrich the microbial community with SRB; Period II served to optimize the 137 operation at the same inlet sulfate concentration set in Period I by providing a higher 138 OLR; Periods III and IV were set to study the sulfate reducing activity at a moderate 139 initial sulfate concentration; Period V served to explore the limits of the system by setting 140 the highest sulfate inlet concentration and, finally, Period VI targeted the recovery of the 141 initial UASB stability when the lowest SLR was set. Table 1 shows average operating 142 conditions and standard deviations obtained from each operational period. SLR and OLR 143 were calculated considering the reaction volume only. 144 During period VI, short-term assays were carried out during 60 h to assess the sulfate 145 elimination capacity (S-EC) in the UASB reactor under variable loading rate conditions 146 typically found in industrial activities. The experiment consisted of a stepwise decrease 147 of the sulfate inlet concentration every 12 h (from 450 mg S L-1 to 120 mg S L-1). The 148 COD/S was also varied since the OLR remained constant during the short-term 149 experimental assays. At each concentration tested, effluent was collected to measure the 6 150 concentration of sulfate, TDS and VFA. In addition, sulfate concentration in the influent 151 was also measured every 12 hours. 152 2.3. Analytical methods 153 Sulfate (SO42-) and thiosulfate (S2O32-) concentrations were analyzed by ion 154 chromatography with conductivity detection using a Dionex ICS-2000 equipment with 155 an Ultimate 3000 Autosampler Column Compartment, and an IonPac AS18 column 156 (ThermoScientific, USA). COD was measured using COD kits and a photometer 157 (Lovibond®). 158 VFA concentration were measured by gas chromatography (7820-A, Agilent 159 Technologies) equipped with a DB-FFA column and using a flame ionization detector 160 (FID) with helium as carrier gas. Prior to VFA analyses, samples were prepared following 161 the procedure described in Baeza et al. (2017) which consisted of pipetting 0.8 mL of 162 filtered samples together with 0.2 mL of a preserving solution (which also contained 163 hexanoic acid as the internal standard) in a glass vial of 1.5 mL. The VFA species 164 analyzed included acetic, propionic, butyric, isobutyric and valeric acids. Only acetic and 165 propionic acids were detected in significant amounts. All samples were filtered at 0.22 166 µm (Millipore, USA). 167 A sulfide selective electrode (VWR International Eurolab, S.L) connected to a benchtop 168 meter (Symphony, VWR) was used for the off-line measurement of TDS concentration. 169 Prior to their measurement, samples were diluted and preserved in sulfide antioxidant 170 buffer (SAOB). The SAOB composition was (g L-1): ascorbic acid (35) and EDTA (67) 171 dissolved in NaOH (2M). 172 CH4, CO2 and H2S in the biogas produced were analyzed by gas chromatography (7820- 173 A, Agilent Technologies, USA). The volume of the gas produced in the UASB reactor 174 was calculated following the Gas Bag Method (GBM) as presented in Ambler and Logan 7 175 (2011). This method is based on 1) measuring the initial composition of the collected gas 176 in the bag, 2) adding a known volume of tracer gas (CO2 in this case) in order to produce 177 an appreciable change in the area of the CO2 peak in the GC chromatogram and 3) 178 analyzing the new composition after the injection. The average methane flowrate was 179 calculated based on the volume of gas collected along variable time periods in which 180 biogas was accumulated in the sampling bag located on top of the UASB and the methane 181 concentration in the gas bag. 182 2.4. Illumina sequencing analysis 183 Microbial diversity analysis was performed using next-generation sequencing. Genomic 184 DNA was extracted from samples of the inoculum and on day 190 of the UASB operation 185 by applying the protocol of MoBio PowerSoil™ DNA extraction kit (MoBio 186 Laboratories, USA). The quantity and quality of the extracted DNA were evaluated by 187 using a NanoDrop 1000 Spectrophotometer (Thermo Fisher Scientific, USA). DNA 188 metabarcoding analysis was performed on an Illumina MiSeq platform by AllGenetics & 189 Biology SL (A Coruña, Spain). For library preparation, a fragment of the bacterial 16S 190 V4-V5 ribosomal RNA gene of around 400 bp was amplified using the primers 515F (5' 191 GTG CCA GCM GCC GCG GTA A 3') and 909R (5' CCG TCA ATT YHT TTR AGT 192 3'). 193 194 3. RESULTS 195 3.1 Long-term performance of the UASB and short-term experiments 196 The UASB performance was evaluated during 400 days of continuous operation in terms 197 of sulfate removal efficiency (S-RE), COD removal efficiency (COD-RE) and sulfate and 198 COD elimination capacities (S-EC and COD-EC, respectively) using crude glycerol as 199 carbon source. Table 2 shows the results obtained from the long-term UASB operation as 8 200 averages and standard deviations of all data acquired in each period. Monitoring results 201 of sulfur species are shown in Figure 2 while COD measurements together with the 202 average flowrate of methane and the concentration of each VFA monitored are presented 203 in Figure 3. 204 As shown in Table 2, the UASB operation was divided into six periods. During the UASB 205 start-up, sulfate inlet concentration and OLR were maintained at 235±17 mg S-SO42- L-1 206 and 12.0±2.1 kg O2 m-3d-1, respectively. As can be observed in Figure 2, sulfate reduction 207 started almost immediately after inoculation and increased steadily during period I. 208 During period II (days 99-115) the OLR was stepwise increased since the organic matter 209 was limiting in Period I for the complete reduction of sulfate. An S-RE up to 99% with 210 an almost complete removal of the COD (Figure 3A) was obtained at the end of this 211 period. Consequently, the SLR and OLR were increased in Period III, day 115 to 197, by 212 doubling the inlet sulfate and COD concentrations in order to reach a higher sulfate 213 reduction capacity in the system while maintaining the COD/S ratio around 5 g O2 g-1 S, 214 which was found to provide the best results in terms of sulfate and COD removal 215 efficiencies during period II. Even if almost complete sulfate removal (S-RE up to 94%) 216 was found during period III, there was a progressive VFA accumulation coupled to a 217 decrease of the COD-RE (Figure 3B). 218 Despite the S and C loads were not changed, a fourth period was defined because the 219 actual crude glycerol (glycerol 1) was replaced by a new batch from the supplier at the 220 beginning of period IV. The new crude glycerol (glycerol 2) contained 35% less water, 221 an average COD of 900 g O2 L-1 (640 g C3H8O3 L-1) and a lower BOD5/COD ratio (Table 222 SI-3 in Supplementary Information). Lower average S-REs were found with the new 223 crude glycerol (Table 2) despite some progressive acclimation of functional bacteria to 224 the crude glycerol towards the end of period IV (S-RE above 85%). Consequently, the 9 225 SLR was increased in Period V (days 238-288) to verify the maximum treatment capacity 226 of the reactor. In Period V the lowest COD/S ratio was tested despite the system was 227 already overloaded. VFAs accumulated to reaching their maximum concentrations 228 (Figure 3B) as described in section 3.2. During Period VI the UASB operated during 112 229 days under the conditions tested during period III-IV to recover the initial stability of the 230 system. 231 In addition, short-term experiments were performed. For that purpose, different sulfate 232 inlet concentrations were tested at the end of period VI (days 360-370) to verify system 233 robustness to face quickly variable inlet loads during the operation. Figure 4 shows sulfate 234 and sulfide concentration profiles as well as the corresponding S-RE and S-EC obtained 235 during the short-term assays. As can be observed in Figure 4, the sulfate RE was almost 236 doubled for the lowest sulfate concentration tested (120 mg S-SO42- L-1) compared to the 237 initial situation before the short-term experiment. 238 239 3.2 Organic matter sink: sulfate reduction and biogas and VFA production 240 Even if traces of other VFA were measured from the biodegradation of crude glycerol, 241 only acetate and propionate were predominant and therefore considered for further 242 analysis (Figure 3B). During periods I and II, inlet COD was completely consumed and 243 no VFA were detected in the effluent while some CH4 was produced and recovered as 244 part of the gas phase. From period III onwards, when an average OLR of 25 kg O2 m-3d- 1 245 was fed (Table 1), the effluent contained mainly acetate. As can be observed in Figure 246 3A, this increase in acetate coincided with a decrease in CH4 production, which ceased 247 75 days after the beginning of period III. During period V, the maximum concentration 248 of acetate in the reactor was reached (1000 mg acetate L-1), which progressively decreased 10 249 until the end of the operation when acetate concentrations below 340 mg L-1 were 250 detected. 251 The conversion of COD resulting from each operating period was also assessed in terms 252 of CH4 composition in biogas, VFA concentrations in the effluent (acetate and 253 propionate), and COD used for sulfate reduction (Table 3). The COD balance was 254 calculated based on measurements of inlet and outlet COD and VFAs and corresponding 255 methane production. According to the methane composition in biogas, the biogas flowrate 256 and the TDS and residual COD in the effluent, COD conversion proportions along the 257 different periods were obtained according to processes stoichiometry (see Supplementary 258 Information for calculation details). Table 3 shows that along Periods I and II around 11% 259 of the inlet COD was directed to methane production while almost no VFA accumulated 260 in the reactor. However, between 33 and 41% of the influent COD ended up in acetic acid 261 from Period III until the end of the operation while between 3 and 6% of the inlet COD 262 was converted to propionic acid from Period III onwards. Concomitantly, the COD 263 fraction converted to methane had the opposite behavior and was around 0% from Period 264 III onwards, which was taken into account as a way of reporting the percentage of 265 electrons utilized by methanogens. The potential use of COD for sulfate reduction was 266 more stable along the operation even if a progressive deterioration could be detected that 267 accounted for a 26.5% less of organic matter calculated for this purpose comparing the 268 last and the first periods. The rest of COD was assumed to be used for growth and CO2 269 formation, although it could not be accurately quantified. 270 3.3 . Illumina sequencing analysis and bacterial community assessment 271 The scope of the microbial analysis was not to describe the evolution of the microbial 272 diversity along the UASB operation but to provide further data to explain the switch from 273 methane production to non-methane production conditions from a microbial perspective. 11 274 Thus, the bacterial community through Illumina analysis of the 16S rRNA gene was 275 applied to compare the methanogenic granular sludge used as inoculum with the biomass 276 developed after 190 days of the UASB operation when no methane production was 277 observed. Results obtained from the microbial analysis are presented in Figure 5. 278 Deltaproteobacteria and Methanomicrobia were the main classes detected in the 279 inoculum with a relative abundance of 20% and 16% respectively (Figure 5). Clostridia 280 was the third class in order of abundance (13.5%). After operating the UASB for 190 281 days, Deltaproteobacteria increased their relative abundance to 49% in the sludge bed 282 sample, clearly the most abundant class of the total reads; while Methanomicrobia 283 decreased to 0% without detecting any other methanogenic microorganism. Clostridia 284 was the second class in order of abundance (12.15%), followed by Gammaproteobacteria 285 and Bacteroidia (11 and 7.3% of total reads, respectively). 286 As the operation proceeded, in the biomass community of day 190 (Tables S1 and S2; 287 Supplementary Information), Desulfovibrio was the most abundant OTU at genus level, 288 with a 35.3% of total retrieved sequences. Oude Elferink et al. (1994) reported that 289 Desulfovibrio had higher affinity for sulfate and higher growth rate than other SRB genera 290 such as Desulfobulbus and Desulfobacter. Interestingly, Desulfatirhabdium sp. accounted 291 for 2% of total operational taxonomic units at genus level which has been described as 292 butyrate-oxidizing bacteria (Balk et al., 2008). In the case of Proteobacteria that were not 293 SRB, the highest relative proportion of microorganisms belonged to the 294 Enterobacteriaceae family (11%). 295 296 4. DISCUSSION 297 4.1 Startup of a sulfate reducing UASB reactor and influence of inoculum 298 In practice, start-up of full-scale UASB reactors for sulfate reduction is handicapped 299 because of the lack of reactors from where inocula with a high density of SRB can be 12 300 withdrawn. Few works have reported the start-up of sulfidogenic reactors with inocula 301 that have not been adapted to sulfidogenic conditions (García-Solares et at., 2014; Omil 302 et al., 1998).. Inoculation with methanogenic sludge from widespread, full-scale 303 mesophilic anaerobic digesters is the most common alternative and, probably, the only 304 alternative in practice at full-scale. The evolution of the UASB performance observed in 305 Figure 2 shows that stable sulfate removal efficiencies higher than 80% were achieved 306 just one month after the continuous operation of the UASB reactor initially inoculated 307 with granular sludge from an anaerobic digester treating wastewater in a pulp and paper 308 industry. The inoculum was not pre-adapted but sulfate reduction started almost from the 309 beginning of the operation since sulfate was present in the pulp and paper industry and, 310 consequently, sulfate-reducing bacteria. This is in agreement with Roest et al. (2005), 311 who stated that anaerobic digesters sludge from paper mill industries are suitable for 312 providing an appropriate process culture to promote sulfidogenesis. 313 Compared to previous works, such short and efficient start-up was remarkable 314 considering the source of the inoculum used. As examples, Gonçalves et al. (2005) needed 315 over 6 months to bioactivate an UASB to obtain anaerobic sulfidogenic sludge able to 316 degrade 400 mg SO42- L-1 using molasses as C source, while Bertolino et al. (2015) needed 317 over 200 days to enrich granular sludge from an UASB treating domestic wastewater 318 during the treatment of 2 g SO42-L-1 influent with pure glycerol. In our work, the granular 319 sludge used as inoculum was mainly methanogenic, which was confirmed through 320 Illumina sequencing. Methanosaeta and Methanobacterium were the most abundant 321 methanogens at genus level in the inoculum, what was expected as they are the most 322 characteristic archaeal sequences found in anaerobic digesters (Leclerc et al., 2004), while 323 a reduced amount of SRB were found. Figure 3A shows that the maximum flow of 324 methane was produced during period I and II due to the influence of the inoculum. During 13 325 period I (OLR of 12.0±2.1 kg O2 m-3d-1) and II (OLR of 15.8±4.6 kg O2 m-3d-1), the 326 average organic matter consumption was 86% and 89% respectively. COD concentrations 327 were below 100 mg O2 L-1 in the effluent, which probably corresponded to the less 328 biodegradable matter contained in crude glycerol considering that the anaerobic 329 biodegradability of crude glycerol due to presence of such inhibitory impurities has been 330 reported to be between 65-85% (Viana et al., 2012). Furthermore, 10% of the oxidized 331 organic matter was used for methane production during period I (Table 3). Similarly, 332 during period II, 11% of the transferred electrons were utilized for methane production. 333 Despite such methanogenic activity, Figure 2 shows that a stable operation in terms of 334 sulfate reduction was reached by the end of period II with almost complete removal of 335 sulfate and COD (S-RE of 96.5% and COD-RE of 89.3%). Despite methane production, 336 results confirmed that organic substrates were available for sulfate reduction and that 337 microbial communities underwent a fast and gradual acclimation to their environment. 338 Our work demonstrates that using methanogenic granular sludge from a paper and pulp 339 industry leads to a fast start-up of sulfidogenic UASBs when moderate inlet sulfate 340 concentrations of 235±17 mg S-SO42- L-1 are treated using crude glycerol as electron 341 donor at C/S ratio of 3.8 g O2 g-1 S. 342 4.2 Shifts in the organic matter sink 343 The sink of the organic matter can shift drastically due to the evolution of the microbial 344 population, which influences the performance of the UASB reactor. During Period III, a 345 high S-RE was reached after few days of operation which allowed obtaining a maximum 346 S-EC of 6.6 kg S-SO42- m-3 d-1 (273 g S-SO42- m-3 h-1) from this period (Period III). 347 Compared to previous periods, the COD-RE dropped drastically (Figure 3A) and an 348 accumulation of VFA was observed (Figure 3B), which indicated that a steady, almost 349 complete sulfate reduction could be reached at a SLR of 4.6 kg S-SO42- m-3 d-1 and COD/S 14 350 ratios lower than 5 g C g-1 S. As previously described by Pol et al. (1998), when a sulfate- 351 rich wastewater is fed into an anaerobic reactor, organic matter will be removed both via 352 methanogenesis and sulfate reduction and when methanogenesis becomes suppressed 353 then a gradual decrease in the organic matter conversion (COD removal) is observed, 354 which was corroborated herein with crude glycerol instead. During this third period (OLR 355 of 24.4 kg COD m-3d-1), the average organic mater consumption was 38 % while only 2.3 356 % of the COD removed ended in methane production when the COD/S was increased to 357 5.4g O2 g S-1 (Tables 2 and 3). Taking into account that the reported cellular yield for 358 acidogenic bacteria (0.14-0.17 gVSS/gCOD) is five times higher than that of acetogenic 359 bacteria (0.025-0.051 gVSS/gCOD) or methanogenic archaea (0.01-0.054 gVSS/gCOD) 360 (Pavlostathis and Giraldo-Gomez, 1991) glycerol will be readily available for acidogenic 361 bacteria, and the limiting step will be the methanogenesis. 362 During the anaerobic digestion of glycerol, some organic acids (acetic, propionic, butyric, 363 valeric and others), produced by fermentative acidogenic bacteria, cannot be consumed 364 by methanogenic archaea at the same rate at which they are produced (Viana et al., 2012). 365 The accumulation of VFA indicated that the slowly growing methanogens could not 366 sufficiently and rapidly metabolize the intermediate products from VFA producers 367 (acidogenic and acetogenic populations). Since acetate is mainly converted by 368 methanogens and no methanogens were found in the sludge sample from day 190 (Figure 369 5), increasing concentrations of acetic acid were found in the reactor between period III 370 and V. This is in agreement with the production of methane measured from the gas phase 371 (Figure 3A) and with some authors statements (Harada et al., 1994; Omil et al., 1998), 372 who pointed out that the predominance of SRB over methanogens in sulfate-rich streams 373 is only achieved after long-term operation (more than 100 days) in UASB reactors. As 374 reported with other electron donors (Dar et al., 2008; Raskin and Rittmann, 1996), SRB 15 375 also out-competed methanogens using crude glycerol. However, more research is needed 376 to understand to which extent is this competition beneficial or, if losing completely the 377 presence of methanogens at such a low up-flow velocity would imply losing S-EC due to 378 other problems. Potentially, diffusional limitations and bed stratification may appear due 379 to the lack of gas bubbles moving upward. 380 4.3 Long-term UASB performance and microbial diversity changes 381 High sulfate reduction efficiencies together with VFA accumulation were also observed 382 by Bertolino et al. (2012). From period III onwards, the acetate concentration remained 383 in the 400-1100 mg O2 L-1 range. Although the S-RE was significant in Period III (80%), 384 it progressively decreased to below 80% (even below 50% in periods V and VI). Despite 385 some SRB are able to oxidize acetate to CO2 (Widdel and Pfennig, 1982; Szewzyk and 386 Pfennig, 1987; Muyzer and Stams, 2008), only incomplete oxidizers were detected in the 387 190-day sample (Supplementary Information). Consequently, promoting acetate- 388 oxidizing SRB may be an alternative to increase the sulfate reduction concomitantly 389 producing a less C loaded effluent. 390 Most of the COD used for methane production lead to acetate and propionate 391 accumulation from Period III onwards (Table 3) together with an evolution of the 392 microbial diversity (Figure 5). Deltaproteobacteria was the most abundant class after 190 393 days of operation. Many genera such as Desulfovibrio, Desulfobacter and 394 Desulfuromonas belong to this class and play a fundamental role in the sulfur cycle. Only 395 4% of reads were not identified at class level, compared to the 31% of reads not identified 396 in the inoculum sample. This result indicated that the microbial community specialized 397 in more specific functions and that populations were selected according to operating 398 conditions. In the presence of sulfate, SRB usually out-compete methanogens, which only 399 dominate in a low-sulfate environment (Oude Elferink et al., 1994). 16 400 As acclimation proceeded under the high TDS concentration reached during the operation 401 of the reactor, methanogens were completely washed out. In general, sulfate-reducing 402 bacteria can grow with a much wider substrate range than methanogens (Muyzer and 403 Stams, 2008). Consequently, methanogenic communities require syntrophic associations, 404 which are not essential in sulfate reducing environments (Janssen et al., 2009). Several 405 SRB are able to use glycerol as an electron donor and some Desulfovibrio species have 406 been reported to grow with glycerol (Stams et al., 1985; Kremer and Hansen, 1987; 407 Esnault et al., 1988). As reported by Hu et al. (2015) and Lens et al. (1998) the complete 408 or incomplete oxidation of organic substrates accomplished by some species of SRB will 409 depend on the COD/SO42- ratio in the influent. However, in this study, it is reasonable to 410 conclude that SRB always performed incomplete oxidation at the ratios tested. The 411 disappearance of methanogens and the concomitant accumulation of acetate in the system 412 suggested that methanogens were probably the only microorganisms consuming acetate 413 at observable rates. 414 It remains an open question how acetate oxidation can be stimulated in order to improve 415 the reactor performance. Despite some promising attempts have been made (Lens et al., 416 1998), further development of strategies for augmentation of acetotrophic SRB are 417 warranted to increase the sulfidogenic capacity of the process. A strong association 418 between two acidophiles, a sulfate reducing bacterium and a non-sulfate reducing 419 bacterium is proposed by Kimura et al. (2006) but further improvements of the operation 420 are still required. Also, despite VFA accumulation has been regarded as a sign of process 421 failure in anaerobic digestion, VFA accumulation can be seen as an opportunity in sulfate- 422 reducing UASBs since VFA have important biotechnological potential as these 423 carboxylates can be used as substrates for production of biofuels and bioplastics, or in 424 other bioprocesses. The loss of organic matter from the UASB reactor is economically 17 425 undesirable since the reducing power supplied with glycerol is only partly used. In 426 addition, further resources must be used to treat the excess COD from the anaerobic 427 reactor. In the sulfur recovery process depicted in Figure S1, a reduction-oxidation 428 bioprocess was proposed. Then, the COD in the effluent could be treated in the CSTR 429 reactor for the partial oxidation of sulfide to elemental sulfur although an extra 430 consumption of oxygen to treat COD would be required. Consequently, optimization in 431 the use of the electron donor is warranted. 432 4.4 Potential process limitations 433 Inhibitory substances are often found to be the main cause of anaerobic reactor 434 disturbance and failure as they cause an adverse shift in the microbial population or 435 inhibition of bacterial growth (Chen et al., 2008). Inhibition of anaerobic digestion is 436 usually diagnosed by a decrease of the steady-state rate of methane gas production and 437 accumulation of organic acids (Kroeker et al., 1979), which was found in the long-term 438 operation of the UASB. The organic acid and methane forming microorganisms differ 439 widely in terms of physiology, growth kinetics, and sensitivity to environmental 440 conditions (Pohland and Ghosh, 1971). At pHs below 7.0, most carboxyl groups are 441 undissociated, thus they pass freely through the membrane and can inhibit the growth of 442 many bacteria. Uncharged molecules such as acetic acid may be inhibitory because they 443 diffuse across the cell membrane and act as an uncoupler (Ghose and Wiken, 1955), 444 whereas acetate ion is not permeant. Inhibition concentrations of 4.68·10-3 mg free acetic 445 acid L-1 (pH=7.5) have been reported to block acetoclastic methanogenesis (Fukuzaki et 446 al., 1990). Despite the buffer used, the pH in the reactor varied from 8.4-8.7 at the inlet 447 to 6.7-7.5 at the outlet due to VFAs accumulation, and particularly acetic acid. 448 Considering the pka of acetic acid (4.76) and the concentrations of acetate found in the 449 early stages of period III (around 375 mg acetate L-1), concentrations above 0.23 mg free 18 450 acetic acid L-1 found in the UASB could have led to significant methanogens inhibition. 451 In addition, free H2S concentrations leading to 50% inhibition of methanogenesis of 250 452 mg S L-1 in the pH range 6.4–7.2 and 90 mgS L-1 at pH = 7.8–8.0 have been reported 453 (Koster et al., 1986), indicating that methanogens were also inhibited by sulfide 454 accumulation in the early stages of Period III. Overall, TDS and acetate accumulation 455 lead to a fast decrease of the methanogenic activity in Period III. No methane production 456 from period IV until the end of the operation indicated that methanogenic communities 457 were more susceptible to dissolved sulfide concentration than SRB as was also observed 458 in Jing et al. (2013). 459 Despite the high concentrations of acetate found in the reactor during period III, IV and 460 V (400-1100 mg O2 L-1), sulfate reduction proceeded at high S-RE during period III, 461 indicating that SRB were not affected by acetic acid. Similarly to methanogens, free H2S 462 may inhibit SRB. However, inhibitory free H2S concentration in literature are often 463 contradictory and confusing probably due to the difference in anaerobic inocula used, the 464 susceptibility of anaerobes and the experimental methods and conditions tested in each 465 study, and particularly the pH in the bioreactor. Anaerobic treatment of sulfate-rich 466 wastewater proceeds successfully at COD/sulfate ratios lower than 10 g COD g-1 SO42- 467 when precautions are taken to prevent sulfide toxicity (Pol et al., 1998). The TDS during 468 the operation reached 460 mg S L-1 by day 240 (Period V). Reis et al. (1992) found that 469 more than 547 mg H2S L-1 can completely inhibit SRB activity at pH 6.2, whereas at pH 470 9.0, dissolved H2S is mainly in the form of HS−, which does not penetrate into cells easily 471 (Mora-Naranjo et al., 2003) and therefore would not have a strong inhibitory effect over 472 SRB. As Reis et al. (1992) observed, sulfate uptake decreased when sulfide concentration 473 in the medium increased, and increased again when it was removed from the medium, 474 which pointed out that sulfide is a reversible inhibitor of SRB. A pH range of 6.7-6.8 at 19 475 the outlet of the reactor points out at a reduction of the potential maximum SRB rates due 476 to inhibition of SRB by hydrogen sulfide. 477 4.5 Overall performance of the sulfidogenic UASB 478 Sulfidogenesis was achieved through adaptation of granular sludge with important 479 methanogenic activity, using electrons derived from substrate towards sulfate reduction. 480 While adapting methanogenic granular sludge to sulfate reduction is one of the most 481 common and widespread procedures to engineer microbial sulfate reduction (García- 482 Solares et al., 2014), the stability of the system during long-term operations is still a cause 483 of concern. During period V, SLR was maintained during 50 days but the high sulfate 484 inlet resulted in a sulfate-reduction failure. The system was overloaded and its maximum 485 capacity, 6.5 kg S m-3d-1, was reached after 165 days of operation at a SLR of 6.7 kg S- 486 SO42- m-3d-1 and a COD/S of 5.6 g O2 g-1 S. Overall, the performance of the UASB is 487 comparable to that obtained by Bijmans et al. (2008) (9.7 kg S m-3 d-1) using formate, 488 which is more biodegradable than crude glycerol. Higher S-ECs were found compared to 489 Boshoff et al. (2004) (600 mg SO4 L-1 d-1) who used tannery effluent as carbon source. 490 After a long-term operation of 360 days, short-term SLR assays (Figure 4) were 491 performed to study the capability of the UASB to reduce sulfate at such a low HRT (2h) 492 and under dynamic conditions with non limiting COD availability. The UASB adapted 493 well to transient load changes and, more interestingly, recovered to the initial load 494 exhibiting a 25% higher S-RE compared to that before the short-term experiment. 495 However, it remains to be investigated why such temporary load decrease resulted 496 apparently beneficial for the UASB performance considering that the same sulfide 497 concentration was found before and after the stepwise decrease of the inlet sulfate 498 concentration. Based on a sulfur balance, a larger C/S ratio during the short-term 499 experiment (up to 19.1 g COD g S-1) could have led to an increase in the production of 20 500 organosulfur compounds. Overall, a sulfur balance of 85-95% along the UASB operation 501 was obtained based on inlet and outlet sulfate (S-SO42-) concentrations and produced TDS 502 (Figure 1). Such imbalance was attributed to other organosulfur compounds such as 503 dimethyl sulfide (DMS) or dimethyl disulfide (DMDS) amongst others that were 504 qualitatively detected in the effluent of the UASB (see Table SI-4 in Supplementary 505 Information). 506 507 4. CONCLUSIONS 508 Long-term operation of a sulfidogenic UASB reactor can be successfully achieved using 509 crude glycerol as carbon source at low up-flow velocities. It was demonstrated that at 510 OLR above 24 kg O2 m-3d-1 and SLR of 4.6 kg S-SO42-m-3d-1 VFA were accumulated. 511 The TDS concentration increase together with VFA accumulation were potential 512 inhibitors of methanogenic activity and when methane production decreased, glycerol 513 was converted mainly to acetic acid and propionic acid. It was not only the COD/S-SO42- 514 ratio, but a sum and combination of factors along the operation that determined the 515 competition between SRB and methanogenic archaea. However, further batch activity 516 tests are warranted to properly validate the results obtained herein. 517 518 5. ACKNOWLEDGMENTS 519 The authors are members of the GENOCOV research group from the Department of 520 Chemical, Biological and Environmental Engineering at UAB (Universitat Autonoma de 521 Barcelona), a unit of Biochemical Engineering of Xarxa de Referencia en Biotecnologia 522 de Catalunya (XRB), Generalitat de Catalunya. Authors acknowledge Nubiola Spain 523 (FERRO Corporation) and the Spanish Government, Ministerio de Economía y 524 Competitividad, through the project CTQ2015-69802-C2-1-R MINECO/FEDER, EU, 21 525 for the financial support provided to perform this research. The authors also acknowledge 526 UIPSA and ECOMOTION for their kind collaboration in this work. 527 528 6. REFERENCES 529 Ambler, J.R., Logan, B.E., 2011. Evaluation of stainless steel cathodes and a 530 bicarbonate buffer for hydrogen production in microbial electrolysis cells using a 531 new method for measuring gas production. Int. J. Hydrogen Energy 36, 160–166. 532 Athanasoulia, E., Melidis, P., Aivasidis, A., 2014. Co-digestion of sewage sludge and 533 crude glycerol from biodiesel production. Renew. Energy 62, 73–78. 534 Baeza, J.A., Martínez-Miró, À., Guerrero, J., Ruiz, Y., Guisasola, A., 2017. 535 Bioelectrochemical hydrogen production from urban wastewater on a pilot scale. J. 536 Power Sources 356, 500–509. 537 Balk, M., Altinbaş, M., Rijpstra, W.I.C., Damsté, J.S.S., Stams, A.J.M., 2008. 538 Desulfatirhabdium butyrativorans gen. nov., sp. nov., a butyrate-oxidizing, sulfate- 539 reducing bacterium isolated from an anaerobic bioreactor. Int. J. Syst. Evol. 540 Microbiol. 58, 110–115. 541 Bertolino, S.M., Rodrigues, I.C.B., Guerra-Sá, R., Aquino, S.F., Leão, V.A., 2012. 542 Implications of volatile fatty acid profile on the metabolic pathway during 543 continuous sulfate reduction. J. Environ. Manage. 103, 15–23. 544 Bertolino, S.M., Silva, L.A.M., Aquino, S.F., 2015. Comparison of UASB and 545 fluidized-bed reactors for sulfate reduction. Brazilian J. Chem. Eng. 32, 59–71. 546 Bijmans, M.F.M., Peeters, T.W.T., Lens, P.N.L., Buisman, C.J.N., 2008. High rate 547 sulfate reduction at pH 6 in a pH-auxostat submerged membrane bioreactor fed 548 with formate. Water Res. 42, 2439–2448. 549 Boshoff, G., Duncan, J., Rose, P.D., 2004. Tannery effluent as a carbon source for 550 biological sulphate reduction. Water Res. 38, 2651–2658. 551 Chen, C., Liu, L., Lee, D.J., Guo, W., Wang, A., Xu, X., Zhou, X., Wu, D., Ren, N., 552 2014. Integrated simultaneous desulfurization and denitrification (ISDD) process at 553 various COD/sulfate ratios. Bioresour. Technol. 155, 161–169. 554 Chen, Y., Cheng, J.J., Creamer, K.S., 2008. Inhibition of anaerobic digestion process: A 555 review. Bioresour. Technol. 99, 4044–4064. 556 Dar, S.A., Kleerebezem, R., Stams, A.J.M., Kuenen, J.G., Muyzer, G., 2008. 557 Competition and coexistence of sulfate-reducing bacteria, acetogens and 558 methanogens in a lab-scale anaerobic bioreactor as affected by changing substrate 559 to sulfate ratio. Appl. Microbiol. Biotechnol. 78, 1045–1055. 560 De Vrieze, J., Raport, L., Willems, B., Verbrugge, S., Volcke, E., Meers, E., Angenent, 561 L.T., Boon, N., 2015. Inoculum selection influences the biochemical methane 562 potential of agro-industrial substrates. Microb. Biotechnol. 8, 776–786. 22 563 Esnault, G., Caumette, P., Garcia, J.L., 1988. Characterization of Desulfovibrio 564 giganteus sp. nov., a Sulfate-reducing Bacterium Isolated from a Brackish Coastal 565 Lagoon. Syst. Appl. Microbiol. 10, 147–151. 566 Fernández, E., Mora M., Gabriel D. (2017). Long-term UASB operation as first stage of 567 an elemental sulfur recovery process from S-rich effluents. Paper presented at the 568 7th International Conference on Biotechniques for Air Pollution Control and 569 Bioenergy. La Coruña, Spain, 19-21 July (pp. 70-71). Reprografía Noroeste S.L. 570 Fukuzaki, S., Nishio, N., Nagai, S., 1990. Kinetics of the methanogenic fermentation of 571 acetate. Appl. Environ. Microbiol. 56, 3158–3163. 572 García-Solares, S.M., Ordaz, A., Monroy-Hermosillo, O., Jan-Roblero, J., Guerrero- 573 Barajas, C., 2014. High Sulfate Reduction Efficiency in a UASB Using an 574 Alternative Source of Sulfidogenic Sludge Derived from Hydrothermal Vent 575 Sediments. Appl. Biochem. Biotechnol. 174, 2919–2940. 576 Ghose, T., Wiken, T., 1955. Inhibition of Bacterial Sulphate-Reduction in Presence of 577 Short Chain Fatty Acids. Physiol. Plant. 8, 116–134. 578 Gonçalves, M.M.M., Leite, S.G.F., Sant’Anna, G.L., 2005. The bioactivation procedure 579 for increasing the sulphate-reducing bacteria in a UASB reactor. Brazilian J. Chem. 580 Eng. 22, 565–571. 581 Harada, H.U.S.M.K., 1994. Interaction Between Sulfate Reducing Bacteria and 582 Methane Producing Bacteria In Uasb Reactors Fed With Low Strength Wastes 583 Containing Different Levels Of Sulfate. Water Res. 28, 355–367. 584 Hu, Y., Jing, Z., Sudo, Y., Niu, Q., Du, J., Wu, J., Li, Y.Y., 2015. Effect of influent 585 COD/SO42- ratios on UASB treatment of a synthetic sulfate-containing 586 wastewater. Chemosphere 130, 24–33. 587 Janssen, A.J.H., Lens, P.N.L., Stams, A.J.M., Plugge, C.M., Sorokin, D.Y., Muyzer, G., 588 Dijkman, H., Van Zessen, E., Luimes, P., Buisman, C.J.N., 2009. Application of 589 bacteria involved in the biological sulfur cycle for paper mill effluent purification. 590 Sci. Total Environ. 407, 1333–1343. 591 Jiang, F., Zhang, L., Peng, G.L., Liang, S.Y., Qian, J., Wei, L., Chen, G.H., 2013. A 592 novel approach to realize SANI process in freshwater sewage treatment - Use of 593 wet flue gas desulfurization waste streams as sulfur source. Water Res. 47, 5773– 594 5782. 595 Jing, Z., Hu, Y., Niu, Q., Liu, Y., Li, Y.Y., Wang, X.C., 2013. UASB performance and 596 electron competition between methane-producing archaea and sulfate-reducing 597 bacteria in treating sulfate-rich wastewater containing ethanol and acetate. 598 Bioresour. Technol. 137, 349–357. 599 Kimura, S., Hallberg, K.B., Johnson, D.B., 2006. Sulfidogenesis in low pH (3.8-4.2) 600 media by a mixed population of acidophilic bacteria. Biodegradation 17, 159–167. 601 Klimont, Z., Smith, S.J., Cofala, J., 2013. The last decade of global anthropogenic 602 sulfur dioxide: 2000-2011 emissions. Environ. Res. Lett. 8. 603 Koster, I.W., Rinzema, A., de Vegt, A.L., Lettinga, G., 1986. Sulfide inhibition of the 23 604 methanogenic activity of granular sludge at various pH-levels. Water Res. 20, 605 1561–1567. 606 Kremer, D.R., Hansen, T.A., 1987. Glycerol and dihydroxyacetone dissimilation in 607 Desulfovibrio strains. Arch. Microbiol. 147, 249–256. 608 Kroeker, E.J., Schulte, D.D., Sparling, A.B., Lapp, H.M., 1979. Anaerobic treatment 609 process stability. J. (Water Pollut. Control Feredation) 51, 718–727. 610 Leclerc, M., Delgènes, J.P., Godon, J.J., 2004. Diversity of the archaeal community in 611 44 anaerobic digesters as determined by single strand conformation polymorphism 612 analysis and 16S rDNA sequencing. Environ. Microbiol. 6, 809–819. 613 Lens, P.N.L., Van Den Bosch, M.C., Hulshoff Pol, L.W., Lettinga, G., 1998. Effect of 614 staging on volatile fatty acid degradation in a sulfidogenic granular sludge reactor. 615 Water Res. 32, 1178–1192. 616 Liamleam, W., Annachhatre, A.P., 2007. Electron donors for biological sulfate 617 reduction. Biotechnol. Adv. 25, 452–463. 618 Liu, Y., Tay, J.H., 2004. State of the art of biogranulation technology for wastewater 619 treatment. Biotechnol. Adv. 22, 533–563. 620 Mora-Naranjo, N., Alamar-Provecho, C., Meima, J., Haarstrick, A., Hempel, D.C., 621 2003. Experimental investigation and modelling of the effect of sulfate on 622 anaerobic biodegradation processes in municipal solid waste. Water Sci. Technol. 623 48, 221–227. 624 Mora, M., Lafuente, J., Gabriel, D., 2018. Screening of biological sulfate reduction 625 conditions for sulfidogenesis promotion using a methanogenic granular sludge. 626 Chemosphere 210, 557–566. 627 Muyzer, G., Stams, A.J.M., 2008. The ecology and biotechnology of sulphate-reducing 628 bacteria. Nat. Rev. Microbiol. 6, 441–454. 629 Nakazawa, M.M., Silva Junior, W.R.S., Kato, M.T., Gavazza, S., Florencio, L., 2015. 630 Anaerobic treatment of crude glycerol from biodiesel production. Water Sci. 631 Technol. 72, 1383–1389. 632 Nghiem, L.D., Nguyen, T.T., Manassa, P., Fitzgerald, S.K., Dawson, M., Vierboom, S., 633 2014. Co-digestion of sewage sludge and crude glycerol for on-demand biogas 634 production. Int. Biodeterior. Biodegrad. 95, 160–166. 635 O’Reilly, C., Colleran, E., 2006. Effect of influent COD/SO42- ratios on mesophilic 636 anaerobic reactor biomass populations: Physico-chemical and microbiological 637 properties. FEMS Microbiol. Ecol. 56, 141–153. 638 Omil, F., Lens, P., Hulshoff Pol, L., Lettinga, G., 1996. Effect of upward velocity and 639 sulphide concentration on volatile fatty acid degradation in a sulphidogenic 640 granular sludge reactor. Process Biochem. 31, 699–710. 641 Omil, F., Lens, P., Visser, A., Hulshoff Pol, L.W., Lettinga, G., 1998. Long-term 642 competition between sulfate reducing and methanogenic bacteria in UASB reactors 643 treating volatile fatty acids. Biotechnol. Bioeng. 57, 676–685. 24 644 Oude Elferink, S.J.W.H., Visser, A., Hulshoff Pol, L.W., Stams, A.J.M., 1994. Sulfate 645 reduction in methanogenic bioreactors. FEMS Microbiol. Rev. 15, 119–136. 646 Pavlostathis, S.G., Giraldo-Gomez, E., 1991. Kinetics of Anaerobic Treatment. Water 647 Sci. Technol. 24, 35–59. 648 Philip, L., Deshusses, M.A., 2003. Sulfur dioxide treatment from flue gases using a 649 biotrickling filter - Bioreactor system. Environ. Sci. Technol. 37, 1978–1982. 650 Pohland, F.G., Ghosh, S., 1971. Developments in anaerobic stabilization of organic 651 wastes - the two-phase concept. Environ. Lett. 1, 255–266. 652 Pol, L.W.H., Lens, P.N.L., Stams, A.J.M., Lettinga, G., 1998. Anaerobic treatment of 653 sulphate-rich wastewaters. Biodegradation 9, 213–224. 654 Qian, J., Lu, H., Jiang, F., Ekama, G.A., Chen, G., 2015. Beneficial co-treatment of 655 simple wet flue gas desulphurization wastes with freshwater sewage through 656 development of mixed denitrification – SANI process. Chem. Eng. J. 262, 109– 657 118. 658 Raskin, L., Rittmann, B.E., 1996. Competition and Coexistence of Sulfate-Reducing 659 and Methanogenic Populations in Anaerobic Biofilms. Appl. Environ. Microbiol. 660 62, 3847–3857. 661 Reis, M. A, Almeida, J.S., Lemos, P.C., Carrondo, M.J., 1992. Effect of hydrogen 662 sulfide on growth of sulfate reducing bacteria. Biotechnol. Bioeng. 40, 593–600. 663 Roest, K., Heilig, H.G.H.J., Smidt, H., De Vos, W.M., Stams, A.J.M., Akkermans, 664 A.D.L., 2005. Community analysis of a full-scale anaerobic bioreactor treating 665 paper mill wastewater. Syst. Appl. Microbiol. 28, 175–185. 666 Santos, S.C., Liebensteiner, M.G., van Gelder, A.H., Dimitrov, M.R., Almeida, P.F., 667 Quintella, C.M., Stams, A.J.M., Sánchez-Andrea, I., 2017. Bacterial glycerol 668 oxidation coupled to sulfate reduction at neutral and acidic pH. J. Gen. Appl. 669 Microbiol. 64, 1–8. 670 Srivastava, R.K., Jozewicz, W., 2001. Flue Gas Desulfurization : The State of the Art. 671 Air Waste Manag. 51, 1676–1688. 672 Stams, A.J.M., Hansen, T.A., Skyring, G.W., 1985. Utilization of amino acids as energy 673 substrates by two marine Desulfovibrio strains. FEMS Microbiol. Lett. 31, 11–15. 674 Szewzyk, R., Pfennig, N., 1987. Complete oxidation of catechol by the strictly 675 anaerobic sulfate-reducing Desulfobacterium catecholicum sp. nov. Arch. 676 Microbiol. 147, 163–168. 677 Viana, M.B., Freitas, A. V., Leitão, R.C., Pinto, G.A.S., Santaella, S.T., 2012. 678 Anaerobic digestion of crude glycerol: a review. Environ. Technol. Rev. 1, 81–92. 679 Weijma, J., Hulshoff Pol, L.W., Stams, A.J.M., Lettinga, G., 2000. Performance of a 680 thermophilic sulfate and sulfite reducing high rate anaerobic reactor fed with 681 methanol. Biodegradation 11, 429–439. 682 Widdel, F., Pfennig, N., 1982. Studies on dissimilatory sulfate-reducing bacteria that 25 683 decompose fatty acids II. Incomplete oxidation of propionate by Desulfobulbus 684 propionicus gen. nov., sp. nov. Arch. Microbiol. 131, 360–365. 685 Zhou, J.M., Song, Z.Y., Yan, D.J., Liu, Y.L., Yang, M.H., Cao, H. Bin, Xing, J.M., 686 2014. Performance of a haloalkaliphilic bioreactor and bacterial community shifts 687 under different COD/SO42-ratios and hydraulic retention times. J. Hazard. Mater. 688 274, 53–62. 689 26