Magnetite accelerates syntrophic acetate oxidation in methanogenic systems with high ammonia concentrations

Summary Ammonia accumulation is a major inhibitory substance causing anaerobic digestion upset and failure in CH 4 production. At high ammonia levels, CH 4 production through syntrophic acetate oxidization (SAO) pathways is more tolerant to ammonia toxicity than the acetoclastic methanogenesis pathway, but the low CH 4 production rate through SAO constitutes the main reason for the low efficiency of energy recovery in anaerobic digesters treating ammonia‐rich substrates. In this study, we showed that acetate fermentation to CH 4 and CO 2 occurred through SAO pathway in the anaerobic reactors containing a high ammonia concentration (5.0 g l−1 NH 4 +–N), and the magnetite nanoparticles supplementation increased the CH 4 production rates from acetate by 36–58%, compared with the anaerobic reactors without magnetite under the same ammonia level. The mechanism of facilitated methanogenesis was proposed to be the establishment of direct interspecies electron transfer (DIET) for SAO, in which magnetite facilitated DIET between syntrophic acetate oxidizing bacteria and methanogens. High‐throughput 16S rRNA gene sequencing analysis revealed that the bacterial Geobacteraceae and the archaeal Methanosarcinaceae and Methanobacteriaceae might be involved in magnetite‐mediated DIET for SAO and CH 4 production. This study demonstrated that magnetite supplementation might provide an effective approach to accelerate CH 4 production rates in the anaerobic reactors treating wastewater containing high ammonia.


Introduction
The anaerobic digestion of organic wastes is regarded as one of important strategies for bio-energy recovery. The concentration of ammonia has been found to be a vital factor affecting the performance and stability of CH 4 production from anaerobic digestion (Rajagopal et al., 2013). High levels of ammonia, either present in proteinrich wastes or accumulated during protein degradation, might cause a severe deterioration in the stability and performance of anaerobic digestion. In general, unstable performance or process failure occurs as the total ammonia nitrogen reaches 1500-7000 mg l À1 (Rajagopal et al., 2013), and the inhibiting ammonia concentrations depend on many factors such as the substrate, inocula, environmental conditions (temperature, pH) and acclimation periods (Chen et al., 2008). The direct inhibitory effects of high-level ammonia on microbial activity, in particular, on acetate-utilizing methanogens are considered as the primary cause of the decline in anaerobic digester performance and stability (Koster and Lettinga, 1984;Robbins et al., 1989).
In the anaerobic digester producing CH 4 , the energy sources for methanogens are limited to acetate, H 2 /CO 2 , CO, formate, methanol, ethanol, isopropanol, methylamines and methysulfides (Conrad, 2007;Borrel et al., 2011), and acetate is a very important precursor for methanogens producing CH 4 . Acetate degradation to CH 4 can occur either via acetoclastic methanogenesis, in which acetate is cleaved into CH 4 and CO 2 by acetoclastic methanogens, or via syntrophic acetate oxidization (SAO), in which acetate is converted to H 2 /CO 2 that is then reduced to CH 4 by hydrogenotrophic methanogens (Westerholm et al., 2011). Although there are some conflicting findings about the sensitivity of acetoclastic and hydrogenotrophic methanogens to ammonia toxicity, major studies have found that hydrogenotrophic methanogens are more tolerant to ammonia stress than acetoclastic methanogens (Angelidaki and Ahring, 1993;Karakashev et al., 2005). Thus, SAO is frequently detected as the dominant acetate degradation pathway in anaerobic digester treating wastes containing high concentrations of ammonia (Karakashev et al., 2006;Li et al., 2014).
To alleviate the ammonia toxicity effect or to efficiently digest ammonia-rich waste in anaerobic digesters, researchers have attempted the strategies of the bioaugmentation/accumulation of ammonia-tolerant microorganisms, especially those involved in SAO metabolism (Chen et al., 2008;Fotidis et al., 2013aFotidis et al., ,b, 2014. Although SAO provides anaerobic digesters with the flexibility to respond to a high level of ammonia that leads to inhibitory activity for acetoclastic methanogens, methanogenesis via SAO is unfavourable in terms of CH 4 production rate (Karakashev et al., 2006;Lee et al., 2015). Thus, how to facilitate CH 4 production through SAO pathway is critical to improve the performance of anaerobic digestion with high levels of ammonia.
Conductive iron minerals (Kato et al., 2012;Viggi et al., 2014;Li et al., 2015;Yamada et al., 2015;Zhuang et al., 2015) and conductive carbon materials such as activated carbon (Liu et al., 2012), biochar (Chen et al., 2014a) and carbon cloth  have been demonstrated to accelerate methanogenesis, in which these conductive materials mediate direct interspecies electron transfer (DIET) in syntrophic consortia. Kato et al. (2012) first reported that both magnetite and hematite stimulated methanogenesis from acetate through SAO in rice paddy soils, and DIET between syntrophic partners of Geobacter and Methanosarcina species occurred in the presence of conductive iron minerals. To the best of our knowledge, the possibility of magnetite supplementation to enhance the rates of CH 4 production through SAO at high ammonia levels has not been previously investigated. Thus, the aim of this study was to examine magnetite supplementation as a potential method to trigger a faster syntrophic methanogenesis in the ammonia-rich anaerobic digesters. CH 4 production rates and microbial communities were characterized and compared in the magnetite-free and magnetite-supplemented anaerobic reactors with different ammonia concentrations. This allowed us to identify the mechanism of accelerated methanogenesis by magnetite supplementation under high ammonia concentrations.

Influence of magnetite on acetate methanogenesis under different ammonia concentrations
In this study, the bioreactors were not previously acclimatized to ammonia. The accumulated CH 4 production from the acetate-fed incubations with different concentrations of NH 4 -N (0, 0.5 and 5.0 g l À1 ) is summarized in Fig. 1. After a short lag phase (6 days), the NH 4 -N-free and 0.5 NH 4 -N bioreactors started to yield CH 4 and proceeded to completion within 25 and 18 days respectively. The high concentration of ammonia (5.0 g l À1 ) significantly extended the lag phase of methanogenesis (20 days), and the magnetite-supplemented and magnetite-free incubations reached the plateau of CH 4 production within 31 and 37 days respectively. Data demonstrated that the presence of magnetite did not substantially affect methanogenesis in both the NH 4 -Nfree and 0.5 NH 4 -N bioreactors. However, in the case of high ammonia concentration (5.0 g l À1 ), CH 4 was generated more rapidly in the presence of magnetite than the unamended incubations. CH 4 production rates during the linear phase of metabolism were estimated from the data in Fig. 1A and compared with respect to the supplementation of magnetite (Fig. 1B). With an ammonium concentration of 5.0 g l À1 , magnetite stimulated methanogenesis from acetate with a rate (258 lmol day À1 ) that was 36% faster than that without magnetite (190 lmol day À1 ). During the second enrichment for the 5.0 NH 4 -N bioreactors, methanogenesis from acetate proceeded at higher rates without any lag phase, and the presence of magnetite enhanced the CH 4 production rates by 58% (Fig. S1).
As shown in Fig. 1B, CH 4 production rates in the 0.5 NH 4 -N bioreactors were higher than those in both the NH 4 -N-free and 5.0 NH 4 -N bioreactors. The lower methanogenic activities in the NH 4 -free bioreactors were likely due to the lack of a nitrogen source caused by the re-suspension pretreatment of the sludge. Previous studies have evidenced that methanogenesis via SAO under high-level ammonia concentrations is markedly slower than acetoclastic methanogenesis under low-level ammonia concentrations (Karakashev et al., 2006;Lee et al., 2015), which is consistent with the lower CH 4 production rates in the 5.0 NH 4 -N bioreactors relative to the 0.5 NH 4 -N bioreactors in this study. The second reason might be related to the direct toxicity of ammonia to methanogens (Chen et al., 2008;Westerholm et al., 2012), explaining the longer lag phase in the 5.0 NH 4 -N bioreactors.
As shown in Fig. 1C  production was 86-94% of the value predicted from the stoichiometry of acetate degradation to CH 4 (CH 3 COOH?CH 4 + CO 2 ) ( Table 1), suggesting that methanogenesis was the predominant terminal electron-accepting process in all of the acetate-fed incubations. The rest fraction of acetate might be consumed by biomass growth (Siegrist et al., 2002) or alternative electron acceptors in the sludge. It is noticed that CH 4 accumulation in the magnetite-supplemented bioreactors was even lower, which might be resulted from acetate consumption by Fe(III) bioreduction. This was supported by Fe(II) production in the magnetite-supplemented bioreactors (Fig. S2), and the more intensive Fe(III) reduction corresponded to the greater difference in CH 4 recovery between the magnetite-free and magnetite-amended incubations ( Table 1). The XRD measurements showed that, although Fe(III) reduction occurred, no new mineral product was formed, and no change occurred to magnetite particles before and after anaerobic incubation (Fig. S3).
To further investigate the stimulatory effect of magnetite on methanogenesis, control bioreactors in the absence of acetate were conducted with the same inoculum and experimental conditions (Table S1). Data showed very low 'endogenous' CH 4 accumulation in all the acetate-free bioreactors, and methanogenesis in the presence of magnetite was even slower (Fig. S4). These results implied that (i) the re-suspension pretreatment had almost removed the endogenous methanogenic substrate in the sludge; (ii) magnetite did not facilitate methanogenesis from acetate as an additional source of trace elements (e.g. iron ions); and (iii) the stimulatory effect of magnetite on methanogenesis under the high ammonia concentration was substrate dependent.

Syntrophic acetate oxidation under a high level of ammonia
The high concentration of ammonia resulted in the deceleration of methanogenic processes in terms of the lag time and production rate, implying that the route of CH 4 production in the 5.0 NH 4 -N bioreactors might be different from the 0.5 NH 4 -N bioreactors. Under high ammonia concentrations, SAO is more competitive than acetoclastic methanogenesis for CH 4 production (Schn€ urer et al., 1999). SAO usually involves interspecies hydrogen transfer between acetate-oxidizing bacteria and hydrogen-utilizing methanogens, which maintaining a low hydrogen partial pressure (typically below 10 À5 atm) to keep anaerobic acetate oxidation energetically favourable (Stams and Plugge, 2009). Disrupting the delicate balance between syntrophic partners by increasing the concentration of hydrogen is an established method for detecting interspecies hydrogen transfer (Ahring and Westermann, 1988;Warikoo et al., 1996;Rakoczy et al., 2011;Viggi et al., 2014). Here, to identify the route of CH 4 production under different ammonia concentrations and the influence of magnetite on the methanogenic pathway, the effects of hydrogen spiking on methanogenesis from acetate were investigated in both 0.5 NH 4 -N and 5.0 NH 4 -N bioreactors in the presence or absence of magnetite.
When acetate was degrading to CH 4 , 20 ml of H 2 was spiked into the headspace of the bottles, which were then shaken for proper mixing. As shown in Fig. 2A and B, the introduction of H 2 did not affect the rates of acetate conversion to CH 4 in the 0.5 NH 4 -N bioreactors with or without magnetite. During the early half of the incubation period, the kinetics of acetate degradation corresponded to the kinetics of CH 4 production in all the 0.5 NH 4 -N bioreactors, suggesting that CH 4 was generated from acetate. The presence of H 2 increased the CH 4 accumulation within the latter half of the incubation period, which was a result of the conversion of H 2 /CO 2 to CH 4 . As expected, the effect of H 2 spiking on acetate methanogenesis in the 5.0 NH 4 -N bioreactors was different with respect to magnetite supplementation ( Fig. 2C and D). After H 2 spiking, acetate degradation was temporarily inhibited for 10 days in the magnetite-free bioreactors, while acetate degradation in the magnetitesupplemented bioreactors was much less affected by H 2 disruption. During the inhibition of acetate degradation, CH 4 production was not affected and was compensated by the conversion of H 2 /CO 2 to CH 4 , which gradually decreased the concentration of H 2 and alleviated the inhibitory effect on acetate degradation. In summary, the results indicated that: (i) H 2 spiking did not affect acetate degradation in the 0.5 NH 4 -N bioreactors but temporarily inhibited acetate degradation in the 5.0 NH 4 -N bioreactors, indicating the involvement of interspecies hydrogen transfer in the 5.0 NH 4 -N bioreactors. This provided an evidence for CH 4 production via acetoclastic methanogenesis under low ammonia concentrations and via SAO under high ammonia concentrations. (ii) For the 5.0 NH 4 -N bioreactors, the process of acetate degradation in the bioreactors containing magnetite was much less affected by H 2 disturbance than the magnetite-free bioreactors. This implied that SAO, in the absence of magnetite, might highly depend on interspecies hydrogen transfer, whereas SAO in the presence of magnetite might not completely reply on interspecies hydrogen transfer.
The alternative strategy for interspecies hydrogen transfer is DIET, in which two syntrophic partners exchange electrons via electrical interactions that may be facilitated by conductive pili and cytochromes (Summers et al., 2010) and abiotic conductive materials (Kato et al., 2012;Liu et al., 2012). Electrically conductive magnetite has been demonstrated to facilitate syntrophic methanogenesis from acetate, ethanol (Kato et al., 2012;Yamada et al., 2015), propionate (Viggi et al., 2014;Yamada et al., 2015), butyrate (Li et al., 2015) in rice paddy soils and anaerobic digesters. The mechanisms of facilitated methanogenesis are consistently proposed as DIET establishment between syntrophic partners through magnetite minerals, which can function as a surrogate for cytochrome OmcS that is important for DIET . The stimulatory effect of magnetite on methanogenic activity in the 5.0 NH 4 -N bioreactors was likely due to the presence of DIET promoting electron transfer between acetate-oxidizing bacteria and methanogens, as depicted in Fig. 3.   rRNA gene sequences per sample with an average length of 307 bp were generated. The sequences were assigned to 1,410 AE 88~1,702 AE 54 OTUs per sample, with a distance limit of 0.03 (Table S3). The major bacterial populations in the bioreactors were associated with OTUs belonging to the phyla Bacteroidetes, Firmicutes and Proteobacteria (Fig. S5). Most of the archaeal OTUs in the bioreactors were assigned to the orders Methanobacteriales, Methanomicrobiales and Methanosarcinales. Distinct clusters of microbial communities are presented by PCoA in Fig. 4A, which showed that the microbial community in the high NH 4 -N bioreactors was significantly separated from the low NH 4 -N bioreactors by principle component 1. PCoA analyses revealed that the concentration of ammonia was an important effect on variation in the community structure; the presence of magnetite was also a determinant for community structure, especially for the 5.0 NH 4 -N bioreactors. Consistently, the UPGMA tree calculated from the weighted UniFrac distances showed that the microbial community in the reactors with high ammonia concentrations was grouped together and not related to other samples (Fig. 4B).
The family level characterization further illustrates the diversity of functional structures, and Fig. 5A shows a heat map of the most abundant bacterial and archaeal taxa with at least >2% relative sequence abundance in one sample. Evidently, the 5.0 NH 4 -N bioreactors were highly enriched in Methanosarcinaceae but deprived in Methanosaetaceae. This is consistent with previous studies on Methanosaetaceae vs. Methanosarcinaceae dominance in the case of high NH 4 -N concentrations (Karakashev et al., 2005;Song et al., 2010), and the higher resistance of Methanosarcina against ammonia is likely due to its higher volume-to-surface ratio (Wiegant and Zeeman, 1986) and its more versatile spectrum of substrates utilization for CH 4 production (Fotidis et al., 2013a,b). The hydrogenotrophic methanogens of Methanobacteriaceae and Methanomicrobiaceae were increased in the 5.0 NH 4 -N bioreactors with and without magnetite, respectively, in good agreement with previous finding that hydrogenotrophic methanogens are more tolerant to ammonia than acetate-utilizing methanogens (Angelidaki and Ahring, 1993;Karakashev et al., 2005).
The influence of ammonia concentration on acetoclastic methanogens and hydrogenotrophic methanogens is vital to the route of CH 4 production under different ammonia concentrations. Fig. 5B presents the fractions of known methanogens at a family level relative to the total methanogens in each bioreactor. The bioreactors with low ammonia concentrations were dominated by acetoclastic methanogens (Methanosarcinaceae and Methanosaetaceae). The fraction of Methanosaetaceae decreased dramatically in the 5.0 NH 4 -N bioreactors, providing a line of evidence for the SAO pathway in the 5.0 NH 4 -N bioreactors because the previous study had found a strong correlation between the absence of Methanosaetaceae and the involvement of the SAO pathway (Karakashev et al., 2006). These microbial analyses and the results derived from the H 2 experiments supported that acetoclastic methanogenesis might be the dominant pathway at low ammonia concentrations, and that SAO methanogenesis was the prevailing pathway at high ammonia concentrations.
In the 5.0 NH 4 -N bioreactors, the acetate oxidizer of Geobacteraceae and methanogens of Methanosarcinaceae and Methanobacteriaceae were enriched by the supplementation of magnetite (Fig. 5A). Geobacter and Methanosarcinaceae have been demonstrated to be capable of performing DIET in iterative environmental and defined culture studies (Kato et al., 2012;Rotaru et al., 2014a, b;Li et al., 2015). Although the syntrophic DIET function has not been demonstrated with Methanobacteriaceae, previous study found that the presence of Methanobacteriaceae coincided with the presence of magnetite and Geobacteraceae in magnetite-facilitated DIET that stimulating syntrophic butyrate oxidation and CH 4 production in paddy soil enrichment (Li et al., 2015). The possibility of Methanobacteriaceae capable of DIET function is very promising and deserves further investigations. Geobacteraceae were enriched in the magnetite-supplemented bioreactors (Fig. 5A, Fig. S6), which might be a result of Geobacteraceae using Fe(III) from magnetite as electron acceptor (Fig. S2). However, this cannot support the relatively higher enrichment of Geobacteraceae in the 5.0 NH 4 -N bioreactors with the lowest extent of Fe (III) reduction, which implying the potential involvement of Geobacteraceae in methanogenesis from acetate. This further enhanced the possibility of methanogenesis through SAO by Geobacteraceae and methanogens of Methanosarcinaceae and Methanobacteriaceae in the presence of magnetite.
In sum, we studied the effect of magnetite supplementation on methanogenesis from acetate under low and high ammonia concentrations. Methanogenesis was dominant by acetoclastic methanogenesis under low ammonia level and SAO was more favourable under the high ammonia level. The supplementation of magnetite had no stimulatory effect on methanogenesis under low ammonia level, while accelerated methanogenesis via SAO by 36-58% under high ammonia level. The bacterial Geobacteraceae and the archaeal Methanosarcinaceae and Methanobacteriaceae were potentially involved in DIET-mediated SAO and CH 4 production, which was proposed as the mechanism of facilitated methanogenesis by the presence of magnetite.  Experimental procedures

Experimental set-up
All of the experiments were performed using fresh anaerobic sludge as inoculum, which was collected from a fullscale mesophilic wastewater treatment facility treating pig manure (Guangdong Province, China). Before experiments, approximately 1200 ml of sludge was centrifuged at 8000 rpm for 10 min, the supernatants were removed, and the centrifuged sludge was characterized in the general parameters (mg l À1 ): TOC:1110; COD cr : 3250; BOD 5 : 465; Organic matter: 2000; total nitrogen: 56; total P: 26. To prepare the bacterial inoculum, the solid sludge was re-suspended in 135 ml of anaerobic basal medium as previously described (Hao et al., 2015). The details of the basal medium were provided in the Supporting Information. Acetate methanogenesis experiments with different ammonia levels were carried out in 12 sets of bioreactors (each set has triplicate replications) as described in Table S1. NH 4 Cl was used as the ammonia source. Magnetite nanoparticles (Fe 3 O 4 , CAS number: 1317-61-9) were purchased from Sigma-Aldrich Shanghai Trading Co. Ltd. (Shanghai, China), and the average diameter of particles is 50-100 nm by scanning electron microscope (SEM) measurements. The bioreactor experiments were conducted in 275-ml anaerobic serum bottles containing 45 ml of basal medium, 5 ml of re-suspended sludge, 10 mM acetate and 25 mM magnetite (as Fe atoms) if necessary. The effect of magnetite supplementation on acetate methanogenesis was investigated under lowlevel (0.5 g l À1 NH 4 + -N) and high-level ammonia (5.0 g l À1 NH 4 + -N). All of the anaerobic bottles were flushed with an 80%N 2 /20%CO 2 (vol/vol) gas mixture for 60 min at a rate of 10 ml min À1 and were then sealed with Teflon â -coated septa and aluminium crimp caps. The Eh, pH and conductivity of anaerobic bioreactors were measured upon the setting-up (Table S2). To evaluate the influence of magnetite supplementation on the metabolism of the endogenous organic carbon, control incubations in the absence of acetate were also performed under the same experimental conditions. The bottles were incubated at a constant temperature of 37°C in the dark.
To verify the involvement of interspecies hydrogen transfer in the acetate conversion to CH 4 (via SAO), additional batch experiments were conducted, in which H 2 was added to the headspace of the bottles in the presence or absence of magnetite. For each bottle (125ml bottle containing 30 ml of medium), 20 ml of H 2 was introduced using a sterile syringe when acetate was degrading to CH 4 . The concentrations of CH 4 and acetate were periodically monitored after H 2 addition.

Chemical analysis
The CH 4 concentrations in the gas samples were monitored using a GC9700 gas chromatograph (Techcomp Instruments, Shanghai, China) equipped with a flame ionization detector (FID). The temperatures of the inlet, oven and detector were 200, 80 and 350°C respectively. The injection volume was 200 ll, which was extracted from the headspace of the anaerobic bottles using a sterile syringe. The minimum detection limit of CH 4 was 2 ppmv. After slightly vortex, three aliquots of liquid samples were extracted, and analysed for acetate, Fe(II) and ammonia. The concentrations of acetate were determined by HPLC (Shimadzu LC-15C, Japan), equipped with a Wondasil C18 reverse-phase column (250 mm by 4.6 mm, 5 lm pore size), and the lower detection limit was 1.0 mg l À1 . The HCl-extractable Fe (II) concentrations were determined via the ferrozine technique as previously described (Lovley and Phillips, 1986). Ammonium-nitrogen was analysed by the Kjeldahl method, following American Public Health Association's Standard Methods (2005). All of the analyses were performed in triplicate, and the averages were presented along with the corresponding standard deviations (SD).

DNA extraction and 16S rRNA gene sequencing
At the end of the incubation period (when CH 4 production approached a plateau), the samples from the bioreactors were stored at -80°C until DNA extraction. The samples were collected by centrifugation (at 8000 g and 4°C for 10 min) and were then extracted using the PowerSoil TM DNA isolation kit (MO BIO Laboratories, USA) in accordance with the manufacturer's instructions. The DNA concentrations were determined using Qubit 2.0 Fluorometer (Invitrogen, NY, USA). Primer sets F515 (5 0 -GTGCCAGCMGCCGCGGTAA-3 0 ) and R806 (5 0 -GGACTACVSGGGTATCTAAT-3 0 ) were used to amplify the V4 hypervariable region of bacterial and archaeal 16S rRNA. Each DNA sample was amplified in triplicate, and the PCR details were provided in the Supporting Information. The PCR products were sent to Novogene (Beijing, China) for amplicon sequencing using an Illumina Miseq platform. The raw reads have been deposited in the sequence read archive section of NCBI with the following accession number: SRR4171267. Data analysis were performed following previously described methods (Caporaso et al., 2011) using the open source software package QIIME.
To assess the phylogenetic disparities between different microbial communities, principal coordinate analysis (PCoA) and UPGMA tree of weighted UniFrac distances were conducted and analysed with UniFrac software (Lozupone and Knight, 2005). Heat maps were created with the relative abundance data of the relevant taxa of the different samples and were normalized by calculating Z-scores, which represent the relative abundance of taxa in a specific sample that differs from the mean relative abundance of that taxa in all samples normalized by standard deviations. and the Science and Technology Planning Project of Guangdong (2017A030303057). Zhuang, L., Xu, J.L., Tang, J., and Zhou, S.G. (2015) Effect of ferrihydrite biomineralization on methanogenesis in an anaerobic incubation from paddy soil. J Geophys Res Biogeosci 120: 876-886.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Fig. S1. (a) Time course of CH 4 accumulation; (b) average CH 4 production rates during the linear phase of metabolism estimated from the data in Figure S4a for 5.0 g l À1 NH 4 -N incubations during the second enrichment. The error bars represent the standard deviations of three independent incubations. Fig. S2. The concentrations of HCl-extractable Fe(II) in the bioreactors under different ammonia concentrations in the presence or absence of magnetite (CK denotes acetatefree bioreactors; Ac denotes acetate). The error bars represent the standard deviations of three independent incubations. Fig. S3. X-ray diffraction spectrum of magnetite particle in the bioreactors before and after anaerobic incubation. Fig. S4. Time course of CH 4 accumulation in the acetatefree bioreactors under different ammonia concentrations in the presence or absence of magnetite. The error bars represent the standard deviations of three independent incubations. Fig. S5. The relative abundance of microbial community at phylum level according to 16S rRNA gene sequence in each incubation and initial sample.  Table S1. Set up of bioreactors. Table S2. pH, conductivity, redox potential inside the initial bioreactors. Table S3. Similarity-based OTUs and species richness and diversity estimates.