Ferrihydrite associated organic matter (OM) stimulates reduction by Shewanella oneidensis MR-1 and a complex microbial consortia

The formation of Fe(III) oxides in natural environments occurs in the presence of natural organic matter (OM), resulting in the formation of OM-mineral complexes that form through adsorption or coprecipitation processes. Thus, microbial Fe(III) reduction in natural environments most often occurs in the presence of OM-mineral complexes rather than 10 pure Fe(III) minerals. This study investigated to what extent the content of adsorbed or coprecipitated OM on ferrihydrite influences the rate of Fe(III) reduction by Shewanella oneidensis MR-1, a model Fe(III)-reducing microorganism, in comparison to a microbial consortium extracted from the acidic, Fe-rich Schlöppnerbrunnen fen. We found that increased OM content led to increased rates of microbial Fe(III) reduction by S. oneidensis MR-1 in contrast to earlier findings with the model organism Geobacter bremensis. Ferrihydrite-OM coprecipitates were reduced slightly faster than ferrihydrites with 15 adsorbed OM. Surprisingly, the complex microbial consortia stimulated by a mixture of electrons donors (lactate, acetate, and glucose) mimics S. oneidensis under the same experimental Fe(III)-reducing conditions suggesting similar mechanisms of electron transfer whether or not the OM is adsorbed or coprecipitated to the mineral surfaces. We also followed potential shifts of the microbial community during the incubation via 16S rRNA gene sequence analyses to determine variations due to the presence of adsorbed or coprecipitated OM-ferrihydrite complexes in contrast to pure ferrihydrite. Community profile 20 analyses showed no enrichment of typical model Fe(III)-reducing bacteria, such as Shewanella sp. or Geobacter sp., but an enrichment of fermenters (i.e. Enterobacteria) during pure ferrihydrite incubations which are known to use Fe(III) as an electron sink. Instead, OM-mineral complexes favoured the enrichment of microbes including Desulfobacteria and Pelosinus sp., both of which can utilize lactate and acetate as an electron donor under Fe(III) reducing conditions. In summary, this study shows that increasing concentrations of OM in OM-mineral complexes determines microbial Fe(III) reduction rates and 25 shapes the microbial community structure involved in the reductive dissolution of ferrihydrite. Similarities observed between the complex Fe(III)-reducing microbial consortia and the model Fe(III)-reducer S. oneidensis MR-1suggest electron shuttling mechanisms dominate in OM-rich environments, including soils, sediments, and fens, where natural OM interacts with Fe(III) oxides during mineral formation. 30


Introduction
Iron is the fourth most abundant element in the Earth's crust, indicating the role of Fe-cycling in nature is both ubiquitous and widespread. The majority of which is dispersed as iron minerals in natural environments, including soils and sediments with ferrous iron [Fe(II)] and ferric iron [Fe(III)] as the primary redox states (Davison, 1993;Stumm and Sulzberger, 1992;Weber et al., 2006a). Microbially mediated and abiotic redox cycling of Fe play crucial roles in global biogeochemistry, for example, 35 preservation and degradation of organic matter, and the fate of nutrients as well as contaminants for living organisms (Borch et al., 2010;Melton et al., 2014). Iron minerals such as Fe(III) oxides exhibit a high reactivity towards dissolved organic matter (OM), thus Fe(III) oxides are either partially or completely covered by OM in natural environments (Eusterhues et al., 2005;Kaiser and Zech, 2000;Lalonde et al., 2012;Torn et al., 1997). Fe(III) oxides enveloped by OM potentially lead to changes in the surface properties in comparison to non-modified Fe(III) oxides, which ultimately may influence mobility, 40 solubility, and aggregation (Eusterhues et al., 2014;Narvekar et al., 2017;Pédrot et al., 2011;Vindedahl et al., 2016). The poorly crystalline ferrihydrite is one of the most common Fe(III) oxides and typically forms aggregates of nanometer-sized individual crystals (Bigham et al., 2002;Cornell and Schwertmann, 2003;Jambor and Dutrizac, 1998). The coprecipitation of OM with Fe results in the adsorption and occlusion of organic molecules within the interstices between individual ferrihydrite crystals, while adsorption of OM occurs on pre-existing ferrihydrite surfaces (Eusterhues et al., 2014). Moreover, 45 the presence of dissolved OM impedes ferrihydrite growth (Cismasu et al., 2011;Eusterhues et al., 2008;Mikutta et al., 2008;Schwertmann et al., 2005), therefore, ferrihydrite with coprecipitated OM have smaller crystal sizes and more crystallographic defects. Since ferrihydrite is formed usually in OM-rich environments coprecipitation is the more common process in nature (Eusterhues et al., 2008). The different properties of pure ferrihydrite and OM-ferrihydrite coprecipitates may lead to different behaviours during microbial reduction. Due to their smaller crystal size and more defective crystal structure, coprecipitates 50 might faster dissolve. The associated organic material will change the mineral´s surface properties, e.g., the surface charge, with consequences for the accessibility of Fe(III) to microbes, redox-active shuttling compounds, or extracellular enzymes.
Not only do the mechanisms through which OM interacts with Fe(III) oxides, e.g. adsorbed or coprecipitated, influence the reactivity of the mineral-OM complex, the composition of OM may also effect its reactivity. The redox reactivity of OM is attributed mostly to quinones. Past experiments have often been performed using humic acids, i.e., alkaline extracts 55 of natural organic material, which usually contain a higher concentration of the aforementioned redox active functional groups, than natural OM from soil (Klüpfel et al., 2014;Lehmann and Kleber, 2015;Piepenbrock et al., 2014). Humic acids should therefore be more effective in electron shuttling than natural soil organic matter. Previous studies have shown that the presence of dissolved humic acids leads to complexation of Fe(II) (Royer et al., 2002), and complexation and dissolution of Fe(III) (Jones et al., 2009), but also potentially enhance Fe(III) reduction via electron shuttling (Hansel et al., 2004;Jiang and Kappler, 60 1996), endogenous electron shuttling compounds, for example flavins (von Canstein et al., 2008;Kotloski and Gralnick, 2013;Marsili et al., 2008;Nevin and Lovley, 2002;Newman and Kolter, 2000), or the use of Fe(III)-chelating compounds, such as siderophores (Kraemer, 2004;Lovley et al., 1994;Nevin and Lovley, 2002).
Shewanella and Geobacter species represent two of the best studied model Fe(III) reducing microorganisms. 75 Shewanella species have been shown to utilize endogenous electrons shuttles, such as flavins, rather than transferring electron via direct contact between the mineral and the microbe (von Canstein et al., 2008;Kotloski and Gralnick, 2013). Not surprisingly, microbial reduction of Fe(III) oxides like nanometer-sized lepidocrocite and ferrihydrite by Shewanella species can be boosted by the presence of humic acids (Adhikari et al., 2017;Amstaetter et al., 2012;Pédrot et al., 2011;Shimizu et al., 2013) and quinone moieties (Newman and Kolter, 2000) when the amount of dissolved organic carbon is greater than a 80 threshold concentration of 5-10 mg C L -1 (Jiang and Kappler, 2008). Conversely, a study by Eusterhues, et al (2014) reported high C/Fe ratios in coprecipitated and adsorbed ferrihydrite complexes inhibited the reduction of ferrihydrite by G. bremensis. Poggenburg et al. (2016) produced ferrihydrite OM complexes with different organic materials (extracted extracellular polymeric substances, water extracts of soil litter) and used S. putrefaciens and G. metallireducens to reduce them. Their results confirmed that ferrihydrite-bound OM increases reduction rates of Shewanella, whereas it protects the mineral surface from 85 reduction by Geobacter. Further, this study found that reduction rates by Shewanella were correlated to the aromaticity of the mineral-bound OM. Geobacter species have been shown to use direct contact and conductive nanowire structures for electron transfer (Boesen and Nielsen, 2013;Malvankar et al., 2011). Unlike Shewanella, Geobacter do not produce endogenous electron shuttling or chelating compounds (Nevin and Lovley, 2000), however, in the absence of Fe(III) oxides Geobacter are capable of reducing both humic acid (Jiang and Kappler, 2008;Lovley et al., 1996) and extracellular quinone moieties (Scott 90 et al., 1998). Furthermore, the decreased rates of Fe(III) reduction by G. bremensis in incubations containing OM-ferrihydrite complexes compared to pure ferrihydrite minerals suggest that Geobacter do not use external electron shuttles in natural environments rich in OM and Fe(III) oxides. The aforementioned results implicate the effects of OM on Fe(III) reduction rates vary between the two model organisms, Shewanella and Geobacter, presumably due to the various mechanisms of electron transfer used. Taken together, these studies indicate (i) that it depends on the microorganisms present and their Fe 95 respiratory pathways, whether OM on Fe(III) oxide mineral surfaces hinders or promotes Fe(III) oxide reduction. (ii) When electron shuttling is the dominant electron accepting process, a high content of redox active groups in the mineral-associated OM speeds up reduction.
In this study, we used the same experimental design described in the earlier study of Eusterhues et al., 2014, but replaced G. bremensis with S. oneidensis MR-1 to test the effect of increasing amounts of adsorbed and co-precipitated OM 100 on microbial Fe(III) reduction. In a parallel set of incubation studies we used a microbial consortia as inoculum derived from an iron-rich peatland where the microbial reduction of Fe(III) is the dominant electron accepting processes for the degradation of OM under anoxic conditions. Thus, examining the respiratory behaviour of S. oneidensis MR-1 compared to a microbial consortia allows us to study the response of a diverse community of Fe(III) reducers living in a habitat rich in Fe(III) oxides and OM. We compared rates of Fe(III) reduction of ferrihydrite adsorbed and co-precipitated with OM with pure ferrihydrite 4 and followed qualitative and quantitative changes in the microbial community during incubation. To our great surprise, rates observed with Shewanella matched nearly exactly those observed with the complex consortia derived from Schlöppnerbrunnen fen, although Shewanella was not the dominant Fe(III)-reducing microorganism detected at the beginning and end of incubation.
Briefly, granite bedrock is covered by a fibric histosol, often referred to as peat, with a thickness between 40 cm and 70 cm. 115 This fen is fed by anoxic, Fe(II)-rich groundwater (Küsel et al., 2008). The sampling site within the fen used in the current study (site "M") is positioned southwest of previously described drying-rewetting experimental plots (Knorr and Blodau, 2009;Reiche et al., 2008 are routinely found at site "M" (Küsel et al., 2008;Reiche et al., 2008). Peat cores were transported on ice under anoxic 120 conditions (N2) and peat samples were either processed or stored at -20°C until use.

Microbial extraction from peat core samples
The peat microbial community was extracted using a standard protocol adapted to soil. Briefly, a 20 mL syringe, with tip removed, was used as a manual short core sampler to remove 20 mL fractions (~25 g) from bulk peat cores and added to sterile, anoxic 125 mL serum bottles containing 100 mL 0.85% NaCl solution and 8 g sterile glass beads. The slurry was flushed with 125 Ar and shaken at 225 rpm at 4 °C overnight. Samples were transferred to 50 mL Falcon tubes and centrifuged at increasing speeds to observe sedimentation rates. The samples were initially centrifuged at 900 x g for 5 minutes. The sample was decanted into a fresh Falcon tube and the supernatant was centrifuged again at 1200 x g for 10 min. Next, the solid phase was resuspended in 10 mL anoxic 0.85% NaCl solution. The extract, containing live organisms subsequently used for cultivation experiments, was transferred to 30 mL serum bottles and flushed with Ar to ensure anoxic conditions ( Supplementary Fig. 1). 130 The presence of bacteria in the peat microbial extract was confirmed via SYTO (50 μM) staining and microscopy. The peat microbial extractions were performed in six replicates to ensure sufficient inoculum was available for subsequent incubations.

Synthesis of ferrihydrite and organic matter (OM)-mineral complexes
A water extract of the forest floor (Oa and Oe horizons) of a Podzol under spruce and 2-line ferrihydrite was produced as previously described by Eusterhues, et al. (2014). Briefly, the 2-line ferrihydrite was produced via titration of 0.01 M Fe(NO3)3 135 solution with 0.1 M NaOH to pH 5 with constant stirring. Ferrihydrites with varying concentrations of adsorbed OM (AFh) were produced by mixing forest floor extract solutions containing increasing C concentrations with suspensions of freshly precipitated 2-line ferrihydrite at pH 5. Molar C/Fe ratios of these initial solutions were AFhA 0.4, AFhB 1.3, and AFhD 4.2.
Coprecipitated ferrihydrites (CFh) were produced by dissolving Fe(NO3)3 in forest floor extract solutions of increasing concentrations and adding 0.1 M NaOH under vigorous stirring until the solution reached pH 5. Molar C/Fe ratios of these 140 initial solutions were CFhA 0.4, CFhB 1.3, and CFhD 4.2. The solid/solution (g/L) ratio for all synthesized minerals ranged between 0.3 and 1.6. The solid products were separated using centrifugation, washed two times with deionized H2O, freeze dried, and stored in Balch-type tubes under anoxic conditions (N2). The C content was measured using a Carbon-Nitrogen (CN)-Analyzer (Vario EL, Elementar-Analysensysteme, Hanau, Germany) and the Fe content by Inductively coupled plasma optical emission spectrometry (ICP-OES) after total digestion (Spectroflame, Spectro, Kleve, Germany). A characterization 145 of the forest floor extract and ferrihydrite-OM complexes by Fourier-transform infrared spectroscopy (FTIR) and X-ray photoelectron spectroscopy (XPS) can be found in Eusterhues et al. (2014).

Microbial OM ferrihydrite reduction experiments
We used S. oneidensis MR-1 as model Fe(III)-reducer since it is well-characterized and able to utilize a variety of carbon substrates as electron donors Nealson, 1988, 1990;Venkateswaran et al., 1999). Liquid aerobic pre-cultures were 150 grown in 50 mL of lysogeny broth (Luria Bertani medium; LB medium) containing 10 g tryptone, 10 g NaCl, 5 g yeast extract.
These cultures were incubated on a rotary shaker at 150 rpm and the cell concentration was determined by optical density (OD) measurements at 600 nm. For the Fe(III) reduction experiments 2 ml of LB-grown pre-culture was harvested after 14 h at late exponential growth phase and centrifuged at 10,000 x g for 5 min. Cells were washed twice with a defined medium, resuspended in 2 ml of medium and diluted to a final concentration of 2 x 10 5 cells ml -1 in culture tubes, containing a defined 155 medium. aFor all incubations a defined freshwater medium (pH 6.8; Geobacter medium ATCC 1957) containing 1.5 g L −1 NH4Cl and 0.1 g L −1 KCl, was used. After autoclaving and cooling under anoxic conditions (N2 /CO2 (80 / 20 v/v) atmosphere), 30 mL L −1 of 1 M NaHCO3 (autoclaved, CO2), 10 mL L −1 Wolfe's vitamin solution (ATCC 1957), and 10 mL L −1 modified Wolfe's minerals (ATCC 1957). A combination of filter-sterilized (0.2 µm) electron donors (sodium-acetate (2 mM), sodium lactate (2mM), and glucose (2mM)) were added as carbon sources to all incubations to ensure standardization. NaH2PO4 was 160 not added to circumvent interaction of PO4 3− with ferrihydrite.
In the current study the microbial community extracted from peat cores collected in the Schlöppnerbrunnen fen as well as S. oneidensis MR-1 were used as the inoculum in incubations containing either ferrihydrite, OM-adsorbed ferrihydrite, or OM-co-precipitated ferrihydrite. For OM-adsorbed ferrihydrite incubations, three different concentrations of organic matter were used: 44 mg/g, 105 mg/g. and 181 mg/g total organic carbon (TOC). Similarly, three different concentrations of organic 165 matter were used in OM-co-precipitated ferrihydrite incubations: 44 mg/g, 98 mg/g. and 182 mg/g TOC. hungate tubes were sealed with butyl rubber stoppers and flushed with sterile N2/CO2 (80%/20% v/v) for ~2 h to ensure anoxic 170 conditions. Either peat microbial extract (5% v/v) or S. oneidensis MR-1 were used as inoculum in separate incubations containing ferrihydrite, OM-adsorbed ferrihydrite (TOC: 44 mg/g, 98 mg/g, 182 mg/g), or OM-co-precipitated ferrihydrite (TOC: 44 mg/g, 105 mg/g, 181 mg/g). All treatments were performed in triplicate and incubated horizontally at room temperature in the dark and shaken periodically. Fe(III) reduction was monitored approximately every 24 hours using the phenanthroline assay (Tamura et al., 1974) (data not shown). Fe(II) formation kinetics were used as analogues for Fe(III) 175 reduction. Apparent initial reaction rates were estimated by fitting linear regression lines to Fe(II) [mM]) versus time [hrs] for all data points. The slope of the line represents the initial reaction rate. At the end of incubation (~12 days), solid remnants were freeze dried and stored under N2 until X-ray diffraction (XRD) measurements were performed (see description below).

DNA extraction
Genomic DNA was obtained from the peat microbial extract and from the incubation studies using the PowerSoil DNA 180 isolation kit (MoBio Laboratories, Catalog # 12888-50, Carlsbad, CA, USA) according to manufacturer's instructions and stored at -20°C. The DNA was extracted from three biological replicates at 3 different time points (T0, Tmiddle -136 h, Tend -296 h) from all incubations inoculated with either peat microbial extract or S. oneidensis MR-1. DNA extracts were subsequently used as a template for quantitative PCR (qPCR) reactions targeting bacterial 16S rRNA genes to quantify abundance of bacteria over time or microbial community analysis via Illumina sequencing. Genomic DNA from samples 185 taken at T0 and Tend (296 h) were used for Illumina sequencing, while DNA from all three time points were used as the template for qPCR. ).

Quantitative PCR
Bacterial 16S rRNA gene copy and transcript numbers of the total peat microbial community were determined by quantitative PCR using a Mx3000P instrument (Agilent) with Maxima SYBR Green Master Mix (Fermentas). Bacterial 16S rRNA genes 190 were amplified using primer pairs Bac8Fmod/Bac338Rabc (Daims et al., 1999;Loy et al., 2002) according to cycling conditions and standards previously described (Herrmann et al., 2012). S. oneidensis MR-1 16S rRNA gene copy numbers were determined with primer pairs She12F/She220R (Himmelheber et al., 2009). 2 to 20 ng of genomic DNA was used as template. Standard curves were produced using serial dilutions of representative plasmid mixtures and were linear for both primer sets from 5x10 8 to 5x10 2 copies with R 2 values of 0.999-1.000 and the qPCR performed with efficiencies between 80-195 90%. To monitor temporal changes in transcriptional activity on subsamples from T0, Tmiddle (136 h), and Tend (296 hr), transcript numbers were normalized to copy numbers of the corresponding gene copy number per genome: 4.02 for bacterial 16S rRNA and 9 for S. oneidensis 16S rRNA gene (https://rrndb.umms.med.umich.edu, as retrieved on June 13 th 2017). Oneway analysis of variance (ANOVA) was used to compare treatments using the calculated quantification results and determine statistical significance (p<0.05). 200

Illumina MiSeq Amplicon sequencing
16S rRNA gene-targeting amplicon sequencing was done using a bacteria-specific primer pair (341f [S-D-Bact-0341-b-S]/805r [S-D-Bact-0785-a-A]) (Klindworth et al., 2013). Amplicon sequencing was performed by LGC Genomics GmbH (Berlin, Germany). PCR products were generated starting from about 5 ng of extracted genomic DNA extract, 15 pmol of forward and reverse primer in 20 μL of 1 × MyTaq buffer containing 1.5 units MyTaq DNA polymerase (Bioline, City, Country) and 2 μl 205 of BioStabII PCR Enhancer (Sigma-Aldrich, Munich, Germany). For each sample, the forward and reverse primers had the same 10-nt barcode sequence. PCRs were carried out for 30 cycles using the following parameters: 2 min 96°C predenaturation; 96°C for 15 s, 50°C for 30 s, 70°C for 90 s. DNA concentrations of amplicons of interest were quantified semiquantitatively via gel electrophoresis. ~20 ng amplicon DNA of each sample were pooled for up to 48 samples carrying different barcodes. If needed PCRs showing low yields were further amplified for an additional 5 cycles. The amplicon pools 210 were purified with one volume AMPure XP beads (Beckmann-Coulter, Krefeld, Germany) to remove primer dimer and other small mispriming products, followed by an additional purification on MinElute columns (Qiagen, Dusseldorf, Germany).
About 100 ng of each purified amplicon pool DNA was used to construct sequencing libraries using the Ovation Rapid DR Multiplex System 1-96 (NuGEN, Leek, Netherlands). Illumina libraries were pooled and size selected by preparative gel electrophoresis. Sequencing was done on an Illumina MiSeq using V3 chemistry (Illumina) in paired-end mode (2 x 300 bp). 215 Amplicon sequencing was performed genomic DNA samples taken at T0 and Tend (296 hr).

Microbiome profiling
Sequence characteristics were checked using prinseq (v. 0.20.4) (Schmieder and Edwards, 2011). Adaptor sequences still present were subsequently removed with cutadapt (v. 1.7) (Martin, 2011). Paired-end assembly and subsequent quality control were done using vsearch (v. 1.10.2) (Rognes et al., 2016) using the implemented -fastq_mergepairs (-fastq_minmergelen = 220 expected amplicon size) and -fastq_filter (-fastq_maxee = 0.5) functions. Chimera removal was done with uchime (v. 4.2.40) (Edgar et al., 2011). OTU clustering with a defined threshold was omitted, instead sequences were dereplicated to streamline the dataset. Deduced dereplicated sequences were treated as being representative for sequence populations featuring an identity of 100% (in the following referred to as OTU). After alpha diversity analyses, the dataset was stripped by removing OTUs that were represented by less than three sequences. Relative abundances were calculated and standardized based on median 225 sequencing depth across all samples. Datasets were not rarefied to prevent a loss of sensitivity in downstream sample-wise comparisons (McMurdie and Holmes, 2014). Sequences from this simplified dataset were taxonomically assigned against SILVA (release 123, as of September 2016) using the uclust algorithm (Edgar, 2010) implemented in vsearch applying a minimum similarity of 90%. An OTU table including determined taxonomic affiliations and available metadata was assembled for downstream analysis in qiime (v. 1.9.2) (Caporaso et al., 2010) and phyloseq (v. 1.10) (McMurdie et al., 2013). Statistical 230 analyses were carried out using the R statistical software framework (R core team, 208) and the following packages, including respective dependencies, phyloseq, vegan (v. 2.4-3) (Oksanen et al., 2017) , and DESeq2 (v. 1.10.1) (Love et al., 2014).
Differentially abundant taxonomic groups were identified in sample-wise comparisons assuming negative binomial distributions for taxa read count data and using Wald significance tests, and parametric fitting for determined dispersions.
These reflect biological variation better as for instance Poisson distributions (Love et al., 2014) and thus prevent overestimation 235 due to false-positives.

Sequence data deposition
Generated amplicon sequencing data was deposited as raw data at NCBI SRA and are available under the following BioProject accession number: PRJNA391563

Microbial Fe(III) reduction by Shewanella oneidensis MR-1 and by the microbial consortia extracted from peat
We compared the Fe(III) reduction capacity of the model organism S. oneidensis MR-1 with a peat microbial consortia extracted from an anoxic iron-rich peatland for incubations containing pure ferrihydrite (Fh control), coprecipitated OMferrihydrite complexes (CFhA (44 mg/g C), CFhB (98 mg/g C), CFhD (182 mg/g C)), or adsorbed OM-ferrihydrite complexes 250 (AFhA (44 mg/g C), AFhB (105 mg/g C), AFhD (181 mg/g C)). The Fe(III) reduction patterns of S. oneidensis MR-1 and the peat microbial consortia were nearly identical. Rates of Fe(III) reduction in incubations with the peat microbial consortia and S. oneidensis MR-1 increased with increasing TOC concentrations (and C/Fe ratios) regardless of whether adsorbed OMferrihydrite complexes or coprecipitated OM-ferrihydrite complexes were used as the terminal electron acceptor (Fig. 1, Table   1). The lowest Fe(III) reduction rate was observed in S. oneidensis MR-1 incubations with pure ferrihydrite with a reduction 255 rate of 0.022 mmol hr -1 compared to 0.031 mmol hr -1 for incubations with our peat microbial consortia (Fig. 1, Table 1). The highest Fe(III) reduction rates were observed in incubations containing coprecipitated OM-ferrihydrite complexes, with increasing rates of reduction corresponding to increased C/Fe ratios (from 44 to 182 mg g C). Similar patterns occurred in incubations with adsorbed OM-ferrihydrite complexes, such that the higher the C content the higher the rate of Fe(III) reduction (from 44 to 181 mg g C). For samples with the same C content rates of Fe(III) reduction were slightly higher in 260 incubations with peat microbial extract compared to S. oneidensis MR-1 (0.037-0.058 mmol hr -1 and 0.032-0.051 mmol hr -1 , respectively) ( Fig. 1, Table 1). At the same time, rates of Fe(III) reduction were slightly higher for OM-ferrihydrite coprecipitates compared to OM-ferrihydrite adsorption complexes (0.038-0.058 mmol hr -1 compared to 0.032-0.050 mmol hr -1 ) when all incubations are considered (Fig. 1, Table 1).

Quantitative changes of 16S rRNA genes in incubation experiments 265
We followed the abundance of bacterial 16S rRNA gene copies over time in incubations inoculated with either S. oneidensis MR-1 or the microbial consortia extracted from peat to determine potential effects of coprecipitated or adsorbed OMferrihydrite complexes on microbial abundances. In general, bacterial 16S rRNA copy numbers increased over time in all incubations with noticeable increases in gene abundances at the end of incubation and slight variations dependent on the TOC concentration of the OM coprecipitated or adsorbed to the surface of ferrihydrite (Fig. 2). In incubations with pure ferrihydrite, 270 S. oneidensis MR-1 16S rRNA gene copy numbers increased four orders of magnitude from 7.90 x 10 7 cells mL -1 at T0 to 1.99 x 10 11 cells mL -1 at Tend (296 h). S. oneidensis MR-1 16S rRNA gene copy numbers increased approximately five orders of magnitude in incubations with coprecipitated or adsorbed OM-ferrihydrite complexes with the highest C/Fe ratios (8.10 x 10 7 cells mL -1 at T0 to 8.47 x 10 12 cells mL -1 at Tend and 8.10 x 10 7 cells mL -1 at T0 to 1.24 x 10 12 cells mL -1 at Tend, respectively).
In incubations with pure ferrihydrite and the peat microbial consortia, bacterial 16S rRNA gene copy numbers only increased 275 two orders of magnitude from 6.23 x 10 10 cells mL -1 at T0 to 2.02 x 10 12 cells mL -1 at Tend. Bacterial 16S rRNA gene copy numbers increased roughly three orders of magnitude in incubations using the peat microbial consortia as inoculum and adsorbed OM-ferrihydrite complexes with the highest C/Fe ratios (6.23 x 10 10 cells mL -1 at T0 to 2.58 x 10 13 cells mL -1 at Tend), however, the gene copy numbers increased approximately four orders of magnitude in incubations with coprecipitated OMferrihydrite with the highest C/Fe ratios (6.23 x 10 10 cells mL -1 at T0 to 1.36 x 10 14 cells mL -1 at Tend, respectively). Thus, 280 increases in gene copy numbers correlate to the increased Fe(III) reduction rates observed across all incubations with OMferrihydrite complexes and pure ferrihydrite.

Microbial community profiling of the microbial consortia in incubation experiments
Incubations with absorbed and coprecipitated OM-ferrihydrite and varying C/Fe ratios led to distinct shifts in microbial community structure when compared to pure ferrihydrite incubations and the initial microbial consortia extracted from peat 285 (Fig. 3, Fig. 4). Alpha diversity indices addressed intra-sample diversity either based on the observed number of taxa according to the applied OTU definition (observed OTUs) or based on extrapolations of observed taxa numbers putting special emphasis on rare taxa (Chao1). Comparing the different incubations with the initial peat microbial extract (Observed OTUs: 52064, Chao1: 801599except AFhB) revealed decreased diversities in all cases (Observed OTUs: 23120-46770, Chao1: 334438-545083, ACE: 266190-455099) except AFhB (Fig. 4a). Intra-sample diversity measures were complemented by assessing the 290 evenness of datasets by determining the entropy of respective datasets (Shannon). Respective values matched observed intrasample diversities. Calculated Shannon indices showed strongly increased evenness for the initial peat microbial extract (Shannon: 9.47) against all incubations (Shannon: 6.77-7.48), excluding AFhB. Beta diversity analysis is a common approach to gain insights about inter-sample diversity when dealing with complex datasets such as microbial community profiles (Fig.  4b). Corresponding analyses based on Jensen-Shannon divergence highlighted a minor clustering of microbial communities 295 linked to coprecipitated OM-ferrihydrite incubations in case of low and medium C/Fe ratios (CFhA, CFhB). Community profiles originating from incubations supplemented with OM-ferrihydrite having a high C/Fe ratio revealed a clustering independent of ferrihydrite complexation (AFhD, CFhD). The observation of community profiles from incubations clustered distinctly from communities present in the initial peat microbial extract matched the abovementioned differences with respect to intra-sample diversity and evenness (Fig. 4a). 300 Looking at underlying community profiles from a taxonomic perspective revealed a dominance of few phyla when considering determined community profiles as a whole (Fig. 3a). Dominant phyla included especially Gammaproteobacteria (57.9-95.4% relative abundance), Firmicutes (3.1-41.6%), and to a lesser extent Bacteroidetes (up to 0.6% [AFhD, CFhD]) and Acidobacteria (4.5% [peat microbial extract]). Looking at the taxonomic composition at a more resolved level (familylevel) showed that carried out incubations led in all cases to a strong increase of Pseudomonadaceae in relative abundance 305 within the Gamma-Proteobacteria (e.g. AFhA 58.1%, in comparison to peat microbial extract 5.3%) (Fig. 3b) while incubations with high C/Fe ratios favoured elevated relative abundances of Enterobacteriaceae, when compared with low and medium C/Fe ratios. Groups that appeared to be enriched in the incubation with pure ferrihydrite, for example Burkholderiaceae, were found to show only minor abundances in any of the other setups.
Considering the results from beta diversity analysis and community profiling, we were interested in assessing the differential abundance of genus-level taxonomic groups to check for the potential enrichment of taxonomic groups that are 315 linked to iron cycling when comparing the pure ferrihydrite incubation and OM-ferrihydrite incubations. Differential abundances were examined by assuming negative binomial distributions for taxa read counts (Fig 3c). It was apparent that numerous genus-level taxonomic groups were underrepresented when CFh and AFh are contrasted against Fh. This was consistent with reduced intra-sample diversities for adsorbed and co-precipitated incubations in comparison to the pure ferrihydrite incubation (Fig. 4a). Most underrepresented genus-level taxonomic groups were linked to Proteobacteria and 320 especially Enterobacteriales/Enterobacteriaceae including: Ewingella, Rahnella, Serratia, and Yersinia. This pattern was similar for CFh against Fh, as well as AFh against Fh. In most cases log2 fold changes ranged between -5 and -10 for underrepresented groups. The number of overrepresented taxonomic groups was in general lower than the number of underrepresented groups. This finding was much more pronounced for AFh vs Fh than CFh vs Fh. Only three overrepresented genus-level taxonomic groups were identified for AFh vs Fh, namely Pelosinus, Pseudomonas, and Telmatospirillum with 325 log2 fold changes above 5. The same groups were also identified to be enriched in case of CFh vs Fh. However, overall the number of enriched taxonomic groups was higher including various Firmicutes such as Bacillus, Clostridium, Lachnoclostridium, and Paenibacillus. Few genus-level groups comprised over-and underrepresented OTUs. This was the case for Bacillus, Paenibacillus, and Pseudomonas. In the case of Pseudomonas, one particular OTU was found to be enriched.
Consequently, this indicates that pronounced increases in the abundance of Pseudomonaceae in adsorbed and co-precipitated 330 OM-ferrihydrite are apparently mediated by a well-defined sub population of Pseudomonas.

XRD measurements
X-ray diffraction patterns of the starting material showed the two very broad peaks of ferrihydrite (Fe10O14(OH)2). After reduction, diffractograms of the solid remnants still revealed the presence of ferrihydrite but additionally showed sharp peaks of different salts such as halite (NaCl), nahcolite (NaHCO3), and salammoniac (NH4Cl), which derived from the medium. 335 Goethite (α-FeOOH), other Fe(III) oxides or Fe(II) minerals were not found. The newly formed Fe(II) was either effectively chelated by the present organic material or dissolved Fe concentrations were too low to exceed the solubility product of Fe oxides or other common secondary Fe(II) minerals. The latter seems likely as the reduction rates of the present experiment were in fact low (0.022 -0.058 mM h -1 ; Table 1). In previous experiments, using the same OM-ferrihydrites, the same medium,

Discussion 350
The ability of OM to serve as a terminal electron acceptor for microbial respiration in anoxic environments is of special importance and has been well characterized (Lovley et al., 1996). However, most experiments utilizing OM as a terminal electron acceptor have been performed with alkaline extracts: Reduced humic acids have been shown to transfer electrons abiotically to terminal electron acceptors, such as metals, thereby functioning as electron shuttles between microbes and electron acceptors (Kappler et al., 2004). Humic acids, therefore, possess a redox-mediating ability which positively affects 355 the cycling of redox-active minerals, including Fe(III) and Mn(IV)/Mn(III), in conjunction with transformation of organic and inorganic compounds. Subsequent studies have shown that reduced forms of humic substances can serve as electron donors for anaerobic organisms growing on a variety of terminal electron acceptors, including nitrate and fumarate (Coates et al., 2002). Under these conditions, microorganisms utilize the reduced humic substances as an energy source, while using a readily degradable carbon source, such as acetate, for growth. This unique ability to use humic substances as an electron donor 12 provides a possible competitive advantage over other heterotrophs in the environment which are only capable of utilizing a single organic compound as both the carbon and energy source. Previous studies have revealed that some Fe(III)-reducing microorganisms, such as S. alga, G. metallireducens, and Geothrix fermentans, have the ability to undergo dissimilatory humic reduction as well as the ability to couple reduced humic oxidation with nitrate reduction. Moreover, Paracoccus denitrificans, a known denitrifier, is unable to utilize humics or Fe(III) as a terminal electrons acceptors, however, these microorganisms are 365 capable of using reduced humics as an electron donor for denitrification (Coates et al., 2002;Lovley et al., 1999).
This Schlöppnerbrunnen fen has been studied in great detail during the last 10 years, however, the impact of OM on the microbially-mediated Fe(III) reduction processes in this site are not fully understood. Earlier incubation studies with peat soil from the Schlöppnerbrunnen fen demonstrated that amendment with each of these individual substrates (lactate, acetate, glucose) stimulates Fe(III) reduction, with glucose showing the most rapid increase of Fe(II) production (Küsel et al., 2008). 370 Glucose was completely consumed within 3 days of incubation, yielding acetate and CO2 as end products, and H2 and ethanol as transient products. Supplemental lactate yielded acetate as a main product (Küsel et al., 2008). Both acetate and H2 are preferred substrates for many Fe(III)-reducing microorganisms, including Shewanella and Geobacter species. Thus, amendment of incubations containing a microbial consortia extracted from Schlöppnerbrunnen fen with a substrate mixture comprised of the carbon sources used in the aforementioned study (i.e. lactate, acetate, glucose) and ferrihydrite as the terminal 375 electron acceptor should augment growth of many peat microorganisms involved in different steps of anaerobic organic matter degradation using Fe(III) as terminal electron acceptor.
The current study provides clear evidence that the microbial community found in the fen not only has the potential to reduce Fe(III)-oxides complexed with organic matter to the same extent as the model Fe(III)-reducer, S. oneidensis MR-1, but the microbes also increase in abundance under the experimental growth conditions used. We utilized a mixture of electron 380 donors (glucose, lactate, and acetate) to stimulate a great variety of peat microorganisms capable of dissimilatory Fe(III) reduction including those that couple Fe(III) reduction with growth, but also fermenting microorganisms and others that just use Fe(III) oxides as electron sink (Hori et al., 2015;Lentini et al., 2012). Differences in the gene copy numbers at the end of the incubation suggested that not only did the supplemented substrates support growth but also OM adsorbed or coprecipitated to the surface of the ferrihydrite minerals. Based on these results, we can infer that the mineral associated OM was at least 385 partially degradable and not totally protected by its association to minerals or by its molecular structure and size. Furthermore, it is likely that a portion of the ferrihydrite associated OM became re-dissolved over the course of the incubation period, which may have facilitated its utilization by the microbes. Thus, the enhanced growth observed in both the S. oneidensis MR-1 and the microbial consortia can be attributed to the presence of additional complex carbon sources in the form of OM-ferrihydrite complexes compared to pure ferrihydrite.

Mechanisms of OM-ferrihydrite reduction by S. oneidensis MR-1 and the microbial consortia
The results of this study demonstrate that OM stimulates growth and Fe(III) reduction by S. oneidensis MR-1 and the microbial consortia. Fe(II) production increased as the C/Fe ratios increased in all incubations. Both the reduction rates and the total amount of Fe(II) produced at the end of incubation (~12 days) increased in the order Fh < adsorbed OM-ferrihydrite < 395 coprecipitated ferrihydrite. (Fig. 1, Fig. 5, and Table 1). Samples containing coprecipitated OM-ferrihydrite complexes were more reactive than samples containing adsorbed OM-ferrihydrite complexes with similar OM loadings. The latter can be explained by a generally smaller crystal size of OM-ferrihydrite coprecipitates compared to pure ferrihydrite (Mikutta et al., 2008;Eusterhues et al., 2008) and was also observed during reduction by Geobacter bremensis (Eusterhues et al., 2014). The finding that mineral associated OM may enhance ferrihydrite reduction supports the results presented in previous studies 400 examining the role of humic acids as potential electron shuttles by S. putrefaciens under Fe(III) reducing conditions, which showed that S. putrefaciens was capable of using humic acids as an electron shuttle to enhance Fe(III) reduction (Amstaetter et al., 2012;Shimizu et al., 2013). However, enhanced Fe(III) reduction was only observed when the concentration of organic material was high enough to overcome any slowing in the reduction processes caused by OM sorption to the surface reduction sites of the mineral (Amstaetter et al., 2012;Shimizu et al., 2013). Our study revealed that a water extract of natural soil organic 405 matter, which has lower contents of aromatic moieties, and usually lower electron shuttling capacity tohumic acids ( Piepenbrock et al., 2014) can also provide enough electron shuttling compounds to stimulate ferrihydrite reduction by S. oneidensis as well as by the peat microbial consortium.
Furthermore, since the S. oneidensis MR-1 pre-culture used in the present study was grown aerobically and the peat microbial consortia was resuspended in salt solution, we can assume no electron shuttling molecules were secreted by the 410 microorganisms themselves prior to inoculating the incubation experiments. Thus, the absence of any electron shuttles in our pure ferrihydrite incubations might explain the lowest Fe(III) reduction rates out of all the experiments conducted. On the contrary, we suggest the presence of electron shuttling compounds, such as quinones, in the incubations with coprecipitated or adsorbed OM. Quinone moieties have been described as electron shuttling compounds in previous studies (Scott et al., 1998).
In our experiments even small contents of OM on ferrihydrite (44 mg g -1 , i.e., small molar C/Fe ratios of 0.4) lead to increased 415 reduction rates. This is in contrast to Shimizu et al. (2013) and Adhikari et al. (2017), who reported surface passivation at low C/Fe ratios, but increased reduction rates at molar ratios C/Fe > 1.8. A study by Poggenburg et al. (2016) reported enhanced Fe(III) reduction rates in incubations containing OM-ferrihydrite complexes with molar C/Fe ratios of 0.9 -2.1 and using S. putrefaciens as model reducer. Here reduction rates differed mostly due to OM-composition, i.e. aromaticity. Lowest reduction rates were found for ferrihydrite coprecipitates produced with extracellular polymeric substances, while highest reduction rates 420 were found for ferrihydrite coprecipitates produced with a water extract from Podzol litter. The latter is likely comparable in composition to the OM of our experiment. However, absolute reduction rates cannot be compared directly because Poggenburg et al. (2016) attributed their high rates of Fe(III) reduction partly to the desorption of NOM from coprecipitates due to large concentration of phosphate in their nutrient solutions. In the present study, our C/Fe molar ratios ranged from 0.4 OM-ferrihydrite complexes compared to incubations with pure ferrihydrite.
In Geobacter sp., direct contact with oxide surfaces is required for Fe(III) reduction (Nevin and Lovley, 2000).
Geobacter produces nanowires, typically described as conductive extracellular appendages, which facilitate electron transfer by serving as an electrical conduit to the Fe(III) oxide surface (Alves et al., 2016;Reguera et al., 2005). Conversely, Shewanella species are capable of using three primary strategies to allow for electron transfer between itself and solid Fe(III) 430 oxide surfaces: via direct contact with the oxide surfaces, production of a complexing ligand (i.e. siderophore), and the utilization of endogenously or exogenously produced electron shuttles (Cooper et al., 2016;Melton et al., 2014;Shi et al., 2012). A previous study by Eusterhues, et al. (2014) showed the rates of Fe(III) reduction by G. bremensis decreased as the concentrations of OM adsorbed or co-precipitated to the surface of ferrihydrite increased (Eusterhues, et al. 2014). However, under identical reduction experiments and the same ferrihydrite material, Fe(III) reduction by G. bremensis resulted in 435 considerably higher reduction rates of (0.088-0.148 mM h -1 ) compared to S. oneidensis (0.022-0.051 mM h -1 ) and the microbial consortia (0.031-0.058 mM h -1 ) ( Table 1, Eusterhues et al., 2014). Although favoured at high concentrations of ferrihydriteassociated OM, electron shuttling is slower than direct electron transfer via nanowires. Similar results were reported in a study by Poggenburg et al. (2016), in which the rates of Fe(III) reduction by G. metallireducens were greatest with pure ferrihydrite compared to ferrihydrite coprecipitates produced with different natural organic materials. These findings suggest that the 440 reduction rates are not only influenced by organic matter composition and the C/Fe molar ratios, instead they also reflect species dependent capabilities. It was hypothesized that OM may passivate Fe(III) oxide surfaces for direct contact requiring reducers such as Geobacter, whereas it may promote reduction by providing exogenous electron shuttles for non-direct contact requiring reducers such as Shewanella sp. (Eusterhues et al., 2014). In the current study, we observed a similar behaviour for S. oneidensis and the complex microbial consortia extracted from peatland suggesting that non direct-contact requiring 445 microorganisms dominate reduction in the anoxic, Fe-rich fen, making use of the exogenous electron shuttling compounds of the natural OM (Fig. 5). The similarity of the patterns observed point to the suitability of S. oneidensis as model organism to study the effect of OM on ferrihydrite reduction in natural environments.

Potential key players of OM-ferrihydrite reduction
Though the microbial consortia behaves similarly to S. oneidensis MR-1 with regards to Fe(III) reduction capacities and growth 450 over time, the community profiles at the end of incubation did not reveal an enrichment of Shewanella species. Differential abundance analyses between adsorbed OM-ferrihydrite incubations and the pure ferrihydrite incubation even revealed reduced abundances for Shewanella in case of adsorbed OM-ferrihydrite incubations. The enriched community is indeed comprised of microorganisms capable of reducing Fe(III) oxides., However, Shewanella sp. but also and Geobacter sp. did not respond to these growth conditions and are noticeably low abundance. But the lack of enrichment of typical dissimilatory Fe(III)-reducer 455 cannot be explained by the lack of suitable electron donors. Previous work using 13 C-labelled acetate in paddy soils revealed that especially Geobacter-related organisms respond to acetate amendment (Hori et al., 2010).
We saw distinct changes in microbial community structure over time as reflected by alpha-and beta diversity analyses which appeared to be linked to several main drivers shaping the microbial communities in the incubations. The unexpectedly high alpha-diversity observed for the adsorbed OM-ferrihydrite incubation featuring a medium C/Fe ratio (AFhB) is a result 460 of over-estimating diversity, due to a high frequency of singletons (OTUs that are only represented by single sequences). The overall anaerobic incubation conditions with ferrihydrite offered as terminal electron acceptor and a mixture of the carbon substrates glucose, lactate, and acetate activated only a subset of the original community extracted from peat. Fermentative Enterbacteriaceae (Ewingella, Rahnella, Serratia) were overrepresented in the pure ferrihydrite incubation which is reflective of their ability to use glucose as carbon source and available Fe(III) as electron sink. Enterbacteriaceae are known for a mixed 465 acid fermentation profile similar to the products observed in earlier peat soil incubation studies amended with glucose (Reiche et al., 2008). The question of whether Fe(III) can be also used for energy conservation by these bacteria remains unanswered.
Early work by Ottow (1970) suggested Serratia to be involved in hematite reduction. The strong reduction in their relative abundance in OM-ferrihydrite incubations is presumably a result of Fe(III) being less accessible.
Community profiles of OM-ferrihydrite incubations differed tremendously from the one of the pure ferrihydrite 470 incubation. Pseudomonadaceae were highly enriched in all OM-ferrihydrite incubations. The enrichment of Veillonellaceae in adsorbed and coprecipitated OM-ferrihydrite incubations with low C concentrations could be mostly attributed to Pelosinus.
Pelosinus sp. have been characterized as fermentative bacteria which utilize Fe(III) as an electron sink (Shelobolina et al., 2007). These bacteria can use H2 as an electron donor and acetate as a carbon source, and are also capable of fermenting citrate, lactate, and organic impurities. Pelosinus has also been shown to reduce small concentrations of anthraquinone-2.6-disulfonate 475 (AQDS), a synthetic quinone, in the presence of citrate or lactate (Shelobolina et al., 2007). The ability of these microorganisms to use AQDS as an electron sink under reducing conditions might explain their enrichment during our incubations. Quinone moieties are ubiquitous in soils and peat, thus providing the perfect breeding ground for these microorganisms.
Interestingly, Fe(III)-reduction was previously found to be correlated with the presence of Desulfovibrionaceae, when 480 lactate is available as carbon source (Lentini et al., 2012). This partially matches our observations for incubations with coprecipitated ferrihydrite where Desulfovibrio was enriched at low and medium C/Fe ratios. However, Desulfovibrio were nearly absent in adsorbed OM-ferrihydrite incubations. Differences in Fe(III) accessibility in adsorbed and coprecipitated OMferrihydrite or the higher solubility of coprecipitates might explain this difference in community structure. The presence and ecological role of sulfate-reducing bacteria in the Schlöppnerbrunnen fen has been documented several times (Loy et al., 2004;485 Pester et al., 2010). This acidic (pH 4), low-sulfate (20-200 µM) peatland shows significant sulfate-reducing capacities (Loy et al., 2004) which is explained by a rapid recycling mechanism (Küsel et al., 2008). Recycling of sulfate can proceed by the aerobic oxidation of sulfide, e.g., in regions where oxygen penetration and anoxic micro-niches overlap (Knorr and Blodau, 2009). The sulfate-reducing activity is mainly attributed to a member of the 'rare biosphere' in the peat, a Desulfosporosinus species (Pester et al., 2010). Our study suggests that other members of the sulfate reducers are involved in Fe(III) reduction

Summary and Conclusions
In the present study mineral-bound soil OM is shown to increase microbial Fe(III) reduction by S. oneidensis MR-1 and by a microbial consortia extracted from Schlöppnerbrunnen fen peat cores. The reactivity of ferrihydrites with adsorbed OM differed from ferrihydrites with coprecipitated OM, such that higher extents of Fe(III) reduction were observed for 495 coprecipitated ferrihydrites in comparison to adsorbed ferrihydrites. This can be explained by the smaller crystal size and more defective crystal structure of coprecipitated ferrihydrites compared to normal ferrihydrites . The fact that reduction rates for both S. oneidensis MR-1 and the microbial consortia increased systematically with increasing contents of OM in the OMferrihydrite complexes tempts us to speculate about how reduction rates are related to electron transfer mechanisms.
Shewanella sp. are capable of reducing Fe(III) oxides using one of three mechanisms known for reductions, including direct 500 contact, electron shuttles, or production of complexing ligands. Thus, the enhanced Fe(III) reduction rates led us to the conclusion that electron shuttling using exogenous moieties of the mineral-bound OM is likely the preferred mechanism used for reduction (Fig. 5). Since coprecipitation of OM to mineral surfaces is a common process in nature, thus, we can hypothesize that electron shuttling mechanisms are the dominant mechanism for microbial Fe(III) reduction in environments with high concentration of OM, such as soils, sediments, and peat. 505 Experiments with the microbial consortia extracted from peat showed that the presence of mineral associated OM Thus, our observations imply OM coverage does not protect the underlying Fe mineral from microbial reduction in soils dominated by microorganisms that predominantly use electron shuttles or ligands for Fe(III) reduction. This study also revealed that natural soil OM on the surface of ferrihydrites provides enough redox active groups to maintain microbial Fe(III) 515 reduction processes, instead of passivating the surface. While at the same time the OM enhanced overall microbial growth and shaped the microbial community structure.
Data availability. Generated amplicon sequencing data can be found as raw data at NCBI SRA and are available under the following BioProject accession number: PRJNA391563. 520   or adsorbed OM-ferrihydrite (AFh) (D). Each sample was measured in triplicate qPCR reactions with applicable standard 805 deviations. One-way ANOVA was applied to compare individual datasets for statistical significant differences between ferrihydrite incubations. Different letters (a-d) indicate statistically significant differences, while horizontal black lines indicate non-significant differences.  OTUs refers to OTU counts based on the OTU definition given in the materials and methods. Chao1 (Chao, 1984) is an index 855 that puts special emphasis on rare taxa to extrapolate the real number of OTUs based on observed OTUs. The Shannon index (Shannon, 1948) describes the entropy in a community profile, and reaches its maximum when taxonomic groups are evenly distributed in a dataset. Beta diversity (= differences in diversity between samples) analysis was based on Jensen-Shannon divergences (Koren et al., 2013)