bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1 Genome-resolved metagenomics and detailed geochemical speciation analyses yield 2 new insights into microbial mercury cycling in geothermal springs 3 4 RUNNING TITLE: Microbial mercury cycling in acid warm springs 5 6 Caitlin M. Gionfriddo1,+*, Matthew B. Stott2, #, Jean F. Power2, Jacob M. Ogorek3, David 7 P. Krabbenhoft3, Ryan Wick4, Kathryn Holt4, Lin-Xing Chen5, Brian C. Thomas5, Jillian 8 F. Banfield1, 5, 6 and John W. Moreau1, ‡ 9 10 1School of Earth Sciences, The University of Melbourne, Parkville, VIC, Australia 11 2GNS Science, Wairakei Research Centre, Taupō, New Zealand 12 3Wisconsin Water Science Center, U.S. Geological Survey, Middleton, WI, USA 13 4Department of Biochemistry and Molecular Biology, Bio21 Molecular Science and 14 Biotechnology Institute, University of Melbourne, Victoria 3010, Australia 15 5Department of Earth and Planetary Science, UC Berkeley, Berkeley, CA, USA 16 6Department of Environmental Science, Policy, and Management, University of California, 17 Berkeley, California 94720, United States 18 19 20 21 22 23 Current affiliations: 24 +Oak Ridge National Laboratory, Biosciences Division, Oak Ridge, TN, USA 25 # School of Biological Sciences, University of Canterbury, Christchurch, New Zealand 26 ‡ School of Geographical and Earth Sciences, University of Glasgow, UK 27 28 *Corresponding author: [email protected] 29

1 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

30 ABSTRACT 31 Geothermal systems emit substantial amounts of aqueous, gaseous and methylated 32 mercury, but little is known about microbial influences on mercury speciation. Here we 33 report results from genome-resolved metagenomics and mercury speciation analysis of 34 acid warm springs in the Ngawha Geothermal Field (<55 °C, pH < 4.5), Northland Region, 35 Aotearoa (New Zealand). Our aim was to identify the microorganisms genetically equipped 36 for mercury methylation, demethylation, or Hg(II) reduction to volatile Hg(0) in these 37 springs. Dissolved total and methylated mercury concentrations in two adjacent springs 38 with different mercury speciation ranked among the highest reported from natural sources 39 (250–16000 ng L-1 and 0.5–13.9 ng L-1, respectively). Total solid mercury concentrations 40 in spring sediments ranged from 1273 to 7000 µg g-1. In the context of such ultra-high 41 mercury levels, the geothermal microbiome was unexpectedly diverse, and dominated by 42 acidophilic and mesophilic sulfur- and iron-cycling bacteria, mercury- and arsenic-resistant 43 bacteria, and thermophilic and acidophilic archaea. Integrating microbiome structure and 44 metagenomic potential with geochemical constraints, we constructed a conceptual model 45 for biogeochemical mercury cycling in geothermal springs. The model includes abiotic and 46 biotic controls on mercury speciation, and illustrates how geothermal mercury cycling may 47 couple to microbial community dynamics and sulfur and iron biogeochemistry. 48 49 IMPORTANCE 50 Little is currently known about biogeochemical mercury cycling in geothermal systems. 51 This manuscript presents an important new conceptual model, supported by genome- 52 resolved metagenomic analysis and detailed geochemical measurements. This work 53 provides a framework for studying natural geothermal mercury emissions globally. 54 Specifically, our findings have implications for mercury speciation in wastewaters from 55 geothermal power plants and the potential environmental impacts of microbially and 56 abiotically formed mercury species, particularly where mobilized in spring waters that mix 57 with surface- or ground-waters. Furthermore, in the context of thermophilic origins for 58 microbial mercury volatilisation, this report yields new insights into how such processes 59 may have evolved alongside microbial mercury methylation/demethylation, and the

2 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

60 environmental constraints imposed by the geochemistry and mineralogy of geothermal 61 systems.

3 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

62 Introduction 63 Geothermal springs and fumaroles emit substantial amounts of aqueous and gaseous 64 mercury (Hg) (1). Aqueous Hg concentrations in these systems often exceed 100 ng L-1 65 and total Hg levels can approach 25 µg L-1 (2-4). Despite these ultra-high mercury levels, 66 few studies have examined biotic and abiotic mechanisms for Hg transformations or Hg 67 speciation in geothermal springs. Specifically, the potential for native thermophiles to 68 mediate mercury transformations, i.e., reduction of Hg(II) to Hg(0) or

+ 69 methylation/demethylation of mercury (to CH3Hg or “MeHg”; or to Hg[II], respectively; 70 (5-7) remains poorly understood. 71 72 Hg species have a high binding affinity to thiols and lipids, causing damage to proteins, 73 enzymes, and nucleic acids. They can inhibit microorganisms at sub-micromolar 74 concentrations (8). Microorganisms living in environments with elevated Hg (>100 ng L- 75 1) commonly possess genes encoding for reduction of Hg(II) to volatile Hg(0), as a means 76 of Hg detoxification (9, 10). The mer operon is used by many bacteria and archaea to 77 detoxify Hg(II), by converting it to volatile Hg(0) (7, 10, 11). Interestingly, mer has a 78 phylogenetic origin in the thermophiles (12, 13). Additionally, microbes equipped with the 79 organomercurial lyase, encoding merB gene as part of the mer-operon, are able to detoxify 80 organic Hg compounds, in tandem with mercuric reductase (MerA), to produce methane

81 (CH4) and Hg(0) (9, 14). Finally, some anaerobic bacteria are suspected to be able to 82 oxidatively demethylate Hg species independent of the mer pathway, producing carbon

83 dioxide (CO2) and Hg(II) (15), although a biochemical pathway has not been identified for 84 mer-independent demethylation. 85 86 The efficiency of microbial methylation of Hg(II) to MeHg is still largely unknown in 87 geothermal spring ecosystems, particularly under acidic conditions (<55 °C, pH < 4.5; (4, 88 5, 16)). However, the hgcAB genes required by microorganisms to methylate Hg (17) have 89 been reported from many environments, including wetland sediments, rice paddy soils, 90 thawing permafrost, hypersaline and hypersulfidic waters, soda lakes, and geothermal 91 systems (16, 18-20). These environments typically host abundant sulfate- and iron- 92 reducing bacteria (Deltaproteobacteria), as well as methanogenic and acetogenic

4 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

93 Methanomicrobia (Euryarchaeota), Chloroflexi, and Firmicutes, all of which contain 94 species capable of Hg methylation (18, 21-23). Metagenomic analyses have also identified 95 hgcAB genes in Chrysiogenetes, Atribacteria (candidate phylum OP9), and candidate 96 phylum ACD79 (16). Although Hg methylation is most often associated with sulfate 97 reduction, the existence of environmental triggers or controls on Hg methylation remains 98 poorly understood, as does the evolution and phylogenetic distribution of hgcAB genes. 99 100 By comparison, in non-geothermal environments, elevated Hg concentrations can inhibit 101 microbial Hg methylation (24), and lead to conditions favoring MeHg demethylation (25) 102 or methylation-demethylation cycles (26-28). However, in-depth understanding of Hg 103 methylation in geothermal springs, similarly to acid mine drainage (AMD) (29), has not 104 been established. Conversely, acidophiles in AMD systems have been well studied with 105 respect to potential for metal and sulfur cycling (30), but with less focus on Hg speciation 106 and methylation. Here, we combined metagenomic and geochemical speciation analyses 107 to understand Hg transformations in the context of biogeochemical cycling (e.g., S, Fe) in 108 an acidic warm spring microbiome. In order to understand physicochemical constraints on 109 microbial Hg transformations, we studied Hg speciation across a gradient of environmental 110 factors that can influence microbiome composition and activity. We compare our findings 111 to those of previous geothermal spring studies, to refine the conceptual model for 112 geothermal Hg cycling. 113 114 Materials and Methods 115 Site description 116 The Ngawha Geothermal Field (NGF) constituted the field site for investigating Hg cycling 117 in a low pH (<4.5), elevated Hg (>100 ng L-1), and sulfide-rich (>0.1 mg L-1) environment 118 (Figure S1). Mercury ore deposits in the NGF occur as cinnabar, metacinnabar and native 119 Hg(0) in association with active hot springs, fumaroles, and mud pools (31). Elemental Hg 120 (mainly gaseous Hg(0)), travels from deep geological sources to the surface, either in 121 hydrothermal fluids or geothermal gases, where it reacts with oxygen in presence of 122 chloride to form Hg(II); Hg(II) in turn reacts with dissolved sulfide (biogenic) to precipitate 123 cinnabar (32). Roughly 33,000 kg of cinnabar ore was mined from the Tiger Springs area

5 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

124 at Ngawha during the first half of the 20th century (31). Tiger Springs and other areas 125 within the NGF still host an active geothermal system that releases approximately 530 kg

126 of HgT annually, ~44% of which is thought to be emitted to the atmosphere (32). The 127 remaining Hg resides in the local surficial waters and sediments (32). Cinnabar 128 precipitation was confirmed by X-ray diffraction analysis of Tiger Springs sediments and 129 nearby topsoil (Fig. S2). 130 131 Sampling techniques 132 Several springs from three areas, Tiger Springs (TS), Ginn Ngawha Spa (GN), and Ngawha 133 Springs Baths (NS), were sampled in April and October of 2011. Many of the springs were 134 edged with boards for use as soaking baths. Water samples for Hg analyses were filtered 135 through 0.45 µm membrane syringe filters, preserved with 1% v/v reagent grade HCl, and 136 stored in acid-washed HDPE bottles in the dark. Filtered water samples (0.45 µm) for anion 137 analysis were stored in sterile 50 mL plastic Falcon tubes. All filtered samples were stored 138 at 4°C. Redox potential (mV) and pH measurements were taken at each sampling site using 139 an Orion Model 250A portable meter with a glass pH electrode; spring water temperature 140 measurements were also taken at the time of sample collection. Sediment samples were 141 collected from the floor or wall of each spring, as well as from bulk water samples, into 142 sterile 50 mL Falcon tubes. Using sediments collected on the bottom of the boarded spring 143 bath and suspended in water allowed us to capture a mix of aerobic and anaerobic members 144 of the microbial community, as well as those that live across a range of temperature 145 gradient from the cooler waters (<45 °C) to the hotter sediments (>55 °C). Samples were 146 stored on ice for ~4 days until they could be transferred to laboratory storage at -80 °C (i.e.

147 transported from NZ to Melbourne, Australia). Sediment samples were used for HgT 148 analyses and for whole community DNA extractions. 149 150 Hg and MeHg analyses 151 Total Hg and MeHg concentrations of filtered waters and freeze-dried sediments collected 152 in April and October 2011 were measured at the Wisconsin Water Science Center, 153 (WWSC, U.S. Geological Survey, Middleton, Wisconsin) on a Perkin-Elmer Elan 9000 154 quadrupole ICP-MS and a Brooks Rand Atomic Fluorescence Spectrophotometer Model

6 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

155 III, respectively. Filtered water samples were analyzed within six months of sampling at 156 the Wisconsin Mercury Research Lab (WMRL) of the Wisconsin Water Science Center

157 (USGS; Middleton, Wisconsin). Filtered water samples analyzed for HgT species were 158 treated with BrCl solution to ensure all Hg species in the sample were oxidized to Hg(II).

159 Prior to analysis, SnCl2 solution was added to the vials which reduced the Hg(II) species 160 to volatile Hg(0). The samples were then ethylated, purged with argon gas, and analyzed 161 by GC (using Brooks Rand Autosampler and Total-Hg Purge and Trap system) in tandem 162 with atomic fluorescence spectrometry. Methylmercury analysis of filtered waters was 163 determined by distillation, gas chromatography separation, and speciated isotope dilution 164 mass spectrometry using ICP-MS, following USGS Method 01-445 and WMRL standard 165 protocols (33). Four blanks and two duplicate spikes were included in each run for quality 166 assurance. Method detection limits for total and methylated mercury were 0.007 ng, and 167 0.03–1.2 ng L-1, respectively (depending on the dilution factor required for each sample). 168 169 Sediment samples (~5 g wet weight) were freeze-dried overnight on a Heto-Drywinner

170 vacuum system before shipment in sterile glass vials to the WMRL. For solid HgT analysis,

171 freeze-dried samples were digested with 3:1 HCl:HNO3 overnight in a Teflon vessel. The 172 digested sample was then oxidized with BrCl solution. Total Hg analysis was then

173 performed by the same procedure as for filtered water. The detection limit for the solid HgT 174 analysis was 0.2 ng. Field blanks for Hg analyses were prepared using ultrapure reaction 175 grade water spiked with 1% v/v ultrapure HCl stored in each type of sampling material -1 - 176 until analysis. The HgT and MeHgT values for field blanks were 0.32 ng L and 0.57 ng L 177 1, respectively. 178 179 Hg(0) analysis 180 Vapor samples were collected in October 2011 at a height of 5-10 cm over Tiger and Cub 181 Baths (Tiger Springs area) on SKC Anasorb Sorbent tubes (Model C300). These sorbent 182 tubes are typically used to measure passive exposure to mercury in industrial settings (34, 183 35). Flow rate (2 L min-1). The volume of air sampled was regulated using an SKC Sample 184 Air Pump (PCXR4). Gas samples were passed through a soda lime trap before collection 185 on the hopcalite sorbent to trap excess condensation and neutralize acid. Once used for

7 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

186 sampling, sorbent tubes were sealed with Teflon tape and sent to ChemCentre (WA, 187 Australia) to be analyzed per NIOSH Method 6009. The sorbent material was dissolved

188 and oxidized in 1:1 HNO3:HCl. This solution was then diluted with DI water, and 189 immediately before analysis on a cold vapor atomic absorption spectrometer (CVAAS),

190 10% SnCl2 was added to reduce all Hg(II) to Hg(0). This Hg(0) was then purged into the 191 CVAAS analyzer (detection limit = 0.01 µg) in an argon gas stream. Sampling of Tiger 192 and Cub Baths was performed under similar conditions. First, a blank Anasorb tube was 193 exposed to ambient levels of Hg (unsealed, with no air pumped through). An additional 194 blank, that remained sealed, was also included in analyses. Blank samples were below the 195 method detection limit (<0.01 µg). Samples taken from the same site, with varying pump 196 times (20-30 minutes) did not yield similar adsorbed Hg concentrations. A positive 197 correlation between total Hg adsorbed versus volume of air pumped through the adsorbent 198 trap (r2=0.976, n =4), indicated that saturation of the adsorbent was not reached. 199 200 Common ion analysis

- - - - 2- 201 Common anion concentrations (F , Cl , Br , NO3 , SO4 ) were measured in unacidified, 202 filtered water samples collected in April 2011, using a Dionex DX-120 ion chromatograph 203 with an IonPAC As14 column (4 x 250 mm) in the Department of Chemistry at The 204 University of Melbourne. Samples were stored in the dark at 4 °C for 4 months between 205 sample collection and analysis. The instrument was set to the following conditions: 1.4 mL 206 min-1 flow rate, the eluent was 4.8 mM sodium carbonate and 0.6 mM sodium bicarbonate, 207 with an injection volume of 25 µL. Chromatographs were viewed on PeakNet Run System 208 I software. Method detection limit for IC analyses was 0.0005 mg L-1. Determination of 209 sulfide in water samples was performed on filtered water samples prepared in the field for 210 analysis in October 2011 using the Methylene Blue Method (36) and the Hach DR 2800 211 spectrophotometer (Method 8131). Total Fe was measured in filtered, acidified water 212 samples collected in October 2011 using the Ferrozine method (37), with prepared reagents 213 coordinating to Hach Method 8147. Method detection limit for sulfide measurements was 214 0.02 mg L-1and for Fe measurements was 0.5 mg L-1. The samples had been stored in the 215 dark at 4°C prior to analysis. Sample blanks were prepared at time of sample collection, 216 using MQ water.

8 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

217 218 DNA extraction 219 DNA was extracted from sediments collected from Tiger (TS1) and Cub (TS2) Baths and 220 surrounding sites (TS3–TS5) in October 2011 using the MO-BIO PowerMax DNA 221 Isolation Kit, using an alteration to the manufacturer’s protocol. Approximately 5 g of 222 sediment was used for extraction. Sediments were first treated with 5 mL of 10 mM TrisCl 223 (solution C6), the samples were then centrifuged at 6,000 g for 5 minutes and the 224 supernatant was decanted. Then, bead solution C1 was added, and the alternative lysis 225 method in the protocol was followed, which replaces the 10-minute bead-beating step with 226 incubation in a water bath at 65 °C for 30 minutes, followed by vortexing the sample for 1 227 minute (38). This alternative lysis method was used to reduce shearing of the DNA. 228 Duplicate DNA extractions were performed to ensure a sufficient mass of DNA was 229 extracted for each sample. Replicate DNA extractions were concentrated onto the same 230 spin filter and cleaned using QIAquick PCR purification kit (Qiagen, California). 231 232 Metagenomic analysis 233 Approximately 113–216 ng of genomic DNA extracted from samples collected in October 234 2011 (TS1A and TS2A) was barcoded by sample and prepared for sequencing with the 235 Nextera®XT kit, and ten samples were run on a single lane of an Illumina® HiSeq 2500 236 with 2 x 100 bp paired-end sequencing (Australian Genome Research Facility, Melbourne, 237 AU). This produced ~9 giga-bases of sequence data for two Ngawha metagenomes, Tiger 238 Bath (NW1) and Cub Bath (NW2), out of 43 giga-bases for all ten samples (Table S1). 239 Sequences were binned by bar code, quality filtered, and trimmed to remove Illumina® 240 adapters using Trimmomatic (39). Metagenomes were examined after assembling short 241 reads into longer contigs with IDBA-UD (40). Assembled contigs were uploaded and 242 binned for analysis using ggKbase, and have been made publicly available 243 (http://ggkbase.berkeley.edu/). 244 245 Detection of Hg cycling genes in metagenomes 246 The metagenomic read sets were screened directly for sequences sharing homology with 247 hgcA and hgcB, using a hidden Markov model (HMM) method described previously (41).

9 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

248 Metagenomic read sets were also screened directly for sequence homology with merA, 249 using an HMM-search built previously (41). Assembled metagenomic sequences were 250 searched for mer-operon, hgcA and hgcB genes using ggKbase. Genes were annotated in 251 ggKbase by BLAST searches against the NCBI database (42). Multiple merA sequences 252 were extracted from ggKbase, translated to amino-acids using the bacterial translation 253 table, and aligned using ClustalW (43) in MEGA6 (44) with MerA reference sequences 254 from confirmed mer operon-possessing microorganisms (13). 255 256 Genomic binning and phylogenetic analyses 257 Genomic binning of metagenomic data was performed using ggKbase, based on scaffold 258 coverage, GC content of scaffold sequences, and common taxonomy of the contigs. Bins 259 were further refined with emergent self-organizing mapping (ESOM) (45). Scaffold 260 coverage was calculated using Bowtie2 to map reads to the assembled sequences (46). 261 Genome completeness was estimated from the presence of bacterial single copy genes (51 262 in total) or archaeal single copy genes (38 in total) (Table S2). Several bins, 263 NW1_Thermoplasmata_unknown_1, NW2_Desulfurella_acetivorans_33_49, and 264 NW2_Rhodospirillales_68_7, are likely multi-genome bins that could not be refined to 265 single genomes from coverage and average GC splits. Information regarding genomic bins 266 for each metagenome, as well as unbinned scaffolds, is provided in Table S2. Taxonomy 267 was assigned to the metagenome-assembled genomes based on consensus classification of 268 contigs. There are 34 ribosomal proteins considered universal among bacteria, archaea, and 269 eukaryotes, and which can be used to infer phylogeny (47). In this study, we used ribosomal 270 protein s3, a single-copy gene present in every genomic bin from this study, to compare 271 phylogenies and coverage across each metagenome-assembled genome. Ribosomal protein 272 sequences were pulled from ggKbase and then aligned to translated ribosomal protein s3 273 sequences from the NCBI non-redundant protein database. 274 275 Hg cycling genes from publicly available hot spring metagenomes 276 To compare Hg-cycling genes from NGF to those of other geothermal settings, assembled 277 metagenomes from Yellowstone National Park (YNP) were searched for hgcA (Table S3) 278 and merA genes (Table S4). Included in the phylogenetic analysis were HgcA protein

10 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

279 sequences extracted from YNP metagenomic datasets (https://img.jgi.doe.gov) (Table S3). 280 Sequences were obtained by querying all assembled hot spring metagenomic datasets for 281 sequences sharing homology to carbon monoxide dehydrogenase (pfam03599). HgcA 282 proteins share homology with carbon monoxide dehydrogenases and are often mis- 283 annotated as such in microbial genomes (17). HgcA sequences were differentiated from 284 carbon monoxide dehydrogenases using the HMM model described above, with an 285 inclusion value cut-off of 1e-7, and only sequences that included the conserved cap-helix 286 region of HgcA were kept for analyses. Out of a total of 234 hot spring metagenomes (of 287 which 216 were from YNP springs) searched (as of July 2016) there were 4,520 matches 288 to sequences annotated as carbon monoxide dehydrogenases (pfam03599) (2,148 in YNP 289 metagenomes). Of these sequences, 102 were identified as HgcA sequences from 18 290 different metagenomes (Table S3). Included in the phylogenetic analyses were 65 HgcA 291 sequences found in 15 YNP metagenomes. The hgcA genes were found in YNP 292 metagenomes sampled from four sites: Mushroom Springs (Gp0111644–45, Gp0057794), 293 Octopus Springs (Gp0057360, Gp0057796, Gp00111632, Gp0111634, Gp0111638, 294 Gp0111642, Gp0111646), Obsidian Pool (Gp0056876), and Fairy Spring (Gp0051404). 295 Environmental metadata were only provided for a small number of these metagenomes. 296 Sequences from Fairy Geyser Spring (Gp0051404) are from an alkaline (pH: 9–9.2) and 297 mesophilic (33.3–36 °C) phototroph-dominated mat, while sequences from Mushroom 298 Spring (Gp0111644–45, Gp0057794) were from the undermat layer (~3–5 mm) of an 299 anoxygenic and phototrophic microbial mat in an effluent channel of the alkaline bath. The 300 water above the mat had a recorded temperature of 60 °C (48). A separate metagenomic 301 study has reported temperature and pH measurements for Mushroom Spring (60 °C; pH, 302 8.2), Obsidian Pool (56 °C; pH, 5.7), and Octopus Spring (80 – 82 °C; pH, 7.9) (49). The 303 additional 37 hgcA sequences were from assembled metagenomes sequenced from Dewar 304 Creek Spring (77 °C; pH, 8.0) and Larsen North spring, in British Columbia, Canada (50). 305 306 RESULTS 307 Water and sediment chemistry 308 Chemical and physical data for each spring are shown in Tables 1 and 2. Chloride 309 concentrations exceeded 400 mg L-1 in all springs except Cub Bath (TS2) at 29.8 mg L-1.

11 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

310 Across the NGF, sulfate concentrations varied from 9.3 – 1200 mg L-1, with the highest 311 concentrations found in springs of pH < 4 (Table 1). In Tiger Springs, Cub Bath had the 312 lowest sulfate concentration (319 mg L-1), while the other TS sites were > 1000 mg L-1. 313 Sulfide concentrations were nearly 3.5x higher in Cub Bath (6.45 mg L-1) than in Tiger 314 Bath (TS1) (1.82 mg L-1). The total iron concentration in Tiger Bath (12.7 mg L-1) in 315 October 2011 was twice that of Cub Bath (6.06 mg L-1). 316

317 Total Hg and MeHgT measurements of filtered waters from NGF hot springs sampled in 318 April 2011 are shown in Table 2. In comparison to background levels for freshwater -1 -1 319 systems, HgT was relatively high at ~250 – 16,000 ng L , compared to 0.4 – 74 ng L for 320 non-geothermal lakes, and 1 – 7 ng L-1 for rivers and streams (51). Non-thermal waters in 321 the Ngawha region are reported to contain 300 – 500 ng L-1 Hg (32). Previously reported 322 values for dissolved Hg in NGF thermal waters range from 1,000 – 350,000 ng L-1 (32). In

323 this study, the highest levels of HgT were found in the Tiger Springs region (all > 1000 ng 324 L-1) and from Cinnabar Bath (GN1) in the Ginn Ngawha Spa (1460 ng L-1) (Fig. S3). 325 Methylmercury levels ranged from 0.5 – 14.0 ng L-1; however, elevated concentrations of

326 MeHgT did not correlate with higher HgT concentrations. Concentrations of HgT and

327 MeHgT in filtered water samples from NGF were compared to reference values (Table 2). 328 MeHg levels in filtered water samples from YNP were substantially lower than those 329 observed at NGF, with most YNP springs having concentrations below the detection limit 330 (0.013 ng L-1). Those with detectable MeHg were still relatively low at 0.026 – 0.080 ng -1 331 L (5). Only three sites sampled in the NGF exhibited a significant proportion of HgT as 332 MeHg (~1% v/v): Cub Bath (TS2), Kotanitanga Bath (NS1B), and a drainage pool adjacent 333 to Kotanitanga Bath (NS3B). Methylmercury levels were nearly an order of magnitude 334 greater in Cub Bath (TS2) than in Tiger Bath (TS1), even though the total mercury levels 335 were roughly the same (~1000 ng L-1). 336

337 Sediments collected from the Tiger Springs area in October 2011 were analyzed for HgT 338 (Table 2). Total Hg concentrations were highest in the hole adjacent to Tiger Bath (TS3) -1 339 at ~7000 µg g . Tiger Bath (TS1) sediments had about half the solid HgT of TS3, at 3467 -1 - 340 µg g , and Cub Bath sediments had about a third the solid HgT of Tiger Bath (1274 µg g

12 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

341 1). Gaseous Hg0 emissions were also recorded from the baths (Table 2). The Hg(0) 342 concentration measured above Tiger Bath (25.0 ng L-1) was greater than that of Cub Bath 343 (4.34 ng L-1), but also varied across the bath (4.38 – 25.0 ng L-1). Furthermore, these values 344 are an order of magnitude lower than previously reported concentrations of fumarolic 345 Hg(0) in the Tiger Springs area 13.5 – 276 µg L-1, and 710 µg L-1 from Tiger Bath 346 specifically (32). 347 348 Microbial diversity from metagenomic datasets 349 Phylogenetic analysis of ribosomal marker protein s3 from assembled Tiger and Cub Bath 350 metagenomes identified members of Deltaproteobacteria, Gammaproteobacteria, 351 Thermotogae, Micrarchaeota, Parvarchaeota, Thermoplasmata, and other Euryarchaeota 352 (Figs. S4 and S5). Furthermore, phylogenetic analysis revealed a greater breadth in the 353 phyla of genomes resolved from Cub Bath, compared to Tiger Bath, with Verrucomicrobia, 354 Acidobacteria, Firmicutes, Planctomycetes, Alphaproteobacteria, and Betaproteobacteria 355 identified in Cub Bath only (Fig. S4). Ribosomal proteins from Thermoprotei were 356 identified only in assembled Tiger Bath metagenomic data (Fig. S5). Genome binning 357 resulted in acidophilic and thermophilic bacteria and archaea dominating the Tiger and Cub 358 Bath metagenomic datasets; with the average coverage of scaffolded contigs within 359 genome bins ranging from 7.1 – 789x (Fig. 1; Table S2). The highest coverage bins in each 360 bath were related to Acidithiobacillus, Thermotogales and Thermoplasma (Fig. 1). 361 362 Genome bins were comprised of aerobes, as well as obligate and facultative anaerobes, 363 capable of sulfur oxidation (Acidithiobacillus spp., Thiomonas), sulfur reduction 364 (Desulfurella acetivorans, Granulicella mallensis), iron oxidation (Ferroplasma, 365 Acidithiobacillus ferrivorans), iron reduction (Acidobacterium capsulatum), methane 366 oxidization (Methylacidiphilum), and acetate oxidation (Desulfurella acetivorans) (Fig. 2). 367 To elucidate important biogeochemical links to Hg cycles mediated by these microbial 368 phylotypes, metagenomes were searched for genes encoding for Hg, sulfur, sulfate, and 369 methane cycling (Fig. 2). We note here that genes involved in methanogenesis (notably 370 mcrA) and methane oxidation (pmoA) were nearly absent from metagenomes (Fig. 2). 371

13 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

372 Mer operon 373 Assembled metagenomes were screened for genes belonging to the mer operon that encode 374 for mercuric reductase (merA), organomercurial lyase (merB), a periplasmic protein 375 (merP), and inner membrane proteins involved in Hg(II) transport (merT, merC, merE, 376 merF, merG), as well as one or more regulatory proteins (merR, merD) (13). At Ngawha, 377 scaffolded mer genes were often encoded in the higher coverage genome bins of each bath, 378 Acidithiobacillus spp., Thiomonas, and Thermotogae (Table S5; Fig. 2). The coverage of 379 mer scaffolds ranged from 3.9 – 836x, with scaffold lengths of 1,021 – 101,352 bp, 380 indicating that a significant number of reads mapped to each sequence. A high fraction of 381 reads (3.24E-04 – 1.94E-04, Table S1) from each metagenome were predicted to encode 382 mercuric reductase (MerA) using the HMM-model. BlastP analysis of HMM-search 383 outputs indicated that the HMM-model was insufficient for filtering sequences that are 384 predicted to encode for MerA paralogs dihydrolipoamide dehydrogenase and pyridine 385 nucleotide-disulfide oxidoreductase. Therefore, the HMM-model likely over-estimates 386 MerA abundance encoded by the raw reads. MerA homologues identified in the assembled 387 Tiger and Cub Bath metagenomes using ggKbase annotations were related to both Archaea 388 (Euryarchaeota), and Bacteria (Proteobacteria and Bacteroidetes) (Fig. 3). Most archaeal 389 MerA were related to Thermoplasma; however, four MerA homologues branched deeply 390 from known archaeal homologues and were related (<43%, BlastP sequence alignment) to 391 MerA from Candidatus Methanoperedens nitroreducens (NW1 scaffolds 675 and 247, and 392 NW2 scaffolds 1319 and 13696; Fig. 3). Most MerA homologues from Cub Bath were 393 related to Acidithiobacillus spp. (Fig. 3). MerA homologues identified in the metagenomes 394 were from operons also encoding for MerP, MerT, and MerR (Table S5). Several scaffolds 395 related to Euryarchaeota (Thermoplasma) encoded just for MerA and MerP 396 (NW1_scaffold_26, NW1_scaffold_675, NW2_scaffold_876; Table S5), consistent with 397 Thermoplasma mer genes sequenced from acid mine drainage (AMD) (52). 398 399 Two similar but distinct merB genes were identified in assembled Tiger and Cub Bath 400 metagenomes, related to Acidithiobacillus caldus and Thioalkavibrio spp., respectively 401 (Fig. S6). The merB gene from Tiger Bath (NW1_scaffold_113) was linked to a genome 402 bin identified as Acidithiobacillus caldus, with an average scaffold coverage of 750x (Table

14 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

403 S5). The same scaffold (NW1_scaffold_113) contained two merR genes that were 404 convergent and divergent, respectively to mer genes that encode for MerT, MerP, and 405 MerA. The merB gene from Cub Bath (NW2_scaffold_17399) was most closely related to 406 merB from Thioalkavibrio spp.; however the scaffold itself was unbinned, with an overall 407 taxonomic identification of Acidithiobacillus ferrivorans (Table S5). The scaffold also 408 contained merR and merA genes related to Acidithiobacillus ferrivorans (WP_035195121). 409 The translated organomercurial lyases (MerB) from Tiger and Cub Baths aligned to 410 conserved cysteines(53), indicative of their true functionality (Fig S7). 411 412 Biological Hg methylation 413 Both Tiger and Cub Bath metagenomic read sets were searched for sequences sharing 414 homology to Hg methylation genes (hgcA and hgcB) (17). A small fraction of reads (2.16E- 415 07) were identified as fragments of hgcA sequences in the Cub Bath metagenome (Table 416 S1). The nucleotide reads were each 100 bp in length, and when translated, aligned to 417 several regions of HgcA from known methylators, including the highly conserved 418 “G(I/V)NVWC” region of the HgcA protein ((17), Fig. S8). Several of the reads aligned to 419 one another, and a composite amino acid sequence (59 AA in length) is shown in Fig. S8. 420 Phylogenetic analyses of the translated reads revealed two distinct hgcA-like genes in the 421 Cub Bath metagenome (Fig. 4). One of the reads closely matched (BlastP search) to pterin- 422 binding regions of HgcA-like proteins from marine bacteria Streptomyces sp. CNQ-509 423 and Nitrospina spp. (E-value: 8e-05; 93% sequence coverage; 55%), and to fused HgcAB 424 proteins primarily found in thermophilic archaeal and bacterial genomes, Thermococcus 425 sp. EP1, Kosmotoga pacific, Pyrococcus furiosis, and Methanococcoides methylutens (E- 426 value: 3E-06 – 8E-08; 100% sequence coverage; 65–75% sequence ID). Whether these 427 microbes with genomes encoding for an HgcAB fused protein are capable of Hg 428 methylation is unknown (16). When tested for Hg methylation capability, both Pyrococcus 429 furiosus and Methanococcoides methylutens were unable to produce MeHg above controls 430 (16, 54). 431 432 The second set of hgcA reads from Cub Bath (reads 2–7; Fig. 4), including the composite 433 sequence, aligned using BlastP to the pterin-binding region of HgcA proteins from known

15 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

434 and predicted Hg methylators Desulfosporosinus youngiae, Clostridium cellobioparum, 435 Desulfosporosinus sp. Tol-M, Desulfosporosinus sp. BRH-c37(E-value: 4E-21 to 7E-22; 436 100% query cover; 71–75% ID). Phylogenetic analyses (Fig. 4) of the composite sequence 437 revealed homology to HgcA from Nitrospirae bacterium HCH-1 (Genbank ID: 438 LNQR00000000.1) and to HgcA from hot spring metagenomes. The hgcA sequences 439 pulled from assembled YNP and British Columbia hot spring metagenomes from JGI-IMG 440 databases span a wide range of diverse phyla and are predicted to encode HgcA that relate 441 closely to those from Deltaproteobacteria, Firmicutes, Nitrospirae, Planctomycetes, and 442 Euryarchaeota (Methanomicrobia and Thermoplasmata) (Fig. 4). BLASTP searches of 443 translated amino acid sequences revealed common closest matches (<83% ID) to 444 Clostridium straminisolvens, Deltaproteobacterium sp. NaphS2, Nitrospirae bacteria 445 SG8-3 and HCH-1, Physcisphaerae bacterium SG8-4, Smithella spp., Methanocella 446 arvoryzae, and Thermoplasmatales archaeon DG-70-1. Of these closest relatives, only 447 Thermoplasmatales archaeon DG-70-1 is from a thermophilic phylum (Thermoplasmata); 448 however, this strain was isolated from an anaerobic, moderately halophilic, and mesophilic 449 aquatic environment (55), rather than a geothermal spring. 450 451 A high fraction of reads from each metagenome aligned to hgcB sequences from known 452 and predicted methylators (2.08E-06 – 3.02E-06; Table S1). However, BLAST analysis of 453 the translated hgcB-like genes resulted in closest matches to ferredoxin-encoding gene 454 sequences in both Tiger and Cub Bath. To differentiate between other ferredoxin proteins 455 and HgcB homologues, the translated hgcB-like genes were searched for the conserved 456 “C(M/I)ECGAC” site in HgcB (17). A complete hgcB gene was identified in the assembled 457 Cub Bath metagenome (NW2_scaffold_10600_2, Supp. Mat.). A translated BlastP search 458 of the sequence revealed closest matches (E-value: 2E-26–3E-29; 91–98% coverage; 53– 459 57% identity) to HgcB from predicted Hg methylators Dehalobacter spp., Clostridium 460 straminisolvens, Nitrospirae bacterium HCH-1, and Syntrophobotulus glycolicus. Other 461 features on the assembled contig that contains the hgcB gene (NW2_idba_contig_10600), 462 include a partial hgcA gene upstream from hgcB, and downstream genes encode for a 463 DsrE/DsrF-like family protein (involved in intracellular sulfur reduction) related to 464 Planctomycetes; a copper-binding protein related to Aquificae; and a putative

16 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

465 transcriptional regulator related to ArsR repressor, and an arsenical pump membrane 466 protein (ArsB), both related to Firmicutes (Fig. S9). 467 468 Biological sulfur species cycling in Tiger and Cub Baths 469 The availability and speciation of sulfur compounds within geothermal environments 470 constrains the strategies employed by microorganisms to deal with Hg in these ecosystems. 471 For example, Hg methylation is impeded in ecosystems with elevated aqueous sulfide 472 concentrations (56). Conversely, the availability of soluble Hg to microorganisms is limited 473 by the solubility and/or microbially-mediated dissolution of cinnabar/metacinnabar (57). 474 To elucidate these interactions, the Cub and Tiger Bath metagenomes were searched for 475 genes related to dissimilatory sulfite/sulfate reduction using dsrAB genes that encode 476 subunits A and B of the dissimilatory bi(sulfite) reductase enzyme (DsrAB); (58), and the

477 sox pathway genes (soxR, soxXA, soxYZ, soxB, sox(CD)2, soxG) that encode enzymes used 478 in hydrogen sulfide and elemental sulfur oxidation (59). 479 480 Complete or near complete dsrAB sequences were recovered from assembled Tiger and 481 Cub Bath metagenomes using ggKbase annotations (Table S6). The genes were often 482 present on scaffolds containing other genes encoding proteins involved in sulfite/sulfate 483 reduction to sulfide. Both dsrC and dsrD genes were present on scaffolds with dsrAB in 484 NW1_Thaumarchaeota_unknown_1, NW2_Deltaproteobacteria_41_12, and 485 NW2_Acidobacterium_capsulatum_related. Phylogenetic analyses of the translated genes 486 indicated six distinct groups of DsrAB sequences in Tiger and Cub Baths (Fig. 5, Table 487 S6), Groups I-VI. Three distinct groups of reductive bacterial type DsrAB groups (I-III) 488 were identified, most closely related to known thermophilic sulfur-reducing bacteria such 489 as Desulfurella acetivorans as well as uncultured Gemmatimonas sp. Sg8-17. Groups IV 490 and V contained sequences related to reductive archaeal DsrAB in genomic bins related to 491 Thermoplasmata, Thermoplasma acidophilum, and Aciduliprofundum. Group VI DsrAB 492 contain sequences distantly related to Candidatus Rokubacteria CSP1-6, Caldivirga 493 maquilingensis and Thermodesulfobacterium spp. (WP_051754629). Group V DsrAB 494 were likely uncultured reductive archaeal type DsrAB related to Vulcanisaeta; unbinned 495 sequences were from scaffolds with a majority of sequences related to Archaea. Group VI

17 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

496 DsrAB most closely related (>50%) to DsrAB from the sulfate-reducer Candidatus 497 Rokubacteria CSP1-6 (60); the sequence was distinct to DsrA from Group IV and is from 498 genomic bin NW2_Planctomycetes_54_12 (Table S6). The DsrAB phylogenies as 499 represented in Fig. 5 agree with the phylogenetic annotations of the scaffolds from which 500 the genes were obtained (see Table S6). Importantly, while there is evidence for bacterial 501 and archaeal sulfate and sulfur reduction within Tiger and Cub Baths, none of the DsrAB 502 groups detected (Groups I-VI, Fig. 5) contain known Hg methylators from the primary 503 sulfate-reducing phyla, Deltaproteobacteria and Firmicutes (18). Nor were any hgcAB 504 genes present in sulfate-reducing genomic bins from Tiger or Cub Bath. 505 506 Genes encoding for sulfur-oxidation (sox) in acidophilic Beta- and Gammaproteobacteria 507 were detected in both Tiger and Cub Bath metagenomes and were related to 508 Acidithiobacillus spp. and Thiomonas spp. (Fig. 2; Table S7). Furthermore, sequences from 509 archaeal and bacterial sox pathways, including from acidophiles Acidiphilium and 510 Acidocella, were present in the unbinned metagenomic data (Table S7). 511 512 DISCUSSION 513 Geothermal systems provide an environment in which relationships between the chemical 514 and physical processes controlling Hg speciation and bioavailability, and microbial Hg 515 transformations (4, 7), remain poorly understood. Temperature and pH constitute major 516 drivers of microbial diversity in geothermal springs, with pH contributing to a greater 517 extent (61, 62). Indeed, previous studies show that acidic geothermal spring communities 518 appear quite distinct from those of neutral and alkali springs, irrespective of temperature 519 (61, 62). In our study, despite their mutual close proximity and broadly similar 520 physicochemical properties and dissolved total Hg concentrations, the Tiger and Cub Baths 521 of the NGF hosted very distinct microbiomes (Fig. S10). The greater diversity in genomic 522 bins representative of the Cub Bath microbiome, compared to Tiger Bath (Fig. 1), may 523 have promoted Fe and S redox cycling to a greater extent, which in turn could facilitate the 524 dissolution of metal sulfides such as pyrite or cinnabar. Subsequently, this dissolution 525 could have increased the bioavailability of Hg(II) for hgcAB+ equipped microorganisms, 526 by oxidizing reduced S and increasing dissolved Hg(II), respectively (Fig. 6).

18 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

527 528 NGF genomes featured similar metabolic capabilities to those recovered from AMD. The 529 oxidation of reduced sulfur species by both aerobic and anaerobic chemolithotrophic

530 microorganisms produces electrons utilized in respiration and CO2 assimilation (63), 531 critical processes for living in AMD and/or geothermal systems (63, 64). Dominant 532 members of the Tiger Springs’ communities, Acidithiobacillus spp. and Thiomonas spp., 533 can utilize various metal and sulfur-oxidizing enzymes, pathways, electron transport 534 mechanisms, and substrates (63) to sustain activity. Acidithiobacillus ferrivorans is an 535 obligate chemolithoautotroph and facultative anaerobe that oxidizes Fe(II); some strains 536 are also able to utilize sulfur, thiosulfate, tetrathionate, and pyrite (65). Acidithiobacillus 537 ferrooxidans can utilize metal sulfides to support growth (63), and Acidithiobacillus caldus 538 is also capable of oxidizing reduced inorganic sulfur species, producing sulfate via the sox 539 pathway (66). A number of other NGF genome bins, including several associated with 540 Thermoplasma and Thiomonas spp., were equipped to respire using either sulfur or organic 541 carbon (30), and Thiomonas can also oxidize arsenite (As(III)) to arsenate (As(V)) (67). 542 The Thiomonas-like genome bin from Cub Bath showed evidence for the presence of both 543 sox and arsenite oxidase pathways (Figure S2). 544 545 The microbiomes of Tiger Springs were also similar to those found in AMD with respect 546 to stress resistance/response mechanisms for acid and heavy metals (e.g., (30, 68)), 547 featuring mechanisms for pH homeostasis, for example. The highest genome coverage in 548 each bath was associated with Acidithiobacillus caldus, a microorganism equipped with 549 multiple heavy metal resistance pathways (ars, mer, czc, and TeR). The mer genes in Tiger 550 Spring metagenomes were predominately found in aerobic bacterial and archaeal genome 551 bins (Fig. 2; Table S5), particularly the mesophilic acidophiles Thermoplasma and 552 Acidithiobacillus (Fig. 3). This finding was in sharp contrast to the diversity of merA genes 553 in YNP metagenomic datasets, which were principally from archaeal taxa (e.g. 554 Sulfolobales, Acidobales, DPANN). Based on the observations of Geesey et al. (4) that 555 archaea dominated acid springs with high Hg content, and given that MerA homologues 556 are often encoded in the genomes of acidophilic archaea (4, 12, 13), we expected that 557 archaeal MerA would dominate in Tiger and Cub Baths.

19 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

558 559 While Tiger and Cub Baths are considered acidic (pH <4), they are substantially lower in 560 temperature than springs studied in YNP and the Western US; these different observations 561 may be due to a number of factors. First, Geesey et al. (4) noted that the number of bacterial 562 MerA homologues detected in acid springs increases with decreasing temperature (from 563 >73 ºC to <55ºC). Thus, the lower temperatures at NGF may explain the presence of a 564 larger number of bacterial MerA homologues. Second, bacterial taxa are also known to be 565 rare in higher temperature (>65 °C) acid (acidophile diversity, rather than the absence of a taxonomic capacity to 568 transform Hg. 569

570 A notable feature of Cub Bath was the higher percentage of HgT present as MeHgT, 571 compared to Tiger Bath. We speculate that this finding reflects a greater degree of 572 microbially-mediated turnover of aqueous Hg(II) to MeHg in the former spring. However, 573 the difference in MeHg levels between the baths could also be indicative of higher 574 demethylation rates in Tiger Bath compared to Cub Bath. While merB genes were 575 identified in both baths, the sequencing coverage of merB scaffolds was higher in Tiger 576 Bath (836x) compared to Cub Bath (18x) (Table S5). While abiotic methylation in 577 geothermal waters is not yet well understood (71), such processes are not likely to account 578 for the higher MeHg levels in Cub Bath relative to Tiger Bath. Furthermore, NGF MeHg 579 values were 1-2 orders of magnitude greater than values recorded at some YNP hot springs 580 (5, 6). An analysis of NGF metagenomes found that hgcA and hgcB genes were only 581 detected in Cub Bath and not Tiger Bath. Cub Bath Hg methylation genes are most likely 582 bacterial, belonging to a sulfate-reducing clade closely related to Firmicutes and 583 Deltaproteobacteria, possibly from the Nitrospirae (Fig. S9). Microbial Hg-methylation via 584 active sulfate-reducing microorganisms is consistent with the greater observed amount of 585 MeHg (~1.3% of dissolved total Hg) alongside higher concentrations of reduced sulfur 586 (3.5x) in Cub Bath. Of particular note was an hgcAB+ sulfate-reducing bacterium in Cub 587 Bath metagenome (NW2_unbinned) that may represent a novel acidophilic Hg methylator 588 equipped with heavy metal resistance. This genome encoded hgcA and hgcB on a scaffold

20 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

589 with a putative copper chaperone (HMA/CopZ), arsenical resistance operon repressor 590 (ArsR), and arsenical pump membrane protein (ArsB) (Fig. S9). There are few reports of 591 arsenate reduction by sulfate-reducing bacteria (72), although putative ars genes have been 592 found in several species of Desulfovibrio, Desulfosporosinus, Desulfomicrobium, and 593 Desulfotalea. Genomes of known, predicted, and non-Hg-methylating bacteria and archaea 594 were searched for homologous proteins to those encoded on NW2_scaffold_10600 (Table 595 S8). Notably, ArsR family transcriptional regulators are encoded directly upstream to 596 HgcA in Desulfovibrio desulfuricans ND132 (DND132_1054) and Desulfomicrobium 597 baculatum (Dbac_0377). While homologous ArsR proteins are common in genomes 598 containing HgcA (Table S8), no genomes encode all of these proteins, although several 599 genomes of known and predicted methylators encode for copper chaperones and arsenic 600 resistance proteins, providing a measure of confidence in genome binning results. Notably, 601 three homologous proteins are encoded in the genomes of Desulfosporosinus spp., often 602 found in sulfate-rich, heavy metal-contaminated, low pH environments (73), and 603 Desulfosporosinus acidiphilus, with an optimum growth pH of 3.6 – 5.5, was the first 604 acidophile observed to methylate Hg (18). Therefore, we infer that the Cub Bath bacterium 605 is likely a similar taxon capable of both Hg-methylation and As(V) reduction. 606 607 One of the strongest influences on Hg speciation and bioavailability (for methylation or 608 volatilization) in acidic sulfidic hot springs is the formation and dissolution of cinnabar

609 (HgS(s)), which in turn is impacted by the activity of sulfur- and iron-oxidizing

610 microorganisms (5, 32, 56, 74, 75). A previous study (32) identified HgS(s) as the most 611 abundant and widespread Hg-bearing mineral in the Ngawha region, an observation we 612 confirmed by XRD analysis from a topsoil sample in the Tiger Springs area (Fig. S2). The

613 observed physicochemical conditions in Tiger and Cub Baths were on the cusp of HgS(s) 614 formation/dissolution, at pH 2-3 and pH > 4 (76). Thus the bioavailability of Hg(II) in the 615 acid warm springs of the NGF reflected Hg solubility in the context of sulfur speciation 616 and acidic conditions (56, 57). In aerobic sediments, sulfur- and iron-oxidizing bacteria 617 and archaea may enhance the dissolution of cinnabar and increase Hg(II) bioavailability. 618 Like sulfide, chloride can also affect Hg bioavailability to methylating microorganisms; 619 methylation rates have been shown to correlate inversely with increasing chloride (Cl-)

21 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

620 concentration (77, 78). Chloride concentrations in Cub Bath were higher than those of 621 typical freshwaters (0.82 mM), and in combination with low pH, could have decreased net 622 Hg methylation rates without impacting the viability of methylating microorganisms (77). 623 Other factors that may influence Hg bioavailability include dissolved organic material 624 (DOM) and thiol-DOM interactions (79-81), and turbidity which likely limits photolytic 625 Hg transformations. 626 627 Together, these findings can account for the nearly 10x amount of filtered MeHg 628 concentrations in Cub Bath, from which hgcAB+ equipped genome bins were recovered,

629 compared to Tiger Bath, despite their near identical HgT concentrations in filtered water 630 samples and the nearly 3x higher solid Hg content of Tiger Bath. In contrast, no hgcA reads 631 were detected in the Tiger Bath metagenome at the sequencing depth of our study. Thus, 632 bioavailable Hg may be getting enzymatically reduced or re-complexed by Hg-binding 633 ligands (e.g., Cl-, DOM, S2-). Indeed, the low flux of Hg(0) from the surface of the bath 634 suggests that most of the microbially-reduced Hg(II) remains dissolved, and potentially is 635 continuously cycled between Hg(0) and Hg(II) redox states. Advective mixing of anoxic 636 and oxic waters would also promote Hg cycling and exchange between aerobes and 637 anaerobes, and likewise mer-equipped and hgcAB-equipped microbes, in the acid warm 638 springs (Fig. 6). 639 640 While acidophilic microbial mats in YNP have been shown to accumulate MeHg and may 641 actively methylate Hg (5, 6), the relative difference in pH between the two NGF springs is 642 minimal, and therefore unlikely to account for the difference in recovery of hgcAB genes.

643 An alternative explanation may be found in the Hg(T) concentration in the sediments of 644 both springs. The total solid Hg concentration in Tiger Bath was nearly three times higher 645 than that of Cub Bath (3467 µg g-1 and 1274 µg g-1 respectively). Similarly, a small nascent 646 hot spring located adjacent to Tiger Bath (TS3) had extremely high concentrations of both 647 total dissolved and solid Hg (16,700 ng L-1 and 7000 µg g-1 respectively). Taken together, 648 these observations suggest that the area immediately adjacent to Tiger Bath is a highly 649 localized Hg “hotspot”, with elevated Hg levels selecting for microbial genomes encoding 650 for Hg resistance. Hg methylation capability apparently does not confer Hg(II) resistance

22 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

651 (82), and with the exception of two Geobacter spp., mer operon genes appear to be absent 652 in genomes of known hgcAB+ microorganisms (14, 27). These observations are further 653 corroborated by spike-in experiments in river water sediments that showed microbial Hg 654 methylation was inhibited by Hg concentrations as low as 15.3 µg g-1 (24). Another 655 plausible control on Hg methylation in the two baths may have been temperature, which 656 was recorded as nearly 10°C different between baths during each sampling season (Table 657 2). Similar differences in temperature were recorded in the sediments (Tiger Bath at 51.1– 658 68.6°C and Cub Bath at 43.5–60.3°C). Principal component analysis of covariance 659 between geochemical parameters measured in all springs across the NGF (Tables 1 and 2) 660 indicates an inverse relationship between MeHg and spring temperature (Figure S11). 661 Previous work shows that known hgcAB+ methylators and predicted methylators are 662 almost exclusively mesophiles with growth optima ~30°C (18). Known exceptions include 663 the thermophile and known Hg-methylator Desulfacinum hydrothermale (83), 664 psychrotolerant and psychrophilic predicted methylators Geopsychrobacter 665 electrodiphilus and Methanolobus psychrophilus R15, and a group of hyperthermophiles 666 possessing a fused hgcAB-like gene of unknown methylation functionality (16). 667 668 Conceptual model of Hg speciation in warm acidic hot springs 669 In the NGF acid warm springs, Hg biogeochemical cycling reflects the detected or inferred 670 microbially-mediated Hg transformations, as well as constraints imposed by metal sulfide 671 solubility and vigorous microbially-mediated reduce S- and metal-oxidation reactions. The 672 main abiotic and biotic controls on Hg cycling in acidic mesothermal springs are depicted 673 in the conceptual model shown in Fig. 6. The bioavailability of Hg to Mer- and HgcAB- 674 equipped microorganisms is controlled by cinnabar precipitation and dissolution, which in 675 turn is influenced by pH and the presence of sulfur- and iron-oxidizing and sulfate- and 676 iron-reducing bacteria and archaea. Bioavailable Hg(II) is methylated to MeHg by 677 microbes, putatively sulfate-reducing bacteria, equipped with HgcAB. This process is 678 limited by demethylation of MeHg (via MerB) and reduction (via MerA) of Hg(II) to 679 Hg(0), by facultative anaerobic and aerobic iron-cycling and sulfur-oxidizing 680 microorganisms. The volatile Hg(0) may evolve from spring waters, or be photo-oxidized 681 and recycled to Hg(II). A significant sink for Hg(II) within the springs involves formation

23 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

682 of solid HgS (metacinnabar, cinnabar), and potentially the adsorption of MeHg to 683 sediments or particulate organic matter (which was not assessed here). However, in the 684 acidic conditions, most MeHg and Hg(II) should remain dissolved, and could be 685 continuously cycled between methylating and demethylating microorganisms.

686 Importantly, temperature (>50 °C) and elevated Hg(T) concentrations will restrict microbial 687 methylation of bioavailable Hg(II). However, as geothermal inputs mix with cooler water, 688 the microbial Hg methylation potential increases. Therefore, surface waters and 689 groundwaters that receive geothermal inputs, such as catchment waterways down-gradient 690 from NGF springs and discharges from hydrothermal power plants, may be important 691 environmental pathways for MeHg mobilization and bioaccumulation. 692 693

24 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

694 REFERENCES 695 1. Gustin MS, Lindberg SE, Weisberg PJ. 2008. An update on the natural sources 696 and sinks of atmospheric mercury. Applied Geochemistry 23:482-493. 697 2. Ball JW, McCleskey RB, Nordstrom DK, Holloway JM. 2008. Water-chemistry 698 data for selected springs, geysers, and streams in Yellowstone National Park, 699 Wyoming, 2003-2005. Geological Survey (US), 700 3. Mariner R, Presser T, Evans WC. 1977. Hot springs of the central Sierra Nevada, 701 California. US Geological Survey, 702 4. Geesey GG, Barkay T, King S. 2016. Microbes in mercury-enriched geothermal 703 springs in western North America. Science of The Total Environment 569:321- 704 331. 705 5. King SA, Behnke S, Slack K, Krabbenhoft DP, Nordstrom DK, Burr MD, Striegl 706 RG. 2006. Mercury in water and biomass of microbial communities in hot springs 707 of Yellowstone National Park, USA. Applied Geochemistry 21:1868-1879. 708 6. Boyd ES, King S, Tomberlin JK, Nordstrom DK, Krabbenhoft DP, Barkay T, 709 Geesey GG. 2009. Methylmercury enters an aquatic food web through acidophilic 710 microbial mats in Yellowstone National Park, Wyoming. Environmental 711 Microbiology 11:950-959. 712 7. Wang Y, Boyd E, Crane S, Lu-Irving P, Krabbenhoft D, King S, Dighton J, 713 Geesey G, Barkay T. 2011. Environmental Conditions Constrain the Distribution 714 and Diversity of Archaeal merA in Yellowstone National Park, Wyoming, U.S.A. 715 Microbial Ecology 62:739-752. 716 8. Robinson JB, Tuovinen OH. 1984. Mechanisms of microbial resistance and 717 detoxification of mercury and organomercury compounds: physiological, 718 biochemical, and genetic analyses. Microbiological reviews 48:95. 719 9. Schaefer JK, Yagi J, Reinfelder JR, Cardona T, Ellickson KM, Tel-Or S, Barkay 720 T. 2004. Role of the bacterial organomercury lyase (MerB) in controlling 721 methylmercury accumulation in mercury-contaminated natural waters. 722 Environmental science & technology 38:4304-4311. 723 10. Simbahan J, Kurth E, Schelert J, Dillman A, Moriyama E, Jovanovich S, Blum P. 724 2005. Community analysis of a mercury hot spring supports occurrence of

25 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

725 domain-specific forms of mercuric reductase. Applied And Environmental 726 Microbiology 71:8836-8845. 727 11. Chatziefthimiou AD, Crespo-Medina M, Wang Y, Vetriani C, Barkay T. 2007. 728 The isolation and initial characterization of mercury resistant chemolithotrophic 729 thermophilic bacteria from mercury rich geothermal springs. Extremophiles 730 11:469-479. 731 12. Barkay T, Kritee K, Boyd E, Geesey G. 2010. A thermophilic bacterial origin and 732 subsequent constraints by redox, light and salinity on the evolution of the 733 microbial mercuric reductase T. Barkay, K. Kritee, E. Boyd and G. Geesey The 734 natural history of the mercuric reductase. Environmental Microbiology 12:2904- 735 2917. 736 13. Boyd E, Barkay T. 2012. The mercury resistance operon: from an origin in a 737 geothermal environment to an efficient detoxification machine. Frontiers in 738 microbiology 3:349. 739 14. Osborn AM, Bruce KD, Strike P, Ritchie DA. 1997. Distribution, diversity and 740 evolution of the bacterial mercury resistance (mer) operon. FEMS Microbiology 741 Reviews 19:239-262. 742 15. Oremland RS, Culbertson CW, Winfrey MR. 1991. Methyl mercury 743 decomposition in sediments and bacterial cultures: Involvement of methanogens 744 and sulfate reducers in oxidative demethylation, vol 57, p 130-137. 745 16. Podar M, Gilmour CC, Brandt CC, Soren A, Brown SD, Crable BR, Palumbo 746 AV, Somenahally AC, Elias DA. 2015. Global prevalence and distribution of 747 genes and microorganisms involved in mercury methylation. Science advances 748 1:e1500675. 749 17. Parks JM, Johs A, Podar M, Bridou R, Hurt RA, Smith SD, Tomanicek SJ, Qian 750 Y, Brown SD, Brandt CC. 2013. The genetic basis for bacterial mercury 751 methylation. Science 339:1332-1335. 752 18. Gilmour CC, Podar M, Bullock AL, Graham AM, Brown SD, Somenahally AC, 753 Johs A, Hurt Jr RA, Bailey KL, Elias DA. 2013. Mercury methylation by novel 754 microorganisms from new environments. Environmental science & technology 755 47:11810-11820.

26 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

756 19. Liu Y-R, Yu R-Q, Zheng Y-M, He J-Z. 2014. Analysis of the microbial 757 community structure by monitoring an Hg methylation gene (hgcA) in paddy soils 758 along an Hg gradient. Applied and environmental microbiology 80:2874-2879. 759 20. Schaefer J, Kronberg RM, Morel F, Skyllberg U. 2014. Detection of a key Hg 760 methylation gene, hgcA, in wetland soils. Environmental microbiology reports 761 6:441-447. 762 21. Compeau G, Bartha R. 1985. Sulfate-reducing bacteria: principal methylators of 763 mercury in anoxic estuarine sediment. Applied and environmental microbiology 764 50:498-502. 765 22. Fleming EJ, Mack EE, Green PG, Nelson DC. 2006. Mercury methylation from 766 unexpected sources: molybdate-inhibited freshwater sediments and an iron- 767 reducing bacterium. Applied and environmental microbiology 72:457-464. 768 23. Kerin EJ, Gilmour CC, Roden E, Suzuki M, Coates J, Mason R. 2006. Mercury 769 methylation by dissimilatory iron-reducing bacteria. Applied and environmental 770 microbiology 72:7919-7921. 771 24. Chen Y, Bonzongo J, Miller G. 1996. Levels of methylmercury and controlling 772 factors in surface sediments of the Carson River system, Nevada. Environmental 773 Pollution 92:281-287. 774 25. Oremland RS, Miller LG, Dowdle P, Connell T, Barkay T. 1995. Methylmercury 775 oxidative degradation potentials in contaminated and pristine sediments of the 776 carson river, nevada. Appl Environ Microbiol 61:2745-2753. 777 26. Pak KR, Bartha R. 1998. Mercury methylation and demethylation in anoxic lake 778 sediments and by strictly anaerobic bacteria. Applied And Environmental 779 Microbiology 64:1013-1017. 780 27. Lu X, Liu Y, Johs A, Zhao L, Wang T, Yang Z, Lin H, Elias DA, Pierce EM, 781 Liang L. 2016. Anaerobic mercury methylation and demethylation by Geobacter 782 bemidjiensis Bem. Environmental science & technology 50:4366-4373. 783 28. Avramescu M-L, Yumvihoze E, Hintelmann H, Ridal J, Fortin D, Lean DR. 2011. 784 Biogeochemical factors influencing net mercury methylation in contaminated 785 freshwater sediments from the St. Lawrence River in Cornwall, Ontario, Canada. 786 Science of the Total Environment 409:968-978.

27 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

787 29. Rytuba JJ. 2003. Mercury from mineral deposits and potential environmental 788 impact. Environmental Geology 43:326-338. 789 30. Baker BJ, Banfield JF. 2003. Microbial communities in acid mine drainage. 790 FEMS microbiology ecology 44:139-152. 791 31. Christie T, Brathwaite B, Institute of G, Nuclear Science L. 2002. Mineral 792 Commodity Report 8 Mercury, p 32. Institute of Geological and Nuclear Sciences 793 Ltd. 794 32. Davey HA, van Moort JC. 1986. Current mercury deposition at Ngawha Springs, 795 New Zealand. Applied Geochemistry 1:75-93. 796 33. Olson ML, DeWild JF. 1999. Techniques for the collection and species-specific 797 analysis of low levels of mercury in water, sediment, and biota. US Geological 798 Survey Water-Resources Investigations Report:99-4018. 799 34. Takaya M, Joeng JY, Ishihara N, Serita F, Kohyama N. 2006. Field Evaluation of 800 Mercury Vapor Analytical Methods: Comparison of the" Double Amalgam 801 Method" and ISO 17733. Industrial health 44:287-290. 802 35. Gionfriddo C, Ogorek J, Butcher M, Krabbenhoft D, Moreau J. 2015. Mercury 803 distribution and mobility at the abandoned Puhipuhi mercury mine, Northland, 804 New Zealand. New Zealand Journal of Geology and Geophysics 58:78-87. 805 36. Cline JD. 1969. Spectrophotometric determination of hydrogen sulfide in natural 806 waters 1. Limnology and Oceanography 14:454-458. 807 37. Stookey LL. 1970. Ferrozine---a new spectrophotometric reagent for iron. 808 Analytical chemistry 42:779-781. 809 38. Zhou J, Bruns MA, Tiedje JM. 1996. DNA recovery from soils of diverse 810 composition. Appl Environ Microbiol 62:316-322. 811 39. Bolger AM, Lohse M, Usadel B. 2014. Trimmomatic: a flexible trimmer for 812 Illumina sequence data. Bioinformatics 30:2114-2120. 813 40. Peng Y, Leung HC, Yiu S-M, Chin FY. 2012. IDBA-UD: a de novo assembler for 814 single-cell and metagenomic sequencing data with highly uneven depth. 815 Bioinformatics 28:1420-1428.

28 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

816 41. Gionfriddo CM, Tate MT, Wick RR, Schultz MB, Zemla A, Thelen MP, 817 Schofield R, Krabbenhoft DP, Holt KE, Moreau JW. 2016. Microbial mercury 818 methylation in Antarctic sea ice. Nature microbiology 1:16127. 819 42. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. 1990. Basic local 820 alignment search tool. Journal of molecular biology 215:403-410. 821 43. Larking M, Blackshields G, Brown N, Chenna R, McGettigan G, McWilliam H, 822 Valentin F, Wallace I, Wilm A, Lopez R. 2007. ClustalW and ClustalX version 2. 823 Bioinformatics 23:2947-8. 824 44. Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. 2013. MEGA6: 825 molecular evolutionary genetics analysis version 6.0. Molecular biology and 826 evolution 30:2725-2729. 827 45. Ultsch A, Mörchen F. 2005. ESOM-Maps: tools for clustering, visualization, and 828 classification with Emergent SOM, vol 46. Univ. 829 46. Langmead B, Salzberg SL. 2012. Fast gapped-read alignment with Bowtie 2. 830 Nature methods 9:357. 831 47. Yutin N, Puigbò P, Koonin EV, Wolf YI. 2012. Phylogenomics of prokaryotic 832 ribosomal proteins. PLoS One 7:e36972. 833 48. Thiel V, Wood JM, Olsen MT, Tank M, Klatt CG, Ward DM, Bryant DA. 2016. 834 The dark side of the mushroom spring microbial mat: life in the shadow of 835 chlorophototrophs. I. Microbial diversity based on 16S rRNA gene amplicons and 836 metagenomic sequencing. Frontiers in microbiology 7:919. 837 49. Inskeep WP, Jay ZJ, Tringe SG, Herrgard M, Rusch DB. 2013. The YNP 838 metagenome project: environmental parameters responsible for microbial 839 distribution in the Yellowstone geothermal ecosystem. Frontiers in microbiology 840 4:67. 841 50. Eloe-Fadrosh EA, Paez-Espino D, Jarett J, Dunfield PF, Hedlund BP, Dekas AE, 842 Grasby SE, Brady AL, Dong H, Briggs BR. 2016. Global metagenomic survey 843 reveals a new bacterial candidate phylum in geothermal springs. Nature 844 communications 7:10476.

29 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

845 51. EPA. 2001. Water quality criterion for the protection of human health : 846 methylmercury : final. Washington, DC : Office of Science and Technology, 847 Office of Water, U.S. Environmental Protection Agency, 2001. 848 52. Yelton AP, Comolli LR, Justice NB, Castelle C, Denef VJ, Thomas BC, Banfield 849 JF. 2013. Comparative genomics in acid mine drainage biofilm communities 850 reveals metabolic and structural differentiation of co-occurring archaea. BMC 851 genomics 14:485. 852 53. Pitts KE, Summers AO. 2002. The roles of thiols in the bacterial organomercurial 853 lyase (MerB). Biochemistry 41:10287-10296. 854 54. Gilmour CC, Bullock AL, McBurney A, Podar M, Elias DA. 2018. Robust 855 Mercury Methylation across Diverse Methanogenic Archaea. mBio 9:e02403-17. 856 55. Baker BJ, Lazar CS, Teske AP, Dick GJ. 2015. Genomic resolution of linkages in 857 carbon, nitrogen, and sulfur cycling among widespread estuary sediment bacteria. 858 Microbiome 3:14. 859 56. Gilmour CC, Henry EA. 1991. Mercury methylation in aquatic systems affected 860 by acid deposition. Environmental Pollution (Barking, Essex: 1987) 71:131-169. 861 57. Jew A, Behrens S, Rytuba J, Kappler A, Spormann A, Brown Jr G. 2014. 862 Microbially enhanced dissolution of H g S in an acid mine drainage system in the 863 C alifornia C oast R ange. Geobiology 12:20-33. 864 58. Müller AL, Kjeldsen KU, Rattei T, Pester M, Loy A. 2015. Phylogenetic and 865 environmental diversity of DsrAB-type dissimilatory (bi) sulfite reductases. The 866 ISME journal 9:1152. 867 59. Friedrich CG, Bardischewsky F, Rother D, Quentmeier A, Fischer J. 2005. 868 Prokaryotic sulfur oxidation. Current opinion in microbiology 8:253-259. 869 60. Hug LA, Baker BJ, Anantharaman K, Brown CT, Probst AJ, Castelle CJ, 870 Butterfield CN, Hernsdorf AW, Amano Y, Ise K. 2016. A new view of the tree of 871 life. Nature microbiology 1:16048. 872 61. Power JF, Carere CR, Lee CK, Wakerley GL, Evans DW, Button M, White D, 873 Climo MD, Hinze AM, Morgan XC. 2018. Microbial biogeography of 925 874 geothermal springs in New Zealand. Nature communications 9:2876.

30 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

875 62. Sharp CE, Brady AL, Sharp GH, Grasby SE, Stott MB, Dunfield PF. 2014. 876 Humboldt’s spa: microbial diversity is controlled by temperature in geothermal 877 environments. The ISME journal 8:1166. 878 63. Ghosh W, Dam B. 2009. Biochemistry and molecular biology of lithotrophic 879 sulfur oxidation by taxonomically and ecologically diverse bacteria and archaea. 880 FEMS microbiology reviews 33:999-1043. 881 64. Huber JA, Butterfield DA, Baross JA. 2003. Bacterial diversity in a subseafloor 882 habitat following a deep-sea volcanic eruption. FEMS Microbiology Ecology 883 43:393-409. 884 65. Barahona S, Dorador C, Zhang R, Aguilar P, Sand W, Vera M, Remonsellez F. 885 2014. Isolation and characterization of a novel Acidithiobacillus ferrivorans strain 886 from the Chilean Altiplano: attachment and biofilm formation on pyrite at low 887 temperature. Research in microbiology 165:782-793. 888 66. Chen L, Ren Y, Lin J, Liu X, Pang X, Lin J. 2012. Acidithiobacillus caldus sulfur 889 oxidation model based on transcriptome analysis between the wild type and sulfur 890 oxygenase reductase defective mutant. PloS one 7:e39470. 891 67. Duquesne K, Lieutaud A, Ratouchniak J, Muller D, Lett MC, Bonnefoy V. 2008. 892 Arsenite oxidation by a chemoautotrophic moderately acidophilic Thiomonas sp.: 893 from the strain isolation to the gene study. Environmental microbiology 10:228- 894 237. 895 68. Chen L-X, Méndez-García C, Dombrowski N, Servín-Garcidueñas LE, Eloe- 896 Fadrosh EA, Fang B-Z, Luo Z-H, Tan S, Zhi X-Y, Hua Z-S. 2018. Metabolic 897 versatility of small archaea Micrarchaeota and Parvarchaeota. The ISME journal 898 12:756. 899 69. Colman DR, Poudel S, Hamilton TL, Havig JR, Selensky MJ, Shock EL, Boyd 900 ES. 2018. Geobiological feedbacks and the evolution of thermoacidophiles. The 901 ISME journal 12:225. 902 70. Ward L, Taylor MW, Power JF, Scott BJ, McDonald IR, Stott MB. 2017. 903 Microbial community dynamics in Inferno Crater Lake, a thermally fluctuating 904 geothermal spring. The ISME journal 11:1158.

31 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

905 71. Nagase H, Ose Y, Sato T, Ishikawa T. 1984. Mercury methylation by compounds 906 in humic material. Science of the total environment 32:147-156. 907 72. Barton LL, Tomei-Torres FA, Xu H, Zocco T. 2015. Metabolism of metals and 908 metalloids by the sulfate-reducing bacteria, p 57-83, Bacteria-Metal Interactions. 909 Springer. 910 73. Pester M, Knorr K-H, Friedrich MW, Wagner M, Loy A. 2012. Sulfate-reducing 911 microorganisms in wetlands–fameless actors in carbon cycling and climate 912 change. Frontiers in microbiology 3:72. 913 74. Baldi F, Olson GJ. 1987. Effects of Cinnabar on Pyrite Oxidation by Thiobacillus 914 ferrooxidans and Cinnabar Mobilization by a Mercury-Resistant Strain. Applied 915 And Environmental Microbiology 53:772-776. 916 75. Benoit JM, Gilmour CC. 1999. Sulfide controls on mercury speciation and 917 bioavailability to methylating bacteria in sediment. Environmental Science & 918 Technology 33:951. 919 76. Barnett MO, Harris LA, Turner RR, Stevenson RJ, Henson TJ, Melton RC, 920 Hoffman DP. 1997. Formation of Mercuric Sulfide in Soil. Environmental 921 Science & Technology 31:3037-3043. 922 77. Barkay T, Gillman M, Turner RR. 1997. Effects of dissolved organic carbon and 923 salinity on bioavailability of mercury. Appl Environ Microbiol 63:4267-4271. 924 78. Boyd ES, Yu R-Q, Barkay T, Hamilton TL, Baxter BK, Naftz DL, Marvin- 925 DiPasquale M. 2017. Effect of salinity on mercury methylating benthic microbes 926 and their activities in Great Salt Lake, Utah. Science of the Total Environment 927 581:495-506. 928 79. Moreau JW, Gionfriddo CM, Krabbenhoft DP, Ogorek JM, DeWild JF, Aiken 929 GR, Roden EE. 2015. The effect of natural organic matter on mercury 930 methylation by Desulfobulbus propionicus 1pr3. Frontiers in microbiology 931 6:1389. 932 80. Graham AM, Aiken GR, Gilmour CC. 2012. Dissolved organic matter enhances 933 microbial mercury methylation under sulfidic conditions. Environmental science 934 & technology 46:2715-2723.

32 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

935 81. Ndu U, Christensen GA, Rivera NA, Gionfriddo CM, Deshusses MA, Elias DA, 936 Hsu-Kim H. 2018. Quantification of Mercury Bioavailability for Methylation 937 Using Diffusive Gradient in Thin-Film Samplers. Environmental Science & 938 Technology. 939 82. Gilmour CC, Elias DA, Kucken AM, Brown SD, Palumbo AV, Schadt CW, Wall 940 JD. 2011. Sulfate-reducing bacterium Desulfovibrio desulfuricans ND132 as a 941 model for understanding bacterial mercury methylation. Applied and 942 Environmental Microbiology 77:3938-3951. 943 83. Malcolm EG, Schaefer JK, Ekstrom EB, Tuit CB, Jayakumar A, Park H, Ward 944 BB, Morel FM. 2010. Mercury methylation in oxygen deficient zones of the 945 oceans: no evidence for the predominance of anaerobes. Marine Chemistry 946 122:11-19. 947

948

33 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

949 Main text Figures and Tables 950

800

700

600

500

400 Ranked CoverageRanked

300

200

100

0 Bacteria Archaea Thiomonas Acidiphilium Ferroplasma Euryarcheota Thermotogales Acidithiobacillus Planctomycetes Rhodospirillales Thermoplasmata Thaumarchaeota Aciduliprofundum Methylacidiphilum Betaproteobacteria Thermoplasmatales Deltaproteobacteria Granulicella Granulicella mallensis Acidithiobacillus Acidithiobacillus caldus ARMAN Micrarchaeum ARMAN ARMAN Parvarchaeum cuprina Metallosphaera Desulfurella Desulfurella acetivorans Aciduliprofundum boonei Aciduliprofundum Candidatus Micrarchaeota Candidatus Micrarchaeota Candidatus Acidithiobacillus Acidithiobacillus ferrivorans Thermoplasma acidophilum Thermoplasma capsulatum Acidobacterium Thermoplasmatales Thermoplasmatales archaeon Candidatus Micrarchaeum acidiphilum Micrarchaeum Candidatus Thermoplasmatales I-plasma Thermoplasmatales archaeon Tiger Bath (NW1) Cub Bath (NW2) 951

952 Figure 1. Rank abundance by scaffold coverage of ribosomal protein S3 within binned and unbinned 953 genomes from A) Tiger (NW1) and B) Cub (NW2) bath metagenomes. Analyses and annotations 954 performed in ggKbase (ggkbase.berkeley.edu/). Genomic bin phylogeny used for ribosomal S3 proteins from 955 genomic bins, while scaffold phylogeny to the lowest common ancestor is given for unbinned ribosomal 956 proteins. When multiple ribosomal protein S3 had the same taxonomic classification the average coverage is 957 shown, with error bars representing standard deviation. Values ranked by NW2 coverage.

34 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

) n ion) tio at c d is) ) redu n ) ion) e fur oxi n t e tio tio e) yla lfat s tion a a ry) s ogenesa o h cta n n cta u io lat u at u yl th ne oxid Ae (methyl ite red a c-m B (methyl f X pathwayA (metha (sul cA c r g Sul O HHg Hg MerR (regMer operonMerA MerB (redproteinsDsr (demet operon (TPC)S (su Mc Meth 8 2 2 2 1 10 NW1_Acidithiobacillus_caldus_61_717

5 2 1 8 NW2_Acidithiobacillus_caldus_56_340

1 NW1_Archaea_54_220

rage 1 1 5 3 NW2_Thermoplasma_acidophilum_43_190

e 1 NW2_Thermotogales_40_160

4 1 1 2 4 NW2_Acidithiobacillus_sp_62_99

1 6 3 NW2_Aciduliprofundum_36_97

reasing cov c NW1_Thermoplasmatales_archaeon_I-plasma_37_96

1 NW2_Thermotogales_39_88

1 11 4 NW1_Thaumarchaeota_unknown_1

1 1 12 1 1 NW1_Unbinned

4 3 1 1 4 NW2_Acidithiobacillus_ferrivorans_54_62 NW2_Thermoplasmatales_archaeon_I-plasma_36_52

Genomic bins: by de NW2_Thermoplasmatales_archaeon_I-plasma_36_52_rel

6 10 8 2 NW2_Desulfurella_acetivorans_33_49_multi

5 2 1 2 2 NW2_Thiomonas_64_37

2 2 5 5 1 NW1_Thermoplasmata_unknown_1

2 1 1 NW2_Granulicella_mallensis_62_33 NW2_unbinned_3_by_ESOM (unknown phylogeny)

1 1 1 4 3 NW2_Acidobacterium_capsulatum_related 3 6 3 1 NW1_Desulfurella_acetivorans_33_25 NW2_Ferroplasma_35_24 1 1 20 15 14 1 41 11 9 2 NW2_Unbinned 3 NW2_ARMAN_Micrarchaeum_46_22 1 1 2 1 NW1_Thermoplasma_acidophilum_45_19 2 1 13 5 NW2_Acidobacterium_capsulatum_62_18 1 NW2_Thermotogales_39_15_partial 2 NW1_Candidatus_Micrarchaeota_2 2 1 6 3 NW2_Planctomycetes_54_12 1 8 4 NW2_Deltaproteobacteria_41_12 3 3 3 NW2_Acidobacterium_capsulatum_60_11 3 1 8 NW2_Acidithiobacillus_caldus_59_11 1 NW2_Bacteria_34_11 NW2_ARMAN_Parvarchaeum_36_9 NW2_Candidatus_Micrarchaeota_49_9 3 1 3 NW2_Methylacidiphilium_9_partial NW1_Candidatus_Micrarchaeota_1 2 NW2_Micrarchaeota_sp_43_8 7 1 NW2_unbinned_2_by_ESOM (Archaea) 1 1 3 1 NW2_Rhodospirillales_68_7 958 3 1 5 1 NW2_Acidiphilium_65_6

959 Figure 2. Heatmap showing functional proteins from various biogeochemical pathways associated 960 with Hg within genomic bins and unbinned (NGAWHA_1_UNK, NGAWHA_2_UNK) scaffolds from 961 Tiger (NW1) and Cub (NW2) bath metagenomes. Intensity of color refers to number of genes from each 962 bin that encode for the enzyme, operon, or pathway involved in mercury, sulfur, or methane cycling; values 963 provided in cell for reference. Genomic bins were ordered by the consensus coverage for all scaffolds 964 within the bin (highest to lowest). Analyses and annotations were performed in ggKbase 965 (ggkbase.berkeley.edu)

35 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

966

YNP (7) Bootstrap values Caldisphaera lagunensis YNP (2) >90% YNP (3) 70-90% YNP (1) 50-70% Sulfolobus islandicus YNP (4) <50% not shown YNP (1) Sulfolobales archaeon Acidianus hospitalis W1 Crenarchaeota YNP metagenome Acidianus hospitalis W1 Tiger Bath metagenome YNP (2) YNP (1) Cub Bath metagenome YNP (2) Metallosphaera sedula YNP (3) Candidatus Micrarchaeum acidiphilum ARMAN-2 YNP (1) DPANN Candidatus Methanoperedens nitroreducens Archaea Candidatus Methanoperedens sp. BLZ1 Candidatus Azambacteria bacterium GW2011_GWC2_45_7b YNP (2) YNP (2) YNP (1) Thermoplasma_acidophilum_43_190 Thermoplasmatales archaeon A-plasma Thermoplasma volcanium Thermoplasma_acidophilum_45_19 Thermoplasma acidophilum Archaea_54_220 Euryarchaeota Metallosphaera yellowstonensis Thermoplasmatales archaeon E-plasma Thermoplasmatales archaeon I-plasma Unbinned_scaffold_2792 Unbinned_scaffold_20509 Unbinned_scaffold_10131 Thaumarchaeota_unknown_1 Unbinned_scaffold_12915 Unbinned_scaffold_22979 YNP (5) YNP (2) YNP (1) YNP (4) Aciduliprofundum_36_97 Unbinned_scaffold_13696 Thermoplasmata_unknown_1 Unbinned_scaffold_22384 Geobacter (3 sequences) Deltaproteobacteria Rhodospirillales_68_7 Gracilimonas tropica Planctomycetes_54_12 Bacteroidetes Firmicutes (10 sequences) Firmicutes Acidithiobacillus_ferrivorans_54_62 Acidithiobacillus ferrivorans Acidithiobacillus thiooxidans Acidithiobacillus_61_717 Acidithiobacillus caldus Gammaproteobacteria Acidithiobacillus_sp_62_99 Unbinned_scaffold_17399 Unbinned_scaffold_8960 Acidithiobacillus_caldus_56_340 & Unbinned_scaffold_21790 Acidithiobacillus_61_717 Bacteria Citrobacter amalonaticus Shewanella sp. ANA-3 Salmonella enterica Unbinned_scaffold_16732 Thiomonas_64_37 Betaproteobacteria Unbinned_scaffold_2349 Acidobacterium_capsulatum_related & Unbinned_scaffold_10566 Unbinned_scaffold_12511 Verrucomicrobia bacterium LP2A Ralstonia pickettii Pseudomonas fluorescens Burkholderia sp. AU4i Proteobacteria (19 sequences); Unbinned_scaffold_16495 Proteobacteria Planctomycetes_54_12 Oceanicola granulosus dihydrolipoamide dehydrogenase gamma proteobacterium HTCC5015 pyridine nucleotide-disulfide oxidoreductase paralogs 967 0.2 Oceanospirillaceae bacterium ASP10-02a pyridine nucleotide-disulfide oxidoreductase 968 Figure 3. Maximum likelihood tree showing MerA phylogeny of 31 sequences pulled from assembled 969 Tiger (NW1) and Cub (NW2) Bath metagenomes using ggKbase. Genomic bins or scaffold ID (when 970 MerA was unbinned) are given in bold. Included in analysis are 113 MerA homologues, including 46 971 sequences from Yellowstone National Park (YNP) metagenomes (Inskeep et al. 2013). Trees were 972 constructed using Le Gascuel amino-acid substitution model with gamma distribution in MEGA6 (Tamura 973 et al. 2013). Bootstrapped with 100 replications. The initial Neighbor-Joining tree was constructed with 974 pairwise distances estimated using a JTT model. Positions with less than 90% site coverage (e.g. alignment 975 gaps, missing data or ambiguous bases) were excluded. A total of 419 positions were used in the final dataset. 976

36 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

977 978 Figure 4. Maximum likelihood tree showing HgcA phylogeny of reads obtained from the Cub Bath 979 (NW2) metagenome using HMM-search. Included in analysis are 183 sequences, including 56 HgcA 980 homologues from YNP metagenomes, pulled from JGI (Table S9). Amino acid sequences are compared to 981 HgcA homologues from known and predicted methylators. Also included are HgcAB fused proteins from 982 hyperthermophilic bacteria and archaea, and carbon monoxide dehydrogenase/acetyl-CoA synthase subunit 983 gamma (HgcA paralogs) from non-methylators. Trees were inferred from the Le Gascuel amino-acid 984 substitution model with gamma distribution in MEGA6 (Tamura et al. 2013). Bootstrapped with 100 985 replications. The initial Neighbor-Joining tree was constructed with pairwise distances estimated using a 986 JTT model. Positions with less than 93% site coverage (e.g. alignment gaps, missing data or ambiguous 987 bases) were excluded. A total of 55 positions were used in the final dataset. Groups (I-IX) designate distinct 988 subtrees that contain HgcA from hot spring metagenomes. Representative reference sequences are labeled 989 within each subtree, while phylogeny of all branches are indicated by color. 990

37 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Bootstraps: >90% Deltaproteobacteria supercluster (76) 70-90% 50-70% <50%, not shown Thermodesulfovibrio (9) Candidatus Magnetobacterium casensis (JMFO00000000) Hydrothermal vent clones (3) NW2_Deltaproteobacteria_41_12 (NW2_scaffold_16) I Uncultured prokaryote clone A24 (JQ304764) Uncultured family-level lineage 12 NW1_Desulfurella_acetivorans_33_25 (NW1_scaffold_265) II NW2_Desulfurella_acetivorans_33_49_multiple (NW2_scaffold_372) NW2_Desulfurella_acetivorans_33_49_multiple (NW2_scaffold_11471) Desulfurella acetivorans A63 (AHF97346; AHF97345) Unclassified Thermophilic Environmental Sequences (4) Wadden sea sediment sequence (2)

Gemmatimonas sp. SG8_17 (KPJ93667; KPJ93666) reductive bacterial type DsrAB Gemmatimonas sp. SG8_38_2 (KPK65896; DsrA only) Uncultured DsrAB lineage 9 (4) Uncultured DsrAB lineage 8 (2) NW2_unbinned (NW2_scaffold_96) Schlöppnerbrunnen fen peat soil clone (GU371960) Uncultured family lineage 8 Schölppnerbrunnen fen clone II lineage 8 (2) NW2_Acidobacterium_capsulatum_62_18 (NW2_scaffold_126) III NW2_Acidobacterium_capsulatum_62_18 (NW2_scaffold_1823) Hot spring clones (4) Hydrothermal clones (2) Uncultured DsrAB lineage 7 (4) Hot spring clone BSP-102 (AY237257) Uncultured family-level lineage 5 Hot spring clone BLV-6 (AY237259) Uncultured family-level lineage 5 Uncultured DsrAB lineage 6 (4) Archaeoglobus (10) Thermodesulfobium narugense (CP002690) Firmicutes supercluster (11) Uncultured DsrAB lineage 2 (3) Caldiserica bacterium (AQSQ01000030) Unclassified Rasner Möser fen soil clone Rm9 (GU127961) Unclassified Uncultured DsrAB lineage 3 (2) Carboxydothermus (2) New York peatland soil clone W20 (DQ855256) Uncultured DsrAB lineage 4 (2) Sporomusaceae (4) Gordonibacter pamelaceae (2) Gracilibacter sp. BRH c7a (KUO62649; KUO62648) Desulfosporosinus (3) Desulfitobacterium (3) Moorella sp. 60-41 (KUK12290; KUK12289) Thermanaeromonas toyohensis ToBE DSM 14490 (Gi04363) Candidate division pSL4 archaeon (ASPF01000004) Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria, Chlorobi (18) oxidative bacterial NW1_Thermoplasmata_unknown_1 (NW1_scaffold_101)

NW1_Thermoplasmata_unknown_1 (NW1_scaffold_477) IV reductive archaeal archaeal reductive NW2_Thermoplasma_acidophilum_43_190 (NW2_scaffold_1422) NW2_Aciduliprofundum_36_97 (NW2_scaffold_4482) Caldivirga (3) Thermocladium sp. ECH-B (KUO93286; KUO93285) Vulcanisaeta (3) NW1_Thaumarchaeota_unknown_1 (NW1_scaffold_219) NW2_unbinned (NW2_scaffold_21371) V NW2_unbinned (NW2_scaffold_10538; DsrA only) Pyrobaculum (16) Candidatus Rokubacteria bacterium CSP1-6 (KRT71374; KRT71373) NW2_Planctomycetes_54_12 (NW2_scaffold_1267) VI Moorella thermoacetica, copy 2 (2)

991 0.5

992 Figure 5. Phylogenetic analysis of DsrAB by Maximum Likelihood method. The 15 DsrAB homologues 993 in this study were pulled from idba-UD assembled Tiger and Cub metagenomes using ggKbase. They were 994 compared to 218 reference DsrAB sequences pulled from the Dome database, including 15 representative 995 DsrAB sequences from thermophilic environments (Muller et al. 2015). There were six distinct groups of 996 DsrAB homologues found at Ngawha. The tree was constructed using MEGA 6 (Tamura et al. 2013) with 997 Le Gascuel 2008 model with gamma distribution, pairwise distances estimated using a JTT model. All 998 positions with less than 95% site coverage were eliminated, including alignment gaps, missing data, and 999 ambiguous bases. There were a total of 492 positions in the final dataset, with 100 bootstrap replicates. 1000

1001

38 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1002 1003 Figure 6. Conceptual model of biogeochemical cycling of mercury (Hg), sulfur (S), and iron (Fe) in Hg- 1004 enriched, sulfidic, low pH mesothermal springs. Gaseous elemental mercury (Hg0) (as well as Hg(II)) 1005 from deep geological sources enters the surface waters of the springs where it becomes oxidized to Hg(II) 1006 (enhanced by chloride (Cl))(32) and then complexes with sulfides (S2-) to produce cinnabar (red

1007 rhombohedral symbols; HgS(s)). Round icons represent microbial-mediated reactions, white are primarily 1008 aerobic-associated mechanisms, and black are primarily anaerobic. Sulfur-oxidizing bacteria (SOB) 1009 equipped with the SOX-pathway along with Fe-oxidizing bacteria and archaea (FeOBA) are able to enhance

1010 dissolution of metal sulfides, such as pyrite (silver rhombohedral symbols; FeS2) and HgS(s). Sulfate- 1011 reducing bacteria (SRB) and Fe-reducing bacteria (FeRB) further mediate the redox chemistry of S, Fe, and 1012 Hg. As Hg(II) becomes bioavailable to microbes it can be reduced to Hg(0) by microbes equipped with 1013 mercuric reductase (MerA) or methylated to MeHg by HgcAB-equipped microbes. MeHg can be

1014 demethylated by MerB-equipped microbes to Hg(II) and CH4, and then reduced to Hg(0) by MerA. At the 1015 surface of the springs, photoreduction can also contribute to Hg(II) reduction to Hg(0), as well as the 1016 degradation of MeHg. Photolytic oxidation may also counter Hg(0) volatilization from surface waters, 1017 keeping Hg(II) in spring water to be transformed by microbes or partitioned to sulfide minerals. Advective 1018 mixing of spring waters ensures that Hg species travel across the redox boundaries that likely partition the 1019 Mer-equipped microbes to oxic surface waters from the HgcAB-equipped microbes that likely occupy the 1020 anaerobic sediment/water boundary. 1021

39 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1022 Table 1. Anion and cation measurements in filtered water from hot springs in the Ngawha 1023 geothermal field.

- - - - 2- 2- F Cl Br NO3 SO4 S FeT Sample Site GPS Coordinates (Lat, Long) mg L-1 mg L-1 mg/L-1 mg L-1 mg L-1 mg L-1 mg L-1

Tiger Bath (TS1) -35.4075155°, 173.8554535° 2.1 470 2.79 b.d.l 1217.8 1.82 ± 0.13 12.7

Cub Bath (TS2) -35.4075150°, 173.8554755° 1.9 30 0.91 0.74 319.0 6.45 ± 0.15 6.06 Hole to right of Tiger Bath (TS3) -35.4075160°, 173.8554315° b.d.l 395 b.d.l b.d.l 1176.9 0.87 ± 0.47 n.m Hole 3m from Tiger Bath (TS4) -35.4075073°, 173.8554202° b.d.l 591 b.d.l b.d.l 1017.0 n.m n.m

Cinnabar Bath (GN1) -35.4049290°, 173.8600079° b.d.l 876 b.d.l b.d.l 10.2 n.m n.m

Jubilee Bath (GN2) -35.4049108°, 173.8600182° b.d.l 1000 b.d.l b.d.l 91.9 n.m n.m

Twin Bath (GN3) -35.4049380°, 173.8600082° 2.6 1073 5.02 b.d.l 119.9 n.m n.m

Kotanitanga (NS1B) -35.4056976°, 173.8583837° b.d.l 661 b.d.l b.d.l 155.1 n.m n.m Adjacent to Favourite Bath (NS3B) -35.4055000°, 173.8583547° b.d.l 639 b.d.l b.d.l 192.5 n.m n.m

Waikato Bath (NS4B) -35.4055887°, 173.8584129° b.d.l 537 b.d.l b.d.l 54.1 n.m n.m 1024 n.m = not measured, b.d.l = below detection limit (0.0005 mg L-1) 1025 1026

40 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.03.933291; this version posted February 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1027 Table 2. Mercury analyses of Ngawha Geothermal Field (NGF) hot springs: methylmercury (MeHg) and 1028 total mercury (HgT) and the fraction of total Hg as MeHg (as %) in filtered water samples, solid total mercury 1029 (THg(s)) in hot spring sediments, and gaseous elemental mercury (Hg(0)) above the springs. Temperature 1030 and pH measurements taken during each sampling campaign are provided as a range. Published reference 1031 and regional background values are included.

T (°C) MeHg HgT %MeHgT THg(s) Hg(0) -1 -1 -1 -1 Water pH (ng L ) (ng L ) of HgT (µg g ) (ng L ) Ngawha Geothermal Field 35.3– 2.9– Tiger Bath (TS1) 40.5 3.0 1.56 1031 0.15 3467 25 24.5– 3.6– Cub Bath (TS2) 33.4 3.2 13.9 1042 1.33 1274 4.3 35.9– 2.9– Hole to right of Tiger Bath (TS3) 40.3 3.8 4.33 16711 0.03 7000 7.5 Hole 3 m from Tiger Bath (TS4) 21.6 3.96 0.53 2160 0.02 - - Cinnabar Bath (GN1) 40.4 6.72 5.36 1461 0.37 - - Jubilee Bath (GN2) 38 6.45 1.28 244 0.52 - - Twin Bath (GN3) 34.1 6.44 1.61 722 0.22 - - Kotanitanga (NS1B) 32.4 6.54 4.28 487 0.88 - - Adjacent to Favourite Bath (NS3B) 26.5 7.14 3.57 253 1.41 - - Waikato Bath (NS4B) 45.5 6.7 0.90 326 0.28 - - Regional values 17000– 0.017– 2000– NGF soils and waters(1) - - - 710000 - 0.35 78500 Tiger Bath (1) 32–57 5–8 - 710000 - - - NGF non-thermal waters (1) - - - 300 –500 - - - 0.61– 35– 0.06– Puhipuhi mine waters (2) - - 1.02 69.6–240 0.3-1.6 1748 0.50 Wairua River, Northland (3) - - - 100–1000 - <100 - Yellowstone National Park Nymph Lake spring #1, YNP (4) - - 0.026 170 0.02 - - Nymph Lake spring #2, YNP (4) - - 0.43 520 0.08 - - Roadside West Spring, YNP (4) 65 6.4 0.08 200 0.04 - - YNP thermal waters and microbial 36.4– 2.2– <0.013– 0.0049 mats (4) 80 8.6 0.43 15-520 - –120 - Acidic thermal springs YNP (5) - - <0.025 38-94 - - - Acidophilic microbial mats, YNP 0.03- (5) - - 1.62 2-71 - 0.16-18 - YNP filtered thermal waters and 21.9– 1.8– 0.38– microbial mats (6) 74.4 9.3 - 7.3–144 - 20.52 - Reference standards and values North American lakes (7,8) - - 0.003–6 0.04–74 8.11 - - North American river & streams 0.078– (7,8) - - 0.55 1–7 7.86 - - NZ Drinking Water Standard - - - 7000 - - - US EPA, acute aquatic life water standard - - - 2400 - - - US EPA, chronic aquatic life water standard - - - 770 - - - 1032 (1) Davey & Moort 1985, (2) Gionfriddo et al. 2014, (3) Hoggins & Brooks 1973, (4) King et al. 2006, (5) Boyd et al. 1033 2009, (6) Wang et al. 2011, (7) Krabbenhoft et al. 2007, (8) USEPA(2001)

41