Environmental formation of methylmercury is controlled by synergy of inorganic mercury bioavailability and microbial mercury-methylation capacity
Abstract
Methylmercury (MeHg) production is controlled by the bioavailability of inorganic divalent mercury (Hg(II)i) and Hg-methylation capacity of the microbial community (conferred by the hgcAB gene cluster). However, the relative importance of these factors and their interaction in the environment remain poorly understood. Here, metagenomic sequencing and a full-factorial MeHg formation experiment were conducted across a wetland sulfate gradient with different microbial communities and pore water chemistries. From this experiment, the relative importance of each factor on MeHg formation was isolated. Hg(II)i bioavailability correlated with the dissolved organic matter composition, while the microbial Hg-methylation capacity correlated with the abundance of hgcA genes. MeHg formation responded synergistically to both factors. Notably, hgcA sequences were from diverse taxonomic groups, none of which contained genes for dissimilatory sulfate reduction. This work expands our understanding of the geochemical and microbial constraints on MeHg formation in situ and provides an experimental framework for further mechanistic studies.
INTRODUCTION
Methylmercury (MeHg) is the most toxic and bio-accumulative form of mercury (Hg) in the environment (Wiener et al., 2003) and poses significant health risks to humans, fish and wildlife worldwide. MeHg formation by microbes in the environment occurs primarily under low-redox conditions and is dependent on the bioavailability of inorganic divalent Hg (Hg(II)i) and the Hg-methylating capacity of the microbial community (Hsu-Kim et al., 2013). The geochemical constraints on Hg(II)i bioavailability for microbial uptake are controlled by ligand complexation of Hg(II)i by primarily organic and inorganic reduced S (Graham et al., 2013; Hsu-Kim et al., 2013; Poulin, Gerbig, et al., 2017), whereas Hg(II)i methylation capacity is conferred by the presence of the hgcAB gene cluster (Gilmour et al., 2013; Parks et al., 2013). Previous studies individually investigated the importance of Hg(II)i bioavailability (Graham et al., 2013; Hsu-Kim et al., 2013; Jonsson et al., 2012) or microbial communities (Christensen et al., 2018; Compeau & Bartha, 1985; Gilmour et al., 1992; Schaefer et al., 2020) to MeHg formation. In amended sediment slurries with simplified ligand chemistries, neither Hg(II)i bioavailability nor overall microbial activity were strictly limiting; rather, each was shown to influence MeHg production under different conditions (Kucharzyk et al., 2015). In anoxic brackish waters, gene abundance or expression of hgcA combined with predicted abundance of Hg(II)i-sulfide species correlated to MeHg production potentials (Capo, Feng, et al., 2022). A critical step in understanding environmental MeHg production requires the simultaneous quantitative examination of the relative importance of geochemical versus microbial factors to MeHg formation in complex environmental systems, paired with comprehensive measurements of the ligand chemistry and microbial Hg-methylators, which has not yet been done.
Ligand complexation and geochemical speciation of Hg(II)i ultimately govern Hg(II)i availability for uptake by microbial cells (Hsu-Kim et al., 2013), which can have long-lasting effects on Hg methylation (Jonsson et al., 2012) and incorporation into the food web (Jonsson et al., 2014). Under environmental conditions lacking inorganic sulfide, Hg(II)i is exclusively bound to thiol groups (SRed) in dissolved organic matter (DOM) (Haitzer et al., 2002). Conversely, under sulfidic conditions common in anoxic sediments, nano-particulate metacinnabar (β-HgS) dominates Hg(II)i speciation (Gerbig et al., 2011; Poulin, Gerbig, et al., 2017). The bioavailability of Hg(II)i associated with nano-particulate β-HgS is greatest at low-to-intermediate sulfide concentrations (≤ ~0.3 mg/L) and in the presence of DOM of high aromaticity (Graham et al., 2013) and thiol content (Graham et al., 2017). Under very high sulfide concentrations (> ~3 mg/L), nano-particulate β-HgS becomes crystalline and aggregates (Poulin, Gerbig, et al., 2017), decreasing Hg(II)i bioavailability for methylation (Zhang et al., 2012). Further, sulfidic conditions enhance the concentration of thiol groups in DOM via sulfurization reactions (Poulin, Ryan, et al., 2017; Vairavamurthy & Mopper, 1987), which enhances the bioavailability of Hg(II)i to methylation (Bouchet et al., 2018; Graham et al., 2017). However, the net effect of sulfide versus DOM composition and concentration on bioavailability of Hg(II)i in complex environmental systems is still unclear. In pure culture, efforts to minimize the geochemical complexity of study systems has relied on the use of cysteine as a low-molecular weight analogue to thiols in DOM, which promotes the bioavailability of Hg(II)i under laboratory conditions (Gilmour et al., 2018; Graham et al., 2012; Schaefer & Morel, 2009). However, the environmental relevance of cysteine controlling the bioavailability of Hg(II)i has yet to be tested.
The environmental factors controlling the microbial Hg-methylation capacity are poorly understood. Sulfate-reducing bacteria (SRB) have long been considered a primary microbial guild affiliated with MeHg production due to field experiments under molybdate inhibition (Compeau & Bartha, 1985) or sulfate amendment (Gilmour et al., 1992). However, using the hgcAB gene cluster as a molecular marker (Parks et al., 2013), we now recognize the high metabolic and phylogenetic diversity of putative Hg-methylating organisms (Gilmour et al., 2013; Gionfriddo et al., 2016; McDaniel et al., 2020; Podar et al., 2015). Several recent field studies in sulfate-enriched environments observed that SRB accounted for only a small percentage of the hgcA abundance, while the majority of hgcA abundance was associated with fermentative and syntrophic bacteria or methanogenic archaea (Bae et al., 2014; Jones et al., 2019, 2020; Peterson et al., 2020). Attempts to link hgcA abundance to MeHg levels or production have documented mixed results, possibly due to Hg(II)i bioavailability, limited methodologies, and/or changes in hgcA expression/HgcA activity (Bae et al., 2019; Bravo et al., 2016; Capo, Feng, et al. 2022; Christensen et al., 2019; Liu et al., 2018; Millera Ferriz et al., 2021; Roth et al., 2021; Tada et al., 2020). Complex biogeochemical conditions and interdependent microbial communities in the environment also make it difficult to extend observations from laboratory culture studies (Gilmour et al., 2013, 2018; Yu et al., 2018) to natural conditions and anticipate which microbial processes are linked to MeHg production. These complexities may explain the varied response of MeHg production to experimental molybdate inhibition (Bae et al., 2014; Bouchet et al., 2018; Cleckner et al., 1999; Gascón Díez et al., 2016; Schaefer et al., 2020) or sulfate amendment (Gilmour et al., 1992; Jones et al., 2020). Overall, the relationships between microbial community metabolism, hgcA gene content and activity, Hg-methylation capacity of the microbial community, and ultimately MeHg production and accumulation are still poorly understood.
To address these knowledge gaps, we quantified the relative importance of Hg(II)i bioavailability and microbial Hg-methylation capacity on MeHg formation across a sulfate gradient in the Florida Everglades and paired this with microbial community and pore water chemistry characterization. First, a full-factorial MeHg formation experiment was performed using pore waters and intact peat cores collected at six sites across a sulfate gradient to quantify the relative methylation potential of both the pore water and microbial communities in the peat. Next, shotgun metagenomic sequencing was performed to quantify and characterize the microbial community fraction carrying the hgcA gene. Together, these complementary approaches facilitated the isolation of geochemical factors governing Hg(II)i bioavailability from the microbial Hg-methylation capacity (i.e., hgcA abundance). Furthermore, genome-resolved metagenomic analyses identified the metabolic potential of microbes with hgcA in the peat cores. This study demonstrates the synergy between geochemical and microbial factors required for environmental MeHg formation, shows that hgcA gene abundance is a reliable marker for the Hg-methylation capacity of the microbial community, and provides a valuable experimental framework to target processes underlying MeHg formation in diverse aquatic environments.
EXPERIMENTAL PROCEDURES
Site information and geochemical gradients
The Florida Everglades is an ideal “field laboratory” to study the impact of sulfate concentration and DOM concentration/composition on MeHg production due to the combination of extensive atmospheric Hg deposition (Krabbenhoft et al., 1998; Orem et al., 2020) with long-term geochemical gradients stemming from release points of agricultural run-off (Figure S1) (Orem et al., 2011). In this study, six field sites in Water Conservation Areas 2 (WCA-2) and 3 (WCA-3) and Arthur R. Marshall Loxahatchee National Wildlife Refuge (LOX), were chosen (Table S1; Figure S1) to span a range of sulfate, sulfide, and DOM concentration and composition (Figure S2). Ambient MeHg concentrations in the peat were lowest in WCA-2, intermediate at the downgradient sites in WCA-3, and highest at 3A-F and LOX8 (Figure 1A). Ambient pore water MeHg concentrations were similarly low at WCA-2, but relatively consistent concentrations were observed across WCA-3 and LOX8 (Figure 1B). Geochemical data and analytical methods are available in Science Base (Tate et al., 2023).

MeHg formation assays
Details for all materials and methods are provided in the Supporting Information. Briefly, at each of the six sites, filtered pore waters and 18 replicate peat cores (7.6 cm diameter) were collected (Figure S1). A suite of water quality and geochemical measurements, including sulfide, sulfate, DOC concentration, and DOM specific ultraviolet absorbance at 254 nm (SUVA254), were made on the pore waters using established methods (Figures S2 and S3) (Poulin, Ryan, et al., 2017). Three laboratory-prepared “pore waters” were prepared using purged ultrapure water, all with a background solution matched to the average ionic concentration of Everglades pore water, including 1 mg/L sulfate: “F1 HPOA DOM”, which contained 90 mg/L of the hydrophobic organic acid fraction (HPOA) of DOM from the F1 site of the Everglades (Poulin, Ryan, et al., 2017); “Cysteine”, with 40 μM of cysteine; and “Control”, which had no additional organic ligands. The molar concentration of cysteine matched the concentration of reduced S in the F1 HPOA DOM in pore water solutions (see Supporting Information). The 201Hg(II)i tracer was pre-equilibrated with each of the filtered natural and lab-prepared pore waters for a minimum of 4 h. From each of the six field sites, duplicate peat cores were injected with one of the nine different pore water-equilibrated 201Hg(II)i tracers in a full-factorial experimental design, for a total of 108 incubations (Figure S4). 1.5 ml of equilibrated tracer was injected every 1 cm from 2 to 10 cm below the top of the core. Injection concentrations were targeted such that the 201Hg(II)i amendments to the peat would be 13% of the ambient HgT. After 24 h, the peat cores were frozen to stop the experiment and shipped back to the laboratory on dry ice. The top 2 cm of the core (mostly biofilm) was removed, and the next 4 cm (solid peat) were homogenized for analysis. This was previously shown to be a highly active zone of MeHg production (Gilmour et al., 1998). Excess Me201Hg was quantified by distillation and isotope dilution with inductively coupled plasma mass spectroscopy (ICP-MS; iCAP, Thermo Scientific) (DeWild et al., 2002; Hintelmann & Evans, 1997), while excess total 201Hg (201HgT) was measured using BrCl oxidation, SnCl2 reduction, and ICP-MS (Hintelmann & Evans, 1997; Olund et al., 2004). Net Me201Hg production (NMP) was defined as follows: NMP = excess Me201Hg/excess 201HgT × 100. Relative methylation potential values were calculated for the pore water (RMPmatrix) and the peat cores (RMPpeat) by normalizing net Me201Hg production to the highest net Me201Hg production value for any incubation using the same peat core or pore water, respectively (Figure S4). A synchronized permutation test using the two-way analysis of variance format (Basso et al., 2009) with log-transformation was done to test for main and interaction effects of the peat core and pore water source on net Me201Hg production. Model selection was done using Akaike Information Criteria on linear models generated using different combinations of factors. Linear models were used to test for relationships between combinations of RMPmatrix, RMPpeat, geochemical parameters, and hgcA abundance. Incubation data are available in Table S2.
Metagenomics workflow
DNA was isolated from the peat by phenol: chloroform extraction and purified by alcohol precipitation (Lever et al., 2015) then sequenced at QB3 Genomics at the University of California, Berkeley. DNA reads from duplicate metagenomes were coassembled using both metaSPADes and MegaHit (Li et al., 2015; Nurk et al., 2017) and open reading frames were predicted from the assembled contigs using Prodigal (Hyatt et al., 2010). HgcA sequences were identified using a custom Hidden Markov Model (Peterson et al., 2020) and manually verified to contain conserved domains (Parks et al., 2013), then dereplicated across assemblies using CD-HIT (Fu et al., 2012). Confirmed HgcA sequences were aligned with the Hg-MATE database (Gionfriddo et al., 2021) and a maximum-likelihood tree was generated using RAxML (Stamatakis, 2014). This, along with a custom workflow (Gionfriddo et al., 2020), was used to assign a taxonomic affiliation to each hgcA gene. Normalized abundance of hgcA was calculated by first determining the average nucleotide coverage over the hgcA-containing contig, then dividing this by the mean coverage of 16 single-copy ribosomal protein genes (Sorek et al., 2007). Thus, the normalized hgcA abundance is presented as a percentage of the total microbial community. Genomic bins containing hgcA were manually binned using CONCOCT (Alneberg et al., 2014) and refined in Anvi'o (Eren et al., 2015). These bins were taxonomically classified (Chaumeil et al., 2019) and their metabolic pathways identified (Zhou et al., 2022). Raw metagenomic reads are available through the National Center for Biotechnology Information under BioProject accession ID PRJNA808433 and the assemblies, bins and HgcA protein sequences are available through the Open Science Framework (https://osf.io/8muzf/). Code for all analyses and figures is stored on Github (https://github.com/petersonben50/Everglades).
RESULTS AND DISCUSSION
Net Me201Hg production in the peat core assays, quantified as the percent of excess 201HgT measured as excess Me201Hg, ranged from 0% to 8% after 24 h across the six different peat cores incubated with nine pore water matrices (n = 108 peat cores total; Figures 1C, S5; Table S2). The inset in Figure 1C shows how the effect of the two variables (peat core vs. pore water matrix source) on net Me201Hg production can be interpreted in the plot. Across all assays, the response of net Me201Hg production to the pore water matrix source, visualized as the spread between differently coloured lines in Figure 1C, was consistent regardless of the peat core source (Figures 1C and S6). Changes in net Me201Hg production in response to the peat core source, visualized as the increase in net Me201Hg production along the x-axis, were less consistent depending on the pore water matrix, following one of two similar but distinct patterns, discussed in detail below (Figures 1C and S7). Synchronized permutation testing (Basso et al., 2009) showed that both the peat core source (p < 0.0001) and the pore water matrix source (p < 0.0001) had significant effects on net Me201Hg production. There was also a statistically significant interaction effect (p < 0.0001). This interaction effect is visible in Figure 1C in the two modestly different trends in the peat core effects depending on the source of the pore water matrix (Figures 1C and S7). Four of the pore water matrices (Everglades F1 HPOA, 2A-N, 3A-O, and LOX8) facilitated a dramatic increase in net Me201Hg production in cores from sites 2A-A to 3A-O, but then net Me201Hg production levelled off or modestly decreased in cores from sites across WCA-3A and to LOX8. In contrast, the other five pore water matrices resulted in modest increases in net Me201Hg production in cores from high to low sulfate, with a notable increase in net Me201Hg production in cores from sites 3A-F and LOX8 (Figures 1C and S7). One possible source of this interaction is demethylation activity, which has been shown in isotopically enriched incubations after 8 h in peat from the Everglades and would increase as Me201Hg concentrations increased (Gilmour et al., 1998). Another possibility is the complete methylation of the bioavailable pool of 201Hg(II)i in the high-producing incubations (Janssen et al., 2016). Either explanation is supported by the observation that pore water matrices that produce the plateau also produced the most Me201Hg and would result in an underestimation of Hg-methylation capacity, particularly at 3A-F and LOX8. Additional possible causes of this interaction effect are discussed in detail in the Supporting Information. Despite this interaction, the relative effects of each pore water matrix and peat core were notably consistent (Figures 1C, S6 and S7). Model selection identified a linear model without the interaction effect as the best fit for the data. Together, this suggests that the independent effects of the peat core and the pore water matrix had a notably larger effect on net Me201Hg production than the interaction between them.
Geochemical controls on Hg(II)i methylation
The pore water matrix source had a significant and consistent influence on net Me201Hg production across the six peat cores (Figure 1C and S6), likely by establishing the bioavailability of the Hg(II)i tracer, as demonstrated in previous studies (Gilmour et al., 1998; Graham et al., 2012, 2017; Jonsson et al., 2012, 2014; Moreau et al., 2015). Thus, the influence of the pore water matrix on net Me201Hg production reflects changes in 201Hg(II)i bioavailability due to ligand chemistry (Figure S6). Regardless of the source of the peat core, the Everglades F1 HPOA DOM solution yielded the most bioavailable 201Hg(II)i, which is consistent with previous observations and attributed to the high aromaticity and thiol content of this DOM (Graham et al., 2013; Moreau et al., 2015; Poulin, Ryan, et al., 2017). Conversely, the control solution always resulted in the lowest net Me201Hg production. Surprisingly, the cysteine solution, which matched the thiol concentration of the Everglades F1 HPOA DOM, also resulted in exceptionally low net Me201Hg production, comparable to the control matrix. The net Me201Hg production of the six natural pore waters were distributed between that of the Everglades F1 HPOA DOM and the control matrix. Those collected from sites closest to where aromatic DOM and sulfate-rich canal water is released to the marshes (Sites 2A-N and 3A-O) consistently promoted the highest net Me201Hg production of the natural pore waters, whereas pore water from sites distant to canal inputs (e.g., sites 2A-A and 3A-F) exhibited notably lower net Me201Hg production levels. LOX8 pore waters resulted in intermediate Me201Hg formation.
To quantify the variation in net Me201Hg production due to pore water matrix source for comparison to geochemical parameters, we calculated a “relative methylation potential” for each of the different pore water matrices (RMPmatrix) as follows. First, incubations were grouped by the source of the peat core; then, net Me201Hg production for each incubation was divided by the highest net Me201Hg production value of any incubation within the group (Figures S4 and S8). Of the measured geochemical properties of the natural and laboratory prepared pore water solutions (DOC, DOM SUVA254, inorganic sulfide, UV absorbance), DOM SUVA254 exhibited the strongest correlation with RMPmatrix (adjusted R2 = 0.494; p < 0.001; Figure 2). Significant correlations with RMPmatrix were also observed for DOC concentration (adjusted R2 = 0.405; p < 0.001; Figure S9a) and UV254 absorbance (adjusted R2 = 0.376; p < 0.001; Figure S9b), the latter being a parameter that captures differences in both DOC concentration and DOM aromaticity. This is consistent with extensive prior work showing that high aromatic DOM increases Hg(II)i bioavailability and facilitates MeHg formation in pure culture experiments (Graham et al., 2012, 2013; Moreau et al., 2015), as more aromatic DOM is not expected to stimulate microbial metabolism in the cores over the short timeframe of the experiments. Sulfide and RMPmatrix were positively correlated, albeit weakly (adjusted R2 = 0.055; p = 0.008; Figure S9c). While it is known that high sulfide concentrations can inhibit MeHg production (Benoit et al., 1999; Graham et al., 2013) due to the formation of crystalline and aggregated β-HgS of low bioavailability (Poulin, Gerbig, et al., 2017; Zhang et al., 2012), aromatic DOM with high SRed content can inhibit crystalline β-HgS formation and promote Hg(II)i availability to methylation (Graham et al., 2017; Poulin, Ryan, et al., 2017). We interpret the high pore water RMPmatrix from site 2A-N to indicate that even the highest sulfide concentration (3.5 mg/L) was insufficient to suppress Hg(II)i methylation under the high DOC concentration and high DOM SUVA254 (Graham et al., 2013). Sulfate was not correlated to RMPmatrix (Figure S9d; R2 = 0.026, p = 0.051). We infer that in this system and during the duration of the experiments, the DOM SUVA254 is a more important variable than sulfide for controlling Hg(II)i bioavailability. This is highlighted by the similarity in RMPmatrix of 2A-N pore water and F1 HPOA DOM, which were collected from proximal locations, albeit several years apart, and have similar DOC concentrations and DOM SUVA254, but very different sulfide concentrations (Figures 2 and S9).

The Hg(II)i-cysteine solution yielded very low net MeHg formation across all six study sites (Figures 1C and S8) despite having thiol concentration equimolar to the F1 HPOA solution, which is inconsistent with previous pure culture laboratory studies (Graham et al., 2012; Schaefer et al., 2011; Schaefer & Morel, 2009). This is particularly striking considering that cysteine levels in the environment are far lower than those used in this study (Zhang et al., 2004). This may be explained by cysteine's lack of aromaticity needed to sterically inhibit nano-particulate β-HgS growth (Gerbig et al., 2011; Poulin, Gerbig, et al., 2017; Zhang et al., 2012), or the rapid degradation of cysteine under environmental conditions (Chu et al., 2016) that allows the 201Hg(II)i tracer to sorb to the peat, thus diminishing its bioavailability. Regardless of the mechanism, the findings support that cysteine-complexed Hg(II)i is unlikely to be environmentally relevant for MeHg formation. In total, the results are in general concurrence with laboratory studies demonstrating that aromatic, thiol-rich DOM plays a key role in promoting Hg(II)i bioavailability (Graham et al., 2013), with the notable disagreement that cysteine did not promote Hg(II)i bioavailability in nature.
Microbial controls on Hg(II)i methylation
The source of the peat cores also had a significant effect on net Me201Hg production. The filtered pore water matrices controlled the bioavailability of the 201Hg(II)i tracer but contained no microbes, whereas the influence of the peat cores on net Me201Hg production reflected the Hg-methylation capacity of the microbial community in the incubation. The net Me201Hg production response to the peat cores was split in one of two similar patterns depending on the pore water matrix used in the incubation, as described above (Figures 1C and S7). However, it was always very low in peat cores from sites with high sulfate and sulfide (2A-N and 2A-A) and increased in peat cores from sites with low to non-detectable sulfate and sulfide.
The relative methylation potential of the peat cores (RMPpeat) was quantified to identify the relationship between the Hg-methylation capacity of the microbes and the abundance of the hgcA gene. RMPpeat was calculated by grouping all incubation assays by the pore water matrix and normalizing net Me201Hg production to the highest level of Me201Hg produced within that group (Figure S4). As observed with the raw net Me201Hg production data (Figure S7), the RMPpeat was lowest in peat cores from high sulfate sites (Site 2A-N, 2A-A) and increased systematically in cores with decreasing sulfate (Figure S10). Eighty-seven unique hgcA genes across the six sites were identified using shotgun metagenomic sequencing of the peat cores (Tables S3–S5; additional details in Supporting Information). Normalized hgcA abundance correlated significantly and positively with RMPpeat (adjusted R2 = 0.494; p < 0.0001; Figure 3A) due to an increase in hgcA abundance from sites with high sulfate to low sulfate (Figure 3B). Previous attempts to correlate hgcA abundance to MeHg levels have documented mixed results (Bae et al., 2019; Bravo et al., 2016; Capo, Feng, et al., 2022; Christensen et al., 2019; Liu et al., 2018; Millera Ferriz et al., 2021; Roth et al., 2021; Tada et al., 2020), possibly due to changes in Hg(II)i bioavailability or methodological constraints of qPCR-based hgcA quantification (McDaniel et al., 2020). Other studies suggest additional genes may confer MeHg production (Bowman et al., 2020; Munson et al., 2018). However, the correlation between hgcA gene abundance and the microbial Hg-methylation capacity suggests that hgcA is the dominant MeHg formation pathway in Everglades peat. Recent work showed decreases in hgcA alpha diversity to coincide with decreases in MeHg production thought to be independent of changes in Hg(II)i bioavailability (Jones et al., 2020). This may have reflected an overall decrease in hgcA abundance, as we also observed an increase in hgcA richness and evenness coincident with an increase in hgcA abundance and Hg-methylation capacity (Figure S11). Transcription of hgcA, while thought to be constitutive based on experiments in culture (Gilmour et al., 2011; Goñi-Urriza et al., 2015), varies between different organisms in the environment (Capo, Broman, et al., 2022; McDaniel et al., 2020). Overall trends in hgcA gene abundance versus expression were consistent in brackish waters (Capo, Feng, et al., 2022), but exhibited divergent trends in sediments from the same site (Capo, Broman, et al., 2022). Collectively, this shows that the controls on hgcA gene expression are not well understood. Additionally, the relationship between hgcA expression and MeHg production by individual cells is unclear. However, the correlation observed here between hgcA and RMPpeat suggests that in this ecosystem at least, the hgcA abundance, independent of hgcA transcription or metabolic activity of the Hg-methylators, is sufficient to identify the Hg-methylation capacity of the microbial community.

Next, the community composition and metabolic potential of the microbes with hgcA (hgcA+) were evaluated to establish potential linkages between biogeochemical processes and MeHg formation (Table S6). Details of the metabolic analyses are provided in the Supporting Information. The trends in the beta diversity of hgcA are not aligned with the trends in the peat core RMPpeat or sulfate levels (Figure 3C). Methanogenic archaea-associated hgcA genes accounted for the largest portion of hgcA abundance (37%–55% of the total hgcA coverage; Figures 3B, S12, and S13). These hgcA sequences were exclusively associated with predicted hydrogenotrophic or methylotrophic methanogens, but not acetoclastic methanogens, which is consistent with previous work (Gilmour et al., 2018) (Figure 3D). Methanogen-associated mcrA genes increased in abundance across the sulfate gradient (Figure S14a). A comparison of methanogen-associated hgcA and mcrA abundances indicates that 50%–100% of methanogens across the sulfate gradient carried hgcA (Figures 3B and S14a). The remainder of the hgcA sequences were from a diverse group of hgcA+ bacteria, including Chloroflexi, Aminicenantes, Spirochaetes, and non-SRB Syntrophobacterales, among other rarer groups (Table S5). Metabolic pathway analysis of reconstructed hgcA+ genomes from Chloroflexi, Aminicenantes, and Syntrophobacterales and comparison of unbinned hgcA to genomes with closely related genes confirmed that all classified non-methanogen-associated hgcA+ microbes in these peat cores are fermentative (Figure 3D). Several hgcA genes were highly divergent from the hgcA sequences in the reference database, resulting in 0%–5% of the hgcA genes (by abundance) being unclassified with no information on the metabolic potential. Importantly, none of the hgcA sequences were expected to be associated with SRB (Figure 3D). This is not due to a lack of SRB, as SRB accounted for up to 4.5% or 7.5% (depending on the marker used) of the microbial population, increasing in abundance across the sulfate gradient (Figure S14b). This surprising finding is discussed in detail below. Although subtle differences in the taxonomic affiliation of hgcA+ community members were observed across the six sites, the relative contribution of organisms from different levels of the microbial food web to the hgcA pool do not differ substantially with respect to sulfate levels (Figures 3D and S13). Thus, we hypothesize that the metabolic pathways directly contributing to MeHg production are likely consistent across the sulfate gradient. This consistency and the linear relationship between RMPpeat and overall hgcA abundance (Figure 3A) suggest that the observed differences in the Hg-methylation capacity are governed by abundance of Hg-methylators rather than their metabolic activity.
While recent studies have shown SRB to account for a small percentage of the microbial community even under sulfidic conditions (Capo, Broman, et al., 2022; Jones et al., 2019, 2020; Peterson et al., 2020), including within the greater Everglades ecosystem (Bae et al., 2014), none of these have confirmed the absence of SRB-associated hgcA sequences. Molybdate inhibition experiments have shown the importance of sulfate reduction for MeHg production in Everglades peat, particularly in the high sulfide sites (Bae et al., 2014; Gilmour et al., 1998). Together, this suggests that SRB play an indirect role in MeHg production in the peat that is not represented by the abundance of SRB-associated hgcA genes. It is possible that rare hgcA-carrying SRB, undetected due to insufficient sequencing depth, influenced MeHg formation; however, this is unlikely given the complete absence of SRB-associated hgcA sequences and the close linear relationship between hgcA and MeHg production capacity (Figure 3A). Alternatively, SRB could indirectly control MeHg formation by controlling carbon and energy flow, both above (fermentation) and below (methanogenesis), through the anaerobic microbial food web, thus influencing the metabolic activity of hgcA+ organisms in the community. For example, under anoxic conditions, fermentative organisms break down and convert large organic molecules into smaller carbon compounds, but they rely on syntrophs or respiratory organisms to consume these products (Figure 3D) (Arndt et al., 2013). SRB can oxidize smaller organic molecules either by reducing sulfate or in syntrophy with methanogens, where they ferment volatile fatty acids (e.g., propionate, butyrate) to methanogenic substrates (acetate, CO2 and hydrogen) (Sieber et al., 2012). The parallel increase in mcrA and dsrAD with decreasing sulfide levels may indicate increasing levels of SRB-methanogenic syntrophy (Figure S14). These syntrophic interactions are known to enhance MeHg formation (Yu et al., 2018), and given the high hgcA abundance within the methanogenic community, may contribute to the observed increase in Hg-methylation capacity (Figure S10). If hgcA-containing methanogens are reliant on SRB through syntrophy, this could explain the inhibition of MeHg formation by molybdate as well (Bae et al., 2014; Cleckner et al., 1999; Gilmour et al., 1998; Newport & Nedwell, 1988). Overall, we hypothesize that terminal respiration is dominated by sulfate reduction and hydrogenotrophic methanogenesis at sulfate-enriched sites, whereas low sulfate sites exhibit greater fermentation of small organic acids by SRB coupled syntrophically to hydrogenotrophic methanogenesis and acetate consumption by acetoclastic methanogens.
MeHg production and accumulation
A major knowledge gap in the field is whether Hg(II)i bioavailability or Hg-methylation capacity is the rate-limiting step for MeHg production in environmental systems. By isolating these two effects, we were able to compare them to each other and to the production of MeHg. There was no correlation between Hg(II)i bioavailability (RMPmatrix) and Hg-methylation capacity (RMPpeat; Figure 4), suggesting that the ability of microbial communities to methylate Hg was not linked to how much bioavailable Hg was present. This supports the hypothesis that MeHg production is not the “native function” of hgcA, as has been proposed in previous work (Parks et al., 2013; Smith et al., 2015). We also compared how each factor influenced MeHg production under in situ conditions, termed “native MeHg production”. Neither factor was solely limiting for native MeHg production; rather, a synergy of the two factors was required. Native MeHg production was only high at sites where both the pore water RMPmatrix and the microbial community RMPpeat were high (Figure 4). For example, peat from sites 3A-O and 3A-N had similar hgcA+ microbial communities (Figure 3B,C) that also corresponded to nearly identical RMPpeat values (Figure 3A). However, native MeHg production at site 3A-O was much higher due to higher RMPmatrix values (Figure 4), which are linked to the higher DOM SUVA254 promoting Hg(II)i bioavailability (Figure 2). Conversely, the pore water RMPmatrix at site 2A-N was similar to that at site 3A-O, but the low RMPpeat at 2A-N was responsible for the very low native MeHg production (Figure 4). This synergistic effect is consistent with work in brackish marine waters that showed predicted concentrations of Hg(II)i-sulfide complexes and gene abundance or expression of hgcA collectively correlated with MeHg production potential (Capo, Feng, et al., 2022). Together, these data suggest that Hg(II)i bioavailability and the Hg-methylation capacity of the microbial community both control MeHg formation under environmental conditions and that either of them can limit MeHg production (Figure 5).


Another major knowledge gap is how MeHg production and the factors that govern it relate to ambient MeHg pools in sediment and pore water which have accumulated over time. In this study, MeHg concentrations in the peat (Figure 1A) and pore water (Figure 1B) increased systematically with decreasing sulfate. However, the pattern in Me201Hg formation under native conditions was much different, showing high net Hg formation rates at 3A-O and LOX8, but low at the other four sites (Figure 1C). Additionally, we observed MeHg production up to 3.4% of the tracer under ambient conditions at 3A-O but the %MeHg values at this site are only 1.5%. These observations may be due, in part, to other biogeochemical processes influencing ambient MeHg levels that were not measured in this study. One likely possibility is that much of the ambient Hg(II)i is sorbed strongly to the peat and is not available for methylation, but it is unclear how this would change across the sulfate gradient. Another likely process is MeHg degradation, which does occur in Everglades peat (Gilmour et al., 1998; Marvin-DiPasquale & Oremland, 1998). The demethylation gene merB was detected at all sites and decreased in abundance with decreasing sulfate concentrations, in opposition to the trend in hgcA (Figure S15); however, demethylation occurs at a consistent rate across the sulfate gradient in Everglades peat (Marvin-DiPasquale & Oremland, 1998). Despite these other potential effects, calculated RMPpeat values and ambient MeHg concentration in the peat were strongly and positively correlated (adjusted R2 = 0.885; p = 0.003; Figure S16a), while RMPmatrix values were not correlated with ambient MeHg concentration in the peat (adjusted R2 = −0.250; p = 0.9759; Figure S16b). We propose that RMPpeat represents the longer-term, site-specific MeHg production potential, whereas RMPmatrix represents the potential shorter-term (seasonal) effects of aqueous ligands promoting Hg(II)i methylation.
Role of sulfate in controlling MeHg production in the environment
This study offers new insights into the long-standing hypothesis that sulfate and sulfide are the master variables controlling MeHg production and add complexity to the well-documented linkages between anthropogenic sulfate loading and MeHg production across the Everglades (Gilmour et al., 1998; Hurley et al., 1998; Orem et al., 2020) and other peatlands worldwide (Coleman Wasik et al., 2012, 2015; Mitchell et al., 2008; Poulin et al., 2019; Tjerngren et al., 2012). The current model is that at high sulfide concentrations, Hg(II)i bioavailability is drastically reduced, due to the formation of crystalline nano-particulate β-HgS of lower bioavailability (Gerbig et al., 2011; Gilmour et al., 2018; Poulin, Gerbig, et al., 2017; Zhang et al., 2012), while low sulfate concentrations result in lowered SRB activity, leading to reduced MeHg production. Collectively, this was used to explain the “Goldilocks curve” observed in the Everglades, where MeHg formation is maximum under intermediate sulfate and sulfide concentrations (Gilmour et al., 2007; Orem et al., 2020). However, we showed that the low MeHg production at high sulfate sites was due to reduced Hg-methylation capacity by the microbial community, despite the Hg(II)i bioavailability being high. For example, 2A-N pore water resulted in high MeHg production when paired with cores containing high hgcA abundance, but hgcA at 2A-N was low, resulting in low MeHg production under native conditions (Figures 1C and 3B). At the low sulfate end of the gradient, microbial Hg-methylation capacity was highest (high hgcA abundance), but the low bioavailability of Hg(II)i led to reduced MeHg production levels. For example, the peat cores from 3A-F produced high MeHg when provided with 201Hg(II)i equilibrated with pore water from 2A-N, LOX8, or F1 HPOA DOM due to the high hgcA content at 3A-F, but the low bioavailability of 201Hg(II)i in 3A-F pore water drove low MeHg production under native conditions (Figure 1C).
Thus, the influence of anthropogenic sulfate levels on the overall redox status of wetlands and SRB activity on Hg methylation in the Florida Everglades and similarly impacted wetlands is more complicated than previously described. Sulfate reduction exerts control on Hg(II)i bioavailability in a number of ways. While sulfide can precipitate Hg(II)i (Poulin, Gerbig, et al., 2017), reducing its overall bioavailability, this is unlikely to be a dominant process in sites with high concentrations of aromatic DOM, given the high Hg(II)i bioavailability at the high sulfide sites (Figure S8). On the other hand, moderate levels of sulfide, in the presence of aromatic DOM, can enhance methylation by promoting the formation of poorly crystalline nano-particulate β-HgS (Gerbig et al., 2011; Poulin, Gerbig, et al., 2017). Enhanced sulfate reduction can also promote peat degradation, increasing the concentration of high-SUVA254 DOM (Aiken et al., 2011; Luek et al., 2017) and DOM SRed content via sulfurization (Poulin, Ryan, et al., 2017) in wetland pore waters; both of these enhance the bioavailability of Hg(II)i to methylation (Graham et al., 2012, 2013, 2017; Jonsson et al., 2012; Zhang et al., 2012). The effects of sulfate loading on the Hg-methylating microbial community are less clear. Overall, both hgcA abundance and RMPpeat decreased with higher overall sulfate concentrations (Figures 3B and S10), consistent with the lack of hgcA+ SRB and previous work showing a decrease in hgcA diversity and estimated Hg-methylation capacity with increased long-term sulfate loading (Jones et al., 2020). However, past work has clearly shown that SRB activity is important for MeHg production in the Everglades (Bae et al., 2014; Gilmour et al., 1998; Orem et al., 2020). Thus, we propose that SRB influence MeHg production indirectly by stimulating overall microbial metabolism, possibly through consuming fermentation products (Arndt et al., 2013) and/or by stimulating methanogenic activity through syntrophy (Sieber et al., 2012). Ultimately, functional assays and the deployment of next-generation physiology experiments (Hatzenpichler, 2020) are needed to further probe how the metabolic activity and interactions of the microbial community influence MeHg production.
CONCLUSIONS AND ENVIRONMENTAL IMPLICATIONS
This study presents a dual examination of microbial and geochemical controls on MeHg production in natural peatlands, providing new insights into both the synergy between the hgcA+ fraction of the microbial community and geochemical controls on Hg(II)i bioavailability, and the direct and indirect roles of sulfate. The abundance of metabolically diverse populations with hgcA confer robust potential for Hg-methylation; when paired with geochemical conditions that promote Hg(II)i bioavailability, one can expect MeHg formation and a high potential for food web uptake and MeHg bio-magnification to toxic levels. Given the widely recognized importance of sulfate on spatial and temporal trends in MeHg formation in wetlands globally (Coleman Wasik et al., 2012, 2015; Mitchell et al., 2008; Orem et al., 2020; Poulin et al., 2019; Tjerngren et al., 2012), a mechanistic understanding of the role of sulfate loading on MeHg production is critical. Peatland ecosystems are experiencing seasonal and long-term increases in sulfate levels in response to increased sulfate use in agricultural practices (Hinckley et al., 2020) and coastal wetland inundation with seawater sulfate (Chambers et al., 2019). The results here suggest that ecosystems with lower sulfate levels but high DOM concentration and SUVA254 quality may be well-poised to form MeHg when sulfate levels increase due to the indirect effects of sulfate on Hg(II)i bioavailability. We postulate that the bioavailability of Hg(II)i in environments with lower DOC levels (e.g., marine waters) may be modulated by inorganic sulfide in addition to DOM (Capo, Feng, et al. 2022). We still have much to learn on how environmental conditions such as sulfate concentrations influence hgcA distribution and how interactions between different metabolic guilds influence overall MeHg formation rates. Notwithstanding, this study provides an important framework by which the individual factors that influence MeHg production can be isolated and highlights the need for more advanced methods to elucidate the mechanism by which these factors drive MeHg formation.
AUTHOR CONTRIBUTIONS
Benjamin D. Peterson: Conceptualization (equal); data curation (equal); formal analysis (lead); investigation (equal); methodology (equal); validation (equal); visualization (lead); writing – original draft (lead); writing – review and editing (lead). David P. Krabbenhoft: Conceptualization (lead); data curation (lead); formal analysis (equal); funding acquisition (lead); investigation (equal); methodology (lead); resources (equal); supervision (equal); validation (equal); writing – original draft (equal); writing – review and editing (equal). Katherine D. McMahon: Conceptualization (equal); formal analysis (equal); funding acquisition (equal); methodology (equal); resources (equal); supervision (equal); visualization (supporting); writing – original draft (supporting); writing – review and editing (supporting). Jacob M. Ogorek: Data curation (equal); formal analysis (equal); methodology (equal); resources (equal); supervision (equal); validation (equal). Michael T. Tate: Data curation (equal); formal analysis (equal); investigation (equal); methodology (equal); validation (equal). William H. Orem: Funding acquisition (supporting); investigation (supporting); methodology (supporting); resources (supporting); writing – review and editing (supporting). Brett A. Poulin: Conceptualization (equal); data curation (lead); formal analysis (equal); funding acquisition (equal); investigation (equal); methodology (equal); project administration (lead); resources (equal); supervision (lead); validation (equal); visualization (equal); writing – original draft (equal); writing – review and editing (equal).
ACKNOWLEDGEMENTS
This work was supported by the U.S. Geological Survey (USGS) Priority Ecosystems Science (PES) program (support to David P. Krabbenhoft and Brett A. Poulin). Katherine D. McMahon and Benjamin D. Peterson were supported by a grant from the U.S. National Science Foundation (CBET-1935173). Benjamin D. Peterson was supported by the U.S. National Science Foundation Graduate Research Fellowship Program. All mercury analyses were performed at the Upper Midwest Water Science Center in the Mercury Research Laboratory (U.S. Geological Survey, Madison, WI). Computational analyses were done on the Wisconsin Energy Institute computing cluster, which is funded by the Great Lakes Bioenergy Research Center as part of the U.S. Department of Energy Office of Science. Any use of trade, product or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.
CONFLICT OF INTEREST
The authors declare no conflicts of interest.
Open Research
DATA AVAILABILITY STATEMENT
All metagenomic, incubation and geochemical data are available as supplemental data or online, as outlined in the methods section.