Where Brimstone Meets Quicksilver: Interactions between Agricultural Sulfur Applications and Methylmercury Production along a Vineyard to Wetland Gradient

Abstract

Mercury (Hg), specifically as methyl Hg (MeHg), is a potent neurotoxin with significant consequences for humans and wildlife. This study examines the interactions between sulfur used as a fungicide in California, U.S., vineyards and Hg methylation. We collected surface soils and soil porewater across a vineyard to wetland gradient situated at the terminus of the Napa Valley winegrowing region for analyses of sulfate, total Hg, and MeHg. We found that sulfate runoff from vineyards is transported into adjacent wetlands via drainage ditches and a grassland transition zone. In addition, concentrations of MeHg and percent of Hg present as MeHg in the wetland and drainage ditch were approximately three to eight times higher than values measured in vineyard and grassland transition soils, while potential net Hg methylation rates (k _meth) over a 48 h time period were approximately ten times higher in the wetland compared to the vineyard. Stimulation of Hg methylation in the wetlands and drainage ditch is likely driven by wet, reduced soil conditions. These results show that while Hg methylation is possible in vineyards and adjacent land covers, downstream drainage ditches and wetlands–which intercept sulfate added in vineyards–may be the most notable zones for MeHg production.

Introduction

Atmospheric sulfur (S) deposition has been decreasing in the United States (U.S.) and elsewhere due to national policies, such as the Clean Air Act and its Amendments, that have reduced anthropogenic S emissions. − Yet, concurrent with decreased atmospheric S deposition, agricultural S inputs have increased in many crop systems around the world to account for the reduced atmospheric S inputs, intensification of agriculture, and elevated risk of crop disease. − Current agricultural S use now far exceeds the peak S load from acid deposition, as S amendments are used as fungicides, fertilizers, and regulators of soil pH for a variety of crops; essentially, S use is ubiquitous in agriculture. − For example, in the Midwestern U.S., S fertilizer use has more than doubled from 1985–2015, resulting in increased average agricultural S additions of 0.10 kg S ha^–1 yr^–1 compared to atmospheric S deposition decreases of 0.12 kg S ha^–1 yr^–1 over this same time period.

In agricultural systems, S is applied in many different forms, from elemental sulfur (S⁰) to sulfate (SO₄ ^2–) and organic S. When applied as S⁰, it is often rapidly oxidized to SO₄ ^2–. Once in the environment, SO₄ ^2– can be leached from agricultural soils into nearby ecosystems, including wetlands, where it can then be reduced to sulfide (S^2–) (Figure ). , While the production of S^2– itself is concerning, since it is toxic to many aquatic organisms, − the microbial process of SO₄ ^2– reduction also has implications for mercury (Hg) biogeochemical cycling. , Mercury in its methylated form is a potent neurotoxin with significant consequences for humans and wildlife. , Sulfate reducing bacteria are one of the primary microbial clades that convert inorganic Hg into the neurotoxic and bioavailable form of methyl Hg (MeHg) under anaerobic conditions. ,

1.

Conceptual diagram of sulfur (S) and mercury (Hg) inputs, transport, and processing across a vineyard through grassland transition to wetland gradient, including the production of methylmercury (MeHg).

Higher anthropogenic inputs of S are, therefore, likely to increase MeHg production, as has been shown in laboratory and field studies. – This linkage between S and MeHg was particularly evident in previous ecosystem-level studies in the Everglades Water Conservation Area, where elevated MeHg production was directly connected to S inputs in the upstream Everglades Agricultural Area. Orem and colleagues (2020) found that moderate SO₄ ^2– concentrations in runoff stimulated high rates of MeHg production, and low levels of SO₄ ^2– in runoff were associated with minimal MeHg production. Yet, while there is often a correlation between SO₄ ^2– concentrations and MeHg production, high S^2– production can also prevent Hg methylation − due to the formation of relatively inert HgS, complexation with dissolved organic matter, − and/or lower microbial community capacity for methylation. This relationship has been previously described as the “Goldilocks curve,” whereby both very low and high SO₄ ^2– concentrations inhibit MeHg production while moderate SO₄ ^2– concentrations stimulate high MeHg production. Most studies describing the relationship between SO₄ ^2– availability and MeHg production have occurred in lowland wetlands (e.g., refs ,,, ); few studies have examined this interaction within upland agricultural soils where S is applied, as well as those wetlands directly downstream. Note also that not all studies examining MeHg and SO₄ ^2– concentrations report this inhibitory relationship, as evidenced by research on seasonal prairie wetland ponds.

The use of S as a fungicide in Napa Valley, California (U.S.) vineyards provides a unique opportunity to examine the environmental consequences of S inputs and their interaction with Hg. In these vineyards, farmers use S in isolation of other nutrient additions (e.g., nitrogen, phosphorus), allowing for the effects of S inputs to be studied without confounding variables. Additionally, the wetlands and bays (San Francisco and San Pablo Bays) in this region have high Hg concentrations in sediments, fish, , and birds. These elevated Hg concentrations are the result of historical gold mining in the Sierra Nevada Mountains that used Hg to extract gold, mining of Hg-containing cinnabar ore, − and atmospheric Hg deposition. This setting allows for studying the interactions of agricultural S additions and soil Hg processing across a gradient from the upland soils to the neighboring wetlands–an environmental context that is common to other areas around the world.

In this study, we asked: (1) Do agricultural S inputs get transported from vineyard soils to nearby wetlands? (2) How do rates of Hg methylation compare between cropland soils and adjacent wetlands? and (3) How do patterns of Hg methylation vary across wet and dry seasons? We address these research questions across a vineyard to wetland gradient within a study area situated at the terminus of the Napa Valley winegrowing region and Napa River Watershed. Between the vineyard and wetland, we include drainage ditches and the intervening grassland transition zone. We expected that SO₄ ^2– would be transported from the vineyards through the drainage ditch or grassland transition and into the nearby wetlands, particularly during initial rainfall in the wet season. Due to the anoxic conditions found in the drainage ditch and wetlands along with the potential transport of SO₄ ^2– from the vineyards into these sites, we expected Hg methylation to be highest in the wetlands and drainage ditch. With higher rates of Hg methylation, we also expected to measure elevated MeHg concentrations and potential MeHg production relative to values measured in vineyard soils. Alternatively, we recognized that if the wetlands have high S inputs from marine-derived SO₄ ^2–, there could be suppressed Hg methylation in the wetlands, according to the Goldilocks curve. We further hypothesized that the middle of the wet season would have enhanced methylation compared to the beginning of the season at each of the three land use/land cover sites due to wet, reducing soil conditions.

Materials and Methods

Study Area

We collected samples along a gradient from vineyard through grassland transition or drainage ditch to wetland in the Carneros region of the Napa River Watershed, California, U.S. (38.21°N,122.35°W; Figure ). The vineyard contains Chardonnay on 101–14 rootstock, planted in rows perpendicular to the edge of the bay. The vineyard ends ∼310 m from the edge of the bay, as determined during the dry season. It is located adjacent to other vineyard blocks and access roads. As of 2021, these vineyards used no-until agriculture with cover crops and organic amendments between rows. During the growing season, winegrowers irrigate with water collected in a reservoir. During the preceding wet season, reservoirs fill with stream runoff and are supplemented with recycled wastewater originating from Sonoma and Napa. The 1103 km² Napa River Watershed contains primarily forest (36% of the watershed), grassland/shrubland (30% of the watershed), and vineyards (13% of the watershed), with the remaining land uses/land covers consisting of agriculture, grazing pastures, and urban-residential centers (Figure ). The wetlands at this study site are part of San Pablo Bay, which is connected to the San Francisco Bay estuary.

2.

Location of study site within Napa Valley, California. Colors represent land use/land cover. Inset shows the sampling schematic, with each point representing a sampling location.

The region has a Mediterranean climate. Rains generally occur in October through March (i.e., “wet” or “dormant” season), with wine grapes grown in April to September (i.e., “dry” or “growing” season). Annual precipitation ranges from 107 to 792 mm, with an average of 482 mm. At times, vineyard soils are saturated to more than 0.5 m in depth. During the growing season, S⁰ is sprayed weekly to biweekly as a fungicide to prevent powdery mildew at this and other vineyards.

Sample Collection

We established 24 sampling nodes distributed across three transects spanning a land use/land cover gradient in the study area (Figure ). On each of the three transects, two nodes were in the vineyard and three nodes each were in the grassland transition and wetland land covers. On one of the transects, the three grassland transition nodes were located within a drainage ditch (hereafter, “drainage ditch” sites). Note that during the dry season and periods of the wet season, the wetland nodes and drainage ditch nodes were dry, while during parts of the wet season, the grassland transition nodes were completely saturated (under several centimeters of water). At each of the 24 sampling nodes across the vineyard, grassland transition, drainage ditch, and wetland, we installed tension lysimeters (SoilMoisture, Inc.) to a depth of 30 cm from the soil/sediment surface in December 2022, immediately after the first rainfall of the season.

We collected porewater samples from the lysimeters using polycarbonate VacLok 60 mL syringes or high-density polyethylene (HDPE) bottles under vacuum. Previous studies demonstrated that the lysimeters do not affect ∂³⁴S-SO₄ ^2– values across space and time, and lysimeter blanks from the lab showed that the lysimeters do not leach Hg (values of 0.09–0.20 ng L^–1 Hg when Milli-Q water was run through the lysimeters, including after passing a 2 ng Hg L^–1 solution).

We collected porewater samples after every rain event and subsequent wet period from December 2022 to February 2023 (10 sampling events; Figure ), when porewater was present. We collected additional porewater samples monthly in March, April, and May 2023, as well as during two sampling periods the following year in December 2023 and February 2024. Porewater samples were collected by filtering through a 0.45 μm filter (Thermo Scientific, sterile poly­(ether sulfone) [PES]) in the field into an acid-washed HDPE bottle for SO₄ ^2– analysis. At several time points in December 2022 and January 2023 (n = 11), we collected unfiltered porewater samples for Hg analysis at nodes where we had previously observed large porewater volumes. We did not observe any particles in these unfiltered porewater samples. All samples were collected using clean hands-dirty hands protocol (EPA Method 1669) into new PETG bottles and preserved with trace metal grade HCl to 0.4%. We also collected bottle and filter blanks in HDPE and PETG bottles. All HDPE bottles were shipped on ice and stored frozen (−20 °C) until analysis. All PETG bottles were shipped on ice and stored at 4 °C until analysis.

3.

Precipitation record for the sampling period. Precipitation data are from ref . The dashed red lines denote soil core and porewater sample collection dates, with two sampling events per year. The shaded blue area denotes the time period during which additional porewater sampling occurred after every rain event. The dashed blue line denotes additional monthly porewater sample collection dates.

We collected soil/sediment (hereafter, soil) samples (0–10 cm depth) from each sampling node in October 2022 (dry season), December 2022 (beginning of the wet season), March 2023 (middle of the wet season), December 2023 (beginning of the wet season), and February 2024 (middle of the wet season) (Figure ). Prior to collecting the soil samples, we removed live plant material from the ground surface. Soil samples were collected using clean hands-dirty hands protocol (EPA Method 1669) with a bulb corer (Berry and Bird, Inc.) into three separate Ziplock bags for laboratory incubations, Hg analyses, and other chemical analyses. The soil core collected for Hg analyses was double-bagged. All bags were placed into a mylar bag with an oxygen scrubber and sealed until processing. Samples were shipped on ice to the University of Colorado, Boulder. Both the laboratory incubation and chemical analysis samples were stored at 4 °C until processing. The Hg analysis samples were stored frozen (−20 °C) until analysis.

Incubation Experiment

Within 48 h of sample collection at the beginning of the wet season (December 2022, 2023) and middle of the wet season (March 2023, February 2024), we began an incubation experiment at the University of Colorado, Boulder. All processing of samples occurred within a polyethylene glovebag purged with nitrogen gas. We added field-moist soil to glass serum jars (10–20 g). Each sample was added to three jars to allow for incubations of length t = 0 (placed in the freezer within 30 min of spike addition), t = 24 h, and t = 48 h. We also prepared duplicates of two soil samples (three time points for each).

We prepared MeHg¹⁹⁸ and MeHg¹⁹⁹ from Hg¹⁹⁸O and Hg¹⁹⁹O obtained from Oak Ridge National Laboratories, following Rodríguez Martín-Doimeadios et al. (2002). We prepared Hg­(II)²⁰⁰ from Hg²⁰⁰Cl₂ obtained from Oak Ridge National Laboratories. We prepared solutions containing inorganic Hg­(II)²⁰⁰ to measure potential methylation rates and MeHg¹⁹⁸ to measure potential demethylation rates; we added Hg isotopes to ultrapure water (18.3 MΩ) and allowed them to equilibrate for a minimum of 1 h. We then spiked incubation jars with the isotope solution at a ratio of 2:1 soil:spike solution and crimped the caps. We added inorganic Hg­(II)²⁰⁰ at a level equal to the ambient concentration of inorganic Hg­(II)²⁰⁰ and added MeHg¹⁹⁸ at 10% of the ambient inorganic Hg­(II)¹⁹⁸ concentration. After 0, 24, and 48 h, we placed the incubation jars in the freezer (−20 °C) to end the experiment, and they were stored frozen until processing.

Laboratory Analyses

We extracted SO₄ ^2– from air-dried soil samples using calcium dihydrogen phosphate. , In this method, 10 g of air-dried soil is shaken with 100 mL of 0.15% calcium dihydrogen phosphate for 24 h, centrifuged at 1900 rpm for 10 min, filtered using Whatman 42 filter paper, and frozen (−20 °C) until analysis. This method extracts inorganic SO₄ ^2– only. These filtered extracts, as well as the porewater and surface water samples, were analyzed for SO₄ ^2–, nitrate, and chloride concentrations following EPA Method 300.1 using an ion chromatograph (IC; Metrohm) at the University of Colorado, Boulder. All standards had recoveries within 10% of the accepted values, and all blanks were below detection limit (BDL; 0.1 mg L^–1 SO₄ ^2–).

We selected a subset of porewater samples (monthly samples for all nodes) and soil samples from December 2022 and March 2023 for δ³⁴S −SO₄ ^2– analysis. We extracted the field moist soil samples with water following the protocol above (using water instead of calcium dihydrogen phosphate). We then reduced the pH to 2–4 of the water extracts and porewater samples with trace-metal grade hydrochloric acid, brought the solutions to a boil on a hot plate, and added barium chloride crystals in excess to precipitate barium SO₄ ^2–. Next, we collected the barium SO₄ ^2– precipitate on preweighed 0.45 μm Whatman cellulose nitrate filters. We washed the filters with ultrapure water and dried them at 60 °C overnight. We analyzed the barium SO₄ ^2– precipitates for δ³⁴S-SO₄ ^2– at the University of California Berkeley on a Flash IRMS elemental analyzer coupled with a Delta V Plus isotope ratio mass spectrometer (Thermo Fisher Scientific EA IsoLink). All values reported in the text use the conventional δ-notation in parts per 1000 (‰). We also analyzed the filtered water extracts for SO₄ ^2– concentration using an IC (Metrohm) at the University of Colorado, Boulder. All standards had recoveries within 10% of the accepted values, and all blanks were BDL.

We determined moisture content (field-wet conditions) in soils by drying samples overnight at 105 °C. We determined loss on ignition (LOI) by combusting samples for 4 h at 550 °C in a muffle furnace. We determined pH using a 1:4 air-dried soil:water slurry or 1:4 air-dried soil:calcium chloride solution at the University of Colorado, Boulder. Soil pH was only measured during the first sampling season. Statistical analyses reported in the text use the pH determined in the water slurry. We determined percent carbon (C), nitrogen, and total S content on soil samples from the first sampling season that were air-dried, pulverized for 48 h on a rolling mill, and analyzed on a Thermo Delta Q isotope ratio mass spectrometer (IRMS) interfaced to a Thermo IsoLink Elemental analyzer at the Cornell Isotope Laboratory (COIL) at Cornell University.

We analyzed porewater and surface water samples for total Hg by oxidation, purge and trap, and cold vapor atomic fluorescence spectroscopy following EPA Method 1631 on a Tekran 2600 automated total Hg analyzer at Cornell University. We lyophilized all soil and incubation samples. Soil samples were then sieved and homogenized using a ball and mill grinder. We analyzed all soil samples for total Hg on a Direct Mercury Analyzer (Milestone DMA-80) at either Michigan State University or Cornell University. All soil, water, and incubation samples were distilled, then analyzed for MeHg using isotope dilution with MeHg¹⁹⁹ on a Tekran 2700 automated MeHg analyzer paired either to a ThermoFisher iCAP inductively coupled plasma mass spectrometer (ICP-MS) at Michigan State University or to an Agilent 7800 ICP-MS at Cornell University (modified USEPA Method 1630). We also analyzed the spike solution for each experiment date for total Hg concentration on the Tekran 2600 as well as MeHg¹⁹⁸ on the Tekran 2700 paired to the ICP-MS. These values were used to determine rate constants for the incubation samples. The concentrations of soil and incubation samples are reported in units of dry mass. Quality control information for all Hg analyses can be found in Supporting Information.

Determination of Rate Constants for Incubation Samples

We determined pseudo-first order potential Hg methylation (K _meth) and demethylation (K _demeth) rate constants for these time periods (0–24 h, 0–48 h) based on the production of MeHg²⁰⁰ and loss of MeHg¹⁹⁸ − on the same sample. We controlled for any abiotic methylation or demethylation that occurred as a result of incubation setup by the use of t = 0 samples. , To calculate k _meth and k _demeth, we assumed first-order kinetics for the 48 h of the incubation experiment, , according to eqs and .

$$K_{\mathrm{meth}}=\frac{(-\ln (1-{[\mathrm{MeH}{\mathrm{g}}^{200}]}_t-{[\mathrm{MeH}{\mathrm{g}}^{200}]}_{t_0}))\times 100}{\frac{{[\mathrm{H}{\mathrm{g}}^{200}]}_{t_0}}{t}}$$ 1

whereby [MeHg²⁰⁰] is the time-corrected concentration (ng g^–1) of MeHg²⁰⁰ produced from the Hg­(II)²⁰⁰ spike over a set period of time, t is the time duration of the incubation (0, 24, and 48 h or 0, 1, and 2 d), [Hg²⁰⁰] is the concentration (ng g^–1) of Hg²⁰⁰ added to the incubation jar, and 100 allows for the conversion to units of percent per day.

$$K_{\mathrm{demeth}}=\frac{(-\ln \left(\frac{{[\mathrm{MeH}{\mathrm{g}}^{198}]}_t}{{[\mathrm{MeH}{\mathrm{g}}^{198}]}_{t_0}}\right))\times 100}{t}$$ 2

whereby [MeHg¹⁹⁸] is the time-corrected concentration (ng g^–1) of MeHg¹⁹⁸ remaining over a set period of time, t is the time duration of the incubation (0, 24, and 48 h or 0, 1, and 2 d), and 100 allows for the conversion to units of percent per day. For methylation and demethylation rate constants that were below 0, a value of 0 was used. For demethylation rate constants that were above 100, a value of 100 was used.

Data and Statistical Analyses

We performed all statistical analyses using RStudio version 2023.06.2 + 561 and R version 4.2.2. All statistical figures were made using the ggplot2 and ggpubr packages. The map figure was made in ArcGIS, and the conceptual figure was made in BioRender. Since the data were not normally distributed, all comparisons between groups (e.g., land use/land cover zone, month) were performed using the nonparametric Kruskal–Wallis test, followed by posthoc analysis with the Dunn’s test using the Bonferroni correction. We determined correlations between two variables using the nonparametric Kendall’s tau. We performed nonmetric multidimensional scaling (NMDS) to determine relationships among samples. We used the vegan package for NMDS with input variables of all chemical analyses performed: chloride, loss on ignition (LOI), MeHg, nitrate, percent moisture, pH, SO₄ ^2–, total percent C, total percent nitrogen, total Hg, and total percent S. Additionally, we performed a three-source mixing model using the simmr package. Sources of SO₄ ^2– were determined as vineyard (using δ³⁴S-SO₄ ^2– values from this study), marine (using δ³⁴S-SO₄ ^2– values from refs , ) and precipitation (using δ³⁴S-SO₄ ^2– from ref ), and the model was performed separately for the set of porewater and set of soil extractant samples. Outliers were determined using the EnvStats package in R, which applies Rosner’s Test. All statistical analyses were evaluated at an α of 0.05 for significance. Values reported in the text represent mean ± 1 standard deviation (SD). All raw data and associated metadata are available online.

Results and Discussion

Sulfate Transport from Vineyards

Despite high S applications in the vineyard, vineyard soil total S (hereafter, S) and SO₄ ^2– concentrations were low (0.03 ± 0.02% S, 0.32 ± 0.66 mg g^–1 SO₄ ^2–). Conversely, soil S and SO₄ ^2– concentrations were elevated in the wetland (1.07 ± 0.68% S, 16 ± 13 mg g^–1 SO₄ ^2–) compared to the other sites (p < 0.0001; Figure A,B). Average wetland soil S was 20, 34, and 45 times higher than soils in the drainage ditch, vineyard, and grassland transition, respectively, while wetland soil SO₄ ^2– concentrations were 17, 49, and 490 times higher, respectively. The drainage ditch soils had higher S and SO₄ ^2– (0.057 ± 0.010% S, 0.94 ± 1.8 mg g^–1 SO₄ ^2–) compared to the grassland transition soil (0.025 ± 0.013% S, 0.04 ± 0.03 mg g^–1 SO₄ ^2–; p = 0.014, p = 0.031). Similarly, porewater SO₄ ^2– concentrations were elevated in the wetland (7500 ± 33,000 mg L^–1) and drainage ditch (8400 ± 26,000 mg L^–1) compared to the vineyard (990 ± 2800 mg L^–1; p = 0.018) and grassland transition (46 ± 47 mg L^–1; p < 0.0001; Figure C). Vineyards also had higher porewater SO₄ ^2– concentrations compared to the grassland transition porewater (p < 0.0001). Collectively, our results across the gradient are consistent with previous studies demonstrating that soil SO₄ ^2– is loosely bound in vineyard soils and readily transported during rain events, thereby leading to SO₄ ^2– depletion within the vineyard soils. Our study also supports that SO₄ ^2– accumulates and then flushes out of vineyard soils into adjacent waterbodies. This result has previously been shown in Napa vineyard soils as a result of subsurface lateral flow following storm events and subsequent transport into the main stem of the Napa River. Ours is the first to follow the transport and downstream accumulation of vineyard-derived SO₄ ^2– along a gradient from source to end point.

4.

Concentrations of chemical analytes across sample type. (A) Sulfate concentrations in soil (vineyard: n = 86, ditch: n = 38, grassland transition: n = 84, wetland: n = 83). (B) Percent sulfur in soil (vineyard: n = 30, ditch: n = 15, grassland transition: n = 15, wetland: n = 45). (C) Sulfate concentrations in porewater (vineyard: n = 30, ditch: n = 15, grassland transition: n = 15, wetland: n = 45). (D) Total mercury concentrations in soil (vineyard: n = 30, ditch: n = 15, grassland transition: n = 15, wetland: n = 45). (E) Methylmercury concentrations in soil (vineyard: n = 30, ditch: n = 15, grassland transition: n = 15, wetland: n = 45). (F) Percent of mercury (Hg) present as methylmercury (MeHg) in soil (vineyard: n = 30, ditch: n = 15, grassland transition: n = 15, wetland: n = 45). Boxplots show the median value as the line, with the box representing the first and third quartiles of the data. Letters above the boxplots represent statistically significant differences. Note the broken axis in panels A and B.

Transport of vineyard-applied S is further supported by differences in porewater SO₄ ^2– concentration seasonally (p < 0.0001). Porewater SO₄ ^2– concentrations were 17–240 times higher in December through March compared to April and May for the vineyard (monthly average range: 11–1400 mg L^–1), drainage ditch (monthly average range: 57–17,000 mg L^–1), and wetland (monthly average range: 21–13,000 mg L^–1), while the grassland transition retained a low porewater SO₄ ^2– concentration across all months (monthly average range: 19–54 mg L^–1). By design, our initial sampling event of the year occurred after the first large precipitation event (6.6 cm in 2022, 9.9 cm in 2023) to ensure the soils were fully saturated. Several smaller precipitation events also occurred prior to this large precipitation event (7.4 cm over 4 events in 2022, 9.1 cm over 4 events in 2023), representing the first rainfall since the previous year’s wet season. Early rainfall events likely provided the hydrologic connectivity to transport most of the SO₄ ^2– from the vineyards, with additional SO₄ ^2– mobilization occurring later in the wet season, consistent with findings from past studies by our research group. ,

Sulfur stable isotopes provide a means to discern the likely sources of SO₄ ^2– along the study gradient, such as to differentiate agriculturally derived sulfur from atmospheric and marine sources. ,, Soil δ³⁴S-SO₄ ^2– did not vary by land use/land cover zone; all had a wide range of values from 8.8 to 33‰ (p = 0.47; Figure ). In contrast, porewater δ³⁴S-SO₄ ^2– was more enriched in the wetlands (22 ± 10‰) compared to the drainage ditch (14 ± 4‰; p = 0.0098) and grassland transition (13 ± 3‰; p < 0.0001) and enriched in the vineyards (17 ± 2‰; p = 0.0042) compared to the grassland transition. Measured vineyard δ³⁴S-SO₄ ^2– values in soil and porewater from this study were consistent with previous values reported in the literature of 3.5–17.2‰, , as well as reported values for S fungicide (3.1 ± 1.8‰). The low δ³⁴S-SO₄ ^2– values in the grassland transition could indicate that most of the S is in the stable, mineralized organic form since sulfate from mineralization of organic S or geologic sources have more depleted δ³⁴S-SO₄ ^2– values. Additionally, δ³⁴S-SO₄ ^2– values did not vary by sampling date for soils (p = 0.61) or porewater (p = 0.78), suggesting that the source and processing of SO₄ ^2– is consistent across seasons.

5.

Comparison of sulfate (SO₄ ^2–) concentrations to δ³⁴S values in (A) porewater and (B) soil. Shaded areas represent reported ranges of δ³⁴S-SO₄ ^2– for marine environments, , this vineyard (this study), and precipitation.

We used δ³⁴S-SO₄ ^2– values and a three-source mixing model including vineyard, marine, and atmospheric (i.e., precipitation) sources to determine the source of SO₄ ^2– within soils and porewater along the study gradient. Soil and porewater SO₄ ^2– in the grassland transition and drainage ditch were sourced approximately equally from marine, precipitation, and vineyard SO₄ ^2– sources (SI Figure 1, and SI Table 2), while SO₄ ^2– in the wetland was sourced primarily from marine (51 ± 23% for soil, 57 ± 20% for porewater) and vineyard (36 ± 24% for soil, 33 ± 23 for porewater) sources, with minimal inputs from precipitation (13 ± 8% for soil, 10 ± 5% for porewater). However, if the outlier samples are included in the porewater mixing model, more than 75% of the SO₄ ^2– in wetlands derives from marine sources, suggesting that those few sample locations have predominantly marine influence.

Spatial Variation in Total Mercury and Methylmercury Concentrations

Soil total Hg concentrations were highest in the wetland (84 ± 13 ng g^–1; p < 0.03 for all), followed by the grassland transition (68 ± 30 ng g^–1; p < 0.02 for all), and lowest in the vineyard (29 ± 10 ng g^–1) and drainage ditch (38 ± 13 ng g^–1; p > 0.05; Figure D). These patterns were consistent, even when land use/land cover zone differences were accounted for by normalizing soil total Hg concentrations to LOI (a proxy for organic matter content), with the exception that wetland soils exhibited lower total Hg:LOI than vineyard (p = 0.019) and grassland transition soils (p < 0.0001). However, when soil total Hg concentrations were normalized to total percent C, total Hg:total percent C was highest in the vineyard and grassland transition soils and lowest in the drainage ditch and wetland soils (p < 0.0001). Like the soil total Hg concentrations, porewater total Hg concentrations were highest in the wetland and grassland transition zones, compared to the vineyard and drainage ditch (p < 0.0001). As expected, soil total Hg concentrations did not vary by sampling date (p = 0.25).

Soil MeHg concentrations were higher in the wetland (0.60 ± 0.70 ng g^–1) and drainage ditch (0.48 ± 0.37 ng g^–1) compared to the vineyard (0.074 ± 0.036 ng g^–1; p < 0.0001) and grassland transition (0.14 ± 0.11 ng g^–1; p < 0.008; Figure E). Soil percent of Hg present as MeHg was higher in the wetland (0.70 ± 0.76%) and drainage ditch (1.4 ± 1.1%) compared to the grassland transition (0.24 ± 0.21%; p < 0.002), and higher in the drainage ditch compared to the vineyard (0.27 ± 0.13%; p = 0.0011; Figure F). Similarly, porewater MeHg concentrations were highest in the wetland and drainage ditch and lowest (near detection limit: 1.25 pg) in the grassland transition and vineyard (p < 0.02). Percent of Hg present as MeHg represents long-term production of MeHg. ,− We also found that soil percent of Hg present as MeHg was 1.4–2.5 times higher during the middle of the wet season compared to the beginning of the season (p = 0.019). These results suggest that net soil Hg methylation increases through the wet season, potentially due to more reduced soil conditions, which promote the activity of anaerobic bacteria. , Note that these soils all have low percent of Hg present as MeHg values (range: 0.03–3.8%), consistent with values previously reported for upland soils , and on the lower end of previous reports for wetlands. ,,

Environmental Drivers of Methylmercury Variation

Higher MeHg accumulation in the drainage ditch and wetland soils are likely the result of several driving factors. The drainage ditch and wetland have frequent changes in soil moisture content (28 ± 10% in the drainage ditch and 54 ± 10% in the wetland compared to 20 ± 6% in the grassland transition and 17 ± 3% in the vineyard), which can increase Hg methylation due to more reduced soil conditions. ,,, Terminal electron acceptors such as SO₄ ^2– along with dissolved organic matter become available during oxic drying events, and then MeHg can be produced microbially under anoxic wetting conditions. , The importance of moisture content and soil reducing conditions as drivers of Hg methylation within our study site soils is supported by the correlation of soil percent moisture with both soil MeHg concentration (p < 0.0001, τ = 0.37) and percent of Hg present as MeHg (p = 0.0007, τ = 0.21). Our NMDS results–which show overlap in ordinal space of soil samples with elevated percent moisture and MeHg content for some wetland and drainage ditch samples (Figure )–further support moisture as a driver of soil Hg methylation patterns. In contrast, the vineyard and grassland transition soils are predominantly oxic (i.e., when not saturated), promoting aerobic microbial demethylation pathways over anaerobic microbial methylating pathways.

6.

NMDS plot using all available data for each site. Circles represent samples that have elevated percent moisture [% moisture] or sulfate.

In addition to moisture, elevated soil SO₄ ^2– concentrationsparticularly in the wetlandare likely driving the observed patterns in soil MeHg concentrations. We found a correlation of both soil S and soil SO₄ ^2– concentrations with MeHg at our sites (p < 0.0001, τ = 0.29 for S; p < 0.0001, τ = 0.27 for SO₄ ^2–). Similarly, we found a correlationalthough weakerof both soil S and soil SO₄ ^2– concentrations with percent of Hg present as MeHg (p = 0.0049, τ = 0.18 for S; p = 0.0084, τ = 0.16 for SO₄ ^2–). Several studies have previously reported the linkage between increased SO₄ ^2– concentrations and MeHg production in laboratory (Gilmour et al. 1992, Harmon et al. 2004, Mitchell et al. 2008b) and field studies, − including studies conducted in the Everglades Agricultural Area, which similarly has wetlands downstream of agricultural land. Previous studies have also reported the inhibition of MeHg formation when S^2– − or dissolved organic matter − are high, resulting in the Goldilocks curve. Our results suggest that SO₄ ^2– concentrations lead to enhanced MeHg concentrations in soil and porewater at all our sites and that high SO₄ ^2– concentrations (i.e., range: 0.92–64.1 mg L^–1)including marine-derived SO₄ ^2– along with presumably high S^2– concentrationsin wetlands do not inhibit MeHg production at our sites, despite the Goldilocks curve. However, SO₄ ^2– concentrations alone are not sufficient to explain the observed patterns in MeHg concentrations and percent of Hg present as MeHg.

Spatial Patterns in Potential Net Methylation Rates

While measurements of soil MeHg provide insight into net soil MeHg accumulation at the time of sampling, laboratory experiments are necessary to investigate differences in the potential for soil MeHg production and loss through the determination of potential net methylation rates (k _meth) and potential net demethylation rates (k _demeth) (e.g., refs , , and ). Therefore, we performed laboratory incubations on soil slurry samples, which increased the moisture content of all samples (2:1 ratio of solution:ambient soil). After 48 h of incubation, we found that soil potential Hg methylation rates were elevated in the wetland compared to the vineyard (p = 0.019), and there was no difference in demethylation rates by land use/land cover zone (p > 0.05 for all pairwise comparisons; Figure ). These results of elevated potential Hg methylation rates in the wetland soils are consistent with our findings of elevated MeHg concentrations and percent of Hg present as MeHg in wetland soils. Collectively, they suggest that wetland soils can more efficiently transform Hg than vineyard soils. Even when the vineyard soils are provided with anoxic conditions (i.e., saturation) in laboratory slurries, these conditions are insufficient to account for differences in k _meth compared to the wetland soils. Our results support findings by Huang and Mitchell (2023), in which upland soils in Canadian boreal forests had lower potential Hg methylation rates compared to wetland soils. Our calculated methylation potentials in upland soils and wetlands, while low, are similar to previously reported values in the literature for these types of environments. , Furthermore, since the percent of Hg present as MeHg represents both net potential methylation and demethylation, it can be compared to the quotient of k _meth/k _demeth. We found a correlation between soil k _meth/k _demeth and percent of Hg present as MeHg (p = 0.031, τ = 0.16), suggesting that longer-term in situ net soil MeHg accumulation (i.e., percent of Hg present as MeHg) corresponds with the short-term production and loss of soil MeHg (k _meth/k _demeth) calculated from the laboratory incubations. This result has been noted in previous studies of forest and wetland ecosystem soils. ,

7.

Results from laboratory incubations conducted over 48 h (vineyard: n = 24, ditch: n = 12, grassland transition: n = 24, wetland: n = 36). (A) Potential Hg methylation rates (k _meth). (B) Potential Hg demethylation rates (k _demeth). Note differences in the scale on the y-axis. Letters above the boxplots represent statistically significant differences.

Similar results of lower potential Hg methylation rates in the vineyard soil compared to the wetland and drainage ditch soils, as well as in the grassland transition soil compared to the wetland soil were found for the first 24 h of incubation (0–24 h; p < 0.04 and p = 0.0032; SI Figure 2). Potential Hg demethylation rates were lower in the in the vineyard soil compared to the wetland soil only for the second 24 h of the incubation (24–48 h, p = 0.040). No differences by land use/land cover zone were found for the second 24 h of incubation (24–48 h) for soil potential methylation rates (p = 0.27), nor for the first 24 h of incubation (0–24 h) for soil potential demethylation rates (p = 0.061 for 0–24 h). These results show that soil potential Hg methylation and demethylation rate constants are dependent on experiment duration, suggesting the importance of unifying the length at which laboratory incubations are performed to compare potential responses. In particular, differences in soil potential methylation rate constants between land cover/land use zones for the first 24 h but not the second 24 h suggest a stimulation of microbial activity. The wetland and drainage ditch soils exhibited elevated net Hg methylation in the field (i.e., elevated MeHg concentrations and percent of Hg present as MeHg); soils from these locations thus likely had the organisms and substrates to methylate Hg under the laboratory conditions. Interestingly, we observed that these land use/land cover zone differences only occurred for the first 24 h. Similarly, in a study of subalpine forest soils, Miller et al. (2025) found that site differences in potential Hg methylation rate constants in S-amended laboratory incubations persisted for only the first 24 h.

Temporal Patterns in Potential Net Methylation Rates

The soil k _meth and k _demeth also differed by sampling date (p = 0.0085 for k _meth 0–48 h; p = 0.033 for k _meth 0–24 h; p < 0.0001 for k _demeth for all time intervals; SI Figure 3). Soil potential Hg methylation rates were higher during the middle of the second field season (February 2024) compared to the middle of the first field season (February 2023) for the entire 48 h incubation and the beginning of the first field season (December 2022) for the first 24 h of the incubation. This pattern could have been driven by ambient air temperature at the time of sampling. Since warmer temperatures promote Hg methylation, , ∼5 °C higher average air temperatures during the 2023–2024 field season compared to 2022–2023 (SI Table 3) may have led to field upregulation of microbial communities, thereby enhancing methylation rates in the laboratory slurries. Demethylation potentials exhibited a different pattern, with higher soil potential rates occurring in the middle of the wet season (February 2023 and 2024) compared to the beginning of the wet season (December 2022 and 2023) for all incubation durations (p < 0.0002 for all pairwise comparisons). Variations in seasonal patterns between soil potential Hg methylation and demethylation rates are likely driven by differences in microbial communities that methylate and demethylate Hg. ,, This pattern in soil demethylation potential could result from warmer temperatures and wetter conditions promoting microbial activity over the wet season. ,

Implications

We found evidence that S applied in a vineyard at the terminus of the Napa River Watershed is being transported as SO₄ ^2– through the drainage ditch and grassland transition into nearby wetlands where it mixes with marine-derived SO₄ ^2–. The SO₄ ^2– applied in vineyards appears to be loosely bound to soils, allowing for its transport during rain events. While SO₄ ^2– is transported across the landscape, we recognize that it is only one of many drivers that can influence Hg methylation. Ambient MeHg concentrations and potential MeHg were elevated in locations of our study gradient that experience changes in soil moisture and reducing soil conditions, specifically wetlands and the drainage ditch, suggesting the importance of moisture-induced redox cycling for environmental patterns of MeHg. Variation in C source, vegetation composition, and nutrient status among the different land use/land cover zones could also influence the observed patterns. Our results suggest that downstream wetlands and periodically saturated drainage ditches are notable locations of MeHg production downstream of California vineyards, as well as other similar upland agricultural areas.

All raw data and associated metadata are available online: Gerson, J. R.; Hinckley, E. Chemistry of soils on a vineyard to wetland gradient in Napa Valley, California, 2022 to 2024, ver 1. Environmental Data Initiative, 2026. https://doi.org/10.6073/pasta/9ef3733813f564deee87a4d881f17445 (accessed 2026-03-30).

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.5c13676.

• Quality control and quality assurance for mercury analyses; include results from the sulfate sourcing mixing model, results for laboratory incubations for different time intervals, and results from laboratory incubations by sampling date; source values of δ³⁴S used in the mixed model for sulfate sources, results from the sulfate sourcing mixing model, and temperature data for each sampling month (PDF)

Conceptualization and funding acquisition: E.S.H.. Formal analysis, visualization, and writingoriginal draft: J.R.G.. Resources and supervision: E.S.H. and J.R.G.. Investigation, methodology: C.I., D.C., E.S.H., J.R.G., and J.R.. Writingreview and editing: C.I., D.C., E.S.H., and J.R.

The authors declare no competing financial interest.