2.1 Site description

Sampling sites were located in the Plum Island Sound on the northeastern coast of Massachusetts, USA (42°45^′10^′′, 70°56^′46^′′), between the Gulf of Maine and the city of Boston. The estuary is the largest marsh-dominated estuary in New England, with a total marsh area of 60 km² and salt marsh area of 40 km² (Hopkinson et al., 2018; Millette et al., 2010). Tides are semidiurnal with an amplitude averaging 2.7 m (NOAA Tide Predictions, 2020). We focused our study on high marsh platforms, with an approximate elevation of 1.4 m above the North American Vertical Datum 88, which dominate tidal marshes in New England and accounts for 75 % of the vegetated area in the Plum Island Sound estuary (Wilson et al., 2014; Millette et al., 2010). The high marsh exhibits poor water drainage (Wilson et al., 2014) and is generally inundated twice per month during spring tides and during major storms (Millette et al., 2010). The two dominant species on the high marsh are C₄ species including Sporobolus pumilus (common name: marsh hay) and Sporobolus alterniflorus (common name: smooth cordgrass), with the latter mainly distributed along tidal channels and also dominant in low marsh platforms (Anjum et al., 2012; Cheng et al., 2006; Curtis et al., 1990; Maricle et al., 2009; Sun et al., 2020). Another C₄ species, Distichlis spicata (coastal saltgrass), is often collocated within S. pumilus-dominated sites on high marsh platforms (Arp et al., 1993), whereas Juncus gerardii (saltmarsh rush) usually dominates the terrestrial boundary of the high marsh (Bertness, 1991).

Although the upland watershed of this salt marsh does not have any known point sources of Hg, previous studies pointed to the possibility of legacy anthropogenic sources present in this marsh estuary, possibly from ocean water import of Hg from nearby riverine sources, and other possible sources include regional atmospheric deposition directly to the marsh and its watershed (Evers et al., 2007; Lane et al., 2011, 2020; Wang and Obrist, 2022). A previous study by our group showed high Hg concentrations observed in marsh soils and evidence that the salt marsh soils currently serve as a source of Hg to tidal water and the nearby ocean-dominated lateral particulate-bound Hg transport (Wang and Obrist, 2022). However, a comprehensive mass balance of Hg inputs and outputs in this system is currently missing.

2.2 Sampling and processing of vegetation and soil

Aboveground biomass of the dominant species, S. alterniflorus and S. pumilus, were collected every 4 to 5 weeks between June and November in 2021, corresponding to the active growing season. Additional senesced biomass was sampled in the following year in April 2022, and two additional salt marsh species, D. spicata and J. gerardii, were sampled in September 2018. For each sampling date, eight 1 m² square plots were selected in the footprint area of a micrometeorological flux tower (Forbrich et al., 2018), of which four squares were dominated by S. alterniflorus and four adjacent squares were dominated by S. pumilus. During vegetation sampling, all aboveground vegetation within the 1 m² squares was clipped close to the ground and stored in plastic Ziploc bags in coolers over ice and subsequently in refrigerators until processing. In the laboratory, wet and dry vegetation mass was determined, and vegetation was carefully separated into live and senesced tissues. Note that in these ecosystems, significant amounts of senesced standing tissues from the previous year are present along with current year green tissues. We then prepared the samples for Hg analysis in both bulk samples and in individual species.

In four of the eight sampling sites, quantitative belowground sampling was performed in July 2021, with two plots dominated by S. alterniflorus and the other two plots dominated by S. pumilus. Soil cores with diameter of 10 cm to a depth of 40 cm were taken and separated into depth increments of 0–20 and 20–40 cm. Belowground components were separated into the following components by washing onto a fine mesh with pore size of 0.25 mm: live roots and rhizomes identified by turgidity and color (e.g., hard and white tissues versus soft and gray/discolored); senesced roots, rhizomes, and soil detritus (not recognizable organic matter); and sediments and fine organic matter that passed through the fine mesh (only analyzed in two subsamples). This process was based on visual separation of tissues (Elsey-Quirk et al., 2011; Valiela et al., 1976). We observed live roots and rhizomes only in the top 20 cm of the soils with no recognizable live roots and rhizomes at 20–40 cm depth. All plant tissues were rinsed with tap water until the water was clean, then thoroughly rinsed three times with Milli-Q water, while a selected number of live aboveground tissues were analyzed, both washed and unwashed, for estimation of washable Hg (see Sect. 4.1). All samples were dried at 65 °C to reach a constant weight and were milled and homogenized using stainless steel coffee grinders (for vegetation samples) and a 8530 ShatterBox (for soil samples) prior to analyses. The coffee grinders and ShatterBox were rinsed with Milli-Q water and dried with Kimwipes between samples.

2.3 Hg concentration and stable isotope analysis of vegetation and soil samples

Total Hg concentrations in all components were measured using a tri-cell Milestone DMA-80 direct mercury analyzer (Milestone Inc., Monroe, Connecticut, USA) through thermal decomposition, catalytic reduction, amalgamation, desorption, and atomic absorption spectroscopy following EPA method 7473 (U.S. EPA, 1998). The system was re-calibrated based on daily performance checks using five-point calibration curves. Standard reference materials, including NIST 1515 Apple Leaves (43.2 µg kg⁻¹) and Canadian National Research Council certified reference material MESS-4 (marine sediment, 91 µg kg⁻¹), were used as continuous calibration verifications after every 10 samples. Percent recoveries of total Hg for certified reference materials averaged 99.9 ± 5.5 % (range of 89.6 % to 111.4 %) and all blanks were below detection limits (0.001 ng). All samples were analyzed in triplicate and results were accepted when coefficients of variation were less than 10 %.

Hg stable isotopes were measured on select samples including live aboveground biomass, live root, live rhizomes, and surface (0–22.5 cm) and deeper soils (97.5 cm). Samples were pre-concentrated with a Nippon direct Hg analyzer (Nippon Instruments) as described in Enrico et al. (2021). A HGX-200 cold vapor generator (Teledyne Cetac Technologies) was used to introduce sample Hg to a Thermo Neptune plus MC-ICP-MS at Harvard University. An Apex-Q nebulizer (Elemental Scientific) was used to nebulize a Thallium (Tl) solution and inject TI aerosols in the HGX-200. NIST3133 (primary standard) and RM8610 (previously UM-Almaden, secondary standard) were used as Hg isotopic standard solutions, and NIST997 (thallium isotopic standard solution) was used as the reference material to correct instrument mass bias. NIST 1515 Apple Leaves and Canadian National Research Council certified reference material MESS-4 were used to verify isotope analysis, and standard recoveries were in the acceptable range (from 82 % to 93 %). For analysis on Neptune, all trapped samples and standards were diluted with trapping solution to either 1 or 2 ng mL⁻¹, the concentrations of which could be matched with NIST 3133 bracketing standard and the UM-Almaden concentrations. Quality control results for MESS-4 and RM8610 were similar to previously published findings (Table S4, Blum and Johnson, 2017; Enrico et al., 2021). Small delta (δ) annotation is used for mass-dependent fractionation (MDF), which is reported as per mil (‰) values relative to NIST-3133 based on Eq. (1):

where ^xxxHg is the mass of each Hg isotope between 199 and 204. Capital delta (Δ) annotation is used for mass-independent fractionation (MIF), describing fractionation away from the expected MDF based on Eq. (2):

where ^xxxHg denotes mass of each Hg isotope 199, 200, 201, and 204, and β_xxx is the constant mass-dependent correction factor (0.252, 0.502, 0.752, and 1.492, respectively; Blum and Bergquist, 2007). To determine Hg sources, a ternary isotope mixing model was used to estimate fractions of Hg in aboveground biomass. Endmember Hg sources used included signatures of salt marsh plants roots measured in this study, published signatures of upland foliage samples as a proxy for atmospheric GEM uptake and its fractionation during plant uptake, and published data of precipitation Hg (see Supplement for details). We estimated uncertainty in the mixing model using the variance in precision estimates using Monte Carlo simulations (10 000 trials), with details of the simulation provided in Supplement (Table S5).

2.4 Data analysis

Data were checked for normality (Shapiro–Wilk test) and homogeneity of variance assumptions of statistical tests. The non-normalized data were subjected to a natural logarithmic transformation to ensure a normal distribution. Unpaired Student t tests were used to assess significant differences between groups (e.g., species), and statistical differences between non-washed and washed aboveground vegetation samples were performed using paired Student t tests. Linear regression analyses were performed to determine the rate of aboveground biomass Hg uptake over time. Hg mass and turnover rates were calculated by multiplication of Hg concentrations by corresponding biomass or biomass growth and other mass components at the level of sampling plots. All statistical tests were performed with STATA (Version 16.0, Statacorps, College Station, Texas), the mixing model and its uncertainty calculation was performed using Python (Python 3.12.0), and all regressions and statistical tests presented in text, tables, and figures were based on statistical differences with p values < 0.05. The variability presented in the text and figures is 1 standard deviation of the means.

3.1 Hg concentrations in aboveground and belowground biomass

Hg concentrations in aboveground tissues showed substantial seasonal variations and species-specific differences, with lowest concentrations in live tissues of S. alterniflorus and D. spicata, followed by S. pumilus, and highest concentrations in J. gerardii (Fig. 1a). Despite species differences in Hg concentrations, concentrations in bulk vegetation of communities dominated by S. alterniflorus versus S. pumilus (Fig. 1b) were not statistically different. This likely occurred because both communities are composed of multiple species. For example, S. alterniflorus communities also have a presence of S. pumilus plants, while S. pumilus communities include large numbers of D. spicata plants. Similarly, Hg concentrations of senesced S. pumilus and S. alterniflorus bulk samples were not statistically significantly different from each other (Fig. S1).

Figure 1Seasonal Hg concentrations of the four dominant salt marsh live plant species in 2018 and 2021 (a), and seasonal Hg concentrations of the S. alterniflorus and S. pumilus communities in 2021 (b). Different colors indicate different plant species. Standard errors indicate four replicates; ^*: standard errors indicate duplicates for a sample.

Download

Hg concentrations in live aboveground biomass strongly increased throughout the growing season between June and November across all species. Figure 2 shows linear increases in Hg concentrations in live aboveground tissues in plots dominated by S. alterniflorus and S. pumilus over time (r²=0.84; p<0.01; n=50), with no significant difference in regressions between the two communities. Based on these linear regression slopes, we estimated daily uptake rates of Hg during the growing season of 0.08 ± 0.01 µg kg⁻¹ d⁻¹ for both Sporobolus communities. After senescence, Hg concentrations in senesced aboveground biomass measured in spring of the following year (April 2022) were not further enhanced compared to live biomass samples collected in fall (November 2021; p=0.19) (Figs. 2 and 4a), suggesting that no statistically significant Hg uptake (or loss) occurred in biomass after senescence.

Hg concentrations in live roots and rhizomes (upper 20 cm) were 2 to 3 times higher in S. pumilus plots (258.9 ± 70.3 and 46.6 ± 14.2 µg kg⁻¹, respectively) compared to S. alterniflorus plots (84.5 ± 47.0 and 27.9 ± 1.1 µg kg⁻¹ , respectively) (Fig. 3, Table S1). In belowground tissues, Hg concentrations were higher in senesced biomass, including dead roots, dead rhizomes, and detritus, compared to live tissues. Belowground tissues also showed higher Hg concentrations than separated soil mineral and humus fractions (although only measured in one S. pumilus sample) (Fig. 3, Table S1). Bulk soil Hg concentrations (i.e., composed of all fractions listed above) averaged 194.6 ± 28.3 µg kg⁻¹ of S. alterniflorus community and 171.2 ± 72.1 µg kg⁻¹ of S. pumilus community in the top 20 cm with no significant difference (p>0.05). Bulk soil Hg concentrations of the 20–40 cm soil in S. alterniflorus (279.1 ± 203.8 µg kg⁻¹) were almost twice that of S. pumilus (159.1 ± 122.7 µg kg⁻¹). Overall, Hg concentrations of live roots (171.7 ± 111.9 µg kg⁻¹) were 11 times higher and live rhizomes (37.3 ± 13.6 µg kg⁻¹) were double the concentrations of live aboveground biomass (16.2 ± 2.0 µg kg⁻¹, Table S2).

Figure 2Hg concentrations of live and senesced aboveground biomass of S. alterniflorus and S. pumilus communities corresponding with sampling dates in 2021. Green circles indicate live S. alterniflorus communities, orange triangles indicate live S. pumilus communities, gray circles indicate senesced S. alterniflorus communities, and gray triangles indicate senesced S. pumilus communities. Standard errors indicate four replicates. Note that data in April 2021 were extrapolated and set to zero based on phenological observations that showed no presence of live biomass at this time.

Download

Figure 3Hg concentrations in live above- and belowground biomass of S. alterniflorus and S. pumilus communities, as well as Hg concentrations of minerals and humus and bulk soils up to depth of 20 cm covered by these two plant species. Dark green bars denote S. alterniflorus community; light green bars denote S. pumilus community. Standard errors indicate multiple sample analysis. ^*: Hg concentration in mineral and humus only shown for one site covered by S. pumilus, and standard errors are duplicates.

Download

3.2 Hg pools sizes associated with aboveground and belowground biomass

Aboveground standing live biomass strongly increased from June through August, when it plateaued at a peak biomass in August (507 ± 208 g m⁻²) and September (498 ± 118 g m⁻², a trend that was consistent among the investigated communities) (Fig. 4b). Hg mass contained in live aboveground biomass peaked later (in November) than standing biomass and showed an 8-fold and near-linear increase between June (0.7 ± 0.4 µg m⁻²) and November (5.7 ± 2.1 µ g m⁻²) (Fig. 4c). Peak Hg pools contained in aboveground biomass were 5.7 ± 2.1 µg m⁻² for live tissue and 3.3 ± 1.7 µg m⁻² for senesced tissue, for a total combined standing aboveground biomass Hg pool of 9.0 ± 3.3 µg m⁻² in November (Figs. 4c and 5). This number represents our best estimate of total annual Hg assimilation by aboveground vegetation through the year. Standing aboveground biomass (live and senescent) in the spring of the following year (April 2022, 357 ± 148 g m⁻²) was 39 % lower than total aboveground biomass in November of 2021 (583 ± 208 g m⁻²) (Fig. 4b), and standing Hg pools were 32 % lower in the subsequent spring (6.1 ± 1.9 µg m⁻²) compared to peak fall levels (9.0 ± 3.3 µg m⁻²) (Fig. 4c), showing substantial losses of standing aboveground biomass and associated Hg pools over winter.

Figure 4Seasonal patterns of Hg concentrations (a), biomass dry weight (b), and Hg mass (c) in live aboveground and senesced biomass from June 2021 to April 2022. The green columns represent live biomass, gray columns represent senesced biomass, and orange columns represent total biomass weight and Hg mass of combined live and senesced biomass. Standard errors indicate four replicates. ^*: no live biomass was visible in April 2022 so that senesced biomass equals total biomass.

Download

Live root biomass in surface soils (top 20 cm) averaged 361 ± 114 g m⁻² and live rhizome biomass were approximately twice as large (792 ± 231 g m⁻²), for a combined live belowground biomass of 1,153 ± 321 g m⁻² (Table S2). Belowground Hg pools associated with these live tissues averaged 70.0 ± 63.7 µg m⁻² for roots, 38.1 ± 22.4 µg m⁻² for rhizomes, and 108.1 ± 83.4 µg m⁻² for the combined live belowground tissue, accounting for less than 0.5 % of the total bulk soil Hg pool (Fig. 5a, b, Table S2). We observed a much larger Hg pool associated with senesced biomass (roots, rhizomes, and detritus) averaging 4116 ± 1141 µg m⁻², accounting for 16.1 % of the total bulk soil Hg pool. We estimated a total soil Hg pool in the top 40 cm using measured bulk densities (range of 0.22 and 0.37 g cm⁻³) exceeding 25 000 µg m⁻², with most of this Hg associated with fine soil mineral and humus fraction (83.5 %), while much smaller pools were contained in live and senesced plant tissues as mentioned above.

Figure 5Hg mass of above- and belowground biomass, including live and senesced biomass, and mineral and humus fractions at a soil depth of 40 cm (a). Percentages of Hg mass contribution from belowground sections to the Hg soil pool of a soil depth of 40 cm (b). Different colors indicate different sections of the marsh.

Download

3.3 Hg stable isotope signatures for determining Hg sources

Live aboveground biomass showed negative mass-dependent fractionation (MDF) values for δ²⁰²Hg between −1.61 ‰ and −1.07 ‰, and mass-independent (MIF) values were consistently positive with Δ¹⁹⁹Hg between 0.20 ‰ and 0.43 ‰ and Δ²⁰⁰Hg values between 0.04 ‰ and 0.11 ‰ (Fig. 6, Tables S3 and S6). These aboveground isotopic Hg signatures of salt marsh vegetation fell outside of the range commonly reported in foliar samples of terrestrial vegetation, both regarding mass-dependent and mass-independent signatures. Specifically, terrestrial vegetation Hg signatures are substantially more negative in δ²⁰²Hg values (ranging from −3.06 ‰ to −2.37 ‰ [inter-quartile range, IQR, n=120)] and both Δ¹⁹⁹Hg and Δ²⁰⁰Hg values in terrestrial vegetation generally show negative values (Δ¹⁹⁹Hg: −0.42 ‰ to −0.27 ‰ IQR, Δ²⁰⁰Hg: −0.05 ‰ to 0.01 ‰, IQR) (Fig. 6, Table S6) (review by Zhou et al., 2021).

Figure 6Hg isotopes in salt marsh plants, soils, and foliage (a) of Δ²⁰⁰Hg and δ²⁰²Hg and (b) of Δ¹⁹⁹Hg and δ²⁰²Hg. Composition of Hg sources in marsh vegetation and soils (surface and deep soil layers), all previously published and currently available isotope data of sources of Hg in vegetation and in terrestrial sinks, and atmospheric Hg(0) and Hg(II) sources (Zhou et al., 2021) plotted as (a) even-mass-independent (Δ²⁰⁰Hg) versus mass-dependent (δ²⁰²Hg) isotopes and (b) odd-mass-independent (Δ¹⁹⁹Hg) versus mass-dependent (δ²⁰²Hg) isotopes.

Download

Stable Hg isotope signatures of salt marsh plant roots were different from aboveground biomass, with less negative values for δ²⁰²Hg (−0.75 ‰ and −0.66 ‰), less positive values for Δ¹⁹⁹Hg (0.11 ‰ and 0.22 ‰), and close to zero values (instead of positive values) for Δ²⁰⁰Hg (−0.01 ‰ and 0.04 ‰) (Fig. 6, Tables S3 and S6). The Hg isotope signatures of roots closely overlapped with signatures in surface marsh soils and deeper marsh soils (δ²⁰²Hg: −0.92 ‰ to −0.29 ‰, Δ¹⁹⁹Hg: −0.09 ‰ to 0.20 ‰, and Δ²⁰⁰Hg: −0.02 ‰ to 0.05 ‰, Tables S3 and S6). Similar to aboveground tissues, salt marsh soil isotopic Hg signatures were largely outside the ranges reported for upland soils, particularly for δ²⁰²Hg values that are much more negative in upland soils (δ²⁰²Hg generally between −0.5 ‰ and −2.9 ‰; review by Zhou et al., 2021). Hg isotope signatures of salt marsh rhizomes were highly variable and in-between the values observed in foliage and soils. Specifically, rhizomes showed δ²⁰²Hg values between −1.41 ‰ and −0.70 ‰, Δ¹⁹⁹Hg values between 0.13 ‰ and 0.22 ‰, and Δ²⁰⁰Hg values between −0.05 ‰ and 0.04 ‰ (Fig. 6, Tables S3 and S6).

4.1 Salt marsh vegetation and soil Hg concentrations

Strong seasonal Hg concentration increases in salt marsh aboveground tissues were consistent with patterns reported from upland ecosystems, such as in forest foliage (Wohlgemuth et al., 2020). In upland systems, foliar Hg increases are attributed in large part to atmospheric GEM uptake, which is taken up during the growing season by stomatal and non-stomatal (i.e., cuticular) leaf uptake (review by Zhou et al., 2021). Hg uptake is controlled by leaf physiological processes and related to photosynthetic capacity, leaf nitrogen concentrations, leaf mass area, and stomatal densities and conductance (Wohlgemuth et al., 2022). In support of a physiologically controlled Hg uptake process in salt marsh plants, we observed that Hg concentrations in senesced biomass in April of 2022 were not significantly enhanced compared to live biomass of the previous November (2021) (Fig. 2), indicating that no significant Hg assimilation occurred during wintertime in senesced biomass. However, some increases in Hg concentrations occurred through November even after peak biomass was reached in August and September, which we attribute to continued active plant physiology through late season, as active photosynthesis was measured at least through October at this site (Forbrich et al., 2018; no data are available for November). In contrast to upland plant foliage, however, stable Hg isotope signatures of marsh aboveground biomass show distinctly different Hg source profiles, indicating that Hg uptake was not dominated by atmospheric GEM uptake (see below).

Estimated daily Hg accumulation rates in Sporobolus-dominated aboveground biomass (0.08 µg kg⁻¹ d⁻¹, Fig. 2) were at the lower range of foliar accumulation rates reported in forest foliage (conifer needle: median of 0.07 µg kg⁻¹ d⁻¹, deciduous leaf: median of 0.23 µg kg⁻¹ d⁻¹; Wohlgemuth et al., 2022). This is consistent with the notion that low-statured grassland plants generally exhibit lower Hg concentrations (5 µg kg⁻¹ [1–31 µg kg⁻¹]) than trees (e.g., forest foliage (20 µg kg⁻¹ [2–62 µg kg⁻¹]) (review by Zhou et al., 2021), although it may also be due to different source origins of Hg (Sect. 4.2.1 below). Both dominant marsh species in this study are also C₄ plants, which previous work shows have lower Hg concentrations compared to C₃ species (e.g., 23 ± 9 µg kg⁻¹ versus 53 ± 12 µg kg⁻¹, Canário et al., 2017). In a laboratory study with upland plants, a suggested reason for this was catalase activity, which is related to leaf uptake of Hg vapor and is about 4 times lower in C₄ plants than C₃ plants (Du and Fang, 1983).

Highest Hg concentrations in S. alterniflorus and S. pumilus were observed in fall (11.7 and 24.0 µg kg⁻¹, respectively). Hg concentrations fell within concentration ranges reported from other uncontaminated marsh halophytes (Table S8) but were much lower than those from contaminated marshes (up to 90 µg kg⁻¹).

In contrast to upland plants, salt marsh plants (including both Sporobolus species) have salt glands which are used for selective and active excretion of sea salt (Kirschner and Zinnert, 2020; Maricle et al., 2009). Salt glands have been linked to excretion of metals (Weis and Weis, 2004), and previous studies reported correlations between leaf surface Hg and sodium (Na) release in S. alterniflorus, suggesting active Hg excretion by salt glands (Weis and Weis, 2004; Windham et al., 2001). Windham et al. (2001) proposed that in the Hackensack Meadowlands, a polluted salt marsh ecosystem, seasonal declines in Hg concentrations in S. alterniflorus leaves between May (90 µ g kg⁻¹) and July (30 µg kg⁻¹) were driven by strong leaf excretion of Hg. By washing leaves of a number of samples, we found that washing removed about 6 % of total leaf Hg in S. alterniflorus and 16 % in S. pumilus, respectively (Table S7, note that we use the average wash-off fraction of 11 % to estimate throughfall deposition in the mass balance estimation below). The relatively small loss of Hg associated with washing showed that most leaf Hg was structural and likely internal Hg, which along with observed seasonal Hg concentration increases does not support seasonal Hg losses nor seasonal concentration declines attributed to salt excretion by Windham et al. (2001).

Hg concentrations in live roots and rhizomes were much higher (11 and 2 times, respectively) than live aboveground biomass concentrations. This is consistent with previously reported data that reported high root Hg concentrations in salt marshes (Anjum et al., 2012; Cabrita et al., 2019; Canário et al., 2017; Garcia-Ordiales et al., 2020; Weis and Weis, 2004; Windham et al., 2003). We measured higher Hg concentrations in live roots of S. pumilus (258.9 ± 70.3 µg kg⁻¹) than in S. alterniflorus (84.5 ± 47.0 µg kg⁻¹) (Fig. 3, Table S1). A possible reason for this is the finer roots in S. pumilus (personal observation), and hence higher surface-to-volume ratios, which may facilitate soil Hg uptake. In upland ecosystems, fine root Hg concentrations were reported to be higher than in coarse roots as well (J. J. Wang et al., 2012; X. Wang et al., 2020). High root and rhizome Hg concentrations compared to aboveground tissues in marsh plants stand in contrast to upland studies, which generally report highest concentrations in foliage (20 µg kg kg⁻¹ [2–62 µg kg⁻¹]) and much lower concentrations in roots (7 µg kg⁻¹ [2–70 µg kg⁻¹]) (Zhou et al., 2021). An exception to this are grassland systems, which also show higher root Hg concentrations than in foliage, although the difference was much smaller (e.g., roots: 41 ± 31 µg kg⁻¹; leaves: 20 ± 10 µg kg⁻¹; Zhou and Obrist, 2021). Our data also showed generally much lower root Hg concentrations in this marsh compared to contaminated sites (Canário et al., 2017; Anjum et al., 2012; Garcia-Ordiales et al., 2020) (Wang et al., 2021) (Table S8). Overall, published studies suggest a large range of Hg concentrations in belowground salt marsh biomass, which likely is due to different soil Hg concentrations as dominant Hg sources to roots (see Sect. 4.2.1 below). The higher concentrations of roots and rhizomes in belowground tissues compared to aboveground biomass also suggest limited translocation between belowground and aboveground tissues (Cavallini et al., 1999; Clemens and Ma, 2016; Graydon et al., 2009; Wang et al., 2012).

4.2 Stable Hg isotope signatures and possible origins of Hg in salt marsh vegetation and soil

4.2.1 Salt marsh vegetation

One of the largest mass-dependent fractional (MDF) processes in the environmental systems is due to preferential uptake of light atmospheric GEM isotopes by vegetation foliage that leads to large negative δ²⁰²Hg signatures (generally below −2 ‰), while mass-independent signatures remain similar to that of atmospheric GEM (Demers et al., 2013; Enrico et al., 2016; Yu et al., 2016). However, a recent study also reported inconsistent MIF between foliage and the atmosphere (Wang et al., 2022b). In terrestrial ecosystems, studies have shown that the vegetation uptake of atmospheric GEM and subsequent litterfall, throughfall, and plant senescence serves as the primary source of Hg loading (Demers et al., 2013; Jiskra et al., 2015; Louis et al., 2001; Obrist et al., 2017; Wang et al., 2016; Zheng et al., 2016; Zhou et al., 2021). Aboveground tissues of salt marsh plants show a distinctly different signature than upland foliage: specifically, MDF values were much less negative and values of odd-MIF (Δ¹⁹⁹Hg) and even-MIF (Δ²⁰⁰Hg) were more positive compared to upland foliage (Fig. 6, Table S6).

Hg signatures of salt marsh aboveground tissue were close to signatures of salt marsh soils, yet with slightly more negative δ²⁰²Hg values and more positive Δ¹⁹⁹Hg and Δ²⁰⁰Hg values (Table S6). We used a ternary mixing model to identify potential Hg sources and further quantify their contributions for salt marsh plant leaves based on MDF (δ²⁰²Hg) and even-MIF (Δ²⁰⁰Hg) (Demers et al., 2013; Jiskra et al., 2017; Obrist et al., 2017; Jiskra et al., 2021). In our model, the dominant three endmember Hg sources include (1) direct uptake from marsh plant roots as represented by measured stable isotopes of roots; (2) atmospheric GEM uptake with typical upland MDF reported in the literature, implemented using published data of upland foliage isotope data; and (3) precipitation Hg(II) deposition obtained from published literature (see Supplement for details). Our best estimate shows that the Hg source in salt marsh vegetation consists of a mixture of about $\mathrm{1}/\mathrm{3}$ each from foliar uptake of atmospheric GEM uptake (about 32 %), root uptake (about 35 %), and precipitation deposition (about 33 %). Note that these estimates assume that marsh foliage shows the same isotopic GEM fractionation as upland foliage and further assumes that translocation of root Hg has the same isotope patterns as measured in root biomass. The percentage estimate derived from the tertiary mixing model shows substantial range of uncertainties (Fig. S2) due to a low number of samples and errors associated with analytical isotope determination. However, two different Monte Carlo methods used to estimate errors both suggest similar source contributions (i.e., about $\mathrm{1}/\mathrm{3}$ from each of the three endmembers).

Most notably, the biggest difference in marsh plants compared to upland plants is much less negative δ²⁰²Hg values, suggesting lower contributions from atmospheric GEM uptake, which normally induces strong negative MDF in upland plants. Still, the presence of atmospheric GEM uptake leads to δ²⁰²Hg values in aboveground tissues that are more negative than in roots and soils. Precipitation, which largely consists of oxidized Hg, shows a typical positive anomaly in Δ²⁰⁰Hg linked to upper atmosphere GEM oxidation processes (Enrico et al., 2016; Zhou et al., 2021; Jiskra et al., 2021). Our results also suggest a more important role of root Hg transport to aboveground tissues in salt marsh vegetation (about $\mathrm{1}/\mathrm{3}$) compared to upland ecosystems, which normally show root origins of less than 5 % (review by Zhou et al., 2021). Previous salt marsh studies suggested inconsistent Hg source patterns: for example, an Hg isotope tracer study suggested minor root-to-leaf transport with soils accounting for a small percentage of Hg in marsh plants (i.e., 2.2 %–2.7 % from Cabrita et al., 2019), while a study based on bioaccumulation factors suggested a wide and inconstant range of soil Hg contribution to leaves (from 1.7 %–9.6 % to as high as 46 %; Castro et al., 2009).

Roots of salt marsh plants show a Hg isotope signature that almost perfectly aligns with the signatures observed in soils, suggesting a dominant soil source. In terrestrial plants, Hg assimilated in belowground biomass also is considered to largely be of soil origin (Obrist et al., 2018; Millhollen et al., 2006; Zhou et al., 2021). This also has been proposed in aquatic plants (e.g., mangroves, sawgrass), where root Hg largely derives from surrounding soils (Huang et al., 2020; Mao et al., 2013; Yin et al., 2013). Finally, the few measured rhizome Hg isotope signatures indicate a mix of aboveground and belowground Hg sources, although rhizomes show a large variation in isotope signatures with some samples being closer to aboveground tissue and others being closer to root signatures. This observation would be consistent with the role of rhizomes as storage organs with a lifetime of over 1 year, where carbohydrates and nutrients are re-mobilized between rhizomes and above- and belowground organs based on plant allocation needs.

Table 1Hg mass balance of the study salt marsh ecosystem.

^*: salt marsh area (vegetated): 40 km².

Download Print Version | Download XLSX

4.3 Hg mass balance and turnover fluxes associated with biomass dynamics

4.3.1 Aboveground

We here estimate a mass balance of Hg sources and sinks associated with aboveground vegetation dynamics and turnover and compare these with previously reported fluxes such as lateral tidal exchanges, published wet and gaseous oxidized Hg, and particulate Hg deposition. Hg inputs to this salt marsh include wet Hg deposition, which, based on interpolated data by the NADP program, is estimated at 5.2 µg m⁻² yr⁻¹ (National Atmospheric Deposition Program, Mercury Deposition Network, http://nadp.slh.wisc.edu/maplib/pdf/mdn/hg_Conc_2017.pdf, last access: 1 March 2023), while a lower estimate of 2.9 µg m⁻² yr⁻¹ has been measured at a nearby coastal site on Cape Cod, Massachusetts (Engle et al., 2010). Combining these two data sets, we estimate a mid-point wet deposition of 4.1 µg m⁻² yr⁻¹. Gaseous oxidized Hg (GOM) and particulate Hg (PHg) deposition in this area was estimated at 1.2 µg m⁻² yr⁻¹ based on measurements by Engle et al. (2010) and at 3.0 µg m⁻² yr⁻¹ at a deciduous forest (Harvard Forest) in Massachusetts (Obrist et al., 2021). Hence, a mid-point dry deposition of combined GOM and PHg is estimated at 2.1 µg m⁻² yr⁻¹ (Table 1, Fig. 7). Aboveground vegetation Hg dynamics yields a total turnover of 9.0 ± 3.3 µg m⁻² yr⁻¹ (combined live and senesced biomass at the end of the growing season), and isotope determination suggests that of this, about $\mathrm{2}/\mathrm{3}$ (representing an “external source” (5.9 ± 2.1 µg m⁻² yr⁻¹) derived from atmospheric GEM and precipitation uptake, while $\mathrm{1}/\mathrm{3}$ is from root uptake (3.1 ± 1.1 µg m⁻² yr⁻¹); this hence represents an internal plant–soil recycling within the ecosystem. Laboratory sample washing showed that in addition, on average about 11 % of foliar Hg concentrations was subject to wash-off. In the absence of field throughfall measurements, we used this fraction to estimate throughfall deposition of Hg of 1.0 ± 0.4 µg m⁻² yr⁻¹, based on aboveground vegetation Hg pools (noting that this estimate is very preliminary). Combined atmospheric Hg deposition attributable to aboveground vegetation hence is 6.9 µg m⁻² yr⁻¹ and is close to the combined wet, GOM, and PHg deposition of 6.2 µg m⁻² yr⁻¹. Combined atmospheric Hg sources in this system are estimated at 13.1 µg m⁻² yr⁻¹ (Table 1; range of 7.7 to 19.5 µg m⁻² yr⁻¹).

Figure 7Hg mass balance of the study salt marsh ecosystem. The values shown mostly represent the median Hg flux values with unit of µg m⁻² yr⁻¹, except for the soil pool (^*), which represents an averaged value, and herbivory (${}^{**}$), which indicates a range. The red arrows indicate the emission of Hg back to the atmosphere and its export out of the salt marsh. The dashed arrows represent Hg fluxes related to aboveground biomass.

Download

Aboveground vegetation also results in lateral exchange of Hg between marsh and tidal water via wrack export, i.e., losses of plants and surface litter through tidal flushing. Although difficult to measure, wrack export in this area is composed primarily of S. alterniflorus plants (Hartman et al., 1983) and has been estimated to constitute 16 %–19 % (mid-point of 17.5 %) of accumulated biomass (Duarte, 2017; Duarte and Cebrián, 1996). Hence, we estimate that of the annual Hg uptake by aboveground biomass of 9.0 ± 3.3 µg m⁻² yr⁻¹, about 1.6 µg m⁻² yr⁻¹ (range of 1.4–1.7 µg Hg m⁻² yr⁻¹) may be subject to wrack export. Scaling to the whole salt marsh area (40 km²), the total wrack export from this marsh is estimated around 0.06 kg yr⁻¹ (range of 0.06–0.07 kg yr⁻¹). In a previous study, we quantified Hg exports from the salt marsh system via tidal exchanges of dissolved and particulate Hg (without plants), and estimated 0.7 µg m⁻² yr⁻¹ of dissolved Hg export and 5.6 µg m⁻² yr⁻¹ of particulate Hg export from the marsh to the tidal water (Wang and Obrist, 2022). Our estimated annual Hg export by wrack is higher than lateral export of dissolved Hg but much smaller than lateral export of particulate Hg. Duarte (2017) and Duarte and Cebrián (1996) estimated that about 27 % of NPP is subject to herbivory (e.g., by marsh periwinkle and mummichog), although another study estimated NPP loss due to herbivory at only 5 % (Mann, 1988). Hence, an estimated 5 % to 27 % of plant Hg pools, or 0.5–2.4 µg Hg m⁻² yr⁻¹, may be subject to herbivory. The largest fraction of NPP (57 % to 80 %), equivalent to 5.1–7.2 µg Hg m⁻² yr⁻¹) likely remains in the system and is subject to net accumulation and decomposition. Considering all these inputs and outputs, we estimate a net present-day Hg mass accumulation in this salt marsh ecosystem of 0–11.5 µg m⁻² yr⁻¹, with a mid-point of 5.2 µg m⁻² yr⁻¹, suggesting that this salt marsh currently represents a small net sink of environmental Hg. Absent in this estimate are gaseous exchange fluxes of GEM between soils/water surfaces and the atmosphere, which are currently being quantified using a tower-based measurement system (Edwards et al., 2005; Obrist et al., 2021).