Sedimentary records of mercury stable isotopes in Lake Michigan

Mercury from both natural and anthropogenic sources can enter aquatic ecosystems from point sources, watersheds and atmospheric deposition (Mason and Sheu, 2002; Fitzgerald et al., 2007). Upon entering aquatic ecosystems, Hg may undergo complex reactions and transformations including oxidation, reduction, evasion to the atmosphere as gaseous Hg(0), binding to organic matter in the water column, and adsorbing to particles and depositing to sediments (Fitzgerald et al., 2007). Mercury can be methylated under anoxic and suboxic conditions by sulfate reducing bacteria (e.g. deep water column and sediments), producing methyl mercury (MeHg) (Benoit et al., 2003; Blum et al., 2013; Jonsson et al., 2014), a highly toxic organic compound that is easily bioaccumulated in aquatic food webs (U.S. EPA, 1997).

Sediment Hg profiles, coupled with high resolution age dating (e.g. ²¹⁰Pb and ¹³⁷Cs), has been broadly used to evaluate historical changes and to predict future trends in Hg deposition (Yang et al., 2010; Engstrom et al., 2014; Pizzuto, 2014). Historical reconstruction has shown an increase Hg deposition since global industrialization that began roughly in the mid-1800s. Estimates predict the global amount of Hg in the atmosphere has increased by a factor of 3 to 5 since the preindustrial era (Yang et al., 2010; Engstrom et al., 2014; Pizzuto, 2014). As Hg emission sources are numerous and emission pathways are highly variable (Pirrone et al., 2010), it is often challenging to differentiate and quantify the historical contribution of specific sources both on local and global scales.

Mercury stable isotope geochemistry allow for tracking ambient Hg sources and processes and the fate of Hg in the environment by utilizing the seven stable isotopes of Hg (¹⁹⁶Hg, ¹⁹⁸Hg, ¹⁹⁹Hg, ²⁰⁰Hg, ²⁰¹Hg, ²⁰²Hg and ²⁰⁴Hg) (Bergquist and Blum, 2009; Blum et al., 2014; Yin et al., 2014a). Technological advances in the past two decades [particularly multiple collector inductively coupled plasma mass spectrometry (MC-ICP-MS)] have enabled high precision analysis of Hg isotope ratios (Foucher and Hintelmann, 2006; Blum and Bergquist, 2007; Yin et al., 2010a). Laboratory experiments have observed both mass dependent fractionation (MDF) and mass independent fractionation (MIF) of Hg isotopes. MDF (reported as δ²⁰²Hg) occurs during various physical, chemical, and biological processes. The utilization of MDF improves our understanding of transport, transformation, and bioaccumulation of Hg (Bergquist and Blum, 2009; Yin et al., 2010b, 2014a; Sonke, 2011; Hintelmann, 2012; Kwon et al., 2012, 2013; Blum et al., 2014). MIF of odd Hg isotopes (reported as Δ¹⁹⁹Hg and Δ²⁰¹Hg), the combined result of nuclear volume effect (NVE) and magnetic isotope effect (MIE), can provide additional insight to Hg biogeochemical cycling (Schauble, 2007; Buchachenko et al., 2007). Laboratory experiments quantifying elemental Hg⁰ volatilization (Estrade et al., 2009; Ghosh et al., 2013), equilibrium Hg-thiol complexation (Wiederhold et al., 2010), and dark aqueous Hg(II) reduction (Zheng and Hintelmann, 2010a) have demonstrated that NVE results in Δ¹⁹⁹Hg:Δ²⁰¹Hg of ∼1.6. Whereas MIE, a byproduct of photochemical reactions of aqueous Hg species and most prevalent mechanism resulting in MIF, results in Δ¹⁹⁹Hg:Δ²⁰¹Hg ranging from 1.00 to 1.30 (Bergquist and Blum, 2007; Zheng and Hintelmann, 2009, 2010b; Rose et al., 2015; Chandan et al., 2014). Large variations (>10‰) of both δ²⁰²Hg and Δ¹⁹⁹Hg values have been reported in environmentally relevant matrices such as rocks (Smith et al., 2008), mineral deposits (Hintelmann and Lu, 2003; Smith et al., 2005, 2008; Stetson et al., 2009; Sonke et al., 2010), coals (Biswas et al., 2008; Lefticariu et al., 2011; Sun et al., 2014; Yin et al., 2014b), soils (Biswas et al., 2008; Zhang et al., 2013; Feng et al., 2013), sediments (Gehrke et al., 2011; Foucher and Hintelmann, 2009; Feng et al., 2010; Liu et al., 2011; Bartov et al., 2013; Donovan et al., 2013; Ma et al., 2013; Yin et al., 2013a, 2015), water (Štrok et al., 2014), air (Demers et al., 2013; Rolison et al., 2013), precipitation (Gratz et al., 2010; Chen et al., 2012; Demers et al., 2013; Sherman et al., 2011, 2015) and biological samples (Bergquist and Blum, 2007; Blum et al., 2013; Demers et al., 2013; Yin et al., 2013b). With well-defined source signatures, researchers may estimate the relative contribution of sources of Hg through the utilization of binary and/or triple mixing models (Foucher and Hintelmann, 2009; Feng et al., 2010; Liu et al., 2011; Bartov et al., 2013; Donovan et al., 2013; Ma et al., 2013; Yin et al., 2013a; Lepak et al., 2016).

Determination of isotopic Hg signatures in sediment profiles have proven useful to better understand historical Hg loading to a variety of systems (Jackson et al., 2004; Feng et al., 2010; Sonke et al., 2010; Cooke et al., 2013; Donovan et al., 2013; Ma et al., 2013; Mil-Homens et al., 2013; Gray et al., 2013, 2015; Balogh et al., 2015). However, this task may be more complicated when attempting historical reconstruction for large lakes receiving loads from many tributaries with varied source signatures, contamination histories, and subject to significant changes in nutrient loading and benthos re-engineering by invasive species (Hecky et al., 2004; Rossmann, 2009). Source signatures of Hg isotopes may be altered following their release into the environment through continuous processing by various external physical, chemical and biological influences (Jackson, 2013; Donovan et al., 2014; Yin et al., 2015). For these reasons, the usefulness of Hg isotopes as tracers in large lakes is still unclear and therefore warrants further investigation.

Lake Michigan (LM), the fifth largest freshwater lake in the world, has received Hg loading from local industry through tributaries and from atmospheric deposition (regional to global) (Hartig and Law, 1994; Mason and Sullivan, 1997; Landis and Keeler, 2002; Hurley et al., 1998a, 1998b; Jeremiason et al., 2009; Rossmann, 2002, 2009; Rossmann and Edgington, 2000). Most point-source discharges have been eliminated as a result of the Clean Water Act (CWA) in the early 1970s (Adler et al., 1993), however residual contamination exists in rivers and nearshore sediments (Hurley et al. 1998a; Rossmann, 2002). In our study, both concentrations and isotopic Hg composition were investigated in sediment cores collected from a heavily impacted embayment of LM (Green Bay), and an offshore depositional basin of northern LM, not influenced by direct point sources. The objectives of this study were to: (1) reconstruct rates of input of Hg to northern LM and Green Bay; (2) compare sources of Hg to these depositional zones. Utilizing Hg stable isotopes allows us to better understand sources and processing pathways that lead to distinct depositional patterns in contrasting sedimentary zones of LM.

LM (Figure 1), the third largest of the Laurentian Great Lakes, is the only Great Lake located entirely within the United States. Atmospheric deposition has been reported to be the primary pathway for Hg input to LM, contributing approximately 84% of total annual input in 1994 to 1995 (Landis and Keeler, 2002; U.S. EPA, 2004). However on a local scale, atmospheric Hg may not be the primary source of Hg especially in regions such as bays and nearshore zones influenced by rivers contaminated with Hg. The Fox River, for example, has a long history of industrialization since the first paper mill was built in 1865, and has been listed as an Area of Concern by the USEPA (Christensen and Chien, 1981; Hermanson et al., 1991; Hartig and Law, 1994). Mercury has been used at during paper production and there are numerous pathways for Hg to enter the river, including direct wastewater discharge (Wisconsin Department of Natural Resources, 1997). Water inputs from the Fox River to Green Bay account for one-third of the entire LM drainage basin (Christensen and Chien, 1981) and of the eleven major tributaries (comprising approximately 55% of the LM drainage basin) selected to assess Hg inputs to LM (Hurley et al., 1998a; 1998b), the Fox River contained the highest mean Hg concentrations (Fox River: 27.9 ng L^-1, >90% in the particulate phase; other tributaries: 1.05 to 10.3 ng L^-1). Elevated THg concentrations (up to 7400 ng g^-1) were measured in the sediments of the Fox River (Redman, 1993) and Hurley et al. (1998a) measured a mean of 970 ng g^-1 THg and a maximum of 4190 ng g^-1 on particles collected at the river mouth in 1994.

In the fall 2013, sediment cores were collected from three sites using a custom-designed polycarbonate tube multicorer on board the USEPA R/V Lake Guardian. Sites MI-116 and MI-50 are located in depositional zones of Green Bay, and site MI-112 is located in the offshore region of the Northern Basin (Figure 1). Four cores were collected at each site with polycarbonate core barrels pre-cleaned following method by Hammerschmidt et al. (2011). One of the four cores at MI-50 and MI-112 was field sectioned using a clean polycarbonate slicer at 1.0 cm intervals from the surface to 5 cm, and then at 2.0 cm intervals to the base of the core. The core from MI-116 was frozen in the field, delivered to USGS Wisconsin Mercury Research Lab (WMRL), and then sectioned at 0.5 cm intervals from the surface to the base of the core using a stainless steel band saw. All samples were freeze-dried and homogenized by a ball mill prior to THg and Hg isotopic analyses. Moisture content (water %) and loss on ignition of carbon for each slice was measured (Table S1, Supplemental Information Text S1).

THg in sediments was analyzed at the USGS-Wisconsin Mercury Research Laboratory with a direct combustion system (Nippon MA-2) and atomic absorbance detection, based on U.S. EPA Method 7473 (SW-846). The method detection limit was 1.4 ng g^-1 dry weight. Recoveries of standard reference material (IAEA SL 1) were within 90 to 100%, and the coefficients of variation of triplicate analyses were less than 10%. Between 0.025 and 0.5 g of ground sample was digested (95 °C, 1 hour) in a 5 mL aqua regia (HCl:HNO₃ = 3:1, v:v). Certified reference materials (NIST-2711 and MESS-1) and a secondary solution (UM-Almadén; in aqua regia) were similarly prepared in 10% of the sample count. Isotopic compositions of sample digests were measured on a Neptune Plus MC-ICP-MS, located at the Wisconsin State Laboratory of Hygiene (WSLH). Details of the method used for MC-ICP-MS analysis are given in Supplemental Information Text S1. Following the convention recommended by Blum and Bergquist (2007), Hg-MDF is expressed in δ²⁰²Hg notation in units of permil (‰) referenced to the NIST-3133 Hg standard (analyzed before and after each sample):

Hg-MIF is reported in Δ notation (Δ^xxxHg), and it describes difference between the measured δ^xxxHg and the theoretically predicted δ^xxxHg value, in units of permil (‰), using the following formula:

β is equal to 0.2520 for ¹⁹⁹Hg, 0.5024 for ²⁰⁰Hg, and 0.7520 for ²⁰¹Hg (Blum and Bergquist, 2007).

UM-Almadén solutions at varying concentrations (0.3, 0.5 and 1 ng g^-1), matching specific sample concentrations of Hg, were measured in every 10 samples. Data uncertainties reported in this study reflect the larger values of either the external precision of the replication of the UM-Almadén or the measurement uncertainty of the sample. The overall average and uncertainty of the UM-Almadén measurements (δ²⁰²Hg: -0.52±0.09‰; Δ¹⁹⁹Hg: -0.01±0.05‰; Δ²⁰⁰Hg: 0.00±0.03‰; Δ²⁰¹Hg: 0.00±0.05‰, 2σ, n = 9) agreed well with previous studies (Blum and Bergquist, 2007). Measurements of UM-Almadén with different Hg concentrations showed no statistical differences in isotopic composition (Table S3). Measurements of replicate digests of NIST-2711 (δ²⁰²Hg: -0.13±0.11‰; Δ¹⁹⁹Hg: -0.16±0.09‰; Δ²⁰⁰Hg: -0.02±0.03‰; Δ²⁰¹Hg: -0.16±0.04‰, 2σ, n = 3) and MESS-1 (δ²⁰²Hg -1.85±0.10‰; Δ¹⁹⁹Hg 0.01±0.06‰; Δ²⁰⁰Hg: 0.01±0.03‰; Δ²⁰¹Hg 0.02±0.05‰; 2σ, n = 3) also agreed well with previous studies (Biswas et al., 2008; Donovan et al., 2013; Yin et al., 2014b).

Spatial variation in sedimentation rate has been reported in both Green Bay and offshore of LM. These studies suggest the sedimentation in a particular site is generally constant, based upon log-linear relationships between excess ²¹⁰Pb activities and the accumulated masses (Klump et al., 1997; Song et al., 2005; Zhu and Hits, 2005; Rossmann, 2002, 2009; Rossmann and Edgington, 2000). Based on the assumption of constant mass sedimentation (g cm^-2 yr^-1) and dry density measurements (g cm^-3) interpolated from loss on ignition and wet density of the sediment core intervals, the dry deposition rate (cm yr^-1) of each interval was calculated (Kadlec and Robbins, 1984). With this information, the sediment ages were estimated using the following:

where i represents the name of each core (i: MI-116, MI-50 and MI 112); N represents the number of slices removed from the surface slice of each core [e.g., MI-116 (1 ≤N≤ 58), MI-50 (1 ≤N≤ 17) and MI 112 (1 ≤N≤ 13)]; $TiN$⁠, $diN$ and $ρiN$ represent the age, thickness (cm) and density (g cm^-3) of each slice (N) in each core (i) (summarized in Table S1), respectively. $Ti0$ and R_i represent the collection time (fall 2013) and the mass sedimentation rate (g cm^-2 yr^-1) of each core. The mass sedimentation rates used in this study for sites MI-116 and MI-50 were previously reported by Klump et al. (1997) as station 21 (0.052 g cm^-2 yr^-1) and station 48 (0.022 g cm^-2 yr^-1), respectively (Figure 1). MI-112 is located within the Algoma central basin, with a mean sedimentation rate of 0.040 g cm^-2 yr^-1 reported for this basin (LM 41; Song et al., 2005). Based on equation 3, the range of dates in cores MI-116, MI-50 and MI 112 were estimated to be from 1873 to 2013, 1799 to 2012 and 1734 to 2012, respectively (Figure 2, Table S1).

THg concentration profiles of the three cores show elevated THg in the uppermost sections (Figure 2A), consistent with lake-wide cores by Rossmann (2009). The oldest slices of MI-116, MI-50, and MI-112 have THg concentrations of 36, 35, and 17 ng g^-1, respectively. This is comparable with the geochemical background of Hg in sediments of Great Lakes (∼25 ng g^-1) (Marvin et al., 2004). The greatest THg concentrations are in recently deposited surface layers of the 1970’s (Figure 2A). Influxes of Hg (mass sedimentation rate × THg) were calculated to account for the differences in sedimentation rate in each of the sediment cores (Figure 2B). Fluxes to sediment are comparable with that reported by previous studies (Pirrone et al., 1998; Rossmann and Edgington, 2000; Rossmann, 2002). Influx of Hg in MI-112 in 1998 was 6.0 ng cm^-2 yr^-1, similar to that reported for Algoma Central (5.2 ng cm^-2 yr^-1) from 1994 to 1996 (Rossmann, 2002). Influxes of Hg in the oldest sections of sediment cores are comparable among deposition regions: MI-116 (1.8 ng cm^-2 yr^-1); MI-50 (0.8 ng cm^-2 yr^-1); and, MI-112 (0.7 ng cm^-2 yr^-1).These values are comparable to the preindustrial Hg influx (0.8 ng cm^-2 yr^-1) reported for sediment cores in southern LM (Pirrone et al., 1998). Hg influx profiles peak around the mid-1900s and then decrease after the 1970’s, potentially coinciding with the Clean Water Act (Figure 2B) (Adler et al., 1993). The peak Hg influxes in MI-116, MI-50, and MI-112 are about 3100%, 1600%, and 1400% enhanced relative to preindustrial backgrounds, respectively.

Our sediment profiles (Figure 2C) represented here show large variations in δ²⁰²Hg in MI-116 (δ²⁰²Hg: -1.75±0.11 to -0.52±0.14‰, 2σ), MI-50 (δ²⁰²Hg: -1.73±0.17 to -0.63±0.14‰, 2σ) and MI-112 (δ²⁰²Hg: -1.29±0.17 to -0.56±0.15‰, 2σ). Variable odd Hg-MIF was also observed (Figure 2D). MI-116 (Δ¹⁹⁹Hg: -0.02±0.05‰ to 0.10±0.05‰, 2σ) appears absent of significant Hg-MIF before anthropogenic influences were introduced, while MI-50 (Δ¹⁹⁹Hg: 0.05±0.05 to 0.18±0.05‰, 2σ) and MI-112 (Δ¹⁹⁹Hg: 0.09±0.05 to 0.27±0.05‰, 2σ) show small but significant Hg-MIF. All sediments in LM yielded a Δ¹⁹⁹Hg:Δ²⁰¹Hg of 0.93±0.08 (2σ) (Figure 3), which is consistent with the photo-reduction of Hg(II) reported by Bergquist and Blum (2007). Evidences for stability of Hg isotope ratios in sediments over time have been document by Bartov et al. (2013). Loss of Hg during microbial Hg reduction may cause changes in Hg isotopic compositions over time (Kritee et al., 2013). However, with sediments containing organic matter at concentrations greater than Hg, such as LM sediments, exceptionally strong complexes of Hg(II) with organic matter may form (Morel et al., 1998). High organic matter content has been shown to inhibit Hg(II) reduction to Hg(0) (Mauclair et al., 2008; Gu et al., 2011). The variation of Hg isotopes in sediments then, is largely the result of varied Hg sources both spatially (Foucher and Hintelmann, 2009; Feng et al., 2010; Liu et al., 2011; Bartov et al., 2013; Donovan et al., 2013; Ma et al., 2013; Yin et al., 2013a; Lepak et al., 2016).

Historically, sources of Hg to LM have included direct discharge of anthropogenic Hg, watershed loading of soil Hg and precipitation (Mason and Sullivan, 1997; Landis and Keeler, 2002; Hurley et al., 1998a, 1998b; Jeremiason et al., 2009; Rossmann, 2002, 2009; Rossmann and Edgington, 2000). Intriguingly, the transition of δ²⁰²Hg from the deep layers is always positive in direction and very similar in magnitude for all three cores (Figure 2C), similar to previous observations on sediment cores from lakes and coastal areas (Donovan et al., 2013; Gray et al., 2013, 2015; Balogh et al., 2015). The deepest layer of core MI-112, presumably a pre-industrial period, has a negative δ²⁰²Hg value of ∼ -1.29±0.17(2σ) and positive Δ¹⁹⁹Hg value of ∼ +0.16±0.05 (2σ) (Figure 4A-B). These values are similar to previous data on deep layers of coastal cores (δ²⁰²Hg: ∼ -1.0‰; Δ¹⁹⁹Hg: ∼ +0.2‰)(Donovan et al., 2013; Balogh et al., 2015). Atmospheric deposition is an important input of Hg to Great Lakes (Landis and Keeler, 2002) and sea waters (Mason and Sheu, 2002). Precipitation scavenging oxidized Hg species (e.g., Hg^II_g and Hg_p) in the atmosphere is most likely the primary carrier for atmospheric Hg signals in offshore lake sediments (Shanley et al., 2015). Previous studies have reported mean δ²⁰²Hg of approximately -0.5‰ and positive Δ¹⁹⁹Hg (∼ +0.4‰) and Δ²⁰⁰Hg (mean: ∼0.25‰) in precipitation collected in the Great Lakes region (Gratz et al., 2010; Chen et al., 2012; Demers et al., 2013; Sherman et al., 2011, 2015). Atmospheric Hg(II) is susceptible to photoreduction in precipitation water droplets (Gratz et al., 2010), which may at least partially explain the positive pre-industrial Δ¹⁹⁹Hg in deep layer sediments of MI-112 and other coastal cores. Precipitation-derived Hg appears to be adsorbed to particulates and settle to the benthos. Adsorption of aqueous Hg(II) by particles containing thiol groups (Wiederhold et al., 2010), goethite (Jiskra et al., 2012) and sulfides (Foucher et al., 2013; Smith et al., 2015) may result in negative δ²⁰²Hg (-0.4 to -0.6‰) shifts in the solid Hg phase. Precipitation Hg with δ²⁰²Hg value of ∼ -0.5‰ plus a negative shift of -0.4 to -0.6‰ during adsorption, may also result in similar δ²⁰²Hg to deep layer sediments in MI-112 and coastal cores.These observations suggest that precipitation is main input of Hg to deep layers of MI-112.

The deepest layer sediments of MI-116 and MI-50 show δ²⁰²Hg values of -1.75±0.11‰(2σ) and -1.73±0.17‰(2σ), respectively, which are lower than that of the deepest layers in MI-112, but higher than previous data on the deep layer sediments from other lakes (δ²⁰²Hg: -3 to -2‰) (Figure 4A, Gray et al., 2013, 2015). The negative δ²⁰²Hg in deep layers of other lakes are similar to that reported for terrestrial soils (δ²⁰²Hg: -2.0±0.6‰, σ, n = 48) (Biswas et al., 2008; Zhang et al., 2013; Demers et al., 2013; Jiskra et al., 2015). Mercury strongly bound to soil matrices maybe subjected to only minimal aqueous processing and therefore result in insignificant isotope fractionation during sedimentation (Foucher and Hintelmann, 2009; Liu et al., 2011). Deep layer sediments from other lakes have shown negative Δ¹⁹⁹Hg values similar to ours (-0.3 to -0.1‰) (Figure 4B; Gray et al., 2013, 2015). Watershed soil particles are likely the primary source of Hg to these lakes as Hg in terrestrial soils has shown to be characterized by negative Δ¹⁹⁹Hg values (-0.2±0.2‰, σ, n = 48) (Biswas et al., 2008; Zhang et al., 2013; Demers et al., 2013; Jiskra et al., 2015; Feng et al., 2010; Liu et al., 2011). If we consider physical mixing of sources containing positive and negative MIF, the lack of significant MIF (either positive or negative) in the pre-industrial sediment of MI-116 may indicate similar proportions of Hg from watersheds and precipitation were introduced (if similar magnitudes of MIF from each source are assumed). In general, Δ¹⁹⁹Hg values in deep layers decreased from MI-112, MI-50 to MI-116 (Figure 2D), suggesting increase of Hg inputs by watershed soils.

The uppermost sections of the MI-116 core show much higher δ²⁰²Hg (-1 to 0‰) and the absence of Hg-MIF (Δ¹⁹⁹Hg ∼ 0) (Figures 2C–D and 4A–B). These values are similar to previous data on major anthropogenic Hg sources, which are in general characterized by higher δ²⁰²Hg values ranging from -1 to 0‰ and insignificant Hg-MIF (Sonke et al., 2010; Liu et al., 2011; Blum et al., 2014). Although there are many industries that may release Hg into LM, Hg used in industry is primarily derived from mining, and therefore should exhibit similar signatures. Previous results on Hg ores other metal deposits exhibited δ²⁰²Hg of ∼ -0.6±0.4‰ (σ) and Δ¹⁹⁹Hg close to 0 (Sonke et al., 2010; Liu et al., 2011; Blum et al., 2014; Hintelmann and Lu, 2003; Smith et al., 2005, 2008; Stetson et al., 2009; Laffont et al., 2011; Yin et al., 2013a). Sediments with high THg concentrations is predominately derived from anthropogenic sources as deposition typically occurs near the source, waters are relatively turbid (thereby reducing photochemical processes) or the Hg from these sources are less susceptible to photo-reduction (Foucher and Hintelmann, 2009; Feng et al., 2010; Gehrke et al., 2011; Sonke et al., 2010; Donovan et al., 2013; Cooke et al., 2013; Ma et al., 2013; Mil-Homens et al., 2013; Gray et al., 2013, 2015; Balogh et al., 2015; Yin et al., 2013a; Lepak et al., 2016). Uppermost sections of MI-50 and MI-112 have similar δ²⁰²Hg values with MI-116, but show slightly positive Δ¹⁹⁹Hg. Δ¹⁹⁹Hg generally decreases with increased THg concentrations, which may indicate a dilution by anthropogenic Hg sources. Overall, we suggest that anthropogenic Hg sources are important inputs of Hg in top layer sediments of LM, certainly within Green Bay.

Evaluation of mixing of different sources can be achieved when plotting δ²⁰²Hg and 1/THg values, when using binary and ternary mixing models of Hg isotopes to trace and quantify source apportionment of Hg in sediments (Foucher and Hintelmann, 2009; Liu et al., 2011; Donovan et al., 2013; Ma et al., 2013; Yin et al., 2013a). In our study, δ²⁰²Hg and 1/THg form a linear correlation, suggesting physical mixing of anthropogenic Hg and geochemical background sources of Hg (Figure 4A). Like other small inland water bodies, the background Hg of MI-116 and MI-50 in Green Bay may represent mixing of precipitation and watershed soils, whereas that of MI-112 in offshore LM may be more similar to coastal regions of the ocean where primary Hg influences are mainly atmospherically derived (as mentioned in Section 3.3). For these reasons, we believe the δ²⁰²Hg and 1/THg mixing lines appear to fall on two different slopes in Figure 4. With the mixing lines shown in Figure 4A, it is possible to estimate the influence of both anthropogenic and background signals using an end-member approach and the following binary mixing model (Equations 5–7):

where F_ant. and F_bac. represent fractions of the anthropogenic and background Hg end-members, respectively. δ²⁰²Hg_ant. and δ²⁰²Hg_bac. represent the δ²⁰²Hg of anthropogenic and background Hg end-members, respectively, and δ²⁰²Hg_sample represents the δ²⁰²Hg of a given sample. Values of δ²⁰²Hg for the samples at the base of the each core (MI-116: -1.75‰; MI-50: -1.73‰; MI-112: -1.29‰) were used to represent the background end-member for each core in this study. A core slice with the highest δ²⁰²Hg (-0.52‰) from core MI-116 was chosen to represent the isotopic signature of the anthropogenic end-member. This was chosen because it correlates with the highest concentration of THg at about 1930. With this information the variation in F_ant of each sediment core was evaluated (Figure 5).

Model output indicates two time periods when the onset of anthropogenic Hg pollution in LM was evident. The first indication of Hg pollution starts in the early 1800s in MI-112 and MI-116, and most likely represents the influence of the global industrial revolution; the second indication of Hg pollution starts in the 1880s in both MI-116 and MI-50, and likely reflects the influence of local industry which started in 1865 when the first paper mill was built. Post 1920’s, all three sites display relatively high degrees of anthropogenic influence ranging from 50 to 100%. In addition, a peak of F_ant. was observed in the 1850s in MI-112. Schuster et al. (2002) observed a similar episodic Hg accumulation event in an ice core from a Wyoming glacier which was correlated to California Gold Rush (1850 to 1878) at such a distance from the source. It is not clear whether this peak is related to the Gold Rush, but sediment cores from other lakes in the Midwest have also recorded episodic events during that time period although that study did not attempt to interpret the origin of the peaks (Engstrom and Swain, 1997). We also observed a decreasing trend of F_ant. following the 1970s in offshore core MI-112, which may be the result of recent remediation efforts, as a result of the CWA.

Due to the lack of Hg isotope data in precipitation throughout time, further estimates of the contribution of Hg from precipitation and watershed soils was not conducted in our study. Mercury isotopic composition in modern precipitation show large variations between pristine and industrial-urban areas (Gratz et al., 2010; Chen et al., 2012; Demers et al., 2013; Sherman et al., 2011, 2015). While variable, modern precipitation has shown average Δ²⁰⁰Hg values of ∼ +0.20‰, whereas most other environmental matrices have shown Δ²⁰⁰Hg values close to zero. Δ²⁰⁰Hg has been used as a tracer for precipitation Hg (Lepak et al., 2016). In this study, the MI-116 profile shows insignificant MIF of ²⁰⁰Hg compared to analytical uncertainty for Δ²⁰⁰Hg (±0.04‰, 2σ) (Figure 2E). Somewhat interestingly, many samples in MI-50 and MI-112 have small but significant positive Δ²⁰⁰Hg (> +0.04‰). These samples are located in the deepest layers (prior-1850s) and the uppermost slices (post-1970s). The absence of ²⁰⁰Hg MIF in deep layers of MI-116 suggests that precipitation-derived Hg is diluted by watershed soils, whereas the small positive Δ²⁰⁰Hg in the deep layers (prior-1850s) of MI-50 and MI-112 may suggest that the proportion of atmospherically derived Hg is larger. Increased inputs of anthropogenic Hg may signify an absence of ²⁰⁰Hg MIF in sediments from 1850s to 1970s in MI-50 and MI-112, when industrialization and Hg loading to this system was the greatest. The slightly positive Δ²⁰⁰Hg in top layers of MI-50 and MI-112 is apparent.

Our results on sedimentary Hg influx profiles confirm previous studies on the timing and extent of Hg pollution in LM. Our isotopic studies, coupled with a binary mixing model, enabled us to differentiate between the historical contribution from anthropogenic Hg sources and background inputs. Isotopic analyses using Hg-MIF further allowed us to compare the extent of Hg(II) photo-reduction processes in the water column among sites with different Hg sources. Hg isotopes can be a useful tracer to reveal sources and fate of Hg in large aquatic ecosystems, such as the Great Lakes.

Questions can be directed to the author’s attention: [email protected] and/or [email protected].

© 2016 Yin et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

We would like to thank Michael Tate, Jacob Ogorek, John DeWild, Morgan Maglio and Charlie Thompson from Wisconsin Mercury Research Lab, U.S. Geological Survey, and Middleton WI for the assistance in sample collection and THg analysis. We also thank professor Xinbin Feng from Institute of Geochemistry, Chinese Academy of Sciences to revise this paper. Two anonymous reviewers and editors Robert P. Mason and Joel D. Blum are acknowledged for their constructive comments and useful suggestions that have largely improved the quality of this paper.