Abstract
The Michigan Basin is composed of geological formations that contain brines and evaporites, and solutes from these geological sources have affected benthic sediment pore-water chemistry in Saginaw Bay (Lake Huron). We hypothesize that there exists similar potential for upward solute transport directly from the Michigan Basin into other Great Lakes areas. To test our hypothesis, we present here previously unpublished pore-water chemistry analyses from sediment cores collected during multiple Lake Michigan sampling events (spanning 1991–1999) and a new evaluation of previously published data. In several box cores, pore-water chloride concentrations increase with depth, and Cl:Br ratios are consistent with a geological formation brine source. In all gravity cores we collected from southern Lake Michigan, pore-water sodium concentrations increase with sediment depth. At one sample station, pore-water sodium concentrations exceed 2000 mg L−1 within 2 m of the sediment-water interface. Given the pore-water chemistry changes reported here, combined with information from previous studies of Lake Michigan bedrock geology, a Devonian formation brine is a plausible solute source. The presence of saline pore water within glaciolacustrine sediments underlying Lake Michigan indicates that this solute flux has been active during the past 10 k.y. However, the origins of this solute flux, including timing (onset) and contributions from advective and/or diffusive transport, are unknown. The specific geological source and solute transport process are important to resolve in order to evaluate potential effects of these Michigan Basin solute sources on the Great Lakes' sediment biogeochemistry and water quality.
INTRODUCTION
Recent monitoring studies have documented increasing dissolved solids levels in Great Lakes waters (Chapra et al., 2012; Dugan et al., 2023). Understanding the sources, transport, and fate of dissolved solids in the Great Lakes is important for constraining quantitative estimates of residence times and water-quality projections (Chapra et al., 2009). These estimates provide a foundation for policy development and implementation plans addressing several issues, including: (1) anthropogenic releases of chloride and other dissolved solids to the environment (Canadian Council of Ministers of the Environment, 2011; Michigan Department of Environment, Great Lakes, and Energy, 2021; Harrington et al., 2022; Mackie et al., 2022), and (2) Great Lakes water-use issues, such as water withdrawals (Kelly et al., 2012).
Groundwater is a substantial component of the Great Lakes water budget (Holtschlag and Nicholas, 1998; Kornelsen and Coulibaly, 2014; Grannemann and Van Stempvoort, 2016). Likewise, groundwater is an important transport vector for dissolved solids from anthropogenic and natural sources (Harrington et al., 2022; Mackie et al., 2022), especially on time scales ranging from years to decades and longer (Kornelsen and Coulibaly, 2014; Grannemann and Van Stempvoort, 2016; Cartwright, 2020). Groundwater flow in the Michigan Basin generally is focused regionally toward topographic low areas, with upward hydraulic gradients (Vugrinovich, 1986, 1988; Westjohn and Weaver, 1996a), indicating the potential for upward flow and direct discharge to the Great Lakes. Although recent estimates (Xu et al., 2021) indicate that direct groundwater discharge is a minor component (~0.8%) of the Great Lakes water budget, the effect of this component on dissolved solids transport is unknown. Given the relatively long residence times of water in the Great Lakes—ranging from years to centuries (Chapra et al., 2009)—understanding solute transport from groundwater directly to benthic sediments could be an important parameter to consider when quantifying dissolved solids exchanges (fluxes) between benthic sediments and the Great Lakes water column.
Several Michigan Basin geological formations underlying the Great Lakes drainage basin contain brines and evaporites (Alling and Briggs, 1961; Wold et al., 1981; Vugrinovich, 1986; Frape et al., 1989; Harrison and Voice, 2017). Previous glacial loading and meltwater recharge events are known to have displaced dense brines and dissolved subsurface evaporites, e.g., halite, within these formations (McIntosh and Walter, 2005; Person et al., 2007). These events have substantially altered salinity profiles and emplaced saline fluids laterally and vertically into relatively near-surface environments in the Michigan Basin (Wahrer et al., 1996; McIntosh and Walter, 2005; McIntosh et al., 2011; Person et al., 2012). Solute transport—via diffusion and/or advection—from Michigan Basin saline groundwater is known to have affected pore-water chemistry at several Great Lakes locations (Desaulniers et al., 1981; Cumming and Al-Aasm, 1999; Kolak et al., 1999).
Given the hydraulic drive for upward groundwater flow directly into Lake Michigan (Cartwright et al., 1979; Vugrinovich, 1986), and the combination of regional-scale processes affecting groundwater flow and salinity gradients within the Michigan Basin, we hypothesize that there exists the potential for mixed and/or saline water underneath Lake Michigan at relatively shallow depths (Kolak and Long, 2023). The purpose of this study was to investigate whether salinity, derived from sources at depth, is detectable in Lake Michigan benthic sediments. To test this hypothesis, we present here previously unpublished Lake Michigan pore-water chemistry analyses from sediment box cores (Fig. 1; Table 1) collected during 1991–1993 (analyzed in 1996), and from sediment gravity cores collected in 1999 (analyzed in 1999–2000). These data, in combination with a new evaluation of data from Callender (1969), underpinned assessments of both vertical and horizontal (regional) changes in pore-water chemistry. When compared with variations in sediment type and in regional bedrock geology, these assessments supported preliminary interpretations regarding potential source(s), timing, spatial extent, and geological factors affecting the extent of communication between Lake Michigan benthic sediments and a saline groundwater source at depth.
Pore-Water Salinization in Great Lakes Benthic Sediments—Previous Studies
Benthic sediment pore-water chemistry can identify areas affected by direct groundwater input and/or solute transport from underlying geological formations because changes in pore-water composition, i.e., “gradients,” would document dissolved solids concentrations that increase with sediment depth. The earliest Great Lakes studies to report dissolved solids concentrations, i.e., chloride and/or sodium, that increase with sediment depth include: Callender (1969; Lake Michigan), Weiler (1973; Lake Ontario), and Nriagu and Dell (1974; Lake Erie). The inferred source or process affecting dissolved solids concentrations varied among these studies. Callender (1969) attributed the measured increases in dissolved sodium and chloride concentrations to clay membrane filtration or to release from organic matter decomposition. Weiler (1973) reported that, at one western Lake Ontario station, dissolved sodium concentrations increased with sediment depth, along with a corresponding change in the Na:Cl molar ratio, and attributed these changes to groundwater flow from shore into the lake. However, at most western Lake Ontario stations sampled (Lerman and Weiler, 1970; Weiler, 1973), dissolved chloride and sodium concentrations decreased with sediment depth; these trends were interpreted to indicate that the overlying water column generally was the dominant source for chloride and sodium to pore water in Lake Ontario benthic sediments (Lerman and Weiler, 1970; Weiler, 1973). Nriagu and Dell (1974, their table 1A) reported results of pore-water analyses from Lake Erie benthic sediments and noted that, at several western Lake Erie sample stations, dissolved sodium concentrations increased with sediment depth, but no source or transport mechanism for these increases was identified. Nriagu and Dell (1974, p. 945, and their table 2A) compared their findings with those from Weiler (1973) and concluded that the setting in Lake Erie—increasing sodium concentrations with increasing sediment depth—was the “reverse” of what generally was found in most western Lake Ontario sample stations and was perhaps unique from the other Great Lakes.
Subsequent research has expanded on these foundational studies and shown that this phenomenon—increasing levels of dissolved chloride and/or sodium with sediment depth—is not unique to a specific location within the Great Lakes drainage basin. More recent studies have identified connections between dissolved solids levels in benthic sediments (or water samples) and deeper, saline groundwater sources at multiple sampling events across the Great Lakes. In Lake Ontario, several studies (Drimmie et al., 1992; Drimmie and Frape, 1996; King, 1997) reported increases in dissolved solids levels with increasing sediment depth. At one Lake Ontario sample station, a pore-water chloride concentration of 4000 mg L−1 was measured at ~5 m sediment depth (Drimmie et al., 1992; Drimmie and Frape, 1996, their figs. 3 and 6), with a likely connection (Drimmie and Frape, 1996) to formation water from bedrock units underneath the lake. Elsewhere, a connection between dissolved solids in pore water, or benthic water-column samples, and a Michigan Basin groundwater source was reported at several locations (Fig. 1) in Lake Erie (Haack et al., 2005) and in Lake Huron (Kolak et al., 1999; Hoaglund et al., 2004; Ruberg et al., 2005; Baskaran et al., 2016). Elevated dissolved solids levels, i.e., chloride, measured in Lake Huron sinkholes (Ruberg et al., 2005; Baskaran et al., 2016) demonstrated the potential for direct groundwater inputs to affect Great Lakes benthic sediment chemistry and subsequently to alter benthic microbial ecosystems (Kinsman-Costello et al., 2017). Taken together, the findings from these more recent studies indicate that salinization of pore-water chemistry in Great Lakes benthic sediments is more extensive and widespread than previously known.
Despite these advances, a possible connection between elevated dissolved solids levels in Lake Michigan benthic sediments and a Michigan Basin source remains elusive. Callender (1969) mentioned clay membrane filtration, a process invoked to explain brine formation in sedimentary basins, as the more probable mechanism affecting chloride and sodium levels. However, given the significant amount of overpressure (Hanor, 1994) required to sustain this process, clay membrane filtration is unlikely to occur within the glacial and benthic sediments underlying Lake Michigan. Rather, a connection, i.e., solute mixing, with a groundwater source at depth seems more plausible. Cartwright et al. (1979) reported such a connection between groundwater and pore water from nearshore Lake Michigan stations. Cartwright et al. (1979) also measured upward hydraulic gradients in benthic sediments from offshore (southern) Lake Michigan stations, but no companion geochemical analyses were made at the offshore stations from which to evaluate a possible connection with a Michigan Basin solute source. The pore-water geochemistry data we present here, along with a new evaluation of the Callender (1969) data set, provide a basis for evaluating possible connections among Michigan Basin salinity sources and the “groundwater–sediment–lake water system” (Cartwright et al., 1979, p. 68).
METHODS
Sample Collection and Pore-Water Extraction
Sediment Box Cores (1991–1993)
Sediment box cores were collected from sample stations in Lake Michigan (LM) aboard the R/V Lake Guardian during multiple sampling events spanning 1991 through 1993 (Table 1). The primary objective of these sampling events was to evaluate trace metal and organic contaminant loading histories recorded in Great Lakes sediments. Several of the sampling stations were collocated with U.S. Environmental Protection Agency (U.S. EPA) long-term monitoring stations (see Heft, 1993; Barbiero et al., 2018, their fig. 2). Aboard the R/V Lake Guardian, sediment box cores were collected using a stainless-steel box-coring device measuring 70 × 30 × 40 cm. Detailed descriptions of the box core subsampling and subcore sectioning procedures are reported elsewhere, including specific measures taken for trace metal analyses (Heft, 1993; Sitarz, 1996) and for trace organic contaminant analyses (Simcik et al., 1996).
Sediment subcores collected for pore-water analyses during these 1991–1993 sampling events were sectioned immediately in the onboard laboratory into finer depth increments (~0.5–2.0 cm thickness), transferred into precleaned (acid-washed and distilled, deionized water [DDW]-rinsed) polyethylene (or polypropylene) centrifuge tubes, and refrigerated at 4 °C until they were homogenized and then frozen onboard. These frozen core sections were transported to Michigan State University (MSU) and stored frozen until 1996. These core sections remained viable for pore-water analyses of dissolved chloride (Cl) and bromide (Br), given that these anions generally are conservative at dilute concentrations. However, the potential for adsorption and/or mineral-water interactions during the long holding time precluded the use of these core sections for major cation pore-water analyses. Likewise, pore-water samples were not analyzed for stable isotope analyses (δ18O and δD in water), given the potential for isotope fractionation during the long holding times and freeze/thaw cycle. In 1996, the frozen core sections were thawed overnight at room temperature and then centrifuged for 10 min at 10,000 rpm to separate the pore water. Each resulting pore-water sample (supernatant) was decanted into a precleaned (acid-washed, DDW-rinsed) plastic syringe and filtered through a Millex-HA® 0.45 µm disposable filter into a precleaned polypropylene bottle. Pore-water samples were stored refrigerated at 4 °C until they were analyzed for dissolved chloride and bromide in 1996.
Sediment Gravity Cores (1999)
Seven sediment gravity cores, measuring up to ~3 m in length, were collected from southern Lake Michigan aboard the R/V Lake Guardian during a sampling event conducted 30 August–1 September 1999 (Table 1; Table S11). No companion box cores and no water-column samples were collected. The gravity cores were collected under the auspices of a “vessel of opportunity” study designed to complement planned U.S. EPA research and monitoring activities for the 1999 Lake Michigan Summer Survey; hence, several gravity coring stations were collocated with U.S. EPA long-term monitoring stations (Fig. 1; Table S1; see also Barbiero et al., 2018, their fig. 2). One key objective of the 1999 sampling event was to collect longer (~3 m) sediment cores near stations with the strongest gradients in dissolved solids concentrations as reported in the Callender (1969) study. Sample stations LM 99MSU1 and LM 99MSU2 (Fig. 1, starred locations labeled “1” and “2,” respectively) were located near station S1-29, where Callender (1969, their appendix III) reported a maximum dissolved sodium concentration of 250 mg L−1 measured at 50–55 cm sediment depth.
The 1999 sediment cores were collected using a Benthos gravity coring device, outfitted with a ball valve and piston corer sleeve, obtained from the Center for Great Lakes and Aquatic Sciences (University of Michigan, Ann Arbor, Michigan). The coring device was loaded with polycarbonate liners measuring 73 mm (outer diameter) and weighted with up to eight 25 kg weights. The coring device was lowered via the onboard winch to maintain vertical orientation during entry into the sediment column; the coring device was not allowed to free fall into the sediment column. Once retrieved and secured on deck, the polycarbonate liner was removed from the core barrel and stoppered at the base, and an incision was cut at the top of the liner to drain off excess water. Sediment cores were immediately cut into sections measuring ~1–2 ft (0.3–0.6 m) in length, capped and sealed, and stored at 4 °C in the onboard laboratory until offloaded. Measurements of mud line depth, observed on the outside of the core barrel, and maximum core recovery were recorded at each station (Table S1). Sediment recoveries generally were good, with only two stations, LM 99MSU1 and LM 99MSU7, exhibiting significant loss (>50%) of sediment.
Gravity core sections were taken to the MSU Environmental Geochemistry Laboratory (East Lansing, Michigan) for pore-water extraction and analysis. All gravity core sections were processed and pore-water samples were collected within one month following the 1999 Lake Michigan sampling event. Core sections were stored at 4 °C until squeezing. Immediately prior to squeezing, subsamples from each core section were cleaned by removing the outermost (~1 cm) sediment to avoid contamination from any disturbed (smeared) sediment along the sediment core–polycarbonate liner interface. Descriptions of cleaned sediment subsamples were recorded (Table S2) prior to pore-water extraction. No companion sediment samples were taken for mineralogical analyses; we refer the reader to findings from Callender (1969) regarding Lake Michigan benthic sediment mineralogy and mineral dissolution-precipitation reactions.
Cleaned sediment subsamples were squeezed in a stainless-steel squeezing apparatus (Patterson et al., 1978; Drimmie and Frape, 1996; Kolak et al., 1999) at room temperature for 2–24 h, depending on the volume of pore water collected. Pore-water samples were collected in plastic syringes directly connected to the squeezing device. After collection, pore-water samples were split into multiple aliquots for geochemical analyses, including anions and stable isotopes (δ18O and δD in water), but only the major cation analyses (presented here) were completed. Pore-water samples collected for major cation analysis, including calcium (Ca), magnesium (Mg), potassium (K), and sodium (Na), were filtered through Millex-HA® 0.45 µm disposable filters into precleaned (acid-washed, DDW-rinsed) polypropylene bottles, acidified to pH <2 with ultrapure nitric acid, and stored at 4 °C until analysis.
Pore-Water Chemistry Analyses
Dissolved Chloride and Bromide
In 1996, pore-water samples separated from the thawed box core sediments (collected during 1991–1993) were analyzed in a laboratory at the MSU W.K. Kellogg Biological Station (Hickory Corners, Michigan). Dissolved chloride and bromide concentrations were measured via high-performance liquid chromatography using a Dionex 4000I ion chromatograph with an AS4A column and suppressed conductivity detector. Calibration standards were prepared fresh daily in class A volumetric flasks using serial dilutions of a commercially available stock solution. Results from pore-water chloride and bromide analyses (Data S1) were plotted using the mean depth of the corresponding sediment interval.
Dissolved Major Cations
Dissolved major cation concentrations in pore-water samples extracted from gravity core sediments (collected during 1999) were measured at the MSU Environmental Geochemistry Laboratory via a Perkin Elmer 5100c spectrophotometer using an air-acetylene flame and standard methods for flame atomic absorption (Ca, Mg, and K) and for flame atomic emission (Na) (Wilson, 1989, their tables 1–3; see also APHA, AWWA, and WCPF, 1971; Table S3). Sodium analyses were completed in September 1999; analyses for calcium, magnesium, and potassium were completed in December 2000. An external check on calibration accuracy was performed through analysis of an aliquot of National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 1643d, Trace Elements in Water, with every batch of samples (National Institute of Standards and Technology, 1999; Table S4). Prior to analysis, aliquots of SRM 1643d were spiked according to the same procedure used for samples and calibration standards; aliquots of SRM 1643d also were diluted (1:10) and analyzed (Table S4) to verify the quality of the dilution procedure. Results from pore-water major cation analyses (Data S2) were plotted using the mean depth of the corresponding sediment interval.
RESULTS
Pore-Water Chemistry: Chloride and Bromide
Dissolved chloride concentrations measured in pore-water samples from sediment box cores ranged from 7 mg L−1 to 42 mg L−1 (Fig. 2A). Nearest the sediment-water interface, pore-water chloride concentrations (Fig. 2A) ranged from 10 mg L−1 to 13 mg L−1. These concentrations generally were similar to dissolved chloride levels measured in Lake Michigan water-column samples (Rockwell et al., 1989; Chapra et al., 2009, their fig. 3; Yurista et al., 2015). Dissolved chloride concentrations within the uppermost 5 cm of sediment generally increased toward the sediment-water interface, likely reflecting the influence of water-column composition on shallow pore-water chemistry. At one sample station from northern Lake Michigan, LM 93-70 M (~24 cm total sediment depth), pore-water chloride concentrations decreased with sediment depth, indicating that the water column likely is the dominant chloride source for the entire core sample.
At several southern Lake Michigan sample stations, sediment cores recorded a distinct transition from a water column–dominated system (uppermost ~5 cm) to one in which a chloride source, located at depth, affected pore-water chemistry (Fig. 2A). At stations LM 91BC2 and LM 91BC3, the maximum chloride concentrations (~30–40 mg L−1) measured at depth were approximately three times greater than corresponding concentrations measured nearest the sediment-water interface. These maximum concentrations also were greater than levels previously measured in open waters from Lake Michigan (Rockwell et al., 1989; Chapra et al., 2009; Yurista et al., 2015). These findings compared favorably with data from Callender (1969), who reported similar increases in pore-water chloride concentrations at southern Lake Michigan sample stations.
Dissolved bromide concentrations ranged from 0.10 mg L−1 to 5.3 mg L−1, with the greatest concentration measured in core LM 91BC4 (Fig. 2B). These pore-water bromide concentrations were greater than a mean water-column bromide concentration, 0.017 mg L−1, reported for Lake Michigan (Tiffany and Winchester, 1969, their table 9), but comparable with pore-water bromide concentrations in Saginaw Bay, Lake Huron (0.03–1.7 mg L−1; Kolak, 2000). Dissolved bromide concentrations from deeper (>20 cm depth) Lake Michigan sediments ranged in value from 0.12 mg L−1 to 0.69 mg L−1. Depth profiles of pore-water bromide concentrations (Fig. 2B) generally differed from chloride profiles in that the highest bromide concentrations occurred nearest the sediment-water interface. The presence of elevated bromide concentrations nearest the sediment-water interface might reflect additional release of bromide to pore water from sediments via postdepositional or early diagenetic processes such as organic matter decomposition (Gerritse and George, 1988).
The ratio of dissolved chloride to dissolved bromide (Cl:Br, expressed here on a mg L−1 basis) in water samples (Fig. 2C) can be a useful parameter for distinguishing among different anthropogenic and natural chloride sources to the environment, including formation brines and road salt (halite) (Panno et al., 2006). In sediment box cores, Cl:Br ratios in pore-water samples nearest the sediment-water interface (uppermost ~5 cm sediment) ranged from 2 to 41. In deeper sediments (>20 cm depth), Cl:Br ratios ranged from 34 to 172. At station LM 93-70 M, Cl:Br ratios in pore water ranged in value from 19 to 54, but there was no discernible trend with depth. In all three southern Lake Michigan sample stations, greater Cl:Br ratios were measured at depth, which could indicate the presence of a different solute source at depth.
Pore-Water Chemistry: Major Cations
Among the 1999 gravity cores collected, there were several instances (Fig. 3; data plotted on a logarithmic scale) in which major cation concentrations increased with sediment depth. Callender (1969, their appendix III) also reported increases in dissolved cations with increasing sediment depth; data from that study (station S1-29, bold, white-filled squares in Fig. 3) are plotted here for comparison. To provide further context for evaluating changes with depth (“gradients”) in pore-water composition, we also plotted data from Saginaw Bay, Lake Huron (core SB4; small black-filled squares in Fig. 3). These data are pertinent given that a salinity source underneath Saginaw Bay has affected benthic sediment pore-water chemistry (Kolak et al., 1999; Hoaglund et al., 2004).
Dissolved calcium concentrations (Fig. 3A) varied with sediment depth among the seven gravity cores, indicating that several processes affect pore-water calcium concentrations. Concentrations ranged from 38 mg L−1 to 90 mg L−1 in the uppermost (<15 cm mean sediment depth) pore-water samples. At two sample stations, LM 99MSU2 and LM 99MSU7, dissolved calcium concentrations measured in shallow pore water were comparable to those reported for Lake Michigan offshore water-column samples (35–36 mg L−1; Rockwell et al., 1989), indicating that the water column might be the dominant calcium source in shallow sediments. In the remaining five gravity cores, shallow pore-water calcium concentrations were greater than those reported for the water column, likely indicating the influence of early diagenesis, e.g., carbonate mineral dissolution, on pore-water chemistry. Callender (1969) reported a similar finding and concluded that, due to carbonate mineral dissolution, Lake Michigan pore-water composition generally was saturated with respect to the mineral phases calcite and dolomite. In deeper (>1 m depth) sediments, pore-water calcium concentrations ranged from 34 mg L−1 to 580 mg L−1. At station S1-29, Callender (1969) reported a maximum calcium concentration of 69 mg L−1 at a depth of 50–55 cm, which compares favorably with the pore-water profile from station LM 99MSU1; both were similar to Saginaw Bay core SB4. Collectively, the seven Lake Michigan gravity cores likely show that water-column chemistry and early diagenetic processes affect pore-water calcium concentrations in the uppermost portions of the sediment column, but at depth, another calcium source affects pore-water composition.
Dissolved magnesium pore-water concentrations (Fig. 3B) varied among the gravity cores. In the uppermost (<15 cm mean sediment depth) pore-water samples, magnesium concentrations ranged from 11 mg L−1 to 24 mg L−1, with concentrations in several gravity cores similar to levels reported for the Lake Michigan offshore water column (11–12 mg L−1; Rockwell et al., 1989). This similarity indicates that, at several sample stations, the overlying water column might be the dominant magnesium source to shallow pore water. However, in deeper pore-water samples (>1 m sediment depth), magnesium concentrations ranged from 12 mg L−1 to 350 mg L−1, with the maximum concentration measured at station LM 99MSU1. Dissolved magnesium concentrations in pore-water samples from station S1-29 (Callender, 1969) also increased with increasing sediment depth, reaching a maximum concentration of 28.6 mg L−1 at 50–55 cm depth. Callender (1969) reported that Lake Michigan pore-water samples generally were saturated with respect to dolomite, indicating that within benthic sediments, this mineral phase might control dissolved magnesium concentrations. Collectively, dissolved magnesium (Fig. 3B) concentrations varied among sample stations in a manner similar to that observed for dissolved calcium, indicating that similar processes might affect dissolved concentrations of both elements.
Except for station LM 99MSU1, dissolved potassium concentrations (Fig. 3C) generally were similar among the gravity cores. In several cores, the maximum potassium concentration occurred nearest (<15 cm depth) the sediment-water interface, with concentrations ranging from 1.7 mg L−1 to 3.3 mg L−1. These concentrations were elevated slightly with respect to mean potassium concentrations measured in Lake Michigan water-column samples (1.2 mg L−1; Rockwell et al., 1989). Dissolved potassium concentrations in the deepest samples ranged from 1.6 mg L−1 to 12 mg L−1, with the highest concentration measured at station LM 99MSU1. Dissolved pore-water potassium concentrations from station S1-29 (Callender, 1969) were comparable with those measured in cores from stations LM 99MSU1 and SB4. Dissolved potassium concentrations in southern Lake Michigan gravity cores, except for station LM 99MSU1, appear to be largely driven by water-column inputs and by potassium release from shallow sediments via early diagenetic processes, e.g., organic matter decomposition and/or silicate mineral dissolution. Callender (1969, their table 6) reported a positive correlation between dissolved potassium and dissolved silica levels in pore-water samples from northern Lake Michigan, indicating that silicate mineral equilibria, e.g., kaolinite dissolution-precipitation, might control dissolved potassium concentrations; however, no such correlation was found in pore-water samples from southern Lake Michigan.
Dissolved sodium concentrations (Fig. 3D) increased with sediment depth in all seven gravity cores. Except for station LM 99MSU1, pore-water sodium concentrations measured in the uppermost samples (<15 cm mean sediment depth) ranged from 8.2 mg L−1 to 11 mg L−1. These concentrations were greater than those measured in offshore Lake Michigan water-column samples (5.4–5.5 mg L−1; Rockwell et al., 1989) but generally comparable with the concentration range (3.8–12.7 mg L−1) reported for a compilation of Lake Michigan nearshore water-column analyses (see Torrey, 1976, their fig. 76, and studies cited therein). This water-column signature, present only in samples nearest the sediment-water interface, is readily overprinted by a sodium source at depth. At station LM 99MSU1, pore-water sodium concentrations increased from 102 mg L−1 (measured near the top of the gravity core) to a maximum of 2370 mg L−1 at a mean sediment depth of 1.65 m; there were no anion data (chloride, bromide) available from this gravity core for comparison and source evaluation. However, the pore-water profile at station S1-29 (Callender, 1969), including a maximum dissolved sodium concentration of 250 mg L−1 at 50–55 cm sediment depth, is comparable to that from station LM 99MSU1, and both compare favorably with the sodium pore-water profile from Saginaw Bay core SB4. The combined pore-water chemistry results from box cores (chloride and bromide) and from gravity cores (major cations) indicate that, similar to what Kolak et al. (1999) reported for Saginaw Bay, Lake Huron, there exists a source for salinity at depth that is affecting pore-water chemistry, especially chloride and sodium, in southern Lake Michigan benthic sediments.
DISCUSSION
Geochemical Evolution of Pore-Water Composition
Pore-water chemistry profiles from Lake Michigan sediments recorded increases in dissolved solids levels, including chloride and sodium, with increasing sediment depth. An evaluation of the collective changes, or evolution, in pore-water composition can provide insight into possible processes and solute source(s) affecting benthic sediments. For example, Cl:Br ratios measured in water samples can be used to distinguish among different natural and anthropogenic chloride sources, including salt (halite) and formation brine (evaporated seawater) (Panno et al., 2006). An important consideration here is that, whereas dissolved chloride generally is regarded as conservative in benthic sediments (Lerman and Weiler, 1970), the variability observed in pore-water bromide profiles (Fig. 2B) might indicate bromide release from shallow sediments to pore water during early diagenesis. Such release, especially in the uppermost (<10 cm depth) sediments, could obscure the application of Cl:Br ratios to discern prospective chloride sources. In contrast, dissolved bromide concentrations in deeper pore-water samples (>20 cm depth) are less variable, indicating that Cl:Br ratios measured in deeper pore-water samples might retain useful information regarding the identity of prospective chloride source(s). When evaluated on a graph (not shown) of chloride versus Cl:Br ratio (see Panno et al., 2006, their fig. 3), the majority of Lake Michigan pore-water samples, including all samples from station LM 93-70 M, plot within the “precipitation” or “pristine aquifer” fields. Only the deeper pore-water samples plot near fields denoting a possible seawater/salt/brine source, but the deeper samples still are sufficiently dilute that this graphical approach alone does not unequivocally identify a chloride source.
A comparison (Figs. 4A and 4B) among chloride concentrations and Cl:Br ratios from another location—Saginaw Bay, Lake Huron—yields more definitive information about prospective chloride sources affecting Lake Michigan sediments. In Saginaw Bay pore-water samples with higher dissolved chloride concentrations, corresponding Cl:Br ratios generally ranged in value from 200 to 600, similar to values reported for brines from the Saginaw Lowland area (Badalamenti, 1992, their fig. 4.19) and for shallow groundwater in adjacent counties (Long et al., 1986; Dannemiller and Baltusis, 1990). These similarities indicate that Saginaw Bay pore-water chemistry is influenced by saline groundwater (Kolak et al., 1999), with solutes derived from a mixed formation brine consisting of evaporated seawater and dissolved halite (Long et al., 1986; Badalamenti, 1992; Ging et al., 1996).
When compared with a subset of Saginaw Bay cores (SB1, SB4, SB4b, and SB21), pore-water chloride profiles (Fig. 4A) from southern Lake Michigan box cores generally are comparable with Saginaw Bay profiles, indicating that both study areas are communicating with a salinity source at depth. However, Cl:Br ratios in Lake Michigan pore water (Fig. 4B) generally are much lower than those measured in Saginaw Bay and in Saginaw Lowland area formation brine. This disparity does not necessarily preclude a formation brine source to Lake Michigan pore water: several Michigan Basin geological formations—including the Devonian Traverse Group, Dundee Limestone, and Detroit River Group (with the Richfield Member and Sylvania Sandstone) (Wilson, 1989; Wilson and Long, 1993a, 1993b)—contain brines with Cl:Br ratios similar to those measured in pore water from southern Lake Michigan sediments.
There also exist saline groundwater and brine within relatively impermeable units, e.g., the Devonian Antrim Shale (McIntosh and Walter, 2005), which could be a solute source for Lake Michigan benthic sediments. McIntosh and Walter (2005) reported that halite dissolution, which likely occurred during glacial recharge along basin margins, contributed to the salinity measured in groundwater samples from the Antrim Shale (Northern and Western Producing Trends) in the northern Lower Peninsula of Michigan. Although groundwater samples from the Western Producing Trend (Fig. 4C) contain a relatively small proportion of salinity derived from halite dissolution, i.e., ~20%–40% (McIntosh and Walter, 2005, their table 2), Cl:Br ratios in Western Producing Trend samples are greater than those measured in Lake Michigan pore-water samples. We interpret the results of these comparisons to indicate that: (1) Lake Michigan box cores are sufficiently deep as to record the geochemical signature of an underlying solute source; and (2) southern Lake Michigan sediments are affected by a high-chloride source at depth, with chloride (and other dissolved solids) possibly derived from a formation brine, but one that appears largely derived from evaporated seawater, with minor contribution from halite dissolution.
Major cation pore-water data, including data from this study and Callender (1969), also provide information about prospective sources and processes affecting pore-water chemistry evolution in Lake Michigan benthic sediments. One process that could affect pore-water composition is cation exchange between pore water and mineral surfaces, especially clay minerals. If cation exchange is a dominant process affecting sodium and calcium pore-water concentrations, then dissolved sodium and calcium concentrations should plot along a 2Na:1Ca molar ratio trend (Long et al., 1986, their fig. 5.51). Pore-water samples from the 1999 and the Callender (1969) sediment cores do not plot along such a trend (dashed line, Fig. S1), indicating that cation exchange is not a dominant process affecting dissolved sodium and calcium concentrations. When major cation concentrations—expressed on a milli-equivalent (meq) basis—are plotted on a modified ternary diagram, shallower pore-water samples from the 1999 Lake Michigan cores (orange-filled circles, Fig. 5) generally plot within the Ca-dominant field. The majority of pore-water samples from the Callender (1969) study (blue-filled squares, Fig. 5) also plot within the Ca-dominant field, likely reflecting the fact that many samples analyzed were from relatively shallow depths (<50 cm), and that pore-water chemistry at these shallow depths is influenced strongly by mineral-water interactions (see Results). In deeper sediments (with increasing dissolved solids levels), pore-water samples from both Lake Michigan data sets plot along a trend from no dominant cation into the (Na + K)–dominant field. This evolution of major cation composition in Lake Michigan pore water follows a similar trend to that observed for Saginaw Bay (white-filled squares, Fig. 5), where inputs of Na-dominant, saline groundwater from the Saginaw Lowland area are affecting Saginaw Bay pore-water chemistry.
The Lake Michigan pore-water chemistry trend is comparable to the Michigan Basin groundwater mixing trend reported in Ma et al. (2005, their fig. 8). The Ma et al. (2005) mixing trend (Fig. 5, arrow labeled “trend 1”) depicts the general evolution of Michigan Basin groundwater composition from shallow glacial drift to the Mississippian Marshall sandstone to an intermediate, (Na + K)–dominant end member. This intermediate end member generally is derived from mixing with deeper brines from Devonian- to Silurian-aged geological formations (Fig. 5, arrow labeled “trend 2”), possibly due to regional cross-formational fluid flow within the Michigan Basin and/or due to glacial recharge–induced fluid mixing (McIntosh and Walter, 2005; McIntosh et al., 2011). Overlain on “trend 2” (Ma et al., 2005) are data from Wilson (1989, their appendix C) for several Michigan Basin brine samples (Fig. 5, green- and yellow-filled shapes). Most brine samples from the Traverse Group and Dundee Limestone plot within the (Na + K)–dominant field, whereas samples from the Detroit River Group (including the Richfield Member and Sylvania Sandstone) plot within the Ca-dominant field.
If the data reported here indicate that pore-water composition largely is controlled by simple mixing and solute transport from a source at depth, then the similarity among (Na + K)–dominant compositions indicates that Traverse Group and Dundee Limestone brines are plausible solute sources. Given that the Antrim Shale might subcrop several sample stations (Fig. 1), the possibility also exists that a mixed brine, e.g., evaporated seawater with minor halite dissolution, or saline groundwater emplaced in the Antrim Shale could be a solute source affecting benthic sediment pore-water chemistry.
Geological Factors Affecting Pore-Water Chemistry
The variability among pore-water profiles (Figs. 2 and 3) raises questions regarding the extent to which geological factors, including sediment type and thickness, bedrock geology, or presence/absence of fractures, might affect the extent of mixing (“communication”) with a salinity source. We use here the proposed names from Lineback et al. (1970) for descriptive and reference purposes, and we acknowledge that subsequent studies (Colman et al., 1994; Hansel and Johnson, 1996) have called into question the viability of these names as formal lithostratigraphic units. At station LM 99MSU1, the abrupt transition in dissolved sodium concentrations from ~500 mg L−1 to 2000 mg L−1 (Fig. S2) occurs within a gray-brown silt-clay that likely is the Winnetka Member. This abrupt transition in pore-water chemistry is not constrained to a particular sediment type or boundary, indicating that solute transport is not stratigraphically controlled. Below the Winnetka Member, both gravity cores from stations LM 99MSU1 and LM 99MSU2 recovered a red-brown silt-clay at ~1–1.5 m depth, similar to the Sheboygan Member that Lineback et al. (1970) recovered across several southern Lake Michigan sample stations. The gray silt-clay recovered from station LM 99MSU2 between 250 cm and 275 cm depth (Table S2) likely is the Wilmette Bed, a relatively extensive marker bed that Lineback et al. (1970) described within the Sheboygan Member. The core from station LM 99MSU1 might have recovered only the uppermost portion of the Sheboygan Member (above the Wilmette Bed). Alternatively, the possibility exists that this core sampled an erosional feature, as Colman et al. (1994, their fig. 8) noted that an erosional feature within the Sheboygan Member might have affected some of their sediment cores. The 1999 gravity core sediment column descriptions and corresponding pore-water sodium profiles (Fig. S2) indicate that upward transport of dissolved sodium is occurring through the Sheboygan Member.
The Sheboygan Member represents glaciolacustrine rock flour deposited in a proglacial lake (Lineback et al., 1974; Wickham et al., 1978; Colman et al., 1994). Estimates regarding the timing of deposition for this glaciolacustrine deposit vary (Vitorello and Van der Voo, 1977; Wickham et al., 1978; Colman et al., 1994), with Colman et al. (1994) reporting a most likely age estimate for the Wilmette Bed (within the Sheboygan Member) between 11.0 ka and 10.5 ka. The significance of this age range is that these 1999 gravity cores have recovered glaciolacustrine sediments that likely were deposited in a cold, freshwater environment. Therefore, the presence of elevated dissolved solids concentrations in pore-water samples from station LM 99MSU1 demonstrates that solute transport, from a source at depth, occurred during past ~10 k.y.
The timing of solute transport, especially with respect to regional glaciation, is an important consideration that could account for the disparity in pore-water chemistry gradients between stations LM 99MSU1 and LM 99MSU2. Glacial loading, which is known to increase hydraulic head and pore pressure within underlying bedrock units, is capable of inducing recharge several hundreds of meters into bedrock and displacing dense brines (McIntosh et al., 2011; Person et al., 2012). The increased pore pressure also can induce mechanical failures in underlying bedrock, with resulting fractures providing enhanced permeability conduits for recharge (Person et al., 2012; Zhang et al., 2018). Isostatic rebound following glacial retreat likely induced, or reactivated, fractures in underlying bedrock units and glacial till units underneath Lake Michigan (Lineback et al., 1974; Wickham et al., 1978), providing pathways for solute transport between brines, or mixed water masses, and overlying benthic sediments. The relatively limited spatial extent of such pathways could also produce differences in pore-water chemistry gradients, similar to those observed between stations LM 99MSU1 and LM 99MSU2. A similar mechanism is proposed to account for variations in pore-water chemistry gradients measured in Saginaw Bay, Lake Huron, benthic sediments (Kolak et al., 1999, their fig. 3; Hoaglund et al., 2004). The similarity between these two study areas indicates that fracture-controlled transport, through glacial-induced fractures and/or existing regional fracture systems within bedrock, is an important process to consider when evaluating solute transport from Michigan Basin sources into Lake Michigan benthic sediments.
Although pore-water chemistry gradients vary among Lake Michigan sample stations, the fact that dissolved solids concentrations, i.e., chloride and sodium, increase with sediment depth at so many sample stations—including stations from the 1991–1993 and 1999 sampling events and from the Callender (1969) study—might indicate the presence of a regionally extensive source. An assessment of spatial variations in pore-water chemistry gradients for chloride (Fig. 6) and for sodium (Fig. 7) provides more insight into the possible extent of this source. In this assessment, sample stations were categorized according to whether a discernible trend (increase or decrease) existed in dissolved solids concentrations versus increasing sediment depth. The existence of a substantive trend at a sample station was defined here as a change of 20% or more (increase or decrease) in dissolved chloride (or sodium) concentrations between the shallowest and deepest pore-water samples.
The assessment of pore-water chloride gradients (Fig. 6) included data from Callender (1969) and from the 1991–1993 Lake Michigan box cores. Callender (1969) noted that pore-water chloride concentrations measured in sediments from stations S1-29, S1-41, S1-45, and E-5 might have been contaminated from the use of HCl (Callender, 1969, p. 160, see appendix III footnote). The assessment presented here was based on the remaining uncontaminated pore-water samples. However, we note that the chloride measurements Callender (1969) reported for the two deepest pore-water samples from station S1-29, originally identified as contaminated, might be valid. For example, Callender (1969) reported that the pore-water chloride sample analysis (379.4 ppm) from station S1-29 (depth interval 50–55 cm) was contaminated, but the corresponding major cation analysis, with a dissolved sodium concentration of 250.0 ppm, compares favorably with the chloride analysis on a charge-balance basis.
Pore-water chloride concentrations increased with sediment depth at many southern Lake Michigan sample stations (Fig. 6). One sample station, S1-1, is located relatively close to the eastern Lake Michigan shoreline, and given this proximity, the possibility that the observed chloride gradient at this station might result from anthropogenic activities cannot be discounted. Several sample stations are located in open waters, away from the shoreline, and likely are not influenced by anthropogenic activities such as road salting, or industrial or municipal discharge, or from groundwater pumping. The increases in pore-water chloride concentrations at these offshore stations indicate the presence of a spatially extensive, high-chloride source at depth.
At many sample stations in northern Lake Michigan, pore-water chloride concentrations decreased with sediment depth. Torrey (1976, fig. 95), in an earlier analysis of the Callender (1969) study, reported a similar finding. Torrey (1976) mentioned that, at sample stations located north of a line approximately between Sturgeon Bay and Manistee, pore-water chloride concentrations generally decreased with sediment depth. Torrey (1976, p. 240) attributed this general tendency to either “geologic phenomena” or to anthropogenic activities such as industrial brine discharge, which would result in elevated chloride concentrations in the uppermost sediments.
Several northern Lake Michigan sample stations, including MI-10 and M1-38, did not follow this general trend. However, these two stations are located in the nearshore environment, and we postulate that increases in chloride concentrations with sediment depth at these stations might result from shoreline groundwater discharge to Lake Michigan. Cherkauer et al. (1992) documented the effect of shoreline groundwater discharge on nutrient and chloride transport to the northern Lake Michigan nearshore environment. There also were a few offshore sample stations, including S2-9, S2-12, and S3-4, at which pore-water chloride concentrations increased with sediment depth, but to a lesser degree than that observed in southern Lake Michigan.
A regional assessment of pore-water sodium gradients (Fig. 7), using data from Callender (1969) and from the 1999 Lake Michigan gravity cores, is useful to corroborate interpretations based on pore-water chloride gradients. Pore-water sodium gradients for the sample stations (Fig. 7) were categorized using the same approach as for dissolved chloride. Pore-water sodium concentrations increased with depth at nearly all sample stations in southern Lake Michigan, indicating the presence of a solute source at depth that is similar, in terms of spatial extent, to that observed for pore-water chloride profiles. Pore-water sodium concentrations increased with depth at a few northern Lake Michigan samples stations, most of which were located in nearshore environments, e.g., adjacent to the Door Peninsula (Cherkauer et al., 1992), where lake sediments might be affected by shoreline groundwater discharge and solute transport from anthropogenic sources to the lake. There were several offshore stations in northern Lake Michigan at which there was either no discernible gradient or at which sodium concentrations decreased with depth. Torrey (1976) did not conduct a corresponding analysis of pore-water sodium concentrations, but the Sturgeon Bay–Manistee “dividing line” that Torrey (1976) proposed for chloride also delineates the spatial trends in pore-water sodium concentrations observed between northern and southern Lake Michigan. The two regional assessments of pore-water chloride and sodium gradients in benthic sediments are sufficiently similar that both elements likely are responding to similar source(s). From these regional assessments, we infer that a Na-Cl–rich, saline-groundwater source at depth is affecting pore-water chemistry of southern Lake Michigan benthic sediments.
To evaluate possible geological source(s) for salinity, a generalized bedrock geology illustration (Fig. 1)—compiled from multiple sources, including Cvancara and Melik (1961), Douglas (1969), Wold et al. (1981), Vugrinovich (1986), Dannemiller and Baltusis (1990), Wahrer et al. (1996), and Westjohn and Weaver (1996b) (see also Table S5)—depicts approximate locations and estimated boundaries of Michigan Basin bedrock geological units underneath Lake Michigan. This compilation (Fig. 1) reveals substantial differences between the bedrock geology underlying Lake Michigan and Saginaw Bay, Lake Huron, likely indicating that saline fluids from different geological formations are affecting these two study areas. The Michigan Formation, a regional confining unit, subcrops much of Saginaw Bay; the Saginaw and Parma-Bayport aquifer units and the Marshall aquifer unit appear to subcrop the inner and outer portions, respectively, of Saginaw Bay (Hoaglund et al., 2004). Except for the Mississippian Marshall aquifer unit, the other lithologies do not underlie the Lake Michigan study area (see Harrell et al., 1991, their fig. 4; Westjohn and Weaver, 1996a, their figs. A1 and A2). The Marshall aquifer unit appears to subcrop a portion of Lake Michigan (Fig. 1) and potentially could affect sediments near sample station LM 99MSU4 (Fig. 1, starred location labeled “4”), but given the limited spatial extent, the Marshall aquifer unit likely is not the solute source affecting pore-water composition across southern Lake Michigan. Furthermore, the Marshall aquifer unit, although known to contain brine at the basin center, generally has low total dissolved solids concentrations, ranging from <500 mg L−1 to 1000 mg L−1 (Ging et al., 1996, their fig. 6), in the area bordering eastern Lake Michigan. These total dissolved solids concentrations are lower than the values we documented here for partial analyses (major cations only) of pore-water samples in southern Lake Michigan sediments. These factors indicate that another geological formation, different from Saginaw Bay, is a more plausible solute source to Lake Michigan benthic sediments.
Geological formations ranging in age from Early Mississippian to Middle Silurian underlie a substantial portion of Lake Michigan (Fig. 1; Alling and Briggs, 1961; Ells, 1971; Wold, 1980; Wold et al., 1981); several of these formations contain brine and evaporites (halite) (Vugrinovich, 1986; McIntosh and Walter, 2005). Vugrinovich (1986) reported that several Devonian- and Silurian-aged aquifer units have, or likely had prior to the onset of substantive hydrocarbon production, significant overpressures that could induce upward, cross-formational flow within the Michigan Basin. Although Late to Middle Silurian formations underlie portions of both northern and southern Lake Michigan, brine compositions from these formations (Wilson and Long, 1993b) generally are Ca-dominant and are not a good match with the pore-water chemistry reported here (Fig. 5) and in deeper pore-water samples from the Callender (1969) study. Silurian-aged evaporite deposits, e.g., the Salina Group, merit consideration; however, Salina Group evaporites generally are absent (or depleted) from bedrock underlying northern Lake Michigan (Wold, 1980; Wold et al., 1981; Aminian et al., 1987; Sonnenfeld and Al-Aasm, 1991; Black, 2012). Portions of these evaporite deposits dissolved during the Devonian, resulting in collapse features (Wold, 1980; Black, 1983, 2003) and culminating with the formation of the Mackinac Breccia (Landes, 1945; Wold, 1980; Aminian et al., 1987; Black, 2003, 2012). If remnants of Salina Group evaporites persist underneath northern Lake Michigan—perhaps as recycled evaporites present in younger (Devonian) formations (Sonnenfeld and Al-Aasm, 1991)—then these remnants could affect pore-water chemistry on a localized scale. This scenario could explain why dissolved solids concentrations increase with depth at only a few offshore northern Lake Michigan locations, but additional high-resolution bedrock geology characterization and sediment sampling are needed to evaluate the viability of such a connection. Salina Group evaporites generally are not a prospective solute source for southern Lake Michigan sediments; these deposits pinch out underneath Lake Michigan and are not present underneath the lake in areas located approximately south of Sheboygan (Wold, 1980). Furthermore, Cl:Br ratios in southern Lake Michigan pore-water samples are not consistent with a Salina evaporite (halite) source. These factors likely indicate that, aside from possible localized contributions, the distribution of Silurian formations underneath Lake Michigan does not provide a geological basis from which to explain the larger-scale (regional) variations in pore-water chemistry gradients between northern and southern Lake Michigan.
Late and Middle Devonian formations also are prospective geological sources for solutes to Lake Michigan benthic sediments. The Cl:Br ratios measured in shallow box cores (this study) are consistent with a formation brine (evaporated seawater) source and generally are similar to the range of Cl:Br ratios measured in brines from the Traverse Group, Dundee Limestone, and Detroit River Group (including the Richfield Member and Sylvania Sandstone; Wilson and Long, 1993a). However, the major cation composition of Detroit River Group brines (Wilson, 1989; Wilson and Long, 1993a) generally ranges from no dominant cation to Ca-dominant compositions (Fig. 5) and thus is not a good match for Lake Michigan pore-water chemistry. The major cation compositions from Traverse Group and Dundee Limestone brines generally are Na-dominant and are similar to the pore-water composition reported here for deeper benthic sediments from southern Lake Michigan. From this analysis, we infer that brine(s) from Devonian-aged formations, e.g., the Traverse Group and Dundee Limestone, which appear to subcrop a significant portion of southern Lake Michigan, are a plausible geological source for elevated dissolved solids concentrations measured in southern Lake Michigan benthic sediments. If these formations are key solute sources to Lake Michigan benthic sediments, then the disparity between northern and southern Lake Michigan, in terms of the relative proportions of these lithologies underneath the lake bottom, provides a geological basis (Torrey, 1976) from which to explain the regional differences (Figs. 6 and 7) in pore-water composition and gradients. Furthermore, depending on the extent of these formations under Lake Michigan (Ells, 1971; Vugrinovich, 1986, their fig. 10a), the possibility exists that pore waters with elevated dissolved solids levels might also be found in deeper sediments located offshore from Grand Traverse Bay and Little Traverse Bay.
Although brines from Traverse Group and Dundee Limestone formations are plausible geological solute sources, the specific end member affecting benthic sediment pore-water chemistry might be a remnant water mass from glacially induced recharge that migrated from these formations during (or after) mixing, and currently is largely confined within an impermeable unit, e.g., Devonian Ellsworth and Antrim Shales (McIntosh and Walter, 2005, 2006) and Early Mississippian Coldwater Shale. Given that the Antrim Shale might underlie several sample stations in southern Lake Michigan (Fig. 1), this recharge process and the possibility for mixed solute sources are important considerations for evaluating pore-water chemistry evolution in benthic sediments. Whether the salinity source affecting southern Lake Michigan contains a component derived from glacial (Pleistocene) meltwater, as observed in Saginaw Bay, Lake Huron (Kolak et al., 1999; Hoaglund et al., 2004), and in Lake Ontario (Drimmie and Frape, 1996), currently is unknown. The Devonian and Silurian formations underneath Lake Michigan likely received glacial meltwater recharge during ice loading and retreat, similar to other locations along Michigan Basin margins (McIntosh and Walter, 2005; McIntosh et al., 2011; Person et al., 2012). Resolving the possible existence of a glacial meltwater component, e.g., through stable isotope analyses (δ18O and δD in water), could assist in determining whether a saline groundwater source is largely a remnant (“stagnant”) water mass from an ice loading–induced mixing event, or whether there is direct communication and solute transport from a formation brine at depth.
Depending on the rate of solute transport relative to sedimentation rates, there exists the potential that Lake Michigan benthic sediments could act as sources, rather than sinks, for dissolved solids to the overlying water column. Earlier studies of pore-water chemistry gradients in western Lake Ontario (Drimmie et al., 1992; Drimmie and Frape, 1996; King, 1997) and in Saginaw Bay (Kolak et al., 1999) reported that solute transport generally is diffusion-dominated; however, at some Lake Erie and Lake Ontario sample stations, an advective transport component is present (Drimmie and Frape, 1996). We did not attempt here to evaluate the potential for solute exchange across the Lake Michigan sediment-water interface because, unlike in Saginaw Bay, a contemporaneous set of water column, box core, and gravity core samples is lacking. A good understanding of solute transport—including relative contributions from advective (if present) and diffusive components—is needed to evaluate the potential for solute exchange between benthic sediments and the overlying water column.
Michigan Basin Solute Fluxes to Great Lakes Sediments—Other Possible Locations
The possibility exists that benthic sediments from other locations within the Great Lakes drainage basin might be affected by dissolved solids sourced from underlying Michigan Basin geological formations. The geochemical signatures reported here were observed in settings adjacent to regional lowland areas (Michigan Lowland area and Saginaw Lowland area; Figs. 6 and 7), supporting the conceptual model of a topographically driven, regionally focused groundwater flow system with the hydraulic potential for upward flow from deeper units within the Michigan Basin (Cartwright et al., 1979; Vugrinovich, 1986, 1988). Vugrinovich (1986, p. 8) reported that within the Devonian D3 aquifer unit, “[f]low is directed to topographic lows and outcrops beneath Lakes Huron, Michigan, and Erie.” Portions of deeper geological units within the Michigan Basin, e.g., the Ordovician St. Peter Sandstone (Bahr et al., 1994), also are overpressured, with the highest hydraulic heads found in topographically low areas; these hydraulic heads might be a relict of glacial loading and an indication that the Michigan Basin currently is not in equilibrium with, and is still responding to, the effects of glacial loading. Based on this conceptual model, one might expect to find benthic sediments in southern Lake Huron and western Lake Erie with elevated dissolved solids levels at depth, given the proximity of these settings to the Saginaw Lowland area and the Erie Lowland area (Figs. 6 and 7).
In southern Lake Huron, pore-water compositions reported (Robbins, 1980) from four sample stations in depositional basins might indicate solutes derived from a source at depth. Robbins (1980) reported that both dissolved calcium and magnesium concentrations increased linearly with increasing sediment depth, reaching a maximum calcium concentration of ~50 mg L−1 at ~65 cm depth (Robbins, 1980, their fig. 145) and a maximum magnesium concentration of ~15 mg L−1 at ~65 cm depth (Robbins, 1980, their fig. 147). Robbins (1980) attributed these pore-water chemistry gradients largely to slow dissolution of carbonate minerals in the uppermost sediments. However, these linear gradients in dissolved calcium and magnesium concentrations are comparable to those measured in several 1999 Lake Michigan cores, raising the possibility that the dissolved calcium and magnesium profiles in southern Lake Huron sediments might derive from a solute source at depth rather than mineral dissolution. In contrast, Robbins (1980) reported that dissolved sodium concentrations generally did not increase with sediment depth, with a maximum pore-water sodium concentration of ~4 mg L−1 (Robbins, 1980, their fig. 152), comparable to that measured near the sediment-water interface. Several Michigan Basin geological formations are known to contain Ca-rich brines (Frape et al., 1989; Wilson and Long, 1993b; Ma et al., 2005) and/or evaporite minerals, i.e., gypsum and anhydrite (Cvancara and Melik, 1961), but the extent to which these formations underlie southern Lake Huron (Fig. 1; Cvancara and Melik, 1961; Vugrinovich, 1986) or whether there exists sufficient hydraulic drive for cross-formational flow in this setting (Vugrinovich, 1988; Bahr et al., 1994) are not well known.
Similar to southern Lake Huron, there exists the potential for upward flow and/or solute transport within the Erie Lowland area and into western Lake Erie. Several Devonian and Silurian aquifers are overpressured, with flow patterns in the direction of Lake Erie (Vugrinovich, 1986). Rheaume (1991) reported that wells drilled near Lake Erie into the Devonian and Silurian bedrock flowed at land surface. Multiple studies documented increases in dissolved solids levels (sodium or chloride) with sediment depth at nearshore and offshore stations in the Western Basin of Lake Erie (Nriagu and Dell, 1974; Haack et al., 2005), but a specific connection to a prospective source was not identified. In the adjacent Sandusky Basin, Drimmie and Frape (1996) reported chlorine stable isotope values in pore-water samples similar to those from the underlying Devonian Kettle Point Formation, indicating upward solute transport into Lake Erie benthic sediments.
Understanding the magnitude, source, and rate of solute transport in these geological settings is necessary to quantify estimates of solute exchange among the “groundwater–sediment–lake water system” (Cartwright et al., 1979, p. 68). These estimates are important considerations for modeling biogeochemical processes within benthic sediments, as well as the fate of sediment-associated contaminants. Furthermore, given the relatively long hydraulic residence times of water in the Great Lakes (Chapra et al., 2009), quantification of the rate of solute transport in these settings, in comparison with recent sediment mass accumulation rates, is needed for understanding the extent to which benthic sediments are sinks, or sources, for dissolved solids, i.e., chloride and sodium, to the overlying water column.
CONCLUSIONS
Evaluation of pore-water composition from Lake Michigan sample stations identified several locations at which dissolved solids levels, especially chloride and sodium, increase with increasing sediment depth. Dissolved pore-water chloride levels increased with depth in several shallow sediment box cores collected during 1991–1993, indicating a chloride source at depth; this finding is consistent with an earlier study of Lake Michigan pore-water chemistry. The Cl:Br ratios measured in pore water did not unequivocally identify the chloride source, but the values reported here are comparable to those found in several Michigan Basin formation brines (evaporated seawater). Gravity cores from southern Lake Michigan documented increases in dissolved major cation concentrations, especially sodium, with increasing sediment depth. Pore-water composition near the sediment-water interface generally is Ca-dominant and shifts to Na-dominant with increasing sediment depth. This evolutionary pathway follows a fluid-mixing trend similar to those previously reported for Saginaw Bay (Lake Huron) and elsewhere in the Michigan Basin. These changes in Lake Michigan pore-water chemistry are consistent with a source of saline groundwater underneath the lake; we infer that the most plausible source is Na-Cl–rich groundwater. These findings demonstrate that pore-water salinization in Great Lakes benthic sediments is more extensive than previously recognized.
The sediments recovered in southern Lake Michigan gravity cores are similar to those reported in earlier studies of glaciolacustrine sediments underneath Lake Michigan. The presence of elevated dissolved sodium concentrations in these glaciolacustrine sediments indicates that solute transport from a saline groundwater source at depth has occurred within the past 10 k.y. The variability in pore-water chemistry gradients among gravity cores indicates that fractures might affect solute transport, but the specific timing (onset) and transport mechanism, including whether there presently is an advective flow component, are unresolved.
Based on an evaluation of the underlying bedrock geology in conjunction with regional analyses of pore-water chemistry gradients, we infer that brines from Devonian formations, e.g., the Traverse Group and Dundee Limestone, are plausible sources for the elevated dissolved solids concentrations measured in southern Lake Michigan benthic sediments. Due to extensive recharge and fluid mixing during past glaciations, the possibility also exists that relatively impermeable units are the direct source for solutes to benthic lake sediments. The Antrim Shale, which appears to subcrop several Lake Michigan sample stations (Fig. 1), might contain saline fluids derived from underlying formations, e.g., the Traverse Group and Dundee Limestone. The relative differences in bedrock geology underlying northern and southern portions of Lake Michigan provide a geological basis from which to explain regional variations in pore-water composition, and why pore-water dissolved solids levels generally decrease with sediment depth in northern offshore Lake Michigan sample stations. There are other areas in the Great Lakes drainage basin with similar geologic settings, i.e., adjacent to regional low-lying areas, in which solute transport from underlying Michigan Basin sources might affect benthic sediment pore-water composition.
ACKNOWLEDGMENTS
We thank Ted Callender (Great Lakes Research Division, University of Michigan; U.S. Geological Survey [USGS]) and Norm Grannemann and Dave Westjohn (USGS) for their encouragement and for many helpful discussions during the early stages of the Saginaw Bay and Lake Michigan research. We thank Bob Drimmie and Shaun Frape (University of Waterloo) for helpful discussions and for use of the stainless-steel squeezing apparatus to extract pore waters from sediment cores. We thank GSA Bulletin staff, science editor Brad Singer, and associate editor Peter Eichhubl for their assistance with the manuscript submission and review process. We thank associate editor Peter Eichhubl, GSA Bulletin peer reviewer Jennifer McIntosh, and an anonymous peer reviewer for their comments and thoughtful recommendations on the initial manuscript submission. We further acknowledge, and greatly appreciate, the contributions and support to this study from numerous individuals and scientific organizations, detailed below.
The 1991–1993 Lake Michigan sediment cores were collected aboard the R/V Lake Guardian during research activities funded in part through the Great Lakes Protection Fund, U.S. EPA Great Lakes National Program Office (grant numbers R995233-01 and R995233-02, awarded to principal investigators Steven J. Eisenreich, Deborah L. Swackhamer, and D.T. Long). We thank the captains and crews, Adam Heft and Bill Sitarz (Michigan State University [MSU]), and Katie Golden, Matt Simcik, and Charles Wong (University of Minnesota) for their assistance with sample collection. J.J. Kolak extracted the pore-water samples and conducted chloride and bromide analyses; we thank Steve Hamilton (MSU) for use of the laboratory facilities at the MSU W.K. Kellogg Biological Station.
The 1994 Saginaw Bay sediment cores were collected aboard the R/V Laurentian; we thank the captain and crew, Tina Beals and Jennifer Wilson (MSU), and Julie Stinson and Michelle Struble (Central Michigan University) for their assistance with sample collection. The Saginaw Bay study was supported through funding from the National Science Foundation (grant number EAR-9317244) to principal investigators D.T. Long, Grahame J. Larson, and Duncan F. Sibley (MSU) and Jane M. Matty (Central Michigan University). We thank Norman Andresen (Center for Great Lakes and Aquatic Sciences, University of Michigan) for assistance with use of the gravity-coring device deployed in Saginaw Bay. Rob Ellis (MSU) assisted J.J. Kolak with sediment squeezing and pore-water collection. Terry Councell (USGS) conducted chloride and bromide analyses on Saginaw Bay pore-water samples; J.J. Kolak and staff and students at the Soil and Plant Nutrient Laboratory (Department of Crop and Soil Sciences, MSU) conducted the corresponding major cation analyses.
The 1999 Lake Michigan sediment cores were collected aboard the R/V Lake Guardian via in-kind support of a “vessel of opportunity” proposal that J.J. Kolak and D.T. Long submitted to the U.S. EPA Great Lakes National Program Office (GLNPO). We thank David Rockwell (U.S. EPA-GLNPO) for continued encouragement, especially during earlier attempts (1997–1998) to collect Lake Michigan samples. We thank Norman Andresen (Center for Great Lakes and Aquatic Sciences, University of Michigan) for assistance with procuring the gravity-coring device deployed during the 1999 Lake Michigan sample collection. We thank the R/V Lake Guardian captain Dave Moser and crew, Marc Tuchman (U.S. EPA-GLNPO), and other U.S. EPA personnel for their assistance with sample collection. J.J. Kolak processed the 1999 sediment cores at the MSU Environmental Geochemistry Laboratory, including sediment description and squeezing, pore-water collection, and major cation analyses. Lastly, J.J. Kolak thanks the University of Michigan Library staff, especially at the Buhr Shelving Facility and the Hatcher Graduate Library, for their assistance with locating and accessing historical research studies on Great Lakes water quality and pore-water chemistry.