DSi as a Tracer for Submarine Groundwater Discharge

Introduction

Submarine groundwater discharge (SGD) is an important source of nutrients and metals for the coastal ocean (Slomp and Van Cappellen, 2004; Moore, 2010), affects coastal ecosystems (Lecher and Mackey, 2018) and is a relevant and important water resource for coastal communities (Moosdorf and Oehler, 2017). SGD represents the discharge of terrestrial (fresh), marine (saline), or a mixture of both (brackish) groundwater (or pore water) into the ocean (Burnett et al., 2003; Moore, 2010).

Geochemical tracers have widely been applied to calculate SGD fluxes. This requires detailed knowledge of the tracer concentration coupled to potential sources and sinks, its residence time in the coastal water column and its endmember concentration in groundwater. Radium (Ra) and Radon (²²²Rn) isotopes are commonly used as geochemical tracers for SGD, as both tracers are in general enriched in groundwater by several orders of magnitude compared to surface waters (Swarzenski, 2007; Moore, 2010). The “radium quartet” is advantageous for SGD quantification, as the wide range in half-lives of the four isotopes (²²³Ra = 11.4 days, ²²⁴Ra = 3.66 days, ²²⁶Ra = 1,600 years and ²²⁸Ra = 5.75 years) can be used to trace flow paths and processes over variable time-scales. However, Ra data should be interpreted cautiously when fluid ionic strength changes, which affects the partitioning of Ra between solid and solution phases (Webster et al., 1995; Beck and Cochran, 2013). Furthermore, Ra analyses typically require large sampling volumes (tens to hundreds of liters) (Moore and Reid, 1973) and rapid measurement after sampling, as the short-lived Ra isotopes decay within days to weeks. Radon (²²²Rn = 3.83 days) is an inert noble gas that is not partitioned between solid and solution phases like its counterpart Ra. ²²²Rn measurements are automated (Burnett and Dulaiova, 2003), which facilitates rapid data collection over large spatial areas (Dulaiova et al., 2010) and at stationary time-series. However, ²²²Rn cannot be accurately applied as a tracer in high energy environments (winds, waves), as it is prone to degassing (Fanning et al., 1982).

Dissolved silicon (DSi) is also used as a geochemical tracer for SGD (Street et al., 2008; Garcia-Solsona et al., 2010a; Hernández-Terrones et al., 2011; Hwang et al., 2016; Lubarsky et al., 2018), but has received less attention than ²²²Rn and Ra isotopes. DSi in groundwater is the product of chemical weathering of rocks and sediments, or dissolution of biogenic opal. Consequently, terrestrial and marine groundwater (or pore water) is often enriched in DSi (Rad et al., 2007; Anschutz et al., 2009) and can further be augmented during the passage through the subterranean estuary (STE) (Moore, 2010; Rahman et al., 2019). Consequently, SGD [including pore water exchange (PEX)] usually contains high DSi concentrations, which qualifies DSi as a useful addition to conventionally used tracers such as ²²²Rn or Ra (Kim et al., 2005, 2008; Street et al., 2008; Garcia-Solsona et al., 2010b; Hwang et al., 2016; Tamborski et al., 2018). The advantages of DSi as a tracer for SGD include small sampling volumes, low detection limits and high precision measurements. Samples for DSi are easy to store and transport, which makes DSi logistically desirable as a tracer in remote areas with limited infrastructure, where it may be challenging to sample for ²²²Rn and Ra isotopes. Furthermore, DSi is not prone to degassing, and may thus be a useful tool to compliment ²²²Rn mass balances in high-energy environments. However, DSi does not behave conservatively in many coastal environments, where it can adsorb onto Fe-oxides or reprecipitates as amorphous Al-Si phases (e.g., Mackin and Aller, 1984). One of the main limitations of DSi as a tracer is that it is used by siliceous plankton (e.g., diatoms, flaggelates, radiolarians and picocyanobacteria) as well as higher trophical levels such as sponges, as a nutrient to build up their hard tissues (Brzezinski, 1985; Krause et al., 2011, 2017; Tréguer and De La Rocha, 2013). The aim of this study is to assess under which conditions DSi can be used as a tracer for SGD. We review the current state of the literature and show two case studies in contrasting geological environments to demonstrate the utility of DSi as a tracer for SGD.

DSi in Groundwater and Surface Water

DSi in Groundwater and the Subterranean Estuary

The concentration of DSi in fresh, brackish and marine groundwater is chiefly governed by dissolution rates of minerals in the aquifer and in the sediment (i.e., weathering; Figure 1, f) which is usually dependent on the lithology, the subsurface residence time of groundwater and biogeochemical transformations (Bluth and Kump, 1994; Horton et al., 1999; Jacobson et al., 2003; Rahman et al., 2019). Combined, these factors control DSi endmember concentrations in SGD. Indeed, different SGD flow paths (Santos et al., 2012) may lead to unique DSi endmember concentrations. For example, terrestrial groundwater (Figure 1, a) can have a different DSi endmember concentration compared to tidally driven marine groundwater (Figure 1, b). In either case, differences in DSi concentrations can be attributed to the above mentioned conditions, which are discussed in detail below.

Figure 1. Conceptual figure illustrating different processes governing DSi concentrations in coastal waters and coastal sediments/aquifer. Red arrows represent sources of DSi and black arrows represent sinks of DSi. Orange filled arrows represent hypothetical SGD flow paths, including terrestrial groundwater (a) and marine groundwater (b). Yellow colors represent fresh-brackish waters and blue colors represent saline water. The mixing zone between fresh and saline groundwater is called subterranean estuary (STE). The green line at the seafloor represents biogenic silica which has deposited e.g., after a phytoplankton bloom (Kowalski et al., 2013). The respective processes which lead to an input or output of DSi are labeled in the lower panel. Sources and sinks can vary with respect to the environmental boundary conditions, the size of the arrows is thus not representative for the quantity of their respective fluxes.

In marine groundwaters, DSi concentrations are dominantly controlled by the dissolution of biogenic silica (e.g., DeMaster, 2002). Based on the dissolution of biogenic silica pore water DSi concentrations can reach 1000–1200 μM under typical pressure and temperature conditions found in marine sediments (e.g., Hurd, 1973; Schink et al., 1975; DeMaster, 2003; Loucaides et al., 2012). Alteration processes like incorporation of Al, adsorption of metals (e.g., Al³⁺ and Fe³⁺) onto the silica substrate surface, formation of metal oxide coatings, or aging of the substrate inhibit the dissolution of biogenic opal (Van Cappellen and Qiu, 1997; Dixit and Van Cappellen, 2002; Van Cappellen et al., 2002; Michalopoulos and Aller, 2004; Khalil et al., 2007; Loucaides et al., 2010).

In most terrestrial groundwaters, the net silica enrichment is primarily controlled by lithogenic mineral dissolution (e.g., Ehlert et al., 2016), and thus depends on the lithology of the aquifer under study (Table 1; Rahman et al., 2019). Silicate mineral dissolution rates vary with pH, temperature, salinity, groundwater flow rates and sediment-water volume ratios (Oelkers and Gislason, 2001; Anschutz et al., 2009; Jeandel and Oelkers, 2015; Morin et al., 2015). Si saturation will be controlled by the type of Si-bearing minerals present. Thus, the maximum attainable DSi concentration in groundwater is a function of the surrounding environment. For example, karstic terrestrial groundwater has relatively low DSi concentration (80 ± 63 μM), reflecting the low Si content of carbonates, whereas extrusive igneous groundwater is extremely enriched in DSi (604 ± 192 μM) (Table 1; Rahman et al., 2019).

Table 1. Globally-averaged terrestrial and marine SGD DSi endmember concentrations categorized after lithology, summarized from Rahman et al. (2019).

Further, DSi concentration in groundwater can be modified during passage though the STE. Non-conservative behavior of DSi in groundwater may be assessed from the observed enrichment or deficit of DSi with respect to two-endmember linear mixing between fresh groundwater and surface marine waters. Whereas carbonate aquifers appear to show no net enrichment of DSi in brackish groundwaters, marine SGD flow paths from extrusive igneous and complex lithologies exhibit average net DSi enrichments of ∼50 μM globally (Table 1; Rahman et al., 2019, and references therein). Though there is evidence of non-conservative DSi behavior in aquifers composed of primarily granite (Onodera et al., 2007; Rengarajan and Sarma, 2015; Wang et al., 2015; Lecher et al., 2016), shale (Kim et al., 2005; Lee et al., 2012; Luo et al., 2014; Ye et al., 2016) and sandstone (Weinstein et al., 2011; Sugimoto et al., 2017) lithologies, the data are yet insufficient to constrain net DSi concentrations in marine SGD in these endmember lithologies globally (Table 1). At the land-sea transition, dissolution rates of terrestrially derived silicate-bearing mineral phases or silica substrates (e.g., phytoliths) are on average 4–5 times higher in seawater than in freshwater (Loucaides et al., 2008). These findings are consistent with other laboratory studies of basaltic glass or lithogenic particle dissolution in brackish waters (Daux et al., 1997; Advocat et al., 1998; Techer et al., 2001; Jones et al., 2012; Oelkers et al., 2012; Morin et al., 2015), as well as static and flow-through incubations of sediment at the freshwater-seawater transition zone in the STE (Anschutz et al., 2009; Ehlert et al., 2016; Tamborski et al., 2018). Dissolution rates of sediments of different lithologies, at solid surface to volume ratios close to those found in the STE, range from ∼3.4 to 49 mol Si m^–2 s^–1 (Techer et al., 2001; Anschutz et al., 2009; Ehlert et al., 2016; Tamborski et al., 2018).

The subsurface residence time of groundwater is directly proportional to the flow velocity and flow path length (i.e., transit time) and is thus tied to DSi enrichments via mineral dissolution kinetics. The longer a parcel of fluid is in contact with Si-bearing minerals, the greater DSi that fluid may obtain from water-rock weathering reactions, until a steady-state or transient steady-state is reached between Si dissolution and reprecipitation (Ehlert et al., 2016). In DSi-poor groundwaters, initial dissolution rates are high before Si concentrations stabilize. With longer pore fluid residence times and higher DSi concentrations, dissolution rates decrease (Techer et al., 2001). Coatings or secondary mineral phases form on the substrate, reducing the reactive surface area (Daux et al., 1997; Gislason and Oelkers, 2003). Higher flow velocities can increase dissolution rates as products of the dissolution reaction (i.e., dissolved Si or silicic acid) are transported away from their site of production and the reactions stay far from equilibrium (Anschutz et al., 2009). Due to continuous DSi enrichment in the pore fluid, some portion of the initially released DSi (30 to 60%) will reprecipitate into a secondary mineral phase (e.g., amorphous alumino-silicates), which is dependent on groundwater residence times (Daux et al., 1997; Staudigel et al., 1998). Terrestrial groundwater residence times can vary from days to thousands of years; similarly, marine groundwater residence times can vary from seconds to centuries (Seidel et al., 2015). Knowledge on the time-scale of the flow path under consideration is thus extremely important to evaluate DSi endmember concentrations.

DSi as an Indicator for Groundwater Transport Processes

Dissolved silicon may be used to quantify groundwater residence times when the factors described in section “DSi in Groundwater and the Subterranean Estuary” are considered. Anschutz et al. (2009) investigated the residence time of beach groundwaters based on the kinetics of quartz dissolution in seawater. Intertidal beach sand was incubated with seawater to determine the change in DSi concentration over time. Experiments can be performed under varying conditions, including seawater: sediment ratios, temperature and pH. In addition, incubations can be performed under “static” conditions or with seawater actively circulating with specified flow rates through the sediment. The dissolution rate of silicate minerals can be simply estimated from the linear increase in DSi concentration over time. This methodology has been successfully applied to several sediment types (Anschutz et al., 2009; Charbonnier et al., 2013; Ehlert et al., 2016; Tamborski et al., 2018). Groundwater residence times can be estimated for marine groundwater by assuming that there is no DSi contribution from terrestrial groundwater or from the dissolution of biogenic silica (Ehlert et al., 2016).

Dissolved silicon can also be used to determine if groundwater infiltration or exfiltration occurs within sediments. If the DSi concentration at two distinct depth horizons is known (e.g., 50 and 100 cm), the concentration gradient between both depths can delineate the advective transport of groundwater, assuming that biogeochemical transformations of DSi are negligible between both horizons. The underlying assumption is that in infiltration (recharge) zones, a positive gradient would indicate increasing release of the constituent (DSi) with increasing depth, while in exfiltration (discharge) zones a negative gradient would indicate increasing accumulation of the constituent (DSi) with decreasing depth. With this method, it was possible to visualize in- and exfiltration patterns in the intertidal zone of the beach system on Spiekeroog Island, southern North Sea (Waska et al., 2019).

DSi as a Tracer for SGD in Coastal Waters

If dissolved silicon is used as a tracer for SGD, several processes which affect DSi concentrations have to be considered (Figure 1). Inputs can occur from river and surface runoff (F_runoff) (Figure 1, d), rainfall (F_rain) (Figure 1, e), dissolution of particles in the water column and surface sediments (F_dissolution) (Figure 1, f), diffusion (F_diff) (Figure 1, c), and SGD (F_SGD) (Figure 1, a,b). Output terms include mixing with offshore waters (F_mix) (Figure 1, i), biological uptake (F_biol) (Figure 1, g), for example by benthic and pelagic siliceous plankton, and reverse weathering (F_rev) (Figure 1, h). Assuming steady-state conditions, and that DSi inputs equal DSi outputs, a simple DSi mass balance can be written (Eq. 1).

In coastal environments many of these processes are often time dependent. Over short time-scales (e.g., tidal cycles) DSi can be considered as a conservative tracer for SGD because dissolution of particles, reverse weathering or biological uptake may be considered negligible with respect to the other noted sources and sinks. In this study, we consequently focus on short time-scales (hours to days). Coastal environments influenced by SGD on such short time-scales include smaller embayments, tidal channels, open beaches or rocky coastlines. Assuming that the previously mentioned processes can be neglected under short time-scales, Eq. 1 simplifies to:

Note that Eq. 2 is similar to commonly employed ²²²Rn and Ra mass balances. DSi source and sink terms outlined in Eqs 1 and 2 are discussed in further detail below.

Dissolved silicon inputs from rainfall are usually low when compared to SGD. Concentrations between 0.5 and 15 μM were measured in rainwater in the Yellow Sea and East China Sea (Zhang et al., 2005). In the Mediterranean Sea, concentrations of DSi in rainfall range from below detection limit up to 33 μM, whereas higher concentrations are linked to episodic Saharan dust inputs (Bartoli et al., 2005).

The importance of molecular diffusion as a source of DSi will vary from site to site. Diffusive fluxes (F_diff) can be measured using ex situ or in situ sediment incubation experiments, if bioturbation and bioirrigation can be neglected, for example by inactivation of benthic fauna by asphyxiation (Rutgers Van Der Loeff et al., 1984; Forster and Graf, 1995). Diffusion can also be calculated based on pore water profiles of DSi from Fick’s first law (Schink et al., 1974; Oehler et al., 2015; Tamborski et al., 2018).

River and surface runoff (F_runoff) can contain significant amounts of DSi as well as amorphous silica and solid phases which can rapidly dissolve upon estuarine mixing (Conley, 1997). Therefore, the amount of DSi transported into the coastal water column by rivers has to be carefully determined, e.g., by measuring riverine discharge and DSi concentrations, which is also required for other tracers such as ²²²Rn or Ra. At the global (Dürr et al., 2011; Tréguer and De La Rocha, 2013), regional (Hartmann et al., 2010; Moosdorf et al., 2011), and local (Rad et al., 2007; Schopka and Derry, 2012) scale, DSi fluxes from rivers are well known and measuring methodologies are established. Obviously, close to a river, these fluxes can dominate coastal DSi fluxes. Globally, riverine DSi fluxes amount to 70 t Si per km of coastline, based on riverine Si fluxes of 5.8 Tmol Si/a (Tréguer and De La Rocha, 2013) and a global coastline of 2.3 mio km (Moosdorf et al., 2015). Offshore mixing (F_mix) of a tracer can be estimated by different approaches. In an embayment or an estuary where mixing is dominated by tidal forces, offshore mixing can be calculated based on a tidal prism approach (Dyer, 1973). Offshore mixing rates can also directly be measured, for example, by using an Acoustic Doppler Current Profiler (ADCP). Numerical models can be developed to estimate mixing rates and residence times. Alternatively, short-lived Ra and Rn isotopes may be used to estimate mixing with offshore waters (e.g., Burnett and Dulaiova, 2003; Moore et al., 2006). When short time-scale processes are considered (Eq. 2), the mixing loss of DSi must be well-constrained in order to properly balance DSi sources.

If the previously mentioned DSi fluxes (F_rain, F_runoff, F_mix, F_diff) into the coastal water column can be quantified or neglected, then the SGD flux can be calculated by dividing the DSi flux from SGD (F_SGD) with the DSi concentration in groundwater. Selection of the groundwater endmember should be based on the considerations outlined in section “DSi in Groundwater and the Subterranean Estuary.”

Case Studies

Case Study 1: DSi as a Tracer for SGD in a Tropical Volcanic-Carbonate Karstic Region (Southern Java, Indonesia)

The coastline of the tropical karstic region of Gunung Sewu (southern Java) is made of strongly karstified massive coral reef-limestone with intercalated clay and volcanic ash lenses (van Bemmelen, 1949; Flathe and Pfeiffer, 1965; Waltham et al., 1983; Haryono and Day, 2004). Toward the hinterland mountain ranges occur which consist mainly of sediments and volcanic deposits (a detailed geological description and geological map can be found in Oehler et al., 2018). Groundwater flows from several kilometers in the hinterland toward the coast and is enriched in DSi most likely due to the weathering of volcanic material in the hinterland, and the weathering of volcanic ash lenses during transport toward the coast. DSi enriched groundwater thus discharges through intertidal and submarine springs into the coastal ocean (Oehler et al., 2018). The regional lithology is a combination of volcanic and carbonate rocks in the hinterland and dominantly carbonate at the coast, with groundwater DSi concentrations between 200 and 400 μM with a salinity ranging from 0 to 11 (Figure 2A, triangles). Surface water samples taken within an embayment which receives groundwater from a single submarine spring and is not affected by surface runoff or river discharge, showed a linear inverse correlation between DSi and salinity (Figure 2A, gray dotted line, R² = 0.84), indicating the input of DSi from terrestrial SGD. An inverse linear correlation was also observed between ²²⁴Ra_ex and salinity (Figure 2B, gray dotted line, R² = 0.60), with more scatter between long-lived ²²⁶Ra and salinity (Figure 2C, gray dotted line, R² = 0.03), also indicating inputs of Ra from SGD. Fresh groundwater was low in ²²⁴Ra with a concentration of 10.3 dpm 100 L^–1, whereas brackish groundwater was enriched in ²²⁴Ra with a concentration of up to 38.7 dpm 100 L^–1 (Figure 2B, triangles). ²²⁶Ra was slightly enriched in brackish groundwater and highly enriched in fresh groundwater compared to seawater (Figure 2C).

Figure 2. Salinity plotted against (A) DSi, (B) ²²⁴Ra_ex and (C) ²²⁶Ra concentrations of water samples taken in an embayment in a karstic region in southern Java. Squares and dots represent surface water column measurements and triangles represent samples taken at coastal and submarine springs. Gray dotted lines show linear correlations of samples taken from the water column, while black dotted lines indicate theoretical conservative mixing between the average groundwater concentration and seawater. Note that only selected spring samples were analyzed for Ra isotopes.

Higher DSi concentrations in the coastal water column (139 μM) were observed during low tide and lower concentrations (65 μM) were observed during high tide (Supplementary Material). The tidal variations can be used in a tidal prism approach to calculate water residence times (Eq. 3 in the Supplementary Material) (Moore et al., 2006), which yielded a water residence time of 0.56 days. The surface water residence time value is reasonable, as the embayment is subject to diurnal tidal variations.

Submarine groundwater discharge fluxes based on the different tracer approaches were calculated based on Eq. 5 (see Supplementary Material) and are shown in Table 2. We were not able to calculate a SGD flux based on ²²⁶Ra, due to the high ²²⁶Ra concentrations in seawater (see Supplementary Material). The maximum SGD flux derived from DSi (72 cm day^–1) was based on the brackish groundwater endmember, and represents brackish SGD. Consequently, these results agree well with the average SGD fluxes based on ²²⁴Ra_ex (65 cm day^–1), where the brackish groundwaters were also used as an endmember.

Table 2. SGD calculations in southern Java based on ²²⁴Ra_ex, salinity and DSi mass balances.

However, coastal water DSi concentrations were slightly lower as it would be expected from theoretical conservative mixing between fresh groundwater and seawater (Figure 2A, black dotted line), which indicates either an unknown source of freshwater in the coastal water column which dilutes the DSi signal, or an unknown DSi sink. Freshwater inputs into the coastal water column from rainfall can be neglected, as there was no rainfall within the sampling period. River and surface runoff is minor and can be neglected as a source of DSi as well, which is typical for karstic regions. DSi concentrations in karstic groundwaters can vary over short time scales and the full range of the groundwater DSi endmember may have not been accurately captured in this study, where samples were taken within days. Coastal water DSi concentrations may have also been partially reduced due to an uptake by marine biota (e.g., diatoms), or reverse weathering, which is however, unlikely due to the short coastal water residence times.

Case Study 2: DSi as a Tracer for SGD in a Temperate Coastal Region Protected by Barrier Islands (Spiekeroog, Southern North Sea)

Diatoms are one of the major algae groups present during spring phytoplankton blooms in the North Sea and its surrounding tidal flat system (Reid et al., 1990; Schoemann et al., 1998; Meier et al., 2015; Wiltshire et al., 2015). Diatom growth and the associated DSi uptake control seasonal DSi dynamics in the water column. For example, in the Spiekeroog tidal basin DSi concentrations are high during winter (20–30 μM) and decrease strongly in the water column during the spring diatom bloom reaching concentrations close to zero (Figure 3; Grunwald et al., 2010; Beck and Brumsack, 2012). After the bloom, diatoms remain one of the dominant algal groups during the entire growing season (Meier et al., 2015). Furthermore, dinoflagellates, which are adapted to low nutrient concentrations and can have a siliceous skeleton, become important in summer (Meier, 2014; Wiltshire et al., 2015). Both phytoplankton species lead to a continuous DSi consumption until late autumn. The DSi pool in the water column is replenished by discharge of DSi-rich marine groundwater from the surrounding tidal flat sediments where the deposited diatom detritus is degraded (Billerbeck et al., 2006; Beck et al., 2008; Kowalski et al., 2012). Pore water advection in surface sediments (Huettel et al., 1998) leads to the release of DSi-rich pore waters, especially after the spring bloom (DSi up to 400 μM Kowalski et al., 2012). Additionally, SGD from tidal flat margin sediments (Riedel et al., 2010; Moore et al., 2011) lead to the release of groundwater enriched in DSi throughout the whole year (DSi up to 1000 μM Reckhardt et al., 2015). The latter process occurs only around low tide when the tidal flat margins are exposed and is reflected in tidal DSi dynamics in the water column, with highest DSi concentrations around low tide (Figure 4; Grunwald et al., 2010). Similar to DSi, SGD transports ²²⁴Ra_ex and ²²²Rn to the open water column resulting in tidal variations as well (Figure 4; Moore et al., 2011; Santos et al., 2015).

Figure 3. Seasonal DSi dynamics in the water column of the Spiekeroog tidal basin. The solid line indicates the moving average of the time-dependent median including data from 2007 to 2017 and reflects typical seasonal DSi variations occurring every year. The dashed lines show the 10 and 90% quantiles.

Figure 4. Tidal DSi and ²²⁴Ra_ex dynamics in the water column of the Spiekeroog tidal basin, with highest concentrations/activities around low tide when submarine discharge of DSi/Ra-rich groundwater is highest. The vertical solid and dashed lines indicate low tide and high tide, respectively. The water level changes smoothly depending on the tidal state. In contrast, tidal variations of DSi and ²²⁴Ra_ex show a less smooth pattern because the tidal flat area consists of several tidal flats, which slightly differ in SGD fluxes and DSi endmember concentrations.

The tidal variations in the water column qualify DSi as tracer for SGD because additional DSi is required to obtain the enrichments measured during low tide. Given the low surface freshwater runoff (i.e., by a flood-gate) (Grunwald et al., 2010), the primary DSi source is SGD. Consequently, mass balance models can be developed to estimate SGD fluxes (Eq. 5). For the SGD estimation, a DSi endmember concentration in marine groundwater of 500 μM (Reckhardt et al., 2015) and a DSi concentration difference between tidal basin and nearshore waters of 8 μM (Figure 4) was assumed. Mean water volume (105 × 10⁶ m³) and flushing time (4 days) of the tidal basin was assessed by Stanev et al. (2003) and Moore et al. (2011), respectively. Based on DSi, the SGD flux to the Spiekeroog tidal basin amounts to about 2 × 10⁸ L tidal-cycle^–1, which is in the same range of 2–4 × 10⁸ L tidal-cycle^–1 as calculated by Moore et al. (2011) using Ra isotopes. This SGD flux is equivalent to 30 cm day^–1, assuming a tidal flat margin length of 59 km for the entire Spiekeroog tidal basin (Moore et al., 2011) and a seepage zone of 25 m (Røy et al., 2008).

We propose that at sites highly controlled by seasonal diatom growth like the coastal region of the southern North Sea, the use of DSi is restricted to certain times of the year. With the onset of the spring diatom bloom until late summer, intensive biological DSi consumption and subsequent dissolution of diatom detritus alter the water column composition (e.g., Figure 3). Therefore, we suggest to restrict the use of DSi as tracer for SGD to time spans with low biological activity. ²²⁴Ra_ex may in turn be used as a tracer for SGD independent from the season (Figure 4).

Discussion

When and Where Is DSi a Useful Tracer for SGD?

Dissolved silicon can be used as a tracer for terrestrial and marine SGD, as any water which is in contact with a sediment or aquifer matrix will be enriched in DSi with time. Some studies have used DSi as a tracer for SGD in regions impacted by a high terrestrial groundwater flow. These regions include volcanic and/or karstified carbonate rocks where a high terrestrial SGD flow occurs along conduits and cracks. For example, DSi was applied as a tracer for SGD in karstic carbonate regions such as Yucatan (Hernández-Terrones et al., 2011) and Castello, Spain (Garcia-Solsona et al., 2010b), in volcanic regions such as Hawaii (Street et al., 2008; Lubarsky et al., 2018) or Jeju island (Hwang et al., 2005) or in a mixture of volcanic and carbonate lithology’s such as in southern Java (this study). Marine SGD is usually depleted in DSi in these regions (e.g., Table 1), and therefore DSi will either reflect brackish or fresh discharging groundwater. If brackish SGD discharges into the ocean, DSi can be used to assess the brackish fraction of SGD, in a similar way as Ra isotopes (Garcia-Solsona et al., 2010b). An SGD flux calculated on the basis of a brackish DSi groundwater endmember may thus represent a brackish SGD flux (e.g., as shown in the case study in southern Java). A fresh SGD flux should theoretically be calculated based on the DSi endmember in fresh groundwater, if conservative mixing between fresh groundwater and seawater occurs.

In tropical regions, large amounts of freshwater inputs from surface runoff and rainfall into the coastal water column further limit the applicability of salinity as a tracer for fresh SGD. DSi can consequently be a useful alternative tracer for fresh SGD. While terrestrial SGD is particularly high on small tropical islands (Zektser and Loaiciga, 1993; Moosdorf et al., 2015), only a few studies have been conducted in these regions (e.g., Matson, 1993; Kamermans et al., 2002; Knee et al., 2016; Oehler et al., 2018; Haßler et al., 2019), because they are often remote. In some instances, DSi concentrations and fluxes are reported (e.g., Johnson et al., 2008), but DSi has only rarely been used as a tracer. In southern Java, for example, the logistically challenging infrastructure (e.g., remote beaches, Ra samples needed to be exported) did not allow us to obtain large datasets for Ra (only three groundwater samples). Using DSi as an additional tracer for SGD in this setting thus allowed us to assess SGD fluxes more accurately. The comparably simple methods for measuring DSi combined with its elevated concentrations in groundwater suggest its usability as tracer for SGD in such remote locations.

In environments dominated by permeable sandy sediments DSi can also be used as tracer for SGD, as marine groundwater (or pore water) is highly enriched in DSi, for example reaching concentrations of up to 1000 μM in the tidal flat area of Spiekeroog (Reckhardt et al., 2015). In these regions DSi shows a similar tidal variability as ²²⁴Ra_ex, indicating a similar source from SGD (Figure 4). SGD fluxes based on DSi are in a similar range as those calculated from ²²⁴Ra, indicating its suitability to quantify flux rates. However, high biological uptake rates of DSi (e.g., from pelagic or benthic siliceous phytoplankton) limit the time when DSi can be used as a tracer for SGD to periods when primary productivity is low, for example due to low light availability (Figure 3).

Dissolved silicon can also be used in combination with other tracers such as ²²⁴Ra_ex and ²²⁶Ra or ²²²Rn to calculate SGD fluxes and water residence times (e.g., Hwang et al., 2005), an approach which was successfully applied in Yeongil Bay (Kim et al., 2008), Geoje Bay (Hwang et al., 2016) and in Bangdu Bay (Hwang et al., 2005) and might be a useful tool in further SGD studies. DSi can also be useful in SGD studies if groundwater transport processes are investigated. DSi concentration gradients in the sediment can be used to identify exfiltration and infiltration patterns (Waska et al., 2019). Furthermore, if only marine SGD is investigated, simple sediment dissolution experiments can be carried out in order to estimate the residence time of marine groundwater in the sediment based on DSi groundwater concentrations (Anschutz et al., 2009).

When and Where Is DSi Not Applicable as a Tracer for SGD?

The use of DSi as a tracer for SGD is complicated if unknown or unquantifiable sources or sinks of DSi exist. Inputs of particulate Si can occur from rivers and atmospheric deposition, which need to be considered as particles can partly dissolve and release DSi into the water column. Reverse weathering, adsorption to Fe-oxides, precipitation of amorphous Al-Si phases and biological uptake can reduce DSi concentrations in surface sediments and in the water column. In SGD studies over larger spatial (e.g., shelf) or temporal scales (weeks to years) (Lee et al., 2009; Tamborski et al., 2018), the quantification of the previously named DSi sources and sinks and likewise its applicability as a tracer for SGD will become difficult.

Biological uptake depends on various factors such as the growing speed and abundance of DSi incorporating organisms, the availability of other nutrients, temperature and light (e.g., Abreu et al., 1994; DeMaster et al., 1996; Ragueneau et al., 2002; Krause et al., 2011). In many coastal regions, algae blooms such as diatoms follow a seasonal pattern (van Beusekom and Diel-Christiansen, 2009; Grunwald et al., 2010). Using DSi as a tracer for SGD may not be appropriate in these regions if biological uptake of DSi from the water column and pore waters in surface sediments is much higher when compared to inputs via SGD. DSi is for example depleted during spring phytoplankton blooms in the southern North Sea, whereby it acts as a limiting nutrient for primary producers. Biological uptake rates of DSi are consequently high during this time and difficult to quantify. During and after the deposition of diatoms at the seafloor, dissolution rates of biogenic opal in surface sediments are usually high which leads to a high efflux of DSi into the water column (Ehrenhauss et al., 2004; Oehler et al., 2015) and high DSi concentrations in groundwater (or pore water) (Reckhardt et al., 2015). During these times, the end-member concentration of DSi in groundwater will be extraordinarily high and has to be carefully determined. In the southern North Sea it will thus be more difficult to apply DSi as a tracer for SGD during and after phytoplankton blooms in comparison to periods when primary productivity is low (Beusekom and Diel-Christiansen, 2007). In many coastal regions the timing of phytoplankton blooms are well known, and thus appropriate times can be selected during which measurements should be carried out. In addition, it can be useful to determine the phytoplankton community composition and abundance if DSi is used as a tracer for SGD. Similar to other geochemical tracers, DSi cannot be used as a tracer of SGD if the DSi endmember is poorly constrained. Total SGD is composed of several different flow paths driven by a variety of different forcing mechanisms (Santos et al., 2012). In volcanic and karstic environments, defining the SGD DSi endmember may be relatively straightforward (e.g., Table 1). But in well-developed carbonate karstic aquifers, very low DSi concentrations may occur in groundwater, especially under high flow conditions. Furthermore, in lithologic complex environments, DSi endmember concentrations can vary between flow paths. For example, SGD driven by water exchange between a lagoon and the sea will have a unique DSi endmember different from that of seawater circulation through the permeable beach-face (Tamborski et al., 2019). Recently recharged groundwaters will have lower DSi concentrations compared to older, deeper groundwaters.

Conclusion

We herein assess DSi as a tracer for SGD and which boundary conditions need to be considered. DSi becomes enriched in groundwater due to biogenic silica dissolution and water-rock interactions, and can reach a transient steady-state equilibrium between dissolution and reprecipitation with time. Therefore, groundwaters and pore waters are usually enriched in DSi when compared to surface waters, which makes DSi a useful tracer for terrestrial and marine SGD. Typical DSi concentrations in terrestrial and marine groundwater can be described depending on lithology of the respective aquifer: extrusive igneous, carbonate, sandstone, granite, shale and complex. Carbonate karst and volcanic regions promote a high terrestrial groundwater flow, where DSi works well as a tracer for terrestrial and brackish SGD. In other regions with complex lithologies (mixtures of sand and clay), such as coastal areas of the North Sea, DSi can be used as a tracer for marine SGD (e.g., forced by tidal pumping), due to the dissolution of lithogenic particles and siliceous ooze in the sediment. One of the largest constraints in using DSi as a tracer for SGD is its non-conservative behavior, especially due to biological uptake by diatoms and other primary producers. Therefore, DSi cannot be used as a tracer for SGD if large amounts of DSi are taken up by algae blooms in the area studied. Determining the phytoplankton community composition and abundance, or a conservative mixing line between salinity and DSi may help researchers to understand if the conditions are suitable to use DSi as a tracer for SGD. Furthermore, DSi can most likely not be used as a tracer for SGD over larger temporal (longer than days) and spatial scales, as it will become difficult to quantify all sources and sinks of DSi over larger scales. DSi can easily be sampled and transported; analyses are in general very cost effective and have a small analytical error. State of the art analytical methods allow precise DSi measurements with small sampling volumes of less than 1 mL. This allows endmember sampling in pore waters and groundwater even on a small vertical scale in the sediment, which has been suggested to be necessary for a representative SGD end-member (Cook et al., 2018). Future SGD studies should consider using DSi as an additional tracer to compliment ²²²Rn or Ra investigations.

Data Availability

All datasets generated for this study are included in the manuscript and/or the Supplementary Files.

Author Contributions

MB, RN, CM, BS, and JA collected data for case study 2 and wrote the section “DSi as a Tracer for SGD in a Temperate Coastal Region Protected by Barrier Islands (Spiekeroog, Southern North Sea).” JT and SR contributed with data and interpretation of sections “DSi as an Indicator for Groundwater Transport Processes” and “DSi as a Tracer for SGD in Coastal Waters” and wrote the section “DSi as an Indicator for Groundwater Transport Processes.” TO, NM, and BS collected data for section “DSi as a Tracer for SGD in a Tropical Volcanic-Carbonate Karstic Region (Southern Java, Indonesia).” All authors contributed with the interpretation, writing and editing of the manuscript. TO compiled everything and wrote largest parts of sections “Introduction,” “DSi as a Tracer for SGD in Coastal Waters,” “DSi as a Tracer for SGD in a Tropical Volcanic-Carbonate Karstic Region (Southern Java, Indonesia),” “Discussion” and “Conclusion.”

Funding

TO, NM, and the presented case study 1 were funded through the BMBF junior research group SGD-NUT (grant #01LN1307A). Open access publication fees are paid by Leibniz-Centre for Tropical Marine Research internal funds. The presented case study 2 was financially supported by the DFG Research Group “BioGeoChemsitry of Tidal Flats”, the Ph.D. Research Training Group “The ecology of molecules” funded by the Ministry for Science and Culture of Lower Saxony, and the Institute for Chemistry and Biology of the Marine Environment, University of Oldenburg.

Conflict of Interest Statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

We thank A.-S. Rupp and T. Wischermann for assistance during sampling as well as J. Freund and P. Hähnel for providing Figure 3. The crew of the FK Senckenberg is gratefully acknowledged.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmars.2019.00563/full#supplementary-material

References