Biogeochemistry and limnology in Antarctic subglacial weathering: molecular evidence of the linkage between subglacial silica input and primary producers in a perennially ice-covered lake

Abstract

We report a 6,000 years record of subglacial weathering and biogeochemical processes in two perennially ice-covered glacial lakes at Rundvågshetta, on the Soya Coast of Lützow-Holm Bay, East Antarctica. The two lakes, Lake Maruwan Oike and Lake Maruwan-minami, are located in a channel that drains subglacial water from the base of the East Antarctic ice sheet. Greenish-grayish organic-rich laminations in sediment cores from the lakes indicate continuous primary production affected by the inflow of subglacial meltwater containing relict carbon, nitrogen, sulfur, and other essential nutrients. Biogenic silica, amorphous hydrated silica, and DNA-based molecular signatures of sedimentary facies indicate that diatom assemblages are the dominant primary producers, supported by the input of inorganic silicon (Si) from the subglacial inflow. This study highlights the significance of subglacial water-rock interactions during physical and chemical weathering processes and the importance of such interactions for the supply of bioavailable nutrients.

Background

Subglacial Antarctic lakes were first identified by radio-echo sounding in the late 1960s (Robin et al. 1970; Oswald and Robin 1973). Since then and especially during the past two decades, researchers have identified numerous subglacial lakes (e.g., Kapitsa et al. 1996; Siegert et al. 1996; Jouzel et al. 1999; Karl et al. 1999; Priscu et al. 1999; Christner et al. 2006) and extensive networks of subglacial meltwater channels in Antarctica (e.g., Anderson et al. 2002; Wingham et al. 2006). Siegert and coworkers compiled an inventory of 145 subglacial lakes beneath the East Antarctic Ice Sheet and the West Antarctic Ice Sheet (EAIS and WAIS) (Siegert 2000; Siegert et al. 2005). The subglacial water, which initially derives from melting caused by geothermal heat (heat flow rates, approximately 50 mW m⁻²; Siegert et al. 2012), is involved in various water-rock interaction processes beneath the ice sheet, and these interactions play important roles in the supply of nutrients, including trace metals, to organisms in Antarctic environments. For example, silicon (Si) is one of the critical elements limiting the growth of Antarctic diatoms (e.g., Nelson and Treguer 1992), likewise nitrogen and other elements (e.g., Hutchins and Bruland 1998). Moreover, glacial input of particulate and dissolved Fe is essential to biological productivity in the Southern Ocean (Raiswell and Canfield 2012). In addition to the nutrient contributed by seasonal snowmelt (Hodson 2006), subglacial meltwater flowing through channels may also influence the productivity and diversity of microbial communities by controlling the concentrations of nutrients and the physicochemical conditions of glacial environments (e.g., Tranter et al. 2005; Esposito et al. 2006; Mikucki et al. 2009; Bentley et al. 2011).

At the margins of the EAIS and WAIS, there are a number of locations where former subglacial lakes have emerged from beneath the ice after the deglaciation since the Last Glacial Maximum (LGM, ca. 20 ka; Yokoyama and Esat 2011: lake settings, e.g., Hodgson et al. 2006; Hodgson et al. 2009; Verleyen et al. 2011). Glacial lakes affected by the input of subglacial water can be observed at the retreating margins of the Antarctic ice sheet. In the Rundvågshetta area on the Soya Coast of Lützow-Holm Bay, East Antarctica (Figure 1a), freshwater flows from the subglacial drainage channels of the EAIS (Anderson et al. 2002, and literature therein). The objective of this study was to examine the interactions between the limnology of the subglacial water input and microbiological responses to the subglacial water discharged into a perennially ice-covered glacial lake at Rundvågshetta (i.e., Lake Maruwan Oike) over the last 6,000 years. By combining geochemical data and the molecular signatures preserved in sediment core samples, this study aimed to clarify the primary factors controlling biological facies patterns in subglacial-water-fed lakes.

Figure 1

Drainage map of the Antarctic Ice Sheet and electric conductivity (mS m ⁻¹ ) of the study site. (a) (upper left) Drainage map of the Antarctic Ice Sheet showing areas where marine and geological surveys are being conducted to record the extent of the Last Glacial Maximum (LGM). (right) Detail of the Soya coast, Lützow-Holm Bay, East Antarctica, showing the locations of ice-covered (white) and ice-free (shaded) areas at Rundvågshetta, Skallen, Skarvsnes, and Langhovde. Arrows show the flow directions of the present outlet glaciers (after Sawagaki and Hirakawa 1997). (lower left) Topography at Rundvågshetta, showing the locations of lakes Maruwan, Maruwan-minami, and Maruwan-kita. Labels a to f indicate the locations of the images in Figure 2. NWWS, northwestern Weddell Sea; BS, Bransfield Strait; AP, Antarctic Peninsula; MB, Marguerite Bay; EB, Eltanin Bay; PIB, Pine Island Bay; BC, Bakutis Coast; WG, Wrigley Gulf; SB, Sulzberger Bay; WRS, Western Ross Sea; CRS, Central Ross Sea; ERS, Eastern Ross Sea; NVL, Northern Victoria Land; WLC, Wilkes Land Coast; PC, Pennell Coast; BC, Budd Coast; WI, Windmill Islands; PB, Petersen Bank; VB, Vincennes Bay; VH, Vestfold Hills; PB, Prydz Bay; MRL, Mac. Robertson Land; LHB, Lützow-Holm Bay; EWS, eastern Weddell Sea; SWS, southwestern Weddell Sea (modified from Anderson et al. 2002). (b) Diagram showing the altitudes of lakes (in meters above the mean sea level; AMSL) on the Soya Coast and electrical conductivities of lake waters (in mS/m) obtained by Imura et al. (2003), combined with data from L. Maruwan (diatom-rich microflora; this study) and L. Skallen (cyanobacteria-rich microflora; Takano et al. 2012). Note that the altitude data for L. Maruwan (8.0 m) and L. Skallen (9.64 ± 0.02 m) were referred from Geographical Survey Institute (1984) and Takano et al. (2012), respectively. See the limnological features for the numbers of lakes (Kudoh and Tanabe 2014). Seawater electrical conductivity (e.g., Cox et al. 1967; Lee et al. 2006), salinity (e.g., Koblinsky et al. 2003), and a variety of electrical conductivity measurements in hypersaline lakes (e.g., Williams and Sherwood 1994) are also given.

Methods

Geological setting of the sampling sites

Since the initial work by Yoshikawa and Toya (1957) and Murayama (1977), studies undertaken in the Lützow-Holm Bay area of East Antarctica have contributed further insight into the limnology of Antarctic lakes and the relative sea-level changes that occurred during the Holocene. Imura et al. (2003) described a range of lake types in the Soya Coast region, including freshwater lakes affected by continental glaciers and saline lakes that evaporated after their isolation from seawater during the Holocene glacio-isostatic uplift. Kudoh and Tanabe (2014) reviewed the limnology and ecology of benthic microbial assemblages from saline to glacial lakes in this area. The Holocene marine limit along the Soya Coast is estimated to have been approximately 18 m above the mean sea level (AMSL), based on radiocarbon analyses of in situ bivalve fossils from raised beach deposits (Laternula elliptica; Miura et al. 1998). Cosmogenic-radionuclide-based dating of raised beach exposures showed that Holocene deglaciation occurred locally on the Soya Coast at ca. 10 ka (Yamane et al. 2011). Bassett et al. (2007) also compared predictive models of glacio-isostatically induced relative sea-level changes at seven other locations around the Antarctic coast, as well as on the Soya Coast.

Rundvågshetta (69°54.5′ S, 39°02′ E; Figure 1a) is a rock headland at the southwest margin of the Rundvåg Glacier (e.g., Sawagaki and Hirakawa 1997; Miura et al. 1998). Figure 1b shows the electric conductivity (mS m⁻¹) of the study site, confirming the origin of the lake water inflow. Two lakes on the headland, Lake Maruwan Oike (hereafter, L. Maruwan), with a water level 8 m AMSL (data from the Geographical Survey Institute 1984), and Lake Maruwan-minami (hereafter, L. Maruwan-minami), with a water level >8 m AMSL (the overflow channel of L. Maruwan-minami flows into L. Maruwan; see Figure 2a), provide interesting localities in which to evaluate the influence of subglacial water input on the biogeochemical and sedimentological characteristics of the lakes into which it is discharged (Table 1). It has been reported that the maximum thickness of the surface ice in the Soya Coast region is less than 2 m (Imura et al. 2003). Figure 2 shows lakes Maruwan and Maruwan-minami and the surrounding features, including Rundvåg Glacier, a lateral moraine close to the glacier, and the outflows of lakes Maruwan and Maruwan-minami. We also observed a number of erratic boulders derived from glacial transport processes.

Figure 2

Terrain in the study area, showing lakes Maruwan, Maruwan-minami, and Maruwan-kita and associated geomorphic features. The locations of the images are shown by labels a to g in Figure 1a (lower left panel). (a) Overall features of the study area, showing the proximity of the lakes to Rundvåg Glacier, and the overflow points of lakes Maruwan and Maruwan-minami. (b) Lateral moraine close to Rundvåg Glacier. (c) Outflow of L. Maruwan. (d) Possible subglacial erosion features (after Sawagaki and Hirakawa 1997). (e) Present-day microbial mat (dark area) and erratic boulders transported and deposited by past glaciers. (f) Exposed stratification in glacial ice. (g) Detail of (a), showing a cave-like structure beneath the glacier.

Table 1 Concentrations of major elements (average values of duplicate analyses), TC, TN, and TS in core Mw5S from L. Maruwan

Lacustrine sediments were collected using push-type corers (diameter, < 8 cm; see the core images in Figure 3) during the 47th Japanese Antarctica Research Expedition (December 2005). The latitude and longitude of the sampling position in L. Maruwan (hereafter, sample Mw5S; core length, 156 cm) were recorded with GPS as 69.54′27.7″ S and 39.02′46.7″ E, respectively. The water depth and the ice thickness at this site were 20.2 and 1.5 m, respectively (cf. surface sediment description; Watanabe et al. 2013). We also collected lake sediments from L. Maruwan-minami (hereafter, sample Ms5S) as a reference site (see Additional file 1). The cores were cut into 3 to 10 cm intervals (to fit into the field refrigeration unit) and stored at 0 to 4°C for later geochemical analysis and at −20°C for later molecular analysis.

Figure 3

Core images of lacustrine sediments and lithological descriptions. (a) Cross-section of a core sample obtained from L. Maruwan (sample Mw5S) and (b) lithological descriptions and uncorrected ¹⁴C ages (years BP) of organic carbon. The color data in (a) were obtained with a digital color meter (Konica Minolta, SPAD 503) and revised Standard Soil Color Charts (Oyama and Takehara 2005).

Geochemical analysis of sedimentary facies

The major elements (MgO, Al₂O₃, SiO₂, K₂O, CaO, TiO₂, MnO, and Fe₂O₃) were analyzed with X-ray fluorescence (XRF; JEOL JSX-3211; JEOL Ltd., Tokyo Japan), following Takano et al. (2012), and calibrated to Geological Survey of Japan (GSJ) standards (JA-1, JA-3, JB-1a, JB-3, JG-2, JG-3, JGb-1, JP-1, JLK-1, JLs-1, JDo-1, JSI-1, JCh-1, JR-3, JMn-1, JSd-2, and JSd-3; Imai et al. 1995, 1999). The data presented represent the average values of duplicate analyses. Amorphous hydrated silica (opal-A) identified with X-ray diffraction (XRD; Mac Science Co., Ltd., Yokohama, Japan) was used as a proxy for amorphous biogenic silica derived from diatoms (siliceous primary producers; e.g., Kastner et al. 1977; Leng and Barker 2006), with DNA-based molecular signatures as supporting data (Takano et al. 2012). Color data were obtained with a digital color meter (Konica Minolta, Tokyo, Japan; SPAD 503) and revised Standard Soil Color Charts (Oyama and Takehara 2005).

The methods used for elemental and bulk isotopic analyses are described by Takano et al. (2012). Briefly, the analyses of carbon, nitrogen, and sulfur were performed using a Micro CORDER JM10 (J-Science Lab Co., Ltd, Kyoto, Japan). Carbon and nitrogen isotopic ratios were determined using an elemental analyzer (Costech 4010 or Flash 2000)-isotope ratio mass spectrometer (Thermo Finnigan, Delta Plus, or Thermo Finnigan Delta V Advantage). Carbon and nitrogen isotopic compositions are expressed as per mil (‰) deviations from the standard as:

$$ {\updelta}^{13}\mathrm{C}=\left[{\left({}^{13}\mathrm{C}{/}^{12}\mathrm{C}\right)}_{\mathrm{sample}}/{\left({}^{13}\mathrm{C}{/}^{12}\mathrm{C}\right)}_{\mathrm{standard}}\hbox{--} 1\right]\times 1,000\left({\mbox{\fontencoding{U}\fontfamily{wasy}\selectfont\char104}}\ \mathrm{v}\mathrm{s}.\ \mathrm{P}\mathrm{D}\mathrm{B}\right) $$

$$ {\updelta}^{15}\mathrm{N}=\left[{\left({}^{15}\mathrm{N}{/}^{14}\mathrm{N}\right)}_{\mathrm{sample}}/{\left({}^{15}\mathrm{N}{/}^{14}\mathrm{N}\right)}_{\mathrm{standard}}\hbox{--} 1\right]\times 1,000\left({\mbox{\fontencoding{U}\fontfamily{wasy}\selectfont\char104}}\ \mathrm{v}\mathrm{s}.\ \mathrm{Air}\right), $$

respectively.

The standard deviations for the carbon and nitrogen isotopic compositions obtained using authentic standard reagents (cf. Tayasu et al. 2011) were as follows: BG-A (n = 12, δ¹³C < ±0.26‰, δ¹⁵N < ±0.25‰), BG-P (n = 6, δ¹³C < ±0.05‰, δ¹⁵N < ±0.24‰), and BG-T (n = 9, δ¹³C < ±0.08‰, δ¹⁵N < ±0.26‰) in the first validation and BG-A (n = 10, δ¹³C < ±0.12‰, δ¹⁵N < ±0.26‰), BG-P (n = 6, δ¹³C < ±0.06‰, δ¹⁵N < ±0.18‰), and BG-T (n = 7, δ¹³C < ±0.11‰, δ¹⁵N < ±0.38‰) in the second validation. Some of the organic carbon fractions extracted after HCl pretreatment were analyzed to obtain radiocarbon age data, corrected for the δ¹³C value, using an accelerator mass spectrometer (AMS) at the University of Tokyo, Japan (Yokoyama et al. 2007), or at Beta Analytic Inc., Florida, USA.

DNA-based analyses

DNA was analyzed in 18 sections (0 to 2, 2 to 4, 4 to 6, 8 to 10, 13 to 14, 14 to 16, 22 to 24, 30 to 32, 38 to 40, 49 to 52, 52 to 55, 55 to 58, 58 to 61, 61 to 64, 78 to 80, 95 to 100, 115 to 120, and 135 to 140 cm) of the core sample obtained from L. Maruwan; DNA was extracted from a 0.5 to 1.0 g sample from each section using an UltraClean Soil DNA Isolation Kit (Mobio, Carlsbad, CA, USA). Polymerase chain reaction (PCR) amplification of the 16S rRNA gene, denaturing gradient gel electrophoresis (DGGE), and sequencing of the resulting DGGE bands were performed as described previously (Muyzer et al. 1996; Takano et al. 2012). The sequences were taxonomically identified with the RDP Classifier from the Ribosomal Database Project-II release 11.

Results

Geochemistry, major elemental compositions, and sedimentary facies

The olive-gray, olive-black, and dark olive-gray laminations in the upper sections of core Mw5S from L. Maruwan are the remains of soft microbial mats. The laminations in the lower sections of the core are largely gray to olive-black (Figure 3a). The distinct laminations represent conditions favorable for preservation (i.e., no bioturbation). Some sediment deformation was observed at depths of 0 to 5, 10, and 50 to 60 cm. The lithological descriptions and the corresponding ¹⁴C ages of organic carbon (years BP) are shown in Figure 3b. Geochemical analyses of carbon, nitrogen, sulfur, and major elements were used to identify the sedimentary facies.

Figure 4a shows depth profiles of the concentrations of the major elements and the carbon and nitrogen isotope ratios. Total carbon (TC) and the carbon isotopic composition (δ¹³C) of bulk organic matter ranged from 0.7 to 3.5 wt% (mean, 2.6 ± 0.6 wt%) and from −17.0‰ to −23.6‰ (vs. PDB), respectively. Total nitrogen (TN) and the nitrogen isotopic composition (δ¹⁵N) of the bulk organic matter were <0.48 wt% (mean, 0.3 ± 0.1 wt%) and from +2.2‰ to +6.6‰ (vs. Air), respectively. Total sulfur (TS) ranged from 0.3 to 1.6 wt% (mean, 1.0 ± 0.3 wt%). The profiles show that the increasing trends in carbon and nitrogen with the movement from marine to freshwater facies in L. Skallen and L. Oyako (Takano et al. 2012; Matsumoto et al. 2014) are not observed in L. Maruwan. The sedimentary facies were not distinguishable on the basis of their nitrogen isotopic compositions because of the relic nitrogen inputs from subglacial water (δ¹⁵N up to +6.6‰ in Mw5S; cf. L. Skallen, in the supporting data of Matsumoto et al. 2010).

Figure 4

Vertical profiles of some major elements, SiO₂-Al₂O₃ and SiO₂-Fe₂O₃ diagrams, bivariate TiO₂-Al₂O₃ and TiO₂-Fe₂O₃ plots. (a) Vertical profiles of lithologies, total carbon (TC), δ¹³C_org (‰ vs PDB), total nitrogen (TN), δ¹⁵N (‰ vs air), total sulfur (TS), ¹⁴C ages, wt% of major elements (SiO₂, Fe₂O₃, Al₂O₃, TiO₂, CaO, MgO, K₂O), and major elemental ratios (SiO₂/TiO₂ and Al₂O₃/TiO₂) in core Mw5S from L. Maruwan, showing the transition between the marine and terrestrial stages via possible brackish conditions (*based on microflora differences at a depth of 60 cm, determined with a DGGE analysis). The raw data are provided in Table 1. (b), (c) Al₂O₃-SiO₂ and Al₂O₃-Fe₂O₃ diagrams for L. Maruwan (core Mw5S) and Maruwan-minami (core Ms5S) sediments. Possible end-members of marine and terrestrial major elements are noted on the axes. (d) The similarity of the source rocks of L. Maruwan (core Mw5S, black symbols: R ² = 0.97 and 0.92) and L. Maruwan-minami (core Ms5S, blue symbols: R ² = 0.85 and 0.89) was confirmed with bivariate TiO₂-Al₂O₃ and TiO₂-Fe₂O₃ plots.

The trends in the contents of some major elements indicate a marked transition from marine to lacustrine conditions in each of the cores (Figure 4a). The sediments deeper than 50 cm depth in the Mw5S core were substantially enriched in SiO₂ (<83%), indicating high SiO₂ end-member values under marine conditions and low SiO₂ values under freshwater conditions. The mixture of these two end-members can be illustrated on SiO₂-Al₂O₃ and SiO₂-Fe₂O₃ diagrams, in which contrasting mixing trends are observed (Figure 4b,c). The mixing profiles in these diagrams are similar to those of the reference sample obtained from L. Maruwan-minami (Ms5S; Additional file 1), which is located nearby. The Al₂O₃/TiO₂ and Fe₂O₃/TiO₂ ratios (Figure 4d) provide information about the source rocks (Young and Nesbitt 1998) episodic hydrological surge events (at 0 to 5 and 50 to 60 cm) and subglacial discharge variations (e.g., Brown 2002).

X-ray diffraction analyses of the sediments from core Mw5S indicate that opal-A is abundant throughout the section, except at 0 to 5 cm, suggesting that the silicate is of biogenic origin (Figure 5). To confirm this observation, we also obtained DNA-based molecular evidence of biogenic signatures (see Figure 5c). The high end-member SiO₂ concentrations, which also suggest large contributions of biogenic silica to the sediments, were probably derived from abundant diatoms (cf. determination of opal-A in the discussion).

Figure 5

Depth profiles of biogenic silica concentrations and X-ray diffraction analyses of sediments from core Mw5S. (a) Vertical profiles of biogenic silica content (wt%), (b) opal-A signatures determined with X-ray diffraction analysis, and (c) denaturing gradient-gel electrophoresis (DGGE) analysis of L. Maruwan sediments (core Mw5S). Note that the direction of electrophoresis is from right to left.

Radiocarbon age determinations of sedimentary organic matter

We determined the AMS radiocarbon (¹⁴C) ages of the organic matter in the core samples (Figure 3b and Table 2). The ¹⁴C values (years BP) at the top of the core (1,350 ± 40 years in Mw5S at a depth of 0 to 2 cm) indicate significant inputs of relic carbon. Therefore, the sediment age at the top of the core of 1,300 years was subtracted from the raw values to calibrate the radiocarbon dates with the corresponding calendar ages (after Berkman and Forman 1996; Miura et al. 2002). Similarly, marine benthic organisms have yielded ¹⁴C ages of 1,010 ± 110 to 1,190 ± 90 years BP along the Soya Coast, as reported by Yoshida and Moriwaki (1979). Using this protocol, we estimated the age at which the first sediments were deposited in the bedrock-scoured glacial basin of L. Maruwan (Figures 1 and 4a) to be at least 4,807 to 5,204 cal BP (2σ). Based on the major elemental compositions (Figure 4b), we also obtained an emergence age (i.e., the transition from marine to freshwater) of 3,382 to 3,560 cal BP (2σ); the median age of 3,471 cal BP occurs at a depth of 46 to 49 cm. Here, we note the possibility that the period from 4,000 to 1,350 cal BP represented a hiatus in the sedimentation processes in L. Maruwan. The sedimentary facies in L. Maruwan-minami, in contrast, did not show a similar profile (Additional file 1). Assuming the continuous inflow of subglacial water, the transitions from marine to freshwater conditions via temporary brackish conditions are indistinct in the L. Maruwan cores compared with the transitions observed at lakes Skallen and Oyako, also located on the Soya Coast (Takano et al. 2012).

Table 2 Radiocarbon ( ¹⁴ C) age data obtained with AMS for the organic carbon fractions from core Mw5S from L. Maruwan

Preliminary results for biological facies and molecular signatures from DGGE analysis

A PCR-DGGE analysis was performed to assess the shifts in the composition of residual DNA in the sediments of L. Maruwan (Figure 5c). Depth-related changes were apparent in regions shallower than 6 cm. In the layers deeper than 4 cm, two predominant bands (labeled with arrows in Figure 5c) were consistently observed throughout the range analyzed. The nucleotide sequences of the two bands were identical, corresponding to the chloroplasts of the marine diatom Chaetoceros, and were also identical to the diatom sequence detected in the deeper sediment layers of L. Skallen (Takano et al. 2012). The other sequenced DGGE bands (labeled with triangles in Figure 5c) represented taxa of nonphototrophic bacteria determined with a 16S rRNA analysis (unpublished data).

Discussion

Primary producers and estimates of biogenic silica

The lake sediments in core Mw5S (L. Maruwan) record a continuous input of relic carbon, nitrogen, and bioavailable silica from subglacial water, indicated by their chemical constituents, including major elements and isotopes. Hodson et al. (2010) experimentally verified the chemical weathering and solute production processes in Antarctic glacial meltwater based on aspects of water-rock interactions and hydrochemistry. Stumpf et al. (2012) and Tranter et al. (2005) also suggested the important influence of subglacial weathering on glacial meltwater chemistry, and the potential roles of meltwater constituents as sources of energy (e.g., organic carbon) and nutrients (e.g., nitrogen, sulfur, and phosphorus). The concentration of bulk sulfur was also significant in the freshwater sedimentary stage (Figure 4). Using a functional gene marker, Watanabe et al. (2013) pointed out the importance of sulfur cycling and microbial interactions in freshwater environments, including in the modern sediments of L. Maruwan. With respect to the input from physical weathering, clay-sized particles of aluminosilicate minerals are generated by subglacial abrasion processes. These solid-phase materials are transported as suspended sediments into the lake, where they liberate dissolved silica, which can be utilized by diatoms (Figure 6).

Figure 6

Inflow of subglacial meltwater containing solid-phase materials and dissolved nutrients, including silica, relic carbon, and nitrogen. Schematic models of (a) past lake setting after glacial erosion (i.e., the initiation of sedimentation processes during the marine phase) and (b) present lake setting (lacustrine stage). The relative sea level is controlled by glacio-isostatic uplift. For the present sea ice conditions during the summer season (December 2005; photograph taken from a helicopter), please see the description of the landscape of Rundvågshetta (Additional file 1).

With respect to the input from chemical weathering, the simple weathering of rock minerals, including hydrolysis processes involving carbonates, silicates, and aluminosilicates, probably control the composition of the inputs (cf. chemical models: Raiswell 1984). These processes include the following:

CaCO₃(s) + H⁺(aq) → Ca²⁺(aq) + HCO₃ ⁻(aq)

Mg₂SiO₄(s) + 4H⁺(aq) → 2 Mg²⁺(aq) + H₄SiO₄(aq)

2NaAlSi₃O₈(s) + 2H⁺(aq) + 9H₂O → 2Na⁺(aq) + 4H₄SiO₄(aq) + Al₂Si₂O₅(OH)₄(s)

where CaCO₃(s), Mg₂SiO₄(s), and NaAlSi₃O₈(s) denote calcite, forsterite, and albite, respectively. We observed increasing trends in the concentrations of CaO, MgO, and K₂O in the lacustrine facies after the marine-freshwater transition (Figure 4). The water-rock interactions involving forsterite (Satish-Kumar and Wada 2000) and albite (cf. Shiraishi et al. 1987) in the Lützow-Holm complex will eventually produce dissolved silica as H₄SiO₄ (aq) and a clay mineral (kaolinite) as Al₂Si₂O₅(OH)₄(s), respectively. It is particularly relevant that the dissolved silica is biologically available as an essential nutrient and growth element for diatoms.

Because biogenic opal-A is a major component of the L. Maruwan sediments, we were able to estimate the concentrations of biogenic silica in the sediment. We calculated the flux of biogenic silica in core Mw5S based on the end-member SiO₂ and Al₂O₃ concentrations, assuming that: (i) Al is only contained in the detrital component (of glacial inorganic origin) because Al₂O₃ concentrations are constant throughout the core at 10.5% (representing the freshwater end-member) and (ii) SiO₂ is contained in both the biogenic and detrital components (of glacial inorganic origin), with SiO₂ concentrations in the detritus constant at 46.6% (representing the freshwater end-member). Therefore,

$$ \mathrm{Total}\ \mathrm{silica}\left(\%\right)=\mathrm{Biogenic}\ \mathrm{silica}\left(\%\right)+\mathrm{Detrital}\ \mathrm{silica}\left(\%\right) $$

where the detrital ratio of the end-member (Figure 4b,c) is as follows:

$$ \mathrm{Detrital}\ \mathrm{silica}\left(\%\right)/\mathrm{Detrital}\ {\mathrm{Al}}_2{\mathrm{O}}_3\left(\%\right)=46.6\left(\%\right)/10.5\left(\%\right) $$

Hence, for L. Maruwan:

$$ \mathrm{Biogenic}\ \mathrm{silica}\left(\%\right)=\mathrm{Total}\ \mathrm{silica}\left(\%\right)\hbox{--} 4.43\ \mathrm{Detrital}\ {\mathrm{Al}}_2{\mathrm{O}}_3\left(\%\right) $$

and in L. Maruwan-minami:

$$ \mathrm{Biogenic}\ \mathrm{silica}\left(\%\right)=\mathrm{Total}\ \mathrm{silica}\left(\%\right)\hbox{--} 4.36\ \mathrm{Detrital}\ {\mathrm{Al}}_2{\mathrm{O}}_3\left(\%\right) $$

Based on the TiO₂-normalized plots of Al₂O₃ and Fe₂O₃, the similarity of the source rocks in the two lakes is high (Figure 4d), suggesting the discharge of homogenous meltwater into both lakes.

Figure 5a shows the depth profiles of the reconstructed biogenic silica concentrations in L. Maruwan. After the marine-terrestrial transition, diatoms were the main primary producers, even in the freshwater/brackish environments. Based on the DGGE and molecular analyses, the major phototrophic producers were identical to those detected in the deeper sediment layers of L. Skallen (i.e., diatoms) (Takano et al. 2012). Therefore, the origin of Si in the Antarctic lakes is of particular interest because dissolved silica is essential for the growth of diatoms (i.e., diatoms are the dominant group of phytoplankton with amorphous opal constituents).

Subglacial abrasion and weathering cause continuous input of silica

Sawagaki and Hirakawa (1997) have described the formation of glacial erosional bedforms in the Rundvågshetta area. The erosional bedforms are accompanied by small erosional marks, which support the interpretation that subglacial meltwater has contributed to the erosion at the base of Antarctic glaciers (cf. Figure 6). It is important to note that some subglacial erosional features may have resulted from subglacial streams, as shown in Figure 2d. Anderson et al. (2002) have extensively reviewed the features of subglacial meltwater drain channels beneath continental ice sheets. Observations and interpretations of these bedforms have been used to reconstruct the historical development of glacial erosional bedforms (e.g., Shaw 2002), and to understand the significance and implications of subglacial water-rock interactions. Further support for this finding comes from the cosmogenic radionuclide dating of local glacial erratics and bedrock (Yamane et al. 2011). Both bedrocks and erratics show similar exposure ages, suggesting the presence of warm basal ice during the last glacial period, which would have included subglacial water channels.

We suggest that the continuous input of subglacial meltwater influenced the chemical compositions of areas marginal to the ice sheet (also, see de Mora et al. 1994; Brown 2002). Nelson and Treguer (1992) reported that silicon is an important limiting nutrient for Antarctic diatom productivity. Total biogenic silica (%) and total carbon contents are positively correlated in L. Maruwan (R = 0.75; this study). When inferring the past constraints on the primary production by diatoms in the ice-marginal lakes, we concluded that the subglacial weathering of silicate and aluminosilicate minerals supplied significant levels of minerals, nitrogen, and relic carbon to subglacial meltwaters (e.g., Hutchins and Bruland 1998). Therefore, dissolved inorganic silicates are constantly available for utilization by the diatom communities that are the primary producers in the perennially ice-covered lakes at Rundvågshetta.

Conclusions

• [1] This study provides the first published data on the subglacial limnology and Si biogeochemistry of L. Maruwan, Rundvågshetta. The reservoir effects of relic carbon derived from the subglacial water beneath the Rundvåg Glacier were estimated at 1,300 years (cf. core-top age 1,350 years in Mw5S). The nitrogen isotopic composition of the bulk sediment was slightly ¹⁵N-enriched relative to the dinitrogen in atmospheric air, even in freshwater environments (cf. sedimentary δ¹⁵N in freshwater at L. Skallen). Based on the similarities in the compositions of their major sedimentary elements, lakes Maruwan and Maruwan-minami share the same meltwater source composition.

• [2] We suggest that the amount of bioavailable silica flowing out from subglacial drainage channels under the EAIS and into subglacial and proglacial lakes in the area is extremely limited, underscoring the biological significance of subglacial physicochemical weathering and water-rock interactions. Inorganic Si is the main source of biogenic silica (up to 83 wt%) for lacustrine diatom communities, indicating the strong relationship between the subglacial material exported from the glacial system and the primary production in these glacial lakes during the Holocene.