The influence of Antarctic subglacial volcanism on the global iron cycle during the Last Glacial Maximum

Abstract

Marine sediment records suggest that episodes of major atmospheric CO₂ drawdown during the last glacial period were linked to iron (Fe) fertilization of subantarctic surface waters. The principal source of this Fe is thought to be dust transported from southern mid-latitude deserts. However, uncertainty exists over contributions to CO₂ sequestration from complementary Fe sources, such as the Antarctic ice sheet, due to the difficulty of locating and interrogating suitable archives that have the potential to preserve such information. Here we present petrographic, geochemical and microbial DNA evidence preserved in precisely dated subglacial calcites from close to the East Antarctic Ice-Sheet margin, which together suggest that volcanically-induced drainage of Fe-rich waters during the Last Glacial Maximum could have reached the Southern Ocean. Our results support a significant contribution of Antarctic volcanism to subglacial transport and delivery of nutrients with implications on ocean productivity at peak glacial conditions.

ARTICLE

Received 25 Aug 2016 | Accepted 27 Mar 2017 | Published 9 Jun 2017

DOI: 10.1038/ncomms15425 OPEN

The inuence of Antarctic subglacial volcanism on the global iron cycle during the Last Glacial Maximum

Silvia Frisia1, Laura S. Weyrich2, John Hellstrom3, Andrea Borsato1, Nicholas R. Golledge4,5, Alexandre M. Anesio6, Petra Bajo7, Russell N. Drysdale7,8, Paul C. Augustinus9, Camille Rivard10 & Alan Cooper2

Marine sediment records suggest that episodes of major atmospheric CO2 drawdown during the last glacial period were linked to iron (Fe) fertilization of subantarctic surface waters.

The principal source of this Fe is thought to be dust transported from southern mid-latitude deserts. However, uncertainty exists over contributions to CO2 sequestration from complementary Fe sources, such as the Antarctic ice sheet, due to the difculty of locating and interrogating suitable archives that have the potential to preserve such information. Here we present petrographic, geochemical and microbial DNA evidence preserved in precisely dated subglacial calcites from close to the East Antarctic Ice-Sheet margin, which together suggest that volcanically-induced drainage of Fe-rich waters during the Last Glacial Maximum could have reached the Southern Ocean. Our results support a signicant contribution of Antarctic volcanism to subglacial transport and delivery of nutrients with implications on ocean productivity at peak glacial conditions.

1 School of Environmental and Life Sciences, The University of Newcastle, Callaghan, New South Wales 2308, Australia. 2 Australian Centre for Ancient DNA (ACAD), The University of Adelaide, Adelaide, South Australia 5005, Australia. 3 School of Earth Sciences, The University of Melbourne, Parkville, Victoria 3010, Australia. 4 Antarctic Research Centre, Victoria University of Wellington, Wellington 6140, New Zealand. 5 GNS Science, Avalon, Lower Hut 5011, New Zealand. 6 Bristol Glaciology Centre, School of Geographical Sciences, University of Bristol, Bristol BS8 1SS, UK. 7 School of Geography, The University of Melbourne, Parkville, Victoria 3010, Australia. 8 Environnements, Dynamiques et Territoires de la Montagne, UMR CNRS, Universit de Savoie-Mont Blanc, 73376 Le Bourget du Lac, France. 9 School of Environment, The University of Auckland, Private Bag, Auckland 92019, New Zealand. 10 European Synchrotron Radiation Facility, 38000 Grenoble, France. Correspondence and requests for materials should be addressed to S.F. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 8:15425 | DOI: 10.1038/ncomms15425 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15425

The amplitude of atmospheric CO2 change from glacial to interglacial periods is difcult to explain by a single mechanism, although during the last glacial maximum

(LGM) atmospheric CO2 concentration was largely regulated by iron (Fe) fertilization and decreased vertical exchange of waters of the Southern Ocean1. The correlation of dust-borne Fe ux with phytoplankton productivity and atmospheric CO2 drawdown during the LGM is clear1,2. However, fertilization experiments indicated that Fe associated with dust particulates is of reduced bioavailability when compared to that associated with organic iron-complexing ligands3.

Subglacial environments are an important source of reactive Fe in dissolved, particulate and organic iron-complexing ligand forms prone to hydrological mobilization4. Today, subglacially sourced Fe ux from the Antarctic Ice Sheet is estimated to be higher than the labile Fe ux from dust, and travels hundreds of kilometres offshore, thereby helping to sustain primary productivity in the Southern Ocean47. Ice sheets contribute potentially bioavailable Fe to the oceans via iceberg transport of Fe (oxyhydr)oxide minerals8, as well as through meltwater discharged from subglacial lake drainage7,9. During glacial maxima, a similar, if not higher contribution from subglacial drainage of meltwaters should be expected from the Antarctic Ice Sheet because a reduced dissipation of elevated geothermal heat due to ice-sheet thickening would favour basal ice melt10,11. The region around the margins of the Ross Sea is characterized by high geothermal heat uxes associated with extensional tectonics10,12 and local volcanism13. This is represented by a suite of 24 tephra spanning in age from 15,370150 to 6707 years BP preserved in the Talos Dome ice core, 17 of which are geochemically linked to volcanic eruptions in the Mount Melbourne province in Northern Victoria Land (NVL)14. Thus, the sector of NVL near the Ross Sea rift is ideal for exploring whether or not volcanically inuenced subglacial processes contributed potentially bioavailable Fe to the Southern Ocean in the past.

Evidence for LGM subglacial processes involved in Fe delivery to the Southern Ocean from the Antarctic Ice Sheets has been elusive until now. Datable material capable of recording the chemical, microbial and hydrological signatures of subglacial environments is required, and previous approaches that involved drilling subglacial lakes and sediments suffered from a lack of precise age control15. Subglacial carbonates found in Boggs Valley (7155S; 16131E; elevation 1,160 m a.s.l., NVL), (Fig. 1) provide the rst radiometrically dated petrographic, geochemical and genomic evidence of thermogenic, subglacial drainage events that potentially delivered Fe and other nutrients to the ice margin in the LGM. Boggs Valley is ideally located in an area that is characterized by volcanic complexes, the ages of which span more than 400 Ma16, in a region characterized by elevated geothermal heat ux.

The presence of subglacial carbonates in an area conrms the existence of steady liquid water production at the ice-bedrock interface17,18. Variations in crystal morphology and chemistry in successive growth layers record changes in hydrology, solute concentration, particulate and microorganism load within the basal uid19. This environmental information can be placed into an absolute chronologic framework because the carbonates can be dated precisely with uranium-series methods20,21. The Boggs Valley calcite data sets suggest a relatively important (but as yet neglected) potential contribution of Antarctic volcanism in inducing basal ice melting and sustaining subglacial discharge at peak glacial conditions. Particulate and solutes released via bio-weathering, in a subglacial aqueous environment supporting microbial communities, could have reached the ice margin and supplied potentially bioavailable

Fe to the Southern Ocean. This would explain some discrepancies observed between dust-borne Fe uxes and ocean productivity in the LGM.

Figure 1 | Boggs Valley location, sampling area and geomorphologic context. (a) Map of Antarctica with location of Boggs Valley and other sites mentioned in the text; Talos Dome is located circa 100 km south of the sampling site. (b) Map created with ASTER GDEM data (by METI and NASA), showing contours. These highlight that Boggs Valley connects the Ice-Sheet plateau with the Alexandria Glacier, whose surface has a lower elevation relative to the valley oor. The Alexandria merges with the Rennick, one of the largest outlet glaciers in NVL. The open square marks the area where subglacial calcite samples were collected from glacially polished rock surfaces.

2 NATURE COMMUNICATIONS | 8:15425 | DOI: 10.1038/ncomms15425 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15425 ARTICLE

ResultsAges and sedimentology. The U/Th ages of subglacial calcites are expressed in thousands of years (ka) before the year 2000 (b2k). Boggs Valley translucent columnar calcite sparite (Cs) crusts yielded ages from 26.9510.824 to 17.0540.332 ka, and their high U concentration (1556 p.p.m., Supplementary Table 1) suggests that they formed in anoxic/suboxic aqueous pools in the presence of organic compounds22. The calcites occur over an area of 1,000 m2 of ice-free metamorphic, non-carbonate bedrock (Fig. 1). Their vicinity to moraines and their internal stratigraphy (Fig. 2) are typical of subglacial calcareous deposits found elsewhere and reect changes in hydrology at the ice/rock interface17,21.

Cs grew in anoxic/suboxic pockets kept lled by steady production of basal ice meltwaters17. Dirty microsparite (Dm) veils interrupt the growth of Cs and can be traced across samples. Notably, a 12 mm thick Cs layer bounded by Dm in BV9a2a and BV9b correlates with a Dm veil in BV9a2 and with sediment consolidated by isopachous (phreatic) calcite cement (Cc) in BV9a2c (Supplementary Fig. 1A). Stratigraphic principles suggest these facies are genetically linked and record synchronous episodes of injection of particulate-rich waters into the pockets where clean sparite precipitated. The ages of the Cs bounded by Dm (Cs Dm facies in sample BV9a2a, Supplementary Table 1)

range from 25.1350.537 ka to at least 23.5240.446 ka

(Dm facies in sample BV9b, Supplementary Table 1). Isopachous cements that indurated sediments consisting of rounded pebbles and granules composed of rocks sourced from up-glacier and deposited in bedrock furrows yielded ages from 22.5280.661 to20.8910.996 ka (Cc fabrics in BV9a2c and BV9a1(i), Supplementary Table 1). As Cc post-dated the deposition of the sediment, it is reasonable to hypothesize that clast cementation eventuated after the precipitation of Dm in BV9b at23.5240.446 ka Dm. This suggests synchronous subglacial transport of sediment with pebble-grain size in furrows and particulate injection in basal depressions (Dm). Non-depositional or dissolution gaps interrupting precipitation of Cs mixed with Dm on bedrock protuberances must also be coeval with sediment transport (Supplementary Fig. 1). Thin Dm layers intermittently coat Cs from 21.1130.283 (BV9a2) to 17.0540.332 ka, but there is no evidence for chronostratigraphic correlation with indurated sediments (Supplementary Table 1).

Isotope and chemical properties. The stable isotope composition of Boggs Valley calcites was measured along transects in BV9a2, consisting of Cs, in BV9b, containing both Cs and Dm, and BV11, exclusively formed of Dm. Overall, the d13CVPDB values average

8.0% (Supplementary Fig. 2) implying that the dissolved inorganic carbon in the parent waters derived from microbial

0 m5001,0001,500

BV9a2c

Mn S Si

Ca Fe

0 m100200300

Cs

Cc

0 m 100 200 300

0 m 100 200 300 0 m 100 200 300 0 m 100 200 300 0 m 100 200 300

Cg

0% 20 40 60 80 100

Dm

Ca S Si

Mn

Fe

Dm

BV9b

Cs

Cs

0 m

200 400 600

0 m 200 400 600

0 m 200 400 600

0 m 200 400 600

0 m 200 400 600

Figure 2 | Boggs Valley subglacial calcites petrography and elemental concentration. Sample BV9a2c is representative of deposits found in furrows consisting of rounded clasts cemented by isopachous calcite rims (Cc), granules coated by micrite (Cg: coated grains) capped by columnar Cs. Synchrotron micro XRF elemental maps for the area designated by a white box highlight that maximum concentration of S (as sulfate) is in the calcite matrix. Fe reaches up to 22,000 p.p.m. (as determined by laser ablation (LA)-ICP-MS) in particulate, and it is likely present in the form of oxy-hydroxides when not associated with Si-rich particles. The polished thick section of BV9b is typical of Boggs Valley calcites consisting predominantly of Cs, interrupted by thin layers of Dm, as shown in the microstratigraphic log to the left. The elemental maps for the area designated by a white box show S concentration increasing toward the contact with the overlying Dm. Mn distribution is similar to S and can be interpreted as reecting a common source or an increase in temperature, as Mn incorporation in calcite is temperature dependent59. In the clean Cs, Fe concentration is below 200 p.p.m., with Fe distribution appearing blue on the colour scale. Note that S and Mn concentration distributions follow crystal terminations, providing robust evidence for preservation of environmental signals in Boggs Valley calcites. Scale bar for both BV9a2c and BV9b polished sections is 5 mm. In all maps, elements are reported in normalized counts ranging from high (red) to low (blue) relative concentration according to the 0100% scale bar.

NATURE COMMUNICATIONS | 8:15425 | DOI: 10.1038/ncomms15425 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15425

a

BV9b

BV8a

0% 10% 20% 30% 40% 50%

0% 10% 20% 30% 40% 50%

Proportion taxa >1%

b

BV9b

BV8a

Proportion taxa >1%

Figure 3 | Microbial associations from two sparite samples. (a) Bacterial taxa present at 41% of the total diversity graphed along with the lowest taxonomic identication possible. (b) Thermophilic bacteria close to known taxa. The proportion of Thermomicrobia (blue) and Thermoleophilia (green)

taxa of the total species determined is displayed.

metabolism23 or from subglacial oxidation of organic matter19. The d18OVPDB values average 30.3% in Dm and 28.5% in

Cs (Supplementary Fig. 2). These values are similar to those of carbonates precipitated from Antarctic meltwaters24.

Facies Cs Dm formed from 25.1350.537 to 23.5240.446

ka are characterized by high sulfur, Fe and manganese concentrations (Fig. 2) and enclose calcium uoride spherules (Supplementary Fig. 3). Fe, at up to 22,000 p.p.m., is present in its ferric form as (oxyhydr)oxide, most likely ferrihydrite and goethite, associated with particulates (Fig. 2; Supplementary Fig. 4a). Divalent Fe identied by mXANES peaks at 7,126 and 7,130 eV is more common in Cs, where elemental Fe detected by inductively coupled plasma mass spectrometer (ICP-MS) reaches concentrations of B100 p.p.m. Sulfur mostly occurs as sulfate (Supplementary Fig. 4b) with concentration up to 3,000 to 5,300 p.p.m. Amino-acid sulfonate was also revealed by mXANES (Supplementary Fig. 4b), suggesting a microbial contribution to the subglacial S pool25. Negative-carbon isotope ratio values and sulfonate are consistent with the hypothesis of a subglacial microbial community whose metabolic activity involved organic matter and sulde oxidation. This is corroborated by ancient DNA sequencing of the subglacial calcites.

Microorganism associations. Ancient DNA was extracted from calcite Cs and Dm layers under strict ancient DNA protocols. The compact nature of the fabrics ensures that the system was not

contaminated by organic compounds or microbes more recent than the LGM. A total of 189 sequences were 497% similar to sequences of known taxa, with the highest proportion in Cs layers pre-dating 21.0470.521 ka in BV8a and 20.1430.431 ka in BV9b (Supplementary Data 1, 2, 3). Only 20% could be classied to a genus, and most were observed without clear phylogenetic resolution.

The microbial taxa found in the Boggs Valley subglacial calcites are typical of glaciers, ice, subglacial environments26, Antarctic lake sediments and springs25,27, deep-sea sediments and deep-sea thermal vents28,29. Some pertain to endolithic communities30,31 and include the phyla Chloroexi (4032%), Actinobacteria (2225%), Cyanobacteria (1819%) and Proteobacteria (1317%). Others are associated to communities with known critical functions for silicate bedrock dissolution, including sulfur reducing and organic matter-oxidising bacteria, while others are known to promote calcite precipitation, including autotrophic, heterotrophic and phototrophic taxa3234 (Fig. 3a). Thermophilic microbial species accounted for 47 and 40% of the DNA sequences determined in clean columnar Cs3537 (Fig. 3b). These thermophilic taxa belong to large phylogenetic groups of micro-organisms that thrive at temperatures above 41 C (refs 38, 39). In contrast to the rich microbial association extracted from Cs crusts (BV8a, BV9b), a glacially deformed, Dm sample (BV11) yielded none of the taxa typical of thermophilic environments. The LGM basal meltwater community in Boggs Valley reconstructed from ancient DNA shares similarities to known Antarctic subglacial

4 NATURE COMMUNICATIONS | 8:15425 | DOI: 10.1038/ncomms15425 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15425 ARTICLE

environments5,26, but maintains a unique signal that is indicative of LGM hydrothermal activity in the area.

DiscussionSubglacial calcite growth in Boggs Valley occurred because wet-based subglacial conditions persisted from 26.9510.824 to17.0540.332 ka (Supplementary Table 1), coinciding with the duration of the LGM as recorded in EPICA Dome C (EDC), Talos Dome (TALDICE) and EPICA Dronning Maud Land (EDML) temperature records40 (Fig. 4).

For basal meltwater production to be sustained in a polar glacier during glacial maxima, the ice must have been thick enough to limit dispersion of geothermal heat ux to the atmosphere. The glacier thickness (H) necessary to sustain basal temperature (Tb) conducive to wet-based conditions during the

LGM was modelled by using a thermal conductivity of ice (Kice)

of 2.4 W m 1 K 1, a mean annual air temperature (Ts) of

29.4 C, 9.5 C (ref. 41) below the present temperature of

19.9 C (ref. 42) and a geothermal heat ux (G) of 120.9 mW m 2 inferred from magnetic data10 or 75 mW m 2 hypothesized from modelling12:

Tb Ts

4

0

4

Tsite (C)

8

5%

12

Age distribution (%)

4%

3%

2%

1%

0%

3

nss Ca2+ flux (g m2 ka1 )

2

1

30

0

8:7 10 4H

20

GH Kice

Fe flux (mg cm2 ka1 )

1

Considering the uncertainties in the ice-thickness calculation in equation (1), the LGM glacier that occupied Boggs Valley had to be 580900 m thick to maintain steady basal melting (Fig. 5). This is consistent with the 800 m estimated for the area obtained from a three-dimensional LGM ice-sheet simulation43. This projection suggests that a nominal ice thickness of 750 m was necessary to ensure a hydrous subglacial environment in Boggs Valley (Fig. 5) throughout the LGM.

A high density of measured ages in Boggs Valley calcites (peaks in the age distribution in Fig. 4b) likely coincides with periods when steady basal melting conditions favoured the growth of Cs throughout the LGM (fabric distribution in Fig. 4b). Cs is known to be the product of incomplete re-freezing of basal meltwater when persistent, slow ow lls a subglacial interconnected pore system17. In contrast, when measured ages are scarce, it is plausible that Cs growth was interrupted by dissolution, erosion and injection of sediment-laden meltwaters sourced from up-glacier (Supplementary Fig. 1).

A decrease in age distribution occurs at ca. 25.10.5 and persists until ca. 22.50.6 ka, which coincides with the ages of isopachous cements Cc. Therefore, the process responsible for the subglacial transport and deposition of coarse grained, rounded sediment and particulates to the subglacial environment of Boggs Valley had declined at ca. 23 ka. By considering a cluster of Dm dates around 23.5 ka (23.7690.296, 23.5240.446,23.1350.537 ka, Supplementary Table 1), it is reasonable to surmise that from 25.5 to 23.5 ka (Fig. 4b), when EDML and EDC record a decrease in dust ux, episodes of discharge of sediment-laden subglacial meltwater with high enough energy to reach the ice margin occurred7. Only high-energy current discharged from a subglacial source, such as a lake could have transported pebblesized sediment sourced from up-glacier into Boggs Valley (Supplementary Fig. 1). Under cold conditions of the LGM in Antarctica, surface melting was unlikely to reach the bedrock through crevasses and sustain subglacial ow7. The ancient DNA data suggested that thermophilic microbial taxa were mostly associated with calcites formed during or soon after discharge. This, and the high level of sulfate and uorine in calcites in layers whose ages cluster around 23.5 ka make plausible the notion that a hydrothermally inuenced subglacial lake, similar to Lake Vostok29,44, discharged into Boggs Valley at, or just before 23.5

ka (Fig. 4b). In addition, the 13C depletion recorded in the subglacial calcites indicates that the reservoir was characterized by organic-matter oxidation, or re-mineralization, that resulted in an isotopically light, dissolved inorganic carbon signature. In Antarctica, subglacial lakes show 13C depletion45, so that it is likely that a subglacial meltwater reservoir was drained following breaching of a hydrological barrier. Pore waters, sustained by steady melting of ice, were ushed through when bursts of discharge waters were injected into the basal interconnected pore system of Boggs Valley.

Transantarctic Mountain (TAM) volcanic activity, from Mount Melbourne to Mount Erebus, to the Pleiades (Fig. 1), is characterized by the release of uorine, sulfur and Fe-rich uids46,47. Critically, the LGM uoride record in the EDC ice core has been tentatively ascribed to subglacial eruptions in the

10

0

0 10 20 30 40 50 60

Age (ka b2k)

Figure 4 | Subglacial calcite age distribution in relation to LGM temperature and dust uxes. (a) EPICA Dome C and EPICA DML temperature anomalies corrected for site elevation41; (b) the age distribution of Boggs Valley calcites produced by using a normalized sum of each individual Gaussian distribution for all the 39 analyses where the [230Th/232Th] ratio is 430, including analytical errors (Supplementary

Table 1). Analyses where the [230Th/232Th] ratio is o30 are excluded from the plot. Horizontal bars above the age distribution highlight the changes of fabrics according to ages. Clear columnar Cs spans all ages. Isopachous Cc of clasts (Cc in Fig. 2, Supplementary Table 1) in the indurated sediment has a narrow range of ages, centred soon after the dust minima as identied by non sea-salt Ca uxes (c). Dm is uncommon in layers younger than ca. 22 ka.; (c) non sea-salt Ca uxes from EDML and EDC records71; (d) the total Th normalized Fe ux in the Subantarctic Ocean reconstructed from ODP10902. The dark grey vertical band identies a discrepancy between Fe ux in ODP1090 and dust uxes from EDC and EDML records, and coincide with low frequency of subglacial calcite formation. The ages of subglacial calcites correspond to the timing of the LGM (highlighted by the light grey vertical band) in both EDC and EDML, conrming that the calcite precipitates record an accurate archive of maximum ice thickness.

NATURE COMMUNICATIONS | 8:15425 | DOI: 10.1038/ncomms15425 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15425

a b

Ts: Ts:

LGM

mod.

LGM

mod.

0

10

20

Basal temp (C)

30

40

50

60

70

70 60 50 40 30 20 10 0 70 60 50 40 30 20 10 0Surface temp (C)

Surface temp (C)

Figure 5 | Estimated LGM ice thickness at Boggs Valley. Modelled LGM ice thickness (equation (1)) necessary to induce basal melt for LGM temperature of 9.5 C lower than present41, and geothermal heat ux of (a) 120.9 mWm 2 and (b) 75 mWm 2.

TAM region48. Since geothermal heating in NVL is higher (B20 mW m 2) than in the Lake Vostok region49, it is reasonable to hypothesize that a similar deep subglacial lake existed up-glacier from Boggs Valley. Meltwaters released by drainage likely reached the ice-sheet margin by owing along the axis of the Rennick outlet glacier9,50,51.

We used Icelandic examples52 to estimate the volume of ice melt produced by subglacial volcanism for a range of areal extents based on an ice overburden of B750 m (Supplementary Fig. 5).

The calculated subglacial ice melt was of sufcient volume to offset the overburden pressure exerted by the ice, thereby overcoming a glacio-hydrostatic barrier and generating drainage with adequate energy to keep subglacial conduits open for long periods of time9,50 and transport pebble-sized particles (Supplementary Fig. 1c)

At ca. 17.10.3 ka (Fig. 4b) calcite ceased to form because ice-thickness reduction caused dispersion of geothermal heat and a change to entirely cold-based conditions. This age agrees with an early onset of the deglaciation (ca. 18 ka) documented in areas of strong ice streaming in East Antarctica53,54. As a result, the connection between hydrothermally inuenced subglacial lakes at the interior of the ice sheet and the margin was severed at the end of the LGM in this sector of NVL.

Under todays global warming scenario, meltwaters discharged from supraglacial environments interacting with subglacial sediments and bedrock, deliver dissolved Fe to the ocean7,55. In contrast, at peak glacial conditions it was a combination of geothermal heat ux, thick ice and volcanism that sustained subglacial melting. In Boggs Valley, a wet-based subglacial environment hosted organisms using autotrophic and heterotrophic metabolic pathways capable of releasing divalent Fe (Fe2) from the dissolution of silicates and pyrite oxidation5658.

This Fe2 may have formed complexes with the natural organic matter in the basal meltwater or remained as dissolved species (oo1 nm), both of which could have been adsorbed onto reactive calcite surfaces. Thus, the amount of Fe2 released can be estimated from the concentration of Fe incorporated in the clean sparite crystals. Using a conservative concentration of 100 p.p.m., the Fe2 dissolved in the LGM basal waters was calculated by extrapolating the experimental distribution coefcient between solid calcite and Fe in solution (DFe2 )59 at an inferred basal temperature of just above freezing (B2 C). The value of DFe2 is 3.6 for precipitation rates of 1.86.6 mmol m 2 h 1

estimated for Cs (whose growth rate is 0.57 mm per year, between 24.6340.221 and 20.1530.431 ka in BVa2 and2.16 mm per year between 20.153 and 17.2720.226 ka in BV9b). Assuming that the LGM subglacial waters had a Ca2 concentration of 0.5 mmol l 1 (ref. 60), the divalent Fe concentration would have been B0.32 mmol l 1, a value similar to that reported for Antarctic basal ice61 and measured in outlet glacier meltwater in Greenland55.

The combination of thick ice, geothermal heat ux, volcanism and microbial metabolism very likely resulted in the delivery of bioavailable Fe to the Southern Ocean, and may ultimately have contributed to driving changes in the global C cycle during the LGM. Our new data enables an alternative interpretation of the discrepancies observed in peak Fe ux recorded by ocean sediments at site ODP1090 at ca. 23 ka that are not paralleled by an increase in dust ux recorded by ice cores (Fig. 4). In addition to soluble and complexed Fe, the basal hydrological system transported Fe as ne-grained material, including the mineral ferrihydrite, which could have provided a source of potentially bioavailable Fe that reached the productive iceberg-associated sediments8. The importance of the new data from Boggs Valley is that they corroborate the hypothesis that subglacial Fe evolved from bio-weathering and was delivered to the ocean by volcanically induced discharge processes in the LGM. Our rst nding of thermophilic microorganisms in a well dated archive of subglacial environments and processes support the concept that volcanism had a signicant role in sustaining both subglacial hydrology and subglacial microbial communities at peak glacial conditions.

Boggs Valley calcites are not a unique record of volcanic processes in the TAM region. Calcite precipitates found elsewhere have been interpreted as the product of subglacial hydrothermalism62, and subglacial eruptions are registered in the sulfate and uorine records of TALDICE and EDC ice cores46. The TAM rift has high levels of eruptive activity related to upwelling and lateral ow of the upper mantle from the Ross Sea Rift16. It is also noteworthy that mantle upwelling has been recently associated with waxing and waning of the ice sheets, thereby providing a link between magmatism and glacial cycles63. Further research on similar calcites is needed to test the quite plausible hypothesis that Antarctic volcanism, by changing subglacial meltwater discharge of micronutrients, could potentially inuence global climate.

6 NATURE COMMUNICATIONS | 8:15425 | DOI: 10.1038/ncomms15425 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15425 ARTICLE

Methods

Sampling. Samples of brown-reddish subglacial calcite crusts were collected from the bedrock, consisting of glacially polished amphibolite, where these were not covered by snow, within a total area of 100 m2 (location shown in Fig. 1b) in the course of a survey targeted to obtain terrestrial cosmogenic nuclide exposure ages on adjacent moraines. The morphology of the crust was typical of similar deposits occurring on exposed carbonate and non-carbonate bedrocks of Alpine glaciers. Thus, the crusts were immediately recognized as potentially bearing information on subglacial processes in Antarctica.

For the present study, the most compact samples characterized by columnar Cs layers (Supplementary Table 1) were selected for dating, chemical and microbial analyses, with the reasonable assumption that the Cs fabrics are compact enough to ensure closed system behaviour relative to trace elements and isotope mobilization. Therefore, all environmental proxy data and ages are likely to reect the original depositional environment, and are unaffected by post-depositional alteration or contamination following de-glaciation. In addition, dating, chemical and ancient DNA analyses were carried out on sample BV11 exclusively consisting of greenish microsparite matrix embedding angular clasts. Two samples, BV9a1 and BV9a2c consisted of clasts cemented by isopachous calcite. The cleanest isopachous cements (Cc in Supplementary Table 1), which are as thin as 0.1 mm, were sampled by microdrilling following assessment of their primary nature under the optical microscope. Similarly, micrite veils coating clastic grains (Cg in Supplementary Table 1), which resemble a microbial structure were also microdrilled to test if their age was consistent with a subglacial genesis.

Microscopy and mineralogy. Thin sections of subglacial calcites were observed by optical and epiuorescence microscopy. One millimetre thick wafers of age-equivalent subglacial crusts were polished and ultrasonically cleaned. Imaging at nm-scale resolution on uncoated surfaces was carried out by eld emission scanning electron microscopy with a ZEISS Sigma Variable Pressure eld emission scanning electron microscopy equipped with Bruker EDS system in backscattered electron mode at the Electron Microscope and X-ray unit of the University of Newcastle, Australia. Semi-quantitative chemical microanalyses were carriedout in situ on spot areas of ca. 500 nm diameter and the normalized weight % calculated with the Quantax software (Bruker nano). Epiuorescence was excited by light at 250350 nm wavelength and the most common emission band was in the 450550 nm.

Powder X-ray diffraction on clean calcite and Dm was carried out on a PANalytical h-h diffractometer equipped with a Cu X-ray tube operating at 40 kV and 40 mA. Scans were performed over the range 380 with an integrated step size of 0.017 and a counting time of 100 s per step. Identication of minerals, and cell parameter determination were performed using High Score Plus and the ICSD database (PANalytical).

Dating. U/Th dating was performed on eight Boggs Valley subglacial samplesby multi-collector ICP-MS at the School of Earth Sciences, the University of Melbourne, following the analytical method described in Supplementary Table 1. From each sample, powders were drilled from top to bottom in Cs and Dm layers, with density of sub-samples depending on the thickness of the material. Ages were obtained for 50 sub-samples, 7 of which showing 2s error 41,000 years (highlighted in red, Supplementary Table 1). In these sub-samples, uncertainties are likely due to detrital Th contamination. In most samples, Dm layers are thinner than 0.1 mm (for example in BV9a2, Supplementary Fig. 1A) and the age had to be reconstructed from that of clean Cs layer below and above. The age of mixedCs and Dm (Cs Dm in Supplementary Table 1) in BV9b, is constrained by a

pre-quem age of 26.3070.204 ka, and a date for the top Dm precipitated on partially dissolved Cs of 23.5240.446 ka. It is reasonable to reconstruct, on the basis of correlation with dated Cs Dm in BV9a2a, an age of ca. 25.5 ka

(25.1350.537 ka) for the commencement of Cs Dm precipitation in BV9b

(Supplementary Fig. 1A). Reconstruction of the timing of deposition of sediments was performed by considering that the ages obtained for the Cc (Supplementary Table 1) pre-date the formation of the overlying Cc layer in BVa2c (see Supplementary Fig. 1), but post-date the deposition of the clasts. Given that Cc is as old as 22.5280.661 ka it is reasonable to infer that transport was coeval with the deposition of Dm in BV9b and BV9a2a and, likely, with thewhole-Cs Dm layer.

The type of age model used to correlate samples is, thus, based on stratigraphic principles, whereby similar, genetically related facies bounded by erosion or non-depositional surfaces (unconformities) or correlative conformities (continuously growing columnar Cs). This implies that the stratigraphy of Boggs Valley subglacial calcites reects accommodation space available on the substrate (or between substrate and ice) and Cc ( Dm in BV9a2a and BV9b) formed on

topographic highs relative to the lows were sedimentation of clast occurred.

Geochemistry. Micro X-ray uorescence (XRF) mapping was carried out at beamline ID21 at the European Synchrotron Radiation Facility (Grenoble, France). Double-polished 300 mm thick wafers were ultrasonically cleaned and loaded in the sample chamber operating at a pressure of 10 5 mbar. Maps were acquired by raster scanning with a step of 2 mm in both directions. XRF was stimulated with a

monochromatic beam and was collected in the horizontal plane, using a large solid state detector. This geometry, together with the linear polarization of the incident X-ray beam and the vacuum conditions, minimize background signals from elastic and inelastic scattering. The micro-XRF mapping was carried out in an area of 20 5 mm to include all microstratigraphic components identied by

microscopy. Excitation energy was set at 7.5 keV and 2.6 keV to maximize the XRF yield of the elements of interest and avoid the excitation of Ca, which would easily saturate the XRF detector. XRF spectra were batch-tted using the PyMca software package.

Micro X-ray Absorption Near Edge Structure (mXANES) investigation was carried out by tuning the exciting energy around the S absorption K-edge (from2.45 to 2.55 keV) and the Fe K-edge (from 7.1 to 7.2 keV) utilizing a Si(220) double-crystal monochromator and an energy resolution of 0.25 eV (ref. 64). Calibration with internal standards allowed comparison with reference spectra.

For quantitative geochemical analysis we used a Varian 810 quadrupole ICP-MS equipped with helium excimer laser system (School of Earth Sciences, the University of Melbourne). The helium excimer produces an ultraviolet light beam (24 mm 8 mm) with wavelength of 193 nm and pulse length of 24 ns. Typical

scan speed is 20 mm s 1. Background measurement and analyses of the NIST612 standard were carried out periodically for drift corrections. Data were calibrated to the NIST612 standard and converted to absolute concentrations using 43Ca as internal standard. Background was subtracted and baseline drift corrected by linear interpolation of the standard counts. Because of mass interferences, S and P were quantied by EDS using an internal standard for the regions where these elements were detected by synchrotron-radiation-based micro-XRF.

Stable-isotope analyses were conducted using a GV Instruments GV2003 continuous-ow isotope ratio mass spectrometer at The University of Newcastle, Australia. Samples of B0.75 mg were acidied in evacuated septum-capped vials with 0.05 ml of 105% phosphoric acid. The CO2 evolved from the reaction was admitted into the ion-source chamber under vacuum in an ultra-high-purity helium gas stream. Sample isotopic ratios were standardized to the VPDB scale using an in-house standard of Carrara Marble (NEW1) and two international reference materials, NBS19 and NBS18. Analytical reproducibility for C and O was better than 0.05 and 0.10%, respectively.

Ancient DNA technique. Bacterial amplicon sequencing was undertaken on BV8a, BV9b and BV11a, representative of clear sparite, a sequence of sparite and Dm, and Dm, respectively (Supplementary Fig. 6). Contaminating DNA on the outer surface of the samples was removed by removing B2 mm of the outer surface with a Dremel tool, exposing each side to ultraviolet irradiation for 15 min, and soaking in 3% hypochlorite solution, as previously described65. DNA was then extracted from the decontaminated microsparite using a modied silica-based protocol developed for low biomass extractions in an ultra-clean, specialized ancient DNA laboratory (The Australian Centre for Ancient DNA at The University of Adelaide)66. Sample controls (extraction blank controls) were also processed simultaneously to monitor laboratory and reagent contamination67. The 16S ribosomal RNA (rRNA) encoding gene regions were amplied using polymerase chain reaction68,69. To account for contamination introduced into these low biomass samples via laboratory environment and reagents, sequences identied in extraction blank controls were ltered from the data set67,70. 16S rRNA libraries were created as previously described69 with 38 cycles performed to achieve amplication of limited ancient DNA fragments65. 16S rRNA amplicon libraries were then pooled and sequenced on an Illumina MiSeq, generating B500,000 reads between the three calcite and one extraction blank control sample.

Sequences were then de-multiplexed using CASAVA, trimmed and quality ltered using CutAdapt, and imported into QIIME for downstream analyses. Bacterial species were determined by aligning operational taxonomic units selected by clustering at 97% similarity using UClust to the Greengenes database.

Data availability. Our DNA data are publically available on the QIITA database: study ID 10766: The Inuence of Antarctic Subglacial volcanism on the global Fe cycle during the Last Glacial Maximum. DNA sample list is supplied in Supplementary Data 1. Processed data are available as Supplementary Data 2 and 3 (OTU tables in BIOM format). Raw data les have also been uploaded to the NCBI SRA database as BioProject ID: PRJNA386567; BioSample IDs: SAMN07139230, SAMN07139231. All other relevant data are available upon request from the authors.

References

1. Jaccard, S. L., Galbraith, E. D., Martnez-Garca, A. & Anderson, R. F. Covariation of deep Southern Ocean oxygenation and atmospheric CO2 through the last ice age. Nature 530, 207210 (2016).

2. Martnez-Garca, A. et al. Iron fertilization of the Subantarctic Ocean during the last ice age. Science 343, 13471350 (2014).

3. Chever, F., Sarthou, G., Bucciarelli, E., Blain, S. & Bowie, A. An iron budget during the natural iron fertilisation experiment KEOPS (Kerguelen Islands, Southern Ocean). Biogeosciences 7, 455468 (2010).

4. Death, R. et al. Antarctic ice sheet fertilises the Southern Ocean. Biogeosciences 11, 26352643 (2014).

NATURE COMMUNICATIONS | 8:15425 | DOI: 10.1038/ncomms15425 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15425

5. Boetius, A., Anesio, A. M., Deming, J. W., Mikucki, J. A. & Rapp, J. Z. Microbial ecology of the cryosphere: sea ice and glacial habitats. Nat. Rev. Microbiol. 13, 677690 (2015).

6. Dierssen, H. M., Smith, R. C. & Vernet, M. Glacial meltwater dynamics in coastal waters west of the Antarctic peninsula. Proc. Natl Acad. Sci. USA 99, 17901795 (2002).

7. Hawkings, J. R. et al. Ice sheets as a signicant source of highly reactive nanoparticulate iron to the oceans. Nat. Commun. 5, 3929 (2014).

8. Raiswell, R. et al. Potentially bioavailable iron delivery by iceberg-hosted sediments and atmospheric dust to the polar oceans. Biogeosciences 13, 3887 (2016).

9. Bell, R. E., Studinger, M., Shuman, C. A., Fahnestock, M. A. & Joughin, I. Large subglacial lakes in East Antarctica at the onset of fast-owing ice streams. Nature 445, 904907 (2007).

10. Maule, C. F., Purucker, M. E., Olsen, N. & Mosegaard, K. Heat ux anomalies in Antarctica revealed by satellite magnetic data. Science 309, 464467 (2005).

11. Fraser, C. I., Terauds, A., Smellie, J., Convey, P. & Chown, S. L. Geothermal activity helps life survive glacial cycles. Proc. Natl Acad. Sci. USA 111, 56345639 (2014).

12. Shapiro, N. M. & Ritzwoller, M. H. Inferring surface heat ux distributions guided by a global seismic model: particular application to Antarctica. Earth Planet. Sci. Lett. 223, 213224 (2004).

13. Kurbatov, A. V. et al. A 12,000 year record of explosive volcanism in the Siple Dome Ice Core, West Antarctica. J. Geophys. Res. Atmos. 111, D12307 (2006).

14. Narcisi, B., Petit, J. R., Delmonte, B., Scarchilli, C. & Stenni, B. A 16,000-yr tephra framework for the Antarctic ice sheet: a contribution from the new Talos Dome core. Quat. Sci. Rev. 49, 5263 (2012).

15. Tulaczyk, S. et al. WISSARD at Subglacial Lake Whillans, West Antarctica: scientic operations and initial observations. Ann. Glaciol. 55, 5158 (2014).

16. Ferraccioli, F. et al. Magmatic and tectonic patterns over the Northern Victoria Land sector of the Transantarctic Mountains from new aeromagnetic imaging. Tectonophysics 478, 4361 (2009).

17. Sharp, M., Tison, J.-L. & Fierens, G. Geochemistry of subglacial calcites: implications for the hydrology of the basal water lm. Arct. Alp. Res. 22, 141152 (1990).

18. Hodson, A. et al. Glacial ecosystems. Ecol. Monogr. 78, 4167 (2008).19. Refsnider, K. A. et al. Subglacial carbonates constrain basal conditions and oxygen isotopic composition of the Laurentide Ice Sheet over Arctic Canada. Geology 40, 135138 (2012).

20. Hillaire-Marcel, C. & Causse, C. The late pleistocene Laurentide glacier:ThU dating of its major uctuations and d 18 O range of the ice. Quat. Res. 32, 125138 (1989).

21. Aharon, P. Oxygen, carbon and U-series isotopes of aragonites from Vestfold Hills, Antarctica: clues to geochemical processes in subglacial environments. Geochim. Cosmochim. Acta 52, 23212331 (1988).

22. Sturchio, N., Antonio, M., Soderholm, L., Sutton, S. & Brannon, J. Tetravalent uranium in calcite. Science 281, 971973 (1998).

23. Mook, W. G. Introduction to Isotope Hydrology. Stable and radioactvie Isotopes of hydrogen, Oxygen and Carbon Vol. 25 (Taylor and Francis Group, 2006).

24. Goodwin, I. D. & Hellstrom, J. Glacio-lacustrine aragonite deposition, meltwater evolution and glacial history during isotope stage 3 at Radok Lake, Amery Oasis, northern Prince Charles Mountains, East Antarctica. Antarctic Sci. 19, 365372 (2007).

25. Mikucki, J. A. et al. A Contemporary Microbially Maintained Subglacial Ferrous Ocean. Science 324, 397400 (2009).

26. Rogers, S. O. et al. Ecology of subglacial Lake Vostok (Antarctica), based on metagenomic/metatranscriptomic analyses of accretion ice. Biology 2, 629650 (2013).

27. Christner, B. C. et al. A microbial ecosystem beneath the West Antarctic ice sheet. Nature 512, 310313 (2014).

28. Murray, A. E. et al. Microbial life at 13 C in the brine of an ice-sealed

Antarctic lake. Proc. Natl Acad. Sci. USA 109, 2062620631 (2012).29. Shtarkman, Y. M. et al. Subglacial Lake Vostok (Antarctica) accretion ice contains a diverse set of sequences from aquatic, marine and sediment-inhabiting bacteria and eukarya. PLoS ONE 8, e67221 (2013).

30. Crits-Christoph, A. et al. Functional analysis of the archaea, bacteria, and viruses from a halite endolithic microbial community. Environ. Microbiol. 18, 20642077 (2015).

31. Hughes, K. A. & Lawley, B. A novel Antarctic microbial endolithic community within gypsum crusts. Environ. Microbiol. 5, 555565 (2003).

32. Ariyanti, D. & Handayani, N. A. Feasibility of using microalgae for biocement production through biocementation. J. Bioprocess. Biotech. 2, 1000111 (2012).

33. De Muynck, W., De Belie, N. & Verstraete, W. Microbial carbonate precipitation in construction materials: a review. Ecol. Eng. 36, 118136 (2010).

34. Bii-Ionescu, M., Amann, R. & Grossart, H.-P. Massive regime shifts and high activity of heterotrophic bacteria in an ice-covered lake. PloS ONE 9, e113611 (2014).

35. Crits-Christoph, A. et al. Colonization patterns of soil microbial communities in the Atacama Desert. Microbiome 1, 28 (2013).

36. Foesel, B. U., Geppert, A., Rohde, M. & Overmann, J. Parviterribacter kavangonensis and Parviterribacter multiagellatus a novel genus and two novel species within the order Solirubrobacterales and emended description of the classes Thermoleophilia and Rubrobacteria and its orders and families. Int.J. Syst. Evol. Microbiol. 66, 652665 (2016).37. King, C. & King, G. Thermomicrobium carboxidum sp. nov., and Thermorudis peleae gen. nov., sp. nov., carbon monoxide-oxidizing bacteria isolated from geothermally heated biolms. Int. J. Syst. Evol. Microbiol. 64, 25862592 (2014).

38. Sylvan, J. B., Toner, B. M. & Edwards, K. J. Life and death of deep-sea vents: bacterial diversity and ecosystem succession on inactive hydrothermal suldes. MBio 3, e0027900211 (2012).

39. Houghton, J. et al. Thiosulfate oxidation by Thiomicrospira thermophila: metabolic exibility in response to ambient geochemistry. Environ. Microbiol. 18, 30573072 (2016).

40. Barbante, C. et al. One-to-one coupling of glacial climate variability in Greenland and Antarctica. Nature 444, 195198 (2006).

41. Stenni, B. et al. The deuterium excess records of EPICA Dome C and Dronning Maud Land ice cores (East Antarctica). Quat. Sci. Rev. 29, 146159 (2010).

42. Comiso, J. C. Variability and trends in Antarctic surface temperatures from in situ and satellite infrared measurements. J. Clim. 13, 16741696 (2000).

43. Golledge, N. R. et al. Glaciology and geological signature of the Last Glacial Maximum Antarctic ice sheet. Quat. Sci. Rev. 78, 225247 (2013).

44. Schiermeier, Q. Claims of Lake Vostok sh get frosty response. Nat. News doi:http://10.1038/nature.2013.13364

Web End =10.1038/nature.2013.13364 (2013).

45. Neumann, K., Lyons, W. B., Priscu, J. C., Desmarais, D. J. & Welch, K. A. The carbon isotopic composition of dissolved inorganic carbon in perennially ice-covered Antarctic lakes: searching for a biogenic signature. Ann. Glaciol. 39, 518524 (2004).

46. Udisti, R. et al. Sea-spray and marine biogenic seasonal contribution to snow composition at Terra Nova Bay, Antarctica. Ann. Glaciol. 29, 7783 (1999).

47. Keys, J. H. & Williams, K. Origin of crystalline, cold desert salts in the McMurdo region, Antarctica. Geochim. Cosmochim. Acta 45, 22992309 (1981).

48. Udisti, R. et al. Atmospheresnow interaction by a comparison between aerosol and uppermost snow-layers composition at Dome C, East Antarctica. Ann. Glaciol. 39, 5361 (2004).

49. Siegert, M. J. Antarctic subglacial lakes. Earth Sci. Rev. 50, 2950 (2000).50. Bell, R. E. The role of subglacial water in ice-sheet mass balance. Nat. Geosci. 1, 297304 (2008).

51. Wingham, D. J., Siegert, M. J., Shepherd, A. & Muir, A. S. Rapid discharge connects Antarctic subglacial lakes. Nature 440, 10331036 (2006).

52. Gudmundsson, M. T., Sigmundsson, F., Bjrnsson, H. & Hgnadttir, T. The 1996 eruption at Gjlp, Vatnajkull ice cap, Iceland: efciency of heat transfer, ice deformation and subglacial water pressure. Bull. Volcanol. 66, 4665 (2004).

53. Mackintosh, A. et al. Retreat of the East Antarctic ice sheet during the last glacial termination. Nat. Geosci. 4, 195202 (2011).

54. Golledge, N. R., Fogwill, C. J., Mackintosh, A. N. & Buckley, K. M. Dynamics of the last glacial maximum Antarctic ice-sheet and its response to ocean forcing. Proc. Natl Acad. Sci. USA 109, 1605216056 (2012).

55. Bhatia, M. P. et al. Greenland meltwater as a signicant and potentially bioavailable source of iron to the ocean. Nat. Geosci. 6, 274278 (2013).

56. Montross, S. N., Skidmore, M., Tranter, M., Kivimaki, A.-L. & Parkes, R. J. A microbial driver of chemical weathering in glaciated systems. Geology 41, 215218 (2013).

57. Telling, J. et al. Rock comminution as a source of hydrogen for subglacial ecosystems. Nat. Geosci. 8, 851855 (2015).

58. Magnabosco, C. et al. A metagenomic window into carbon metabolism at 3 km depth in Precambrian continental crust. ISME J. 10, 730741 (2015).

59. Dromgoole, E. L. & Walter, L. M. Iron and manganese incorporation into calcite: Effects of growth kinetics, temperature and solution chemistry. Chem. Geol. 81, 311336 (1990).

60. Welch, K. A. et al. Spatial variations in the geochemistry of glacial meltwater streams in the Taylor Valley, Antarctica. Antarct. Sci. 22, 662672 (2010).61. Raiswell, R., Benning, L. G., Tranter, M. & Tulaczyk, S. Bioavailable iron in the Southern Ocean: the signicance of the iceberg conveyor belt. Geochem. Trans. 9, 7 (2008).

62. Faure, G., Hoefs, J., Jones, L. M., Curtis, J. B. & Pride, D. E. Extreme 18O depletion in calcite and chert clasts from the Elephant Moraine on the East Antarctic ice sheet. Nature 332, 352354 (1988).

63. Crowley, J. W., Katz, R. F., Huybers, P., Langmuir, C. H. & Park, S.-H. Glacial cycles drive variations in the production of oceanic crust. Science 347, 12371240 (2015).

64. Frisia, S., Borsato, A., Fairchild, I. J. & Susini, J. Variations in atmospheric sulphate recorded in stalagmites by synchrotron micro-XRF and XANES analyses. Earth Planet. Sci. Lett. 235, 729740 (2005).

8 NATURE COMMUNICATIONS | 8:15425 | DOI: 10.1038/ncomms15425 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15425 ARTICLE

65. Adler, C. J. et al. Sequencing ancient calcied dental plaque shows changes in oral microbiota with dietary shifts of the Neolithic and Industrial revolutions. Nat. Genet. 45, 450455 (2013).

66. Brotherton, P. et al. Neolithic mitochondrial haplogroup H genomes and the genetic origins of Europeans. Nat. Commun. 4, 1764 (2013).

67. Salter, S. J. et al. Reagent and laboratory contamination can critically impact sequence-based microbiome analyses. BMC Biol. 12, 87 (2014).

68. Caporaso, J. G. et al. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335336 (2010).

69. Caporaso, J. G. et al. Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 6, 16211624 (2012).

70. Cooper, A. & Poinar, H. N. Ancient DNA: do it right or not at all. Science 289, 1139 (2000).

71. Fischer, H. et al. Reconstruction of millennial changes in dust emission, transport and regional sea ice coverage using the deep EPICA ice cores from the Atlantic and Indian Ocean sector of Antarctica. Earth Planet. Sci. Lett. 260, 340354 (2007).

Acknowledgements

Access to the European Synchrotron Radiation Facility (ESRF) was supported by the Australian Synchrotron AS-IA104-ESRF-EC794 grant. Murielle Salom, Giulia Veronesi and Vanessa Johnston provided valuable assistance at beamline ID21 of ESRF. Alfredo Martin-Garcia provided the ODP 1090 Fe uxes, Barbara Stenni provided the EDC and EDML synchronized temperature records and Roberto Udisti shared knowledge onF records and measurements in EDC. Olivier Rey-Lescure is acknowledged for cartography. The Boggs Valley calcites were collected by P.A. during Antarctica New Zealand Event K015, with logistical support from Antarctica New Zealand and the Italian Antarctic Program (PNRA). University of Auckland Research Committee grants provided funding to assist the eldwork.

Author contributions

S.F., J.H. and P.C.A. conceived the original project. J.H. and P.B. performed the U/Th dating. L.S.W. extracted and analysed the microbiological content of the calcites. A.M.A.,L.S.W. and A.C. interpreted the microorganism associations and their environmental

signicance. A.B., S.F. and C.R. performed synchrotron elemental mapping. A.B. performed the mXANES analyses. S.F. carried out all petrographic observations and interpretations. N.R.G. modelled the ice thickness and meltwater volumes. R.N.D. analysed stable isotope ratios. P.C.A. conducted eldwork, collected samples, and contributed geomorphological and geological data. The manuscript was put together and written by S.F. with signicant contributions from all. P.C.A., R.N.D. and A.C. carefully edited the nal text.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Frisia, S. et al. The inuence of Antarctic subglacial volcanism on the global iron cycle during the Last Glacial Maximum. Nat. Commun. 8, 15425 doi: 10.1038/ncomms15425 (2017).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2017

NATURE COMMUNICATIONS | 8:15425 | DOI: 10.1038/ncomms15425 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9