Full Text

Turn on search term navigation

ARTICLE

Received 17 Feb 2015 | Accepted 17 Jun 2015 | Published 23 Jul 2015

DOI: 10.1038/ncomms8850 OPEN

Constraints on soluble aerosol iron ux to the Southern Ocean at the Last Glacial Maximum

T.M. Conway1,2,w, E.W. Wolff1,2, R. Rthlisberger2,w, R. Mulvaney2 & H.E. Eldereld1

Relief of iron (Fe) limitation in the Southern Ocean during ice ages, with potentially increased carbon storage in the ocean, has been invoked as one driver of glacialinterglacial atmospheric CO2 cycles. Ice and marine sediment records demonstrate that atmospheric dust supply to the oceans increased by up to an order of magnitude during glacial intervals.

However, poor constraints on soluble atmospheric Fe uxes to the oceans limit assessment of the role of Fe in glacialinterglacial change. Here, using novel techniques, we present estimates of water- and seawater-soluble Fe solubility in Last Glacial Maximum (LGM) atmospheric dust from the European Project for Ice Coring in Antarctica (EPICA) Dome C and Berkner Island ice cores. Fe solubility was very variable (142%) during the interval, and frequently higher than typically assumed by models. Soluble aerosol Fe uxes to Dome C at the LGM (0.010.84 mg m 2 per year) suggest that soluble Fe deposition to the Southern

Ocean would have been Z10 modern deposition, rivalling upwelling supply.

1 Department of Earth Sciences, University of Cambridge, Cambridge CB2 3EQ, UK. 2 British Antarctic Survey, High Cross, Madingley Road, Cambridge CB3 0ET, UK. w Present addresses: Department of Earth Sciences, Institute for Geochemistry and Petrology, ETH Zrich, 8092 Zrich, Switzerland (T.M.C.);

Federal Ofce for the Environment, Bern, Switzerland (R.R.). Correspondence and requests for materials should be addressed to T.M.C. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 6:7850 | DOI: 10.1038/ncomms8850 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8850

Changes in atmospheric iron (Fe) supply to the oceans (described here as aerosol Fe) during glacial intervals have been suggested as one of a number of possible drivers

for changes observed in atmospheric CO2 between glacial and interglacial climate states (The Iron Hypothesis)13. Fe fertilization of the presently Fe-limited Southern Ocean during glacial times could be responsible for up to 40 p.p.m.v.35 of the regular 80120 p.p.m.v. cycles in atmospheric CO2 that coincide with glacialinterglacial cycles over at least the last 800,000 years6. Increased atmospheric Fe supply to the glacial Southern Ocean, where today low dissolved Fe concentrations limit phytoplankton growth4,7, may have enhanced the efciency of the biological pump, retaining more carbon in the deep ocean and driving a reduction in pCO2 (refs 1,2). While changes to ocean circulation and deep ocean carbon storage may play a dominant role in glacialinterglacial cycling3,810, atmospheric Fe may also strongly inuence the efciency of the biological pump on both glacial and millennial timescales3,11,12, with potentially large effects on the carbon cycle. Indeed, a recent study showed that increased nutrient utilization and greater productivity in the Sub-Antarctic are linked to periods of increased dust (and Fe) supply over the past 160,000 years13, highlighting atmospheric Fe fertilization in the Sub-Antarctic surface ocean as a candidate for explaining millennial-scale uctuations in atmospheric CO2.

Through this mechanism, Fe fertilization may play an important role in glacialinterglacial cycling, in combination with other processes9,13,14. Evidence for higher dust uxes comes from Antarctic ice cores, which record marked increases in atmospheric dust deposition to Antarctica during colder intervals over the past 800,000 years15, most likely due to expanded dust source areas in South America and enhanced atmospheric lifetime and transport16. Sedimentary records also show that the temporal pattern of Antarctic ice-core dust uxes is representative of dust reaching the Southern Ocean, with the dustclimate coupling extending back at least 4 million years5. However, modelling efforts to understand the sensitivity of the oceans and climate to Fe (for example, ref. 17) are still limited by poor constraints on aerosol Fe solubility and the uxes of bioavailable (the Fe that can be taken up for growth by phytoplankton) aerosol Fe to oceans during glacial times.

Converting ice and sedimentary dust records to bioavailable aerosol Fe uxes is not straightforward. Previous studies have used dust ux records with xed composition and Fe solubility18, or non-seasalt Ca (nss-Ca) as proxies for dissolved Fe (ref. 11). Recently, measurements of Fe concentration and uxes in ice cores have become available1923; however, these were made under acidic conditions (pH 12), and are thus far-removed from dissolution in seawater (pH B8) where Fe solubility can be o1%

(ref. 18). Fe solubility is strongly pH dependent (Supplementary Discussion), and acid leaches may access a variable fraction of the total Fe present21,24. Equally, dissolved Fe concentrations within melted Antarctic ice of glacial age (for example, 170 ng g 1 in this study) are typically well above the low-equilibrium saturation concentration of Fe(III) (o0.1 ng g 1; Supplementary Fig. 1), meaning that Fe loss due to precipitation/adsorption of Fe hydroxides is likely under weakly acidic pH conditions (Supplementary Discussion). While Fe(II) is more soluble than Fe(III) under these conditions25, a recent study suggests that a large proportion of Fe within ice-core material is present as Fe(III) (ref. 26). Any loss of Fe(III) via precipitation or adsorption to plastic equipment during processing that is not recovered during acidication will lead to underestimation of soluble Fe, which might affect studies analysing Fe after melting using discrete bottled samples or ow-through systems (for example, refs 19,20,23,26). Thus, while previous ice-core records provide valuable information about broad-scale changes in aerosol Fe

uxes, they provide incomplete information about soluble or bioavailable aerosol Fe supplied to the oceans. We were therefore motivated to seek techniques for more representative estimates of soluble Fe ux to the Southern Ocean during glacial intervals to allow models to more accurately assess the effect of changing atmospheric Fe supply on the carbon cycle.

Here we present high-resolution uxes of total and soluble atmospheric Fe, as well as calculated aerosol Fe solubility, from the Last Glacial Maximum (LGM; 2126 kyr before present (BP)) of the EPICA Dome C ice core15,20 from East Antarctica (EDC; 75 060 S 123 210 E, 3,233 m.a.s.l.; Fig. 1) and from Marine

Isotope Stage (MIS) 2-3 (2252 kyr BP) from an ice core from Berkner Island (79 32.90 S 45 32.90 E; Fig. 1). Soluble Fe was determined using a novel rapid-ltration technique, in meltwater at pH B5.3 (the natural pH of the melted ice) or following leaching of sublimated dry dust with natural seawater, analogous to instantaneous techniques used with modern aerosols27. Our data are much more representative of dissolution of atmospheric dust in the ocean than previous ice-core studies, avoiding both unrepresentative leaching by strong acids19,20,22,23 or precipitation of insoluble Fe(III) hydroxides during melting. These methods are relevant for the possible modes of deposition of dust into the ocean, either deposition within rain, where dust has already been exposed to liquid at pH B5.3, or deposition of dry dust or solid ice/snow containing dust to the ocean where the rst liquid encountered is seawater. We nd that both total aerosol Fe uxes and aerosol Fe solubility were very variable throughout the LGM in EDC ice (142% soluble; pH B5.3 mean 10%, median 6%) leading to highly variable soluble aerosol Fe uxes of 0.010.84 mg m 2 per year at Dome C throughout the interval. At Berkner Island, although aerosol Fe solubility was lower than Dome C (mean B3%), higher dust uxes resulted in similar mean soluble aerosol Fe uxes to those at Dome C during the LGM. While these calculated soluble Fe uxes are not fully representative of bioavailable Fe uxes to the Southern Ocean, they enable an improved estimate of the atmospheric Fe that would have been deposited and dissolved in seawater at the LGM compared with previous studies. Our measured aerosol Fe solubilities and soluble aerosol Fe uxes are higher than typically used in models of ocean biogeochemistry, and thus we suggest that models should consider higher and more spatially and temporally variable soluble atmospheric Fe uxes to the Southern Ocean during glacial intervals.

80 S

70S

60S

30 W 0

30 E

60 W

60 E

Berkner Island

90 W

120 W

90 E

120 E

150 W 180

EPICA Dome C

150 E

Figure 1 | The locations of EPICA Dome C and Berkner Island coring locations on Antarctica.

2 NATURE COMMUNICATIONS | 6:7850 | DOI: 10.1038/ncomms8850 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8850 ARTICLE

ResultsFigure summary. Aerosol Fe solubility and calculated total and soluble Fe uxes from EDC and Berkner ice cores are shown compared with other ice-core parameters in Figs 24, with a comparison of Fe solubility in EDC ice of LGM age at pH B5.3 and in seawater shown in Fig. 2b. The data are also available in Supplementary Data.

Aerosol Fe solubility in EDC and Berkner ice cores. Fe solubility in atmospheric dust deposited at Dome C across the LGM had a mean of 10% (n 23) and 17% (n 11) at pH B5.3

and in seawater, respectively, and a median of 6% at pH B5.3 (Fig. 2 and Supplementary Data). These aerosol Fe solubilities are higher than solubility of loess and desert dust (o0.5%; refs 18,28)

or glacial our (23%; ref. 28), but consistent with high-end solubility estimates for modern atmospheric dust, where transport-induced size/compositional fractionation and chemistry enhance solubility away from source regions29,30. Indeed, dust dry-deposited20 at Dome C during the LGM represents the distal ne-fraction of dust leaving the source regions on Patagonia31. EDC dust is enriched in both Al and Fe compared with both the crust and Patagonian topsoil (Supplementary Data), suggesting that high Fe solubility at Dome C could be driven by a compositional bias towards smaller clay particles containing more readily soluble Fe32 during atmospheric transport30,33. By contrast, mean Fe solubility at Berkner Island, closer to South American dust sources18, was only 3% during 2350 kyr BP (Fig. 2c) and 15% at the LGM.

Fe solubility in EDC samples was very variable across the interval (Fig. 2a; 142%). At several points within the LGM, we measured samples of very similar ages (Fig. 3), showing that variability sometimes also occurs at annualdecadal scales (Figs 2a and 3); for example, Fe solubility in samples at 23,350 and 23,352 kyr BP were 4 and 14%, and samples at 23,391 and 23,396 kyr BP were 17 and 6%, respectively (Supplementary Data and Fig. 3). Variability in Fe solubility through the LGM could be driven24,29 by seasonal changes in dust supply or discrete dust events, but is notable in ice samples, which by their nature represent combined dust deposition over several years. The observed variability in Fe solubility has no clear relationship with total Fe, dust loading or particle size (Fig. 4), suggesting that, if real, it is driven either by particle composition or transport-induced changes in Fe speciation. Similar EDC temperature (Fig. 2a) and assumed atmospheric transport pathways16, combined with a homogenous grain-size distribution, throughout the interval (Fig. 4), suggest that variability is more likely the result of compositional and Fe-speciation changes than transport-induced changes in Fe chemistry. Mineralogy exerts a strong control on Fe solubility, with Fe in clay and silicate minerals much more soluble than Fe hydroxides32, and glacial our enriched in both Fe(II) and clays compared with loess28. While higher solubility may be driven by relative enrichment of clay minerals with transport29,33, variable solubility could be the result of changing contributions from different types of South American dust (for example, loess or glacial material34). Contributions from local dust sources is unlikely at Dome C,

0.1

0.001

EDC

Flux (mg m2 per year)

Fe solubility SW(%)

0.01

dFe(pH 5)

Total Fe

EDC

dFe(sw)

dFe(pH 1)

Fe solubilitypH 5.3 (%)

Fe solubility pH 5.3(%)

0.1

Flux (mg m2 per year)

Berkner

dFe(pH 5)

Total Fe

Fe solubility pH 5.3(%)

EDC Temp. (C)

Mean

0.01

EDC dust (g g1 )

1.2

Mean

0.4

225

205

215

195

pCO 2(p.p.m.v.)

0.8

A1 A2

240

Berk. D

46 48

pCO 2(p.p.m.v.)

LGM

210

EDML 18O

50 52

190

260 280 300

170

22 25 26

21 24

23 22

28 30 34 38 50

EDC3 age (kyr BP) EDC3 age (kyr BP)

Figure 2 | Variability in uxes and chemistry of aerosol Fe deposited to Antarctica shown with other climatic parameters. (a) Fluxes of total and soluble (pH B5.3, seawater, pH B1 (ref. 20)) aerosol Fe, and Fe solubility across the LGM interval from the EPICA Dome C core. (b) Comparison of aerosol

Fe solubility at pH B5.3 and in natural seawater (pH B8) from EDC ice of identical LGM age. Dashed line indicates 1:1. (c) Fluxes of total and soluble (pH B5.3) aerosol Fe and Fe solubility from MIS 23 of the Berkner Island Core. The grey bar denotes the LGM and the orange bars denote Antarctic warm events. Data points represent two measurements of a single sample, and errors are smaller than the size of points. All data are shown on the

EDC3 age scale47. See Methods section for details of other climatic parameters, and details of conversion of Berkner age to EDC3.

NATURE COMMUNICATIONS | 6:7850 | DOI: 10.1038/ncomms8850 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8850

Fe solubility pH 5.3(%)

LGM mean

LGM median

0 23.20 23.25 23.30 23.35 23.40

EDC3 age (kyr BP)

Figure 3 | Fe solubility at pH B5.3 in EPICA Dome C ice of LGM ageat high resolution. Ice of identical/similar age was subsampled (Supplementary Data) to investigate variability in Fe solubility. Ice samples of different age are shown in different colours, with three ice samples sampled twice (red 23.280 kyr BP, orange 23.282 kyr BP and purple23.398 kyr BP; Supplementary Data). Data points represent two measurements of a single sample, and errors are smaller than the size of points. Dashed lines denote the mean and median of all EDC LGM measurements from Fig. 2.

where most glacial-age dust is considered to be from Patagonia31, but they could play a role at Berkner Island.

Comparison of solubility at pH B5.3 and in seawater (Fig. 2b) also shows scatter about a 1:1 line, similar to that observed in modern studies comparing ultrapure water and seawater rapid-leaching techniques27. This scatter may reect heterogeneous particle distribution and chemistry or could partly be the result of differences between the two methods, such as temperature, number of leaches, or solution chemistry enhancing dissolution in seawater. Similarly, while comparison of pH B5.3 leaching between three samples of identical age showed broadly similar solubility, there was also some variability (6 and 6%, 2 and 5% and 3 and 7%; Fig. 3). It is therefore possible that the full observed variability between samples in this study is not real; however, if anything, we would expect methodological concerns to result in underestimation of Fe solubility due to incomplete leaching or precipitation, which does not change our primary observation of high aerosol Fe solubility at times within LGM ice.

LGM aerosol Fe uxes at Dome C and Berkner Island. Decadal to centennial variability in both Fe solubility and total Fe uxes throughout the LGM at Dome C (Fig. 2a; 0.32 mg m 2 per year;

mean 0.760.4 mg m 2 per year) combines to mean that soluble Fe uxes show great variability (Fig. 2a; 0.010.84 mg m3 per year), with mean values of 0.090.17 mg m 2 per year (1 s.d.)

for pH B5.3 soluble Fe and 0.150.13 mg m 2 per year (1 s.d.) for seawater-soluble Fe. Despite the lower Fe solubility at Berkner

Island compared with Dome C, higher atmospheric Fe uxes at the LGM (0.950.07 mg m 2 per year), probably attributable to greater dust deposition owing to closer proximity to South

American dust sources18, result in similar mean soluble Fe uxes(0.030.02 mg m 2 per year) compared with those at Dome C. Total Fe uxes at Berkner Island were almost an order of magnitude lower during Antarctic Warm Events A1 and A2 than during colder periods (Fig. 2c), reecting lower regional dust uxes during warmer intervals15.

DiscussionThe aerosol Fe ux data presented here can be compared with previous ice-core estimates of Fe uxes to Antarctica and the nearby Southern Ocean, during the LGM. We nd that our total EDC aerosol Fe uxes (5.536 mmol m 2 per year; mean13.7 mmol m 2 per year) are considerably higher than previous estimates based on acidic (pH B12) leaching or mineral dust proxies at several Antarctic locations (4.38 mmol m 2 per year;

refs 22,23), including those based on acidic leaches at EDC with a conversion to total Fe (refs 19,22; 6.6 mmol m 2 per year). The difference may partially reect the high-resolution variability in

Fe ux at EDC, but also the fact that, similar to modern dust24, an acidic leach represents a variable portion of the total Fe, dependent on dust composition and leaching timescale. While pH B1 Fe EDC uxes are generally higher than estimates at pH 58 (Fig. 2a), the two ranges overlap, meaning that acidic leaches of ice core do not represent either total or soluble aerosol Fe, although they may capture broad-scale trends. This offset between acidic leaches and total digested Fe may be partially to do with methodology; the pH 1 EDC Fe records were measured in samples that were melted, acidied and only left to dissolve for 424 h (refs 19,20), while other studies have shown that Fe continues to dissolve at pH 12 for up to 3 months22,35. As a result, these short timescale acidic leaches in EDC ice samples may not have fully captured the pH 1 leachable fraction that is present. Previous comparison of pH 1 leached methods with digestion procedures on EDC ice suggested that 3065% of total Fe was recovered by pH 1 leaching methods19,21, but it seems likely that the proportion accessed by these short-term acidic leaches would vary with dust composition and mineralogy. Also, even with 13 month leaching times, acidic leaches are unlikely to fully dissolve dust particles.

Our data also allow evaluation of the use of other proxies for ice core Fe (Fig. 5). We nd that both nss-Ca11 (Ca with Na-calculated seasalt component removed) and total Al are good proxies for the ux of total Fe (Fig. 5). However, taking the EDC dust ux and calculating total Fe concentrations, assuming a constant crustal Fe percentage of B4% in dust18 underestimates total Fe for EDC samples, which are enriched in Fe (mean 8 wt%; Supplementary Data) compared with crustal values of 23% (ref. 36) or Patagonian topsoil (26%; ref. 37). EDC samples are similarly enriched for Al (mean 15 wt%; Supplementary Data), compared with crustal values of 8% (ref. 36) and Patagonian topsoil of 610% (ref. 37). This difference in the dust composition presumably relates to compositional-fractionation processes during transport, such as relative enrichment in silicate minerals, such as clays and feldspars33. As such, changes in composition and mineralogy may be spatially variable and site specic for ice-coring locations, complicating efforts to use dust, Ca or Al uxes as proxies for total Fe. In addition, despite the possible usefulness of these proxies for providing information about timing and magnitude of changes in atmospheric Fe11, the lack of relationship between total and soluble Fe (Fig. 5c) means that they do not approximate soluble or bioavailable Fe. Instead, direct measurements of soluble Fe, together with biological experiments, are needed for constraining changes in soluble and bioavailable Fe supply to the oceans during climatic change and especially within stable climate intervals. However, we do note the strong covariance between ice-core dust and Fe uxes and both Sub-Antarctic productivity and heavier Sub-Antarctic surface nitrogen isotope ratios (indicative of increased nutrient utilization) during the last 160 kyr (ref. 13), as well as a similar covariance between ice-core dust and (pH 1) Fe uxes, sediment Fe mass accumulation rates and Sub-Antarctic export productivity over the last eight glacial cycles5,38; these relationships suggest that both dust or (pH 1 leached) EDC

4 NATURE COMMUNICATIONS | 6:7850 | DOI: 10.1038/ncomms8850 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8850 ARTICLE

Fe solubility pH 5.3(%)Fe solubility pH 5.3(%)

Fe solubility pH 5.3 (%)

Fe solubility pH 5.3(%)

[dFe pH 5.3] (ng g

0 21 22 23 24 25 26

0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3

EDC3 age (kyr BP)

EDC dust 55 cm mean concentration (g g1)

Mean

0 21 22 23 24 25 26

0 20 40 60 80 100 120 140 160

Total Fe concentration (ng g1)

100

Fraction particle size (%) and Fe solubility (%)

Fe solubilitypH ~5.3

FPP (12 m) CPP (35 m)

10.0 9.5 9.0 8.5 8.0Dome C Temperature (C)

0 21 22 23 24 25 26

10.5

Figure 4 | Variability in EPICA Dome C over the LGM. (a) pH B5.3 dissolved Fe concentration, (b) pH B5.3 Fe solubility and (c) pH B5.3 Fe solubility shown with grain size (ne particle, FPP; coarse particle, CPP) from coultercounter analysis15,70. The relationship between pH B5.3 Fe solubility and (d)

dust concentration15, (e) Total Fe concentration (this study), (f) Dome CD air temperature68. Data points represent two measurements of a single sample, and errors are smaller than the size of points.

ice-core Fe uxes do capture the order of magnitude changes in Fe supply to the Southern Ocean that are important for understanding millennial-scale climate and pCO2 variability.

By comparison with estimates of Holocene Fe uxes to Antarctica22, our mean total Fe ux estimates (13.7 mmol m 2 per year) represent a B3050 increase over EDC

(0.30.4 mmol m 2 per year), and a B1020 increase over

coastal sites such as Law Dome (ref. 22; 0.81.1 mmol m 2 per year). The modern gradient in dust deposition between coastal sites and EDC is thought to be largely inuenced by wet deposition, which dominates at coastal sites and is a minor factor inland, and thus might be somewhat smoothed out during drier glacial intervals22. In terms of soluble Fe, the mean EDC seawater-soluble Fe ux of 2.7 mmol m 2 per year represents a 410-fold increase over modern soluble Fe uxes (even assuming

a similar high solubility in modern dust), and is B10 higher

than modern calculated Southern Ocean atmospheric Fe uxes(0.2 mmol m 2 per year; ref. 4). This suggests that the soluble atmospheric Fe ux to the Southern Ocean increased by at least a factor of 10 at the LGM. Such an Fe ux would be broadly equivalent to the expected upwelling supply of deep Fe, which would be reduced from calculated modern values of 816 mmol m 2 per year due to decreased circulation and stratication due to increased sea ice2,4. The range in soluble Fe uxes observed in this study (0.215 mmol m 2 per year) is also similar to modelled modern upwelling Fe ux values for entrainment (deep winter mixing; 9.133 mmol m 2 per year)

and vertical diffusion (1.23.6 mmol m 2 per year), which have been suggested to sustain surface water Fe concentrations in the modern Southern Ocean39.

NATURE COMMUNICATIONS | 6:7850 | DOI: 10.1038/ncomms8850 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8850

EDC

nss-Ca Total Ca

r2=0.857

r 2=0.652

nss-Ca Total Ca

r2=0.981

r 2=0.930

160

[Ca] (ng g1 )

120

50 100 150 200

0 0 10 20

30 40

300

EDC

r 2=0.973

r 2=0.992

250

[Total AI] (ng g1 )

[dFe (pH 5)] (ng g

80 70 60 50 40

200

150

100

0 50 100 150 200 250 300

0 30 40 50 60 70 80

0 0 10 20

80 70 60 50 40

0 0 30 40

0 10 20

EDC

1 )

40 80

0 120 160

Figure 5 | Direct comparison of different parameters as proxies for aerosol Fe in ice cores. Total Fe concentration, in EDC ice of LGM age (left panel) and Berkner Island (BI) ice from MIS 23 (right panel), is shown plotted against (a) nss-Ca and total Ca concentration, (b) total Al concentration and (c) pH B5.3 dissolved Fe concentration. Best-t linear regression lines with r2 values are shown. Data points represent two measurements of a single sample, and errors are smaller than the sizeof points.

[Total Fe] (ng g1) [Total Fe] (ng g1)

Of course, upwelling of deep water supplies both Fe and carbon, and it is the proportion of this carbon that is utilized by primary productivity in surface waters, and thus retained in the ocean, which determines the surface Southern Oceans role in setting pCO2. While the surface ocean is Fe-limited (insufcient bioavailable Fe to fully utilize all the upwelled nutrients), surface waters can act as a source for pCO2. Deposition of atmospheric dust only supplies Fe and thus provides the means to increase utilization of upwelled carbon and nutrients in surface waters. A full understanding of the role of atmospheric Fe supply in carbon cycling of the past thus requires not only knowledge of upwelling and dust deposition Fe uxes but also the resulting change in the carbon to bioavailable Fe concentration ratio in surface waters, as well as how both Fe and carbon are recycled and/or taken to depth. However, models do suggest that an increase in bioavailable atmospheric Fe supply of 510 could

double export production40, meaning that the LGM EDC soluble Fe ux could be large enough to drive big changes in primary productivity, nutrient utilization and carbon sequestration, especially in the Sub-Antarctic zone of the Southern Ocean9,13. Although a recent ocean Fe biogeochemical modelling study17 suggested that pCO2 only has 2 p.p.m.v. sensitivity to dust

globally, lower than the up to 40 p.p.m.v. suggested by recent Sub-Antarctic productivity/nutrient utilization records13, the model used a Fe solubility of only 2% (ref. 17), highlighting the need for constraints on soluble and bioavailable uxes during glacial intervals.

In conclusion, our EDC soluble Fe uxes suggest that soluble Fe uxes to the Southern Ocean during the LGM and glacial times were likely higher than previously considered. Both Fe content (8 wt%) and mean Fe solubility (pH B5.3 10%; seawater 17%) in EDC dust are greater than typically assumed by models (B110%; refs 17,18,4144), suggesting that models may underestimate soluble (and therefore bioavailable) Fe uxes to the Southern Ocean at the LGM. This might especially be the case in modelling studies where 0.52% Fe solubility is used17,4143. To address this, the median LGM pH B5.3 Fe solubility from this study (6%) is simplest for consideration in models, but the rapid variability in Fe uxes on short timescales and higher solubilities should also be considered. We attribute variability in aerosol Fe solubility to changes in the composition and uxes of atmospheric dust arriving at EDC, even on decadal timescales. Therefore, to fully capture the inuence of atmospheric Fe on glacialinterglacial cycling, models must take into account both temporal and spatial variability in dust composition and Fe solubility. Closer to source regions, higher dust uxes may offset lower Fe solubility, and short-term changes in the contribution of different dust types could have important effects on surface ocean biogeochemistry. We anticipate that these better general constraints on the soluble Fe ux to the Southern Ocean will be useful for modellers determining Fes role in marine biogeochemical cycles over glacialinterglacial changes, as well as providing insight into the response of the oceans to anthropogenic CO2 release.

Methods

General laboratory procedures and equipment cleaning. All ice-preparation was carried out in a cold room ( 20 C) at the British Antarctic Survey,

Cambridge, and all sample processing and analysis were carried out in clean laboratories at the University of Cambridge. Seawater analysis was carried out at Old Dominion University, USA. All clean work was carried out within a Class 1,000 HEPA-ltered clean laboratory and/or under Class 100 laminar ow air.

All water used was ultrapure (18.2 MOhm cm) and acids were either quartz-distilled (QD) to high purity from reagent grade or purchased as ultrapure (Fisher Scientic). Strict trace metal protocols were used, including wearing polyethylene (PE) gloves to minimize metal contamination during cleaning, processing and sampling. All plastic equipment was rigorously acid-cleaned following the procedures outlined below, which were based on previously published work45 or adapted from routine cleaning procedures in the Department of Earth Sciences at Cambridge.

New peruroalkoxy (PFA) or uorinated ethylene propylene (FEP) Teon lter rigs, vials, jars, bottles and micro centrifuge tubes were cleaned by sequential immersion in weak (B5%) Decon detergent (24 h), warm 50% (v/v) reagent-grade

HNO3 (2448 h) and warm 50% (v/v) reagent-grade HCl (2448 h). Teon equipment was re-cleaned by immersion in either warm 50% (v/v) HNO3 or HCl (24 h). All Teon equipment was thoroughly rinsed with ultrapure water following each step and then left to dry on a clean drying rack under laminar ow. For any Teon equipment that was to be in contact with ice or samples at near-neutral pH, the plastic was lled with ultrapure water to condition the surface. Nalgene low-density PE and polycarbonate (PC) bottles were cleaned following previously published cleaning methods for preparation of equipment for seawater sampling45: bottles were sequentially immersed in weak (B5%) Decon detergent (48 h), 1 M reagent-grade HCl (48 h), 6 M reagent-grade HCl (48 h) and B1% (v/v) QD-HCl (43 months), with extensive rinsing with ultrapure water between steps. Bottles were conditioned to neutral pH by lling with ultrapure water for 1 week before use. Whatman Nucleopore PC membranes for ltration (0.2 mm) were cleaned by immersion in 6 M QD-HCl (48 h) and then rinsed extensively with ultrapure water. Membranes were always handled with clean PFA tweezers to minimize contamination. Membranes were stored in and ushed with ultrapure water before use to condition the plastic to near-neutral pH for contact with ice samples. Pipette tips were either acid-cleaned45 with 50% (v/v) reagent-grade HCl (24 h) and rinsed with ultrapure water (seawater samples), or rinsed twice with 10% (v/v) QD-HNO3 and twice with ultrapure water before use (non-seawater samples).

For handling and decontamination of ice samples, Kyocera ceramic knives and polytetrauroethylene (PTFE) Teon boards were used. Knives were cleaned by

6 NATURE COMMUNICATIONS | 6:7850 | DOI: 10.1038/ncomms8850 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8850 ARTICLE

sequential immersion in 6 M reagent-grade HCl (24 h) and 1 M reagent-grade HCl (24 h), followed by extensive rinsing with ultrapure water. PTFE boards were cleaned by immersion in 6 M reagent-grade HCl and then rinsed extensively with ultrapure water. Arm-length PE gloves for handling ice were cleaned by immersion in weak (B0.01% v/v) QD-HCl (24 h) and rinsed thoroughly with ultrapure water to ensure no transfer of acid to samples.

Ice samples and cutting and decontamination. Detailed information about the age of ice sampled in this study is shown in Supplementary Data. EDC samples from 21 to 26 kyr BP were chosen to cover the very high dust portion of the LGM (B0.61.2 mg kg 1 dust; ref. 15) based on previous work15,46. A cuts (or B14 of the core) from 15 separate bags of ice (55 cm vertical section; B50 years) from the

EDC ice core were supplied by EPICA (see EPICA cutting plan in Supplementary Fig. 2a). For EDC sampling, B3 cm pieces were typically cut from either the bottom or top of each bag, using a steel bandsaw within the cold room ( 20 C) at

the British Antarctic Survey. Two vertical cuts were then made to create three subsamples of identical age (Supplementary Fig. 2b), which were then taken for decontamination and processing to generate subsamples of identical age for total or soluble analysis. The sampling resolution was therefore typically 50 years, although each sample represents only B3 years. Four bags (959, 984, 986 and 987) were sampled at higher resolution (two to four 3 cm subsamples per 50 years) to investigate higher-resolution variability as well as reproducibility between ice subsamples of identical age. Precise age for each 3 cm sample was calculated from depth within each bag and the EDC3 age scale47. Berkner samples were chosen to cover MIS 23, with two samples within the LGM (23 and 26 kyr BP) and 10 samples across the A events48 of MIS 3 (3650 kyr BP) based on the Berkner dD record (Mulvaney, unpublished data). Ice was supplied as 14 pieces of Berkner core (diameter 50 mm), with 3 cm vertical pieces cut and subsampled analogous to EDC samples.

The outside of ice core material is typically unavoidably contaminated with trace metals from drilling, cutting with bandsaws and handling with gloves49. Thus, the outside of the ice must be decontaminated to remove outer layers where contamination overwhelms the natural signal. Previously, this has been carried out using a melt-head50, polyethylene lathe51, stainless steel blade52, ceramic blade or chisels53. Decontamination may require removal of several cm of core22,51,52,54 during low-dust or less-compacted intervals, or when elements of interest are only present at pg g 1 levels. Preliminary testing in this study demonstrated that, as expected, ultrapure water ice was severely contaminated following cutting with a bandsaw (20 ng g 1 Ca, 4 ng g 1 Fe and 8 ng g 1 Al). The small size of ice samples used in this study prevented the use of techniques that remove several cm of core, and so we developed a technique using Kyocera ceramic blades that sequentially removed contaminated ice-core material. Each piece of ice was placed on a PTFE board and held with a clean PE glove, while the outside few mm were scraped off with a clean ceramic blade. This was repeated three times using separate clean equipment. To achieve the lowest contamination levels, all ice sampling equipment and ice were only handled with clean PE gloves, and all ice decontamination was carried out under Class 100 laminar ow air within the cold room ( 20 C) at the British Antarctic Survey.

Decontamination blanks were 47 pg g 1 for Fe and 116 pg g 1 for Al (o0.1% of natural conc.), and testing established that using acid-cleaned equipment had no effect on the natural pH of ice (pH B5.3). Our simpler technique did not achieve the full decontamination of more rigorous chiselling methods22 (46 pg g 1 c.f. 4 pg g 1 Fe). However, our method can be applied to smaller pieces of ice, and the procedure reduces Fe contamination by factor of

B100 , resulting in insignicant contamination that represents o0.1% of typical

natural Fe concentrations within ice from the high dust portion of the LGM.

Total digestion and pH B5.3 rapid-ltration procedures. Following decontamination, ice samples for total digestion were allowed to melt within PFA vials and acidied with 50 ml of QD-HNO3 (conc.) per 5 ml. A 5-ml aliquot was evaporated to dryness, reuxed overnight at B80 C with 100 ml QD-HNO3 (conc.) and 200 ml Teon-distilled conc. HF, evaporated to dryness, redissolved in 100 ml of 6 mol l 1 QD-HCl, reuxed at B80 C overnight and evaporated to dryness.

Finally, samples were redissolved in 250500 ml of 0.1 mol l 1 QD-HNO3 for analysis by inductively coupled plasma optical emission spectrometry (ICP-OES).

Procedural blanks (n 26) were 1.9 ng g 1 (Fe), 5.7 ng g 1 (Al) and 2.8 ng g 1

(Ca), all o0.5% of natural concentrations.

Following decontamination, ice samples for soluble analysis were processed by either rapid-ltration (melting at natural pH, referred to as pH B5.3) or seawater-leaching methods (see later sections). Ice samples for rapid ltration were allowed to melt at room temperature (B2 h) on a 0.2-mm Whatman polycarbonate membrane inside a 47-mm PFA Savillex lter rig, under vacuum. Rigs were covered to prevent contamination. Liquid that formed was quickly ltered and collected in a PFA reservoir containing an appropriate volume of conc. HNO3 to acidify samples to pH 0. On completion of melting, dust was leached with three 25 ml aliquots of ultrapure water, similarly acidied. This procedure, analogous to instantaneous leaches of modern aerosols27, was designed to minimize any FeOH3 precipitation or wall adsorption, which are concerns with ow-through and long leaching procedures (Supplementary Discussion). Acidied samples were evaporated to dryness and redissolved in 300 ml of 0.1 mol l 1

QD-HNO3 for analysis by ICP-OES. Fe concentration from the meltwater and subsequent ultrapure leaches were combined in nal calculations. Procedural blanks (n 78) were 17 pg g 1 (Fe), 244 pg g 1 (Ca) and 83 pg g 1 (Na), all

corresponding to 02% of natural concentrations.

Elemental analysis. Samples from total digestion or rapid ltration were analysed for dissolved Al, Ca, Fe and other cations using a Varian Vista ICP-OES atthe University of Cambridge. Each sample was analysed six times and a mean concentration was determined by reference to a quadratic calibration curve based on ve standards. Single-session accuracy on a 250-ng per gram internal multi-element reference standard for Al, Ca and Fe was 499.6% with precision of0.41% (1 s.d.; n 10). Long-term reproducibility of the standard was 1001% for

all elements (n 60). Detection limits for Fe were 0.10.4 ng g 1. We assign 5%

uncertainty on data to account for weighing, pipetting and %RSD of six ICP-OES analyses (typically o5%).

Sublimation. Sublimation allows the extraction of gases or dust from ice-core material while preventing melting, which could compromise the sample delity, and is achieved by lowering the pressure and temperature of ice below the triple point (0.01 C, 6.1173 mbar). Sublimation has been previously used to investigate concentrations and isotope ratios of gases (for example, CO2 and CH4) released from bubbles trapped in ice-core material5558, as well as having been utilized to cross-check CO2 measurements obtained using other techniques in Bern59. In the present study, we were motivated to develop a sublimation technique to extract dust from ice cores without exposing the dust to liquid water at a pH lower than that of seawater (pH B8), which may alter the speciation or solubility of Fe within dust before analysis (Supplementary Discussion). The major mode of deposition of dust at Dome C during the LGM is thought to be dry20, meaning that EDC dust extracted by sublimation should be representative of dry dust deposition at Dome C during the LGM.

We designed a sublimation apparatus based on previous studies on gases58, adapted to cleanly extract large masses of dust57 from ice without melting. The apparatus consists of a Schott borosilicate glass desiccator, sealed with Apiezon N vaccum grease, a vaccum pump with oil cleaner and pressure gauge and a 100 C

freeze trap for removal of water vapour (Supplementary Fig. 3). To allow exibility depending on the mass of dust/ice required by experiments, two congurations of the system were designed (Supplementary Fig. 3). In the present study, ice samples were placed directly on membranes within lter rigs within a Savillex PFA Teon Jar with top ports open (2550 g; Supplementary Fig. 3a). A second conguration, not used in the present study, allows for the collection of large dust masses, with larger pieces of ice (450 g) placed within 1 l bottles within a 4-l bottle with holes drilled in the lid (Supplementary Fig. 3b). Although ice sublimes under any pressures and temperatures below the triple point, previous studies have shown that quasi- and surface-melting can occur on ice or iceglass surfaces at temperatures ( 1 to 4 C) and pressures below the triple point6062. To avoid

this, it is necessary to keep the ice cold until the pressure is sufciently lowered, and to operate at pressures much lower than the triple point (o2 mbar). To achieve this, rst the glass desiccator was taken into a Class 100 laminar ow bench within the cold room and allowed to cool before being cleaned with propanol. Jars or bottles were then placed within the cleaned desiccator and the desiccator was closed and sealed before being transferred to the warm laboratory and connected to the pumps. This ensured that samples remained clean and that the ice remained cold for B10 min while the pressure was reduced below the triple point to the typical running pressure of 0.81.6 mbar.

Sublimation was allowed to proceed until all ice was gone, which was assessed visually or when the pressure had dropped to the background of the empty chamber (0.26 mbar). Rates of sublimation were relatively slow, dependent onice mass and volume, with a maximum rate of B250 g per 24 h. Small samples (510 g) sublimed in several hours, but were typically left overnight to ensure all ice was gone.

Once all ice had sublimed, the desiccator was sealed, disconnected and returned to the laminar ow bench to slowly allow clean air to enter the chamber; this prevented both dust being blown away by a sudden increase in pressure and the entry of dirty air from the pump. The bottles containing the dry dust samples were then carefully bagged and taken for seawater-leaching procedures (see next section). Blank contamination for the whole procedure over 24 hours was found to be acceptably low (B40 pg Fe per ice sample).

Seawater-leaching procedures and Fe analysis. Ice samples for seawater-soluble Fe analysis (seawater-leaching) were placed on a membrane on a PFA rig within a 1 l PFA jar within a glass desiccator at 20 C, and quickly placed cleanly under

vacuum (before moisture could form) at B18 C. Samples remained under vacuum until all the ice had sublimed, leaving dry dust on the membrane (see Sublimation section). The dust on the membrane was then leached with four 60-ml aliquots of low-Fe Ross Sea seawater at 2 C. The Ross Sea seawater used in leaching was previously collected63 from 10 m depth in the Southern Ross Sea (76 020 S, 169 530 E) under clean conditions in December 2003. The water was ltered with0.2 mm polypropylene cartridges and stored in 125 l polyethylene barrels, before being subsampled for this work using clean procedures63. We determined the

NATURE COMMUNICATIONS | 6:7850 | DOI: 10.1038/ncomms8850 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8850

background dissolved Fe concentration in this seawater to be 4.7 pg g 1(0.110.02 nmol l 1, 2 s.d.; n 4), consistent with previous determination63,64

of 0.070.02 nmol l 1. Each aliquot of seawater was allowed to leach the dust momentarily, vacuum was applied and then the liquid was pulled through and acidied to pH 1.6 with 120 ml of conc. ultrapure HCl. Samples were left acidied for 42 weeks before being analysed for dissolved Fe by FIA-8HQ, following published techniques45. This method has low detection limits (o5 pg g 1) and analytical precision is typically 510% during routine analysis; we therefore applied 10% uncertainty on seawater Fe concentration. Fe concentrations from four leaches were combined in nal calculations.

Berkner Deuterium and parameters and age scales. The ice core at Berkner Island was drilled to bedrock, and completed in 2005 (ref. 65 and Fig. 2). Deuterium (dD) ratios (expressed relative to the international standard Vienna

Standard Mean Ocean Water) were measured at the NERC Isotope Geosciences Laboratory using standard methods66 and have a typical precision of 1.0%. In

Fig. 2b, the Berkner dD record across A events A1 and A2 is shown overlain on the EPICA Dronning Maud Land (EDML) d18O record67, which is shown on the EDML1 age model that has been synchronized to EDC3 (refs 47,67). In the absence of an ofcial age model for this part of the Berkner Island ice core, Berkner dD is approximately visually tuned to the EDML d18O record so that EDML d18O, Berkner dD and Berkner Fe data are all shown on approximately the EDC3age scale. Dust and temperature in Fig. 2a are from the EDC core15,68, while atmospheric CO2 concentration (pCO2) in both Fig. 2a,b is taken from the composite CO2 record of Schilt et al.69

Fe solubility and ux calculation. Elemental uxes were calculated:

Fe flux mg m 2 per year

AR 1

where [Fe] is the concentration of Fe (mg kg 1) in the ice, AR is the calculated ice-accumulation rate (kg m 2 per year; ref. 47, Mulvaney, unpublished data).

Fe solubility was dened:

Fe solubility %

100 pH 5:3 or seawater soluble Fe

= total Fe

2 where [pH B5.3 or seawater-soluble Fe] and [total Fe] are the concentration in ice meltwater (ng g 1), combined from multiple leaches.

Calculating nss-Ca. nss-Ca in each ice-core sample was calculated in this study from measured dissolved Ca and Na concentration data using paired simultaneous equations50:

ss-Na

dNa

nss-Ca2

=Rt

nss-Ca2

dCa

Rm ss-Na

4 where ss denotes from seasalt, nss from non-seasalt sources, [dNa] dissolved Na concentration at pH B5.3, [dCa] dissolved Ca concentration at pH B5.3, Rt and

Rm are the Ca/Na ratios in crustal material (1.78) and seasalt (0.038), respectively50.

References

1. Martin, J. H. Glacial-interglacial CO2 change: The Iron Hypothesis. Paleoceanography 5, 113 (1990).

2. Sigman, D. M. & Boyle, E. A. Glacial/interglacial variations in atmospheric carbon dioxide. Nature 407, 859869 (2000).

3. Kohfeld, K. E. & Ridgwell, A. J. in Surface Ocean - Lower Atmosphere Processes. (eds le Qur, C. & Saltzman, E. S.) 251286 (AGU, 2009).

4. Watson, A. J. et al. Effect of iron supply on Southern Ocean CO2 uptake and implications for glacial atmospheric CO2. Nature 407, 730733 (2000).

5. Martnez-Garcia, A. et al. Southern Ocean dust-climate coupling over the past four million years. Nature 476, 312315 (2011).

6. Lthi, D. et al. High-resolution carbon dioxide concentration record 650,000-800,000 years before present. Nature 453, 379382 (2008).

7. Boyd, P. W. et al. Mesoscale iron enrichment experiments 1993-2005: synthesis and future directions. Science 315, 612617 (2007).

8. Anderson, R. F. et al. Wind-driven upwelling in the Southern Ocean and the deglacial rise in atmospheric CO2. Science 323, 14431448 (2009).

9. Sigman, D. M., Hain, M. P. & Haug, G. H. The polar ocean and glacial cycles in atmospheric CO2 concentration. Nature 466, 4755 (2010).

10. Adkins, J. F. The role of deep ocean circulation in setting glacial climates. Paleoceanography 28, 539561 (2013).

11. Rthlisberger, R. et al. Ice core evidence for the extent of past atmospheric CO2 change due to iron fertilisation. Geophys. Res. Lett. 31, L16207 (2004).

12. Ziegler, M., Diz, P., Hall, I. R. & Zahn, R. Millennial-scale changes in atmospheric CO2 levels linked to the Southern Ocean carbon isotope gradient and dust ux. Nat. Geosci. 6, 457461 (2013).

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

14. Jaccard, S. L. et al. Two modes of change in Southern Ocean productivity over the past million years. Science 339, 14191423 (2013).

15. Lambert, F. et al. Dust-climate couplings over the past 800,000 years from the EPICA Dome C ice core. Nature 452, 616619 (2008).

16. 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).

17. Tagliabue, A., Aumont, O. & Bopp, L. The impact of different external sources of iron on the global carbon cycle. Geophys. Res. Lett. 41, 920926 (2014).

18. Mahowald, N. M. et al. Atmospheric global dust cycle and iron inputs to the ocean. Global Biogeochem. Cycles 19, GB4025 (2005).

19. Gaspari, V. et al. Atmospheric iron uxes over the last deglaciation: climatic implications. Geophys. Res. Lett. 33, L03704 (2006).

20. Wolff, E. W. et al. Southern Ocean sea-ice extent, productivity and iron ux over the past eight glacial cycles. Nature 440, 491496 (2006).

21. Ruth, U. et al. Proxies and measurement techniques for mineral dust in Antarctic ice cores. Environ. Sci. Technol. 42, 56755681 (2008).

22. Edwards, R., Sedwick, P. N., Morgan, V. & Boutron, C. Iron in ice cores from Law Dome: a record of atmospheric iron deposition for maritime East Antarctica during the Holocene and Last Glacial Maximum. Geochem. Geophys. Geosyst. 7, Q12Q01 (2006).

23. Vallelonga, P. et al. Iron uxes to Talos Dome, Antarctica, over the past 200 kyr. Clim. Past 9, 597604 (2013).

24. Tian, Z., Ollivier, P., Vron, A. & Church, T. M. Atmospheric Fe deposition modes at Bermuda and the adjacent Sargasso Sea. Geochem. Geophys. Geosyst. 9, Q08007 (2008).

25. Morgan, B. & Lahav, O. The effect of pH on the kinetics of spontaneous Fe(II) oxidation by O2 in aqueous solution - basic principles and a simple heuristic description. Chemosphere 68, 20802084 (2007).

26. Spolaor, A. et al. Iron speciation in aerosol dust inuences iron bioavailability over glacial-interglacial timescales. Geophys. Res. Lett. 40, 16181623(2013).

27. Buck, C. S., Landing, W. M., Resing, J. A. & Lebon, G. T. Aerosol iron and aluminum solubility in the northwest Pacic Ocean: results from the 2002 IOC cruise. Geochem. Geophys. Geosyst. 7, 21 (2006).

28. Schroth, A. W., Crusius, J., Sholkovitz, E. R. & Bostick, B. C. Iron solubility driven by speciation in dust sources to the ocean. Nat. Geosci. 2, 337340 (2009).

29. Trapp, J. M., Millero, F. J. & Prospero, J. M. Trends in the solubility of iron in dust-dominated aerosols in the equatorial Atlantic trade winds: Importance of iron speciation and sources. Geochem. Geophys. Geosyst. 11, Q03014 (2010).

30. Baker, A. R. & Croot, P. L. Atmospheric and marine controls on aerosol iron solubility in seawater. Mar. Chem. 120, 413 (2010).

31. Basile, I. et al. Patagonian origin of glacial dust deposited in East Antarctica (Vostok and Dome C) during glacial stages 2, 4 and 6. Earth Planet. Sci. Lett. 146, 573589 (1997).

32. Journet, E., Desboeufs, K. V., Caquineau, S. & Colin, J.-L. Mineralogy as a critical factor of dust iron solubility. Geophys. Res. Lett. 35, L07805 (2008).

33. Glaccum, R. A. & Prospero, J. M. Saharan aerosols over the tropical North Atlantic - Mineralogy. Mar. Geol. 37, 295321 (1980).

34. Sugden, D. E., McCulloch, R. D., Bory, A. J.-M. & Hein, A. S. Inuence of Patagonian glaciers on Antarctic dust deposition during the last glacial period. Nat. Geosci. 2, 281285 (2009).

35. Koffman, B. G., Handley, M. J., Osterberg, E. C., Wells, M. L. & Kreutz, K. J. Dependence of ice-core relative trace-element concentration on acidication.J. Glaciol. 60, 103112 (2014).36. Taylor, S. R. & McLennan, S. M. The Continental Crust: its Composition and Evolution (Blackwell Scientic Publishing, 1985).

37. Gaiero, D. M. et al. Iron and other transition metals in Patagonian riverborne and windborne materials: geochemical control and transport to the southern South Atlantic Ocean. Geochim. Cosmochim. Acta 67, 36033623 (2003).

38. Martnez-Garcia, A. et al. Links between iron supply, marine productivity, sea surface temperature, and CO2 over the last 1.1 Ma. Paleoceanography 24,

PA1207 (2009).39. Tagliabue, A. et al. Surface-water iron supplies in the Southern Ocean sustained by deep winter mixing. Nat. Geosci. 7, 314320 (2014).

40. Ridgwell, A. J. & Watson, A. J. Feedback between aeolian dust, climate, and atmospheric CO2 in glacial time. Paleoceanography 17, 1059 (2002).

41. Moore, J. K., Doney, S. C. & Lindsay, K. Upper ocean ecosystem dynamics and iron cycling in a global three-dimensional model. Global Biogeochem. Cycles 18, GB4028 (2004).

42. Fung, I. Y. et al. Iron supply and demand in the upper ocean. Global Biogeochem. Cycles 14, 281295 (2000).

43. Tagliabue, A., Bopp, L. & Aumont, O. Evaluating the importance of atmospheric and sedimentary iron sources to Southern Ocean biogeochemistry. Geophys. Res. Lett. 36, L13601 (2009).

8 NATURE COMMUNICATIONS | 6:7850 | DOI: 10.1038/ncomms8850 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8850 ARTICLE

44. Zhang, Y. et al. Modeling the global emission, transport and deposition of trace elements associated with mineral dust. Biogeosci. Discuss. 11, 1749117541 (2014).

45. Sedwick, P. N., DiTullio, G. R. & Mackey, D. J. Iron and manganese in the Ross Sea, Antarctica: seasonal iron limitation in Antarctic shelf waters. J. Geophys. Res. 105, 11321 (2000).

46. Delmonte, B. et al. Comparing the Epica and Vostok dust records during the last 220,000 years: stratigraphical correlation and provenance in glacial periods. Earth-Sci. Rev. 66, 6387 (2004).

47. Parrenin, F. et al. The EDC3 chronology for the EPICA Dome C ice core. Clim. Past 3, 485497 (2007).

48. Blunier, T. & Brook, E. J. Timing of millennial-scale climate change in Antarctica and Greenland during the last glacial period. Science 291, 109112 (2001).

49. Boutron, C. F. & Batifol, F. M. Assessing laboratory procedures for the decontamination of Polar snow or ice samples for the analysis of toxic metals and metalloids. Ann. Glaciol. 7, 711 (1985).

50. Rthlisberger, R. et al. Dust and sea salt variability in central East Antarctica (Dome C) over the last 45 kyrs and its implications for southern high-latitude climate. Geophys. Res. Lett. 29, 1963 (2002).

51. Candelone, J. P., Hong, S. & Boutron, F. C. An improved method for decontaminating polar snow or ice cores for heavy metal analysis. Anal. Chim. Acta 299, 916 (1994).

52. Gabrielli, P. et al. Determination of Ir and Pt down to the sub-femtogram per gram level in polar ice by ICP-SFMS using preconcentration and a desolvation system. J. Anal. At. Spectrom. 19, 831837 (2004).

53. Chambers, R. Trace elements in an Antarctic ice core as potential indicators of marine biogenic activity (PhD Thesis, Univ. East Anglia, 2006).

54. Gabrielli, P. et al. Variations in atmospheric trace elements in Dome C (East Antarctica) ice over the last two climatic cycles. Atmos. Environ. 39, 64206429 (2005).

55. Wilson, A. T. & Long, A. New Approaches to CO2 Analysis in polar ice cores.J. Geophys. Res. 102, 2660126606 (1997).56. Wilson, A. T. Application of AMS 14C dating to ice core research. Radiocarbon 37, 637641 (1995).

57. Wilson, A. T. & Donahue, D. J. AMS carbon-14 dating of ice: progress and future prospects. Nucl. Instrum. Methods Phys. Res. 52, 473476 (1990).

58. Gllk, T., Slemr, F. & Stauffer, B. Simultaneous measurements of CO2, CH4, and N2O in air extracted by sublimation from Antarctic ice cores: conrmation of the data obtained using other extraction techniques. J. Geophys. Res. 103, 1597115978 (1998).

59. Siegenthaler, U. CO2 konzentrationsmessungen an polaren eisbohrkernen und tests mit einer neuen extraktionsmethode. Dip. Thesis (Univ. Bern, 2002).60. Elbaum, M., Lipson, S. G. & Dash, J. G. Optical study of surface melting on ice.J. Cryst. Growth 129, 491505 (1993).61. Furukawa, Y. & Ishikawa, I. Direct evidence for melting transition at interface between ice crystal and glass substrate. J. Cryst. Growth 128, 11371142 (1993).

62. Furukawa, Y., Yamamoto, M. & Kuroda, T. Ellipsometric study of the transition layer on the surface of an ice crystal. J. Cryst. Growth 82, 665677 (1987).

63. Garcia, N. S., Sedwick, P. N. & DiTullio, G. R. Inuence of irradiance and iron on the growth of colonial Phaeocystis antarctica: implications for seasonal bloom dynamics in the Ross Sea, Antarctica. Aquat. Microb. Ecol. 57, 203220 (2009).

64. Sedwick, P. N., Garcia, N. S., Riseman, S. F., Marsay, C. M. & DiTullio, G. R. Evidence for high iron requirements of colonial Phaeocystis antarctica at low irradiance. Biogeochemistry 83, 8397 (2007).

65. Mulvaney, R., Alemany, O. & Possenti, P. The Berkner Island (Antarctica) ice-core drilling project. Ann. Glaciol. 47, 115124 (2007).

66. Abram, N. J., Mulvaney, R. & Arrowsmith, C. Environmental signals in a highly resolved ice core from James Ross Island, Antarctica. J. Geophys. Res. 116, D20116 (2011).

67. Ruth, U. et al. EDML1: a chronology for the EPICA deep ice core from Dronning Maud Land, Antarctica, over the last 150 000 years. Clim. Past 3, 475484 (2007).

68. Jouzel, J. et al. Orbital and millennial Antarctic climate variability over the past 800,000 years. Science 317, 793796 (2007).

69. Schilt, A. et al. Atmospheric nitrous oxide during the last 140,000 years. Earth Planet. Sci. Lett. 300, 3343 (2010).

70. Delmonte, B. et al. Aeolian dust in East Antarctica (EPICA-Dome C and Vostok): Provenance during glacial ages over the last 800 kyr. Geophys. Res. Lett. 35, L07703 (2008).

Acknowledgements

We thank Mervyn Greaves, Jason Day, Susan Foord (deceased), Hazel Chapman, Paul Geissler and Jack Triest for their technical assistance with sampling and analysis, and Peter Sedwick for providing Ross Sea seawater and facilities for analysis of dissolved Fe in seawater samples. This work is a contribution to the European Project for Ice Coring in Antarctica (EPICA), a joint European Science Foundation/European Commission (EC) scientic programme. This study was funded by a Natural Environment Research Council (NERC) studentship to T.M.C., and E.W.W. is funded by a Royal Society professorship.

Author contributions

All authors devised the study; T.M.C performed sample processing and analysis, R.M. supplied samples and data, and the paper was written by T.M.C. and E.W.W. with contributions from other authors.

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 nancial 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: Conway, T. M. et al. Constraints on soluble aerosol iron ux to the Southern Ocean at the Last Glacial Maximum. Nat. Commun. 6:7850 doi: 10.1038/ncomms8850 (2015).

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/

NATURE COMMUNICATIONS | 6:7850 | DOI: 10.1038/ncomms8850 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

Word count: 9722

Show less

Abstract

Translate

Relief of iron (Fe) limitation in the Southern Ocean during ice ages, with potentially increased carbon storage in the ocean, has been invoked as one driver of glacial-interglacial atmospheric CO₂ cycles. Ice and marine sediment records demonstrate that atmospheric dust supply to the oceans increased by up to an order of magnitude during glacial intervals. However, poor constraints on soluble atmospheric Fe fluxes to the oceans limit assessment of the role of Fe in glacial-interglacial change. Here, using novel techniques, we present estimates of water- and seawater-soluble Fe solubility in Last Glacial Maximum (LGM) atmospheric dust from the European Project for Ice Coring in Antarctica (EPICA) Dome C and Berkner Island ice cores. Fe solubility was very variable (1-42%) during the interval, and frequently higher than typically assumed by models. Soluble aerosol Fe fluxes to Dome C at the LGM (0.01-0.84 mg m^-2 per year) suggest that soluble Fe deposition to the Southern Ocean would have been ≥10 × modern deposition, rivalling upwelling supply.

Details

Title

Constraints on soluble aerosol iron flux to the Southern Ocean at the Last Glacial Maximum

Author

Conway, Tm; Wolff, Ew; Röthlisberger, R; Mulvaney, R; Elderfield, He

Pages

7850

Publication year

2015

Publication date

Jul 2015

Publisher

Nature Publishing Group

e-ISSN

20411723

Source type

Scholarly Journal

Language of publication

English

DOI

https://doi.org/10.1038/ncomms8850

ProQuest document ID

1698024583

Constraints on soluble aerosol iron flux to the Southern Ocean at the Last Glacial Maximum

Jump to:

Full Text

Abstract

Details

Suggested sources