Pb and Ba concentrations and Pb isotopes
Elemental concentrations and Pb isotopic compositions measured in the innermost parts of individual samples are illustrated in Figure 1, together with the profiles of dust flux or deuterium (δD) as a function of the age of the ice, and all data are listed in Table S1. Figure 1 also shows published post-MBE data for the EDC ice core samples, dated from 2 kyr (MIS 1) to 220 kyr B.P. (MIS 7.3), by Vallelonga et al. (2010)18.
Both the Pb and barium (Ba) concentrations show strong variability, with mean concentrations of Pb and Ba that are approximately 7 times higher during cold periods (δD < –405‰) than during interglacials (δD > –405‰) (Table S1), primarily dependent on the dust fluxes (Figures 1 and S1 of the supporting information). Note that glacial and interglacial periods were defined with a threshold δD value (–405‰), below which Antarctic temperature and dust flux show a clear correlation5,23. The Pb concentrations are positively well correlated with Ba (a conservative crustal reference element) during colder periods (δD < ~–420‰) with Pearson’s correlation coefficient of 0.924 that is significant at the 0.01 significance level (2-tailed) (p = 0.01), while there is a lack of significant correlation during less cold periods (δD > ~–420‰) (Pearson’s correlation coefficient of 0.344 at p = 0.117) (Figure S2). This reflects that dust was the main source of Pb in the EDC ice during cold climatic conditions21,24.
Furthermore, the Pb concentrations show a sharp decrease when the δD values increase during colder glacials and remain very low when the δD values are above –420‰ (Figure S3). The very high Pb concentration (108 pg/g) in the deepest ice at 3,189.45 m (sample no. 40, ~801 kyr B.P., MIS 20.2) greatly exceeds the range of measured concentrations (< 20 pg/g) when the δD values are below –430‰ (Figures 1 and S3). Despite the stratigraphic continuity of multi-parametric climatic records above 3,200 m as described previously2, the very high Pb content may suggest non-climate-related influences, possibly bedrock, and thus the sample is not included when interpreting climate signals. The Pb concentrations of sample nos. 25, 27, and possibly 20 also show apparent deviations from concentration levels during the intermediate and interglacial climatic stages (Figure S3), possibly due to Pb contributions from large volcanic eruptions25.
The Pb isotope ratios vary from 1.1824 to 1.2332 for 206Pb/207Pb and from 2.4556 to 2.4939 for 208Pb/207Pb with mean values of 1.2022 and 2.4715, respectively (Table S1). Temporal changes in the pre-MBE 206Pb/207Pb ratios are less well characterized by temperature in Antarctica, while the post-MBE values during 2–220 kyr B.P. show a general trend associated with climatic conditions with higher values during warm or less cold periods and lower values during very cold climatic stages (Figure 1). This difference may be partly due to larger age intervals (~90–250 years) integrated by individual pre-MBE samples relative to the post-MBE ones (~10–60 years), resulting in the smoothing of pre-MBE source-specific Pb isotopic signatures.
Comparison of Pb isotopic compositions before and after the MBE
Figure 2 shows the plot of 206Pb/207Pb versus 208Pb/207Pb for the pre-MBE and post-MBE glacials and interglacials. A perspective to consider is that natural Pb in Antarctic ice could originate from both dust and volcanoes17,18,26. An estimate of the dust contribution was made based on the Pb/Ba ratio of upper continental crust, ~0.0318,27. According to the fraction of Pb of dust origin, isotopic signatures in each sample were divided into dust-dominant Pb (> 60% with Pb/Ba < 0.05) and non-dust dominant Pb (< 60% with Pb/Ba > 0.05), as done for post-MBE isotopic signatures18, which facilitated comparison of the isotopic compositions between the pre-MBE and post-MBE periods. A smaller estimated dust contribution indicates an increased non-dust (that is, volcanic) contribution.
In Figure 2, the pre-MBE Pb isotopic compositions show the distribution of the wide ranges of 206Pb/207Pb and 208Pb/207Pb with a general trend of less radiogenic compositions for dust-dominant Pb during glacial periods. Recently, Gili et al. (2016)19 constrained well-defined potential source areas (PSAs) of dust in southern South America (SSA) by coupling Pb isotopic compositions in the EDC samples from Vallelonga et al. (2010)18 with new Pb isotopic data from unexplored PSAs in SSA. In this study, however, Pb isotopic data by Vallelonga et al. (2010)18 were not categorized into dust-dominant and non-dust dominant Pb for individual samples. In Figure 2, the post-MBE dust-dominant Pb isotopic signatures, combined with the PSAs fields defined by Gili et al. (2016)19, indicate the central-western Argentina (CWA) and Patagonia as the primary sources of dust in East Antarctic ice during glacial periods. This is compatible with previous Sr-Nd isotopic constraints on the Antarctic dust provenance of Patagonia9–16 and CWA16 during the post-MBE glacials. Gili et al. (2016)19 suggested the Puna-Altiplano Plateau (PAP) in the Andean Cordillera as the secondary source during glacials, based on a new perspective of Pb isotopic compositions falling on a distinct isotopic field of the PAP, characterized by higher 208Pb/207Pb relative to a given 206Pb/207Pb (Figure 2), consistent with the Sr-Nd isotopic constraints15,16,28. In Figure 2, part of the post-MBE glacial dust-dominant Pb isotopes also shows a shift toward the PAP field, supporting previous suggestions of the PAP provenance of glacial dust to the EAP.
Compared to the post-MBE glacial Pb isotopic compositions, the pre-MBE glacial dust-dominant Pb isotopes are characterized by lower radiogenic values (Table S2), exhibiting a tendency (except for sample no. 27, ~712 kyr B.P., Figure S4) to converge toward the partially overlapping field between Patagonia, Tierra del Fuego (TdF), and the southern and middle CWA (S-CWA and M-CWA, respectively) (Figure 2). Considering the Sr-Nd isotopic evidence for a dominant dust contribution from northern Patagonia relative to southernmost Patagonia (including TdF) in the post-MBE cold climates16, the pre-MBE glacial dust Pb isotopic compositions emerging within the overlapping field between Patagonia and TdF may be signals of northern Patagonia. As a result, the ratios distributed in a relatively narrow range between Patagonia, S-CWA, and M-CWA are thought to be a consequence of equally significant contributions from these potential dust sources to the EAP glacial dust during the pre-MBE glacials. The isotopic signature of sample no. 27 moves toward the most radiogenic McMurdo-Erebus volcanic field (Figure S4), likely associated with the effects of volcanic eruptions, probably Antarctic, as mentioned before (Figure S3). Interestingly, our new dust isotopic data show no dust transport from the PAP to the EAP during glacials prior to the MBE, which contrasts with previous observations for the post-MBE glacial dust as described above.
The pre-MBE interglacial dust-dominant Pb isotopes distribute within the Patagonian field (sample no. 11) or the overlapping field between Patagonia and TdF (sample nos. 3 and 34) (Figures 2 and S4), limiting the discrimination between the two potential sources. However, the isotopic ratios in the interglacial samples (nos. 1, 2, and 4) with relatively high fraction of dust Pb (~55 to 58%) show the signatures of Patagonia (Figures 2 and S4), providing insights into dominant dust supply from Patagonia during the pre-MBE interglacials. The isotopic composition plotted within the M-CWA field that partially overlaps with the southern Puna field is observed for sample no. 38 (Figure S4). Because the other pre-MBE interglacial and glacial isotopic data of dust-dominant Pb tend to fall on the Patagonian and CWA fields above the PAP compositional trend (Figure 2), this isotopic composition may reflect the M-CWA isotopic signature. A single data point would restrict the interpretation of the M-CWA dust provenance. However, the pre-MBE glacial dust-dominant isotopic compositions for sample nos. 26 and 30, with respective δD values of –407 and –411‰ near to the interglacial threshold δD value (–405‰) (Table S1), distribute within the M-CWA field (Figures 2 and S4). Taken together, our Pb isotope data suggest that the most important dust source was Patagonia, with an additional contribution from CWA, during the pre-MBE interglacials, although further studies with larger sample sizes are required to ensure the validity of our findings. For comparison, previous isotopic and geochemical studies suggested multiple sources of Patagonia14,16,19,29, CWA29, PAP16,19, and Australia13,14,30,31 for the post-MBE interglacial dust in the EAP ice.
A large difference in the isotopic compositions before and after the MBE is also observed for non-dust dominant Pb. In Figure 2, both the post-MBE glacial and interglacial isotopes of non-dust dominant Pb vary over a wide range of the mixing line between the non-radiogenic Patagonian dust field and the very radiogenic McMurdo-Erebus volcanic field. In contrast, the pre-MBE non-dust dominant Pb isotopes remained less radiogenic, except for four samples (sample nos. 20, 28, 29 and 40, Figure S4), relative to the post-MBE values (Figure 2 and Table S2). The distinctive non-radiogenic compositions of the pre-MBE volcanic Pb are likely due to extra-Antarctic volcanic contributions in association with the reduction of wet removal efficiency, coupled with an equatorward shift of the SWW belt (see next sections).
Dust provenance and its relevance to a shift of the SWW before the MBE
New isotopic evidence for Patagonian dust in the EDC ice core during the pre-MBE glacials indicates that Patagonia was the major dust source throughout the glacial periods over the past 800 kyr, consistent with previous Sr-Nd isotopic fingerprints of Patagonian glacial dust in the EAP ice prior to the MBE13,28. This suggests that the SWW remained over Patagonia (particularly northern Patagonia) under both pre-MBE and post-MBE Antarctic cold climates.
The dominance of Patagonian dust during glacial periods over the past 800 kyr would be linked to the glacial advances of the Northern and Southern Patagonian Ice Sheet (PIS), stretched from 37°S to 56°S, and the associated increase of fluvioglacial outwash deposits, allowing the enhanced dust entrainment associated with an increase in the vigour of atmospheric circulation12,32,33. Despite the lack of detailed records of glacier fluctuations in the PIS far back to MIS 20, the PIS glaciers likely reached its greater extent during glacial periods since the Great Patagonian Glaciation (~1 Ma ago)34. This suggests that the periodic expansion of the PIS glaciers in response to cold climate conditions, together with the northward expansion and intensification of the SWW6,16,33,35, may have played an important role in the persistence of Patagonian dust provenance in the EAP ice during both the pre and post-MBE glacial periods. Along with an important role of the PIS, the persistent Patagonian glacial dust would be attributed in part to the larger increase of the exposed Argentine Shelf during the relative sea level low-stands between –80 m and –150 m below present sea level12,33,36, when fine-grained shelf sediments were delivered from Patagonia37, which in turn retained a dominant Patagonian dust signature33.
Apart from a dominant and persistent Patagonian origin of glacial dust, our isotopic signatures indicate that CWA (S-CWA and M-CWA) emerges as an important dust supplier to the EAP during the pre-MBE glacials, as in the Sr-Nd isotopic constraints on the CWA origin of dust in recent glacial periods16. This suggests a northward extension of the SWW belt between northern Patagonia (39°–42°S) and CWA (27°–39°S), far north of its present position of the strong zonal winds between ~45°S and ~60°S38, during both pre-MBE and post-MBE glacial periods. Although an equatorward displacement and strengthening of the SWW belt during glacial periods remains under debate (e.g., Kohfeld et al., 201339 and references therein), our explanation is consistent with the hypotheses that during the glacial conditions, an equatorward shift in the SWW was as large as 7–10° relative to its interglacial position6 and the strengthening of the northern margin of the SWW occurred at 33°–40°S35. The northward shift of the northern edge of the SWW belt would have induced more vigorous northwesterly winds and the consequent increase in dust emissions over S-CWA and M-CWA16, coupled with drier conditions in the SSA north of 40°S39,40, enhancing the input of pre-MBE glacial dust from these areas to the EAP.
Another prominent feature of pre-MBE glacial isotopic compositions is the absence of a PAP dust signature, which contrasts with the hypothesis of its potential contribution during the post-MBE glacials, attributed to an equatorward movement of the subtropical westerly jet stream (SJT) over the PAP, a high elevation basin (~4,000 m a.s.l.)15,16,19,28. We attribute the non-contribution of PAP to the pre-MBE glacial dust in the EAP to environmental conditions that reduced either dust production in the PAP or efficient transfer of PAP dust to the EAP under cold climates of Antarctica prior to the MBE, probably linked to shorter pre-MBE glacials than younger ones, which consequently reduced the dust productivity in the major dust sources at lower latitudes in the Southern Hemisphere41. Although hypothetical, a less pronounced increase in the EDC glacial dust fluxes prior to the MBE (notably MIS 16, 18 and 20) relative to the post-MBE glacials (e.g., MIS 8 and 10), as noted previously36, would be partly due to the absence of any glacial dust supply from PAP (Figure 1), assuming the similarity of a strengthening of SSA PSAs and the transport efficiency of dust from SSA PSAs to the EAP between the pre-MBE and post-MBE glacial conditions5. A tentative assessment of the relative dust contribution in the EDC glacial ice, using the Nd isotopic composition between a Patagonian and a PAP end-member, suggested that when glacial dust input to the EAP increased, the contribution from Patagonia decreased, while the contribution from a secondary source, PAP, increased, and vice versa, during Pleistocene glacial times28, supporting the above hypothesis.
During the pre-MBE interglacials, our Pb isotopic signatures characterize a dominance of dust from Patagonia with the existence of additional input of dust from CWA, suggesting favorable atmospheric circulation for the persistence of dust production and transfer of dust from these source regions. Patagonia is situated in a large region in SSA, extending from ~39°S down to the southern tip (~55°S) of South America including TdF in the southernmost part, and the climate is currently controlled by the dynamics of the SWW belt16. Satellite and surface observations near the sources and in Antarctica, combined with model simulations, identified modern dust transport from the southern Patagonia and TdF as far south as ~55°S to the EAP in austral summer42. Compared to this current feature of the southernmost dust source, the dominant contribution of dust from Patagonia, excluding TdF, during the pre-MBE interglacials may reflect slightly northward shifted SWW belt under cooler pre-MBE interglacial climates. Meanwhile, the presence of dust from CWA suggests an influence of the SWW belt exerting over this region, following a systematic movement of the SWW with the mean climate6,43. Today, the atmospheric circulation in CWA is dominated by the northeasterly wind in austral summer and northwesterly wind in austral winter44, and by the katabatic wind (called Zonda wind), an extremely dry wind blowing from west to east, injecting dust aloft, from May to August16. Taken together, our isotopic constraints on the provenance of dust in the pre-MBE interglacial ice can be explained by a northward shift and/or extension of the SWW belt in response to cooler pre-MBE interglacial climates relative to the post-MBE interglacials. This may have enhanced dust supply and strong wind uptake as a consequence of more frequent cyclonic influence and more steady zonal winds in SSA as in the case for winter conditions in the current climate45. Interestingly, the EDC ice core record shows a substantial increase of dust fluxes during cooler pre-MBE interglacials (MIS 15.1, 15.5, 17.3 and 19.3), compared to the warmer post-MBE interglacials (MIS 1 and 5.5)3,5: the average (± SD) dust flux during the pre-MBE MIS 15.1, 15.5, 17.3 and 19.3 interglacials is 0.85 ± 0.61 mg/m2/yr, which is ~ 2 times the average flux (0.45 ± 0.20 mg/m2/yr) of the post-MBE MIS 1 and 5.5 interglacials. Furthermore, the EDC interglacial ice CO2 levels prior to the MBE were ~30–40 ppm lower than those after the MBE46,47. This may be partly related to the weakened strength of the Antarctic Circumpolar Current in response to a northward shift in the mean position of the SWW, resulting in reduced ventilation of respired CO2 in the deep ocean to the atmosphere6,7. However, any speculation must be treated cautiously due to the lack of a comprehensive understanding of processes and mechanisms for the lower interglacial CO2 levels prior to the MBE48.
Pre-MBE volcanic isotopic signatures and atmospheric implications
An interesting difference in the Pb isotopic signatures between the pre-MBE and post-MBE periods is observed when the 206Pb/207Pb ratios are plotted as a function of relative dust-Pb proportions (Figure 3). The post-MBE 206Pb/207Pb ratios with a dust-Pb fraction of < 60%, reflecting an increase of volcanic Pb source, show a clear compositional trend moving toward the most radiogenic McMurdo-Erebus volcanic field. This trend was previously interpreted as a result of an increasing mixture of volcanic Pb from degassing volcanoes within Antarctica18,19. For comparison, the pre-MBE 206Pb/207Pb ratios show no distinct increasing trend (except for sample nos. 20, 28, 29 and 40) (Figures 3 and S5), approaching the isotopic compositions characterized by extra-Antarctic volcanoes north of 60°S (Figure S6).
A possible clue to explain this difference can be seen from the climatic conditions for pre-MBE and post-MBE individual data. In Figure 3, the post-MBE volcanic Pb appears to increase during very cold (δD < –425‰) or intermediate (–405‰ < δD < –425‰) climates, while the pre-MBE samples typically show an increasing volcanic contribution mostly during warm (δD > –405‰) or intermediate climates. Note that the pre-MBE samples (nos. 19, 24 and 25) classified into the intermediate climate have δD values between –408 and –411‰, (Table S1). Discrete volcanic tephra layers identified in the EDC ice core do not match with the post-MBE samples18,49 and they tend to disappear in the EDC ice older than 358 kyr BP, probably due to the variations in the frequency of large volcanic eruptions50. This suggests no impact of sporadic explosive volcanic eruptions on the compositional trend of volcanic Pb isotopes observed for the pre-MBE periods, except for sample nos. 20, 25 and 27, showing abnormally enhanced Pb concentrations, probably due to large volcanic eruptions (Figure S3). The trend of climate-related pre-MBE 206Pb/207Pb ratios does not change after excluding these three data points (Figure S5).
Given the above, the a priori guess would be that the dominant isotopic signatures of Antarctic volcanoes during the post-MBE glacials are likely due to a stronger dynamical isolation of the polar vortex area over Antarctica under cold climate conditions12, reducing the advection of extra-Antarctic volcanic degassing Pb to the EAP through the low levels of troposphere, which differs from the dust transport from South American sources via the upper tropospheric pathways5,23. Interestingly, the pre-MBE isotopic compositions for sample nos. 28 and 29 with much lower δD values (–438‰ and –422‰, respectively) show an approach to the McMurdo-Erebus volcanic signature (Figure S5), supporting the above speculation. On the other hand, the lack of extra-Antarctic volcanic Pb signature during the post-MBE interglacials is most likely due to the enhanced en route wet removal resulting from either a more intense hydrological cycle5,23,39,40 or the poleward shift of intense cyclones in the mid-latitudes under warm climate conditions51. This hypothesis, however, contradicts a previous study proposing that high super-chondritic (volcanic) platinum (Pt) and iridium (Ir) fluxes in the EAP during the post-MBE interglacials (MIS 1 and 5) were probably due to the enhanced advection of air masses from lower latitudes to the EAP according to a weakened polar vortex52. This contradiction may arise from differences in the removal efficiency of individual volcanogenic metals from the atmosphere primarily via wet deposition processes of within-cloud scavenging and below-cloud scavenging (wash-out). The modeling study of volcanic emissions showed that volcanic PbCl2, a highly soluble species enriched in degassing volcanic emissions, decreased exponentially in the atmosphere with the distance from the volcano due to the rapid wet deposition53. In contrast, refractory and chemically inert Ir and Pt that are enriched in volcanic emissions in association with the high fluorine content, compose water insoluble fluoride species (e.g., iridium hexafluoride, IrF6)54, which would have the potential for further atmospheric transport over long distances. Recent studies identified well-defined Pt and Ir peaks in the volcanic layers of Antarctic and Greenland snow deposits without Pb enrichments coincident with these peaks24,55,56, supporting our hypothesis.
As a result, we can infer that extra-Antarctic volcanic Pb signatures in the pre-MBE interglacial climates are most likely related to a longer lifetime for volcanic aerosols from degassing volcanoes, associated with the reduction of wet removal efficiency. Together with a prolonged lifetime of volcanic aerosols, a northward shift and/or extension of the SWW belt during cooler pre-MBE interglacials could have enhanced an advection of volcanic Pb from volcanoes outside Antarctica to the EAP. Our results point to a substantial weakening of the hydrologic cycle in line with cooler sea surface temperature of the Southern Ocean during pre-MBE lukewarm interglacials relative to the post-MBE interglacials8,57.