A Radiometrically Dated Record of Antarctic Ice Sheet Response to Millennial-Scale Climate Cycles during Glacials and Interglacials

Gavin Piccione (  gpiccion@ucsc.edu ) University of California Santa Cruz https://orcid.org/0000-0002-7637-9796 Terrence Blackburn Santa Cruz Slawek Tulaczyk University of California, Santa Cruz https://orcid.org/0000-0002-9711-4332 Troy Rasbury Stony Brook University https://orcid.org/0000-0003-0591-4461 Mathis Hain University of California, Santa Cruz https://orcid.org/0000-0002-8478-1857 Daniel Ibarra Brown University https://orcid.org/0000-0002-9980-4599 Katharina Methner Senckenberg https://orcid.org/0000-0002-3753-8256 Chloe Tinglof University of California Santa Cruz Brandon Cheney University of California Santa Cruz Paul Northrup Kathy Licht Indiana University-Purdue University


40
The key link between this ocean thermal forcing and ice sheet mass lies in the delivery of heat to the marine-41 terminating ice sheet margins, affecting ice shelves and grounding lines. Ice sheet stability is regulated by deliver CDW to the base of ice shelves and grounding lines, triggering enhanced basal melting and retreat 66 during SH millennial warm periods 5 .

67
Although ice sheet models 10,18 and modern observations 2-4 indicate that the AIS is susceptible to 68 ice loss through ocean thermal forcing, regional differences in ice bed topography, drainage geometry, and 69 ice thickness 19 in peripheral sectors of Antarctica may lead to geographic differences in grounding line 70 vulnerability, adding spatiotemporal complexity to ice sheet response. Millennial-scale SO upwelling also 71 varies in intensity depending on the background climate state, with the slowest AMOC causing increased 72 SO upwelling during glacial terminations, and relatively strong AMOC resulting in weaker SO upwelling 73 during interglacial periods 20,21 . Therefore, geologic evidence of AIS evolution across a wide geographic 74 range and diverse climate states is necessary to ground truth simulations of suborbital changes in ice mass.

75
However, existing geologic records documenting millennial-scale AIS mass loss 22-24 are limited to bipolar 76 seesaw events during the last two glacial terminations, are constrained by low-resolution age models, and 77 are restricted spatially to ice shelf systems and offshore sediments. This leaves the regional extent and 78 magnitude of AIS response to suborbital climate change unconstrained. Here, we present observations from 79 an archive of subglacial hydrologic evolution recorded by chemical precipitates that formed >900km apart 80 beneath the EAIS over a combined >100kyr period during the Pleistocene. This dataset provides a novel 81 sequence of high-resolution U-series age constraints on ice sheet evolution in response to millennial-scale 82 climate change. Mineralogic and geochemical variations in subglacial precipitates provide evidence for 83 basal hydrologic changes that are correlated with bipolar seesaw-related SH climate cycles. Combining 84 precipitate data with a reduced-complexity model of ice sheet thermodynamics, we demonstrate a link 85 between subglacial hydrologic conditions and millennial-scale changes in ice sheet velocity.

89
In this study, we report geochronological and geochemical results collected from two subglacial 90 precipitates: PRR50489, which formed in the David Glacier catchment and was collected at Elephant 91 Moraine; and MA113, which formed beneath ice feeding Law Glacier and was collected at Mount Achernar 92 moraine (Fig. 2). These samples were precipitated over tens of thousands of years in subglacial aqueous 93 systems on the EAIS side of the Transantarctic Mountains (TAM), and were subsequently eroded and 94 transported in exhumed sections of basal ice to be finally deposited on the surface within supraglacial moraines 25 . PRR50489 and MA113 are 3 and 9cm thick respectively, with alternating layers of calcite and 96 opal (Extended Data Fig. 1) implying cyclic change in the subglacial environment.

97
We measured 234 U-230 Th ages on eleven opal layers from PRR50489 that constrain the timeframe 98 of precipitation from 230 to 147ka (Fig. 1i), and ten opal and calcite layers from MA113 ranging in age 99 from 55 to 42ka (Fig. 1f). We construct a stratigraphic age model for each sample using a Bayesian Markov 100 chain Monte Carlo model in which the principal of superposition is imposed on each dated layer to refine 101 age estimates based on stratigraphic order 26 (Extended Data Fig. 2). Depth profiles of Si and Ca 102 concentration collected using Energy Dispersive X-ray Spectroscopy provide a continuous representation 103 of sample mineralogy: with high Ca areas representative of calcite and low Ca areas representative of opal.

104
We pair stratigraphic age models with Ca concentration spectra to create timeseries describing the 105 oscillations of precipitate mineralogy (Fig. 1a,d). These mineralogic timeseries reveal a temporal cyclicity 106 in opal deposition, with opal layers in PRR50489 precipitated every 8-10kyr between marine isotope stages 107 (MIS) 7 and 6, and opal layers in MA113 precipitated every 2-4kyr during MIS 3 (Fig. 1a, d). To investigate 108 a possible link between cycles of precipitate mineralogy and climate, we compare mineralogic timeseries 109 for each precipitate with climate proxies in both Antarctic 27,28 (Fig. 1b,e,f) and Greenland 29 (Fig. 1g) ice 110 cores. Comparison between Ca-spectra and Antarctic temperature proxies reveals a highly regular linear 111 relationship between climate cycles and precipitate mineralogy, with calcite formation (high Ca wt%) 112 during warm AIM peaks, and opal formation (low Ca wt%) during Antarctic cold periods (Fig. 1). Moving 113 window correlation calculations between MA113 and PRR50489 mineralogic spectra and SH climate 114 proxies from Antarctic ice core records demonstrate a significant temporal correlation between the two 115 records throughout the sampling period (Extended Data Fig. 3). This synchrony indicates that bipolar-116 seesaw-driven climate change trigger variability in EAIS basal environments.

119
To understand the link between ocean-atmosphere-cryosphere teleconnections and the mineralogic 120 composition of subglacial precipitates, we first utilize geochemical and isotopic measurements to

138
Calcite layers in both precipitates exhibit trends in δ 13 C and δ 18 O compositional space that suggest 139 mixing between two isotopically distinct fluids with different carbon concentrations (Fig 3). To match the 140 trends in the calcite data for PRR50489 and MA113, the δ 13 C and δ 18 O depleted, endmember calcite forming 141 water must have 40-fold and 5-fold higher of the total carbon concentration respectively, relative to the 142 low-carbon endmember opal forming water. Though δ 13 C of the opal forming water cannot be directly 143 measured, for the mixing curve to fit calcite compositions and opal δ 18 O values requires the opal forming 144 endmember to be isotopically heavy (δ 13 C >-5 ‰): a composition comparable to that of sub-AIS brines 36,37 .

145
Similar mixing relationships are observed between the 87 Sr/ 86 Sr and δ 18 O composition of opals and calcite ( Fig. 3c,d), requiring endmember waters to be distinct in both Sr concentration and isotopic composition.

147
In both samples, the opal forming waters have more radiogenic (higher) 87 Sr/ 86 Sr and heavier δ 18 Ο than the 148 calcite forming waters. Sr:O mixing curves show that 20-fold and 50-fold of the total Sr in the system 149 originates in the opal forming endmember (Fig. 3c, d). Due to their similar geochemical behavior, strontium

181
To test if endmember mixing is a plausible mechanism for the observed opal and calcite layers, we

201
The key finding from our subglacial precipitate archive is that millennial-scale ocean-atmosphere-

282
The basal thermal regime of AIS outlet glaciers is highly complex, with models demonstrating 283 along-flow transitions between frozen and unfrozen basal conditions resulting from variations in bed 284 topography, ice thickness, and flow rate along the ice sheet path 58 . While we acknowledge that localized 285 basal temperature change could affect precipitate mineralogy, the collective geochronological and 286 geochemical dataset presented here strongly favors hydrologic cycles driven by regional, rather than local, 287 ice response. The consistent relationship between subglacial transitions from freezing to melting recorded 288 at two different locations requires a highly regular triggering mechanism that is linked to the cryosphere 289 and to the broader climate system. On both a glacier and regional scale, fluctuations between basal freezing  (Fig. 3) suggests that ice sheet acceleration in response to ice shelf perturbation enhances hydrologic connectivity between subglacial waters separated by hundreds of kilometers. Given these 300 spatiotemporal constraints, we conclude that opal-calcite transitions in subglacial precipitates result from 301 millennial-scale migration of the regional freeze-melt boundary beneath grounded ice around the Ross 302 Embayment.

303
Subglacial water has a significant effect on ice sheet motion, where ice sheet acceleration is often 304 tied to the presence 59 , flushing 60 , and distribution 61 of subglacial water systems. Yet, the interaction between 305 ice sheet dynamics and long-term changes in the subglacial hydrologic system in Antarctica are virtually 306 unconstrained, largely due to the here-to-fore missing records documenting their long-term climate 307 sensitivity. Here we have presented data from subglacial precipitates that provide evidence for millennial-

330
against Ca concentration spectra. Within the error bounds of our precipitate age models and the ice core 331 climate records, we are not able to estimate sub-millennial leads or lags between the AIS response and 332 climate cycles. Nonetheless, stratigraphic consistency between dated layers, the regular frequency of mean 333 ages, and the significant correlation between our mineralogic timeseries and climate proxy records supports 334 our conclusion of a link between climate teleconnections and subglacial hydrology. Furthermore, as was 335 previously mentioned, calcite layers form rapidly upon introduction of carbon-rich, alkaline waters from 336 the EAIS interior to the marginal system, and the system then slowly transitions back to opal precipitation 337 after hydrologic connectivity is shut off and the waters freeze. Therefore, it is possible that there is missing 338 time between calcite layers that is not accounted for in stratigraphic age models. However, based on the 339 regularity of opal depositional cycles, and the similarity between precipitate opal-calcite cycles and climate 340 proxies, these unconformities do not represent enough time to disrupt the millennial-scale cyclicity of the 341 precipitate mineralogy.

342
Correlation between opal-calcite timeseries from both samples and ice core climate records is 343 assessed both visually, by plotting stacked spectra, and by moving window correlation coefficient 62 344 (Extended data Fig. 3). Although the link between subglacial hydrologic events and SH climate cycles is 345 the result of a complex ocean-atmosphere-cryosphere feedback, stacked records reveal a clear overlap 346 between the mineral transitions in precipitates and ice core climate proxies (Extended data Fig. 3a, c).

347
Moving window Pearson r correlation coefficients show a significant linear correlation between precipitate 348 and SH climate timeseries for both samples. The one acceptation to this synchrony is in sample MA113 349 during the period between 54 and 50 ka. The lack of a clear relationship between the two records during this timeframe could be the result of dating uncertainties in both MA113 and the ice core records, or poor 351 millennial-scale climate resolution in the WDC during this period. Indeed, inspection of the TDC between 352 54ka and 50ka shows small scale climate cycles that do not appear in WDC, and have a similar pattern to 353 opal-calcite cycles in MA113 (Fig. 1d, e, f)

400
Geochemical models of mineralogic cyclicity in subglacial precipitates: and warm Antarctic climate periods respectively, we integrate geochemical and isotopic characterization 403 of the precipitates to inform simulations run using the aqueous geochemical modeling program 404 PHREEQC 46 . The high ion concentrations of subglacial fluids necessitates the use of the Pitzer specific ion 405 interaction approach, which allows PHREEQC to model the aqueous speciation and the mineral saturation 406 index of brines, and has been shown to yield results 39 consistent with the subzero database FREZCHEM 74 .

407
Our modeling approach to simulating opal-calcite transitions can be describe in three parts: 1) Identify the 408 water composition and conditions under which opal will precipitate and calcite will not; 2) Identify the 409 composition and volume of water required to mix with opal forming fluids to produce calcite; 3) Utilize the 410 δ 18 O and δ 13 C isotopic composition of calcite and opal, along with the known or inferred composition of 411 mixing waters (Fig. 3)

474
PHREEQC mixing models successfully produced discrete pulses of calcite with mixtures between 30% and 475 80% meltwater; conditions under which the admixture is undersaturated with respect to opal because the 476 solution is too dilute with respect to Si and is supersaturated with respect to calcite leading to precipitation 477 (Extended Data Fig. 8). Therefore, the addition of carbon-rich, alkaline meltwaters to opal precipitating,

492
Calcite data, in δ 13 C vs δ 18 O space, define a trend that suggests they form through admixture with 493 an isotopically lighter water with respect to both carbon and oxygen. The carbon concentration dependent 494 mixing curve that best fits that calcite data alone requires that the isotopically light endmember, what we'll 495 refer to here as the "meltwater" endmember carries a higher DIC than the isotopically heavier water, which 496 we'll now refer to as the brine endmember. In figure 3a and 3b, we assume that isotopically heaviest opals 497 record the δ 18 O composition of the brine endmember and that the δ 13 C composition matches marine carbon 498 derived from the substrate (-0 ‰), the latter of which is recorded by sodic carbonates suspected of forming 499 from brines in the Lewis Cliff area 96 . Under such assumptions the carbon ratio between meltwater and brine 500 is 97:3 for PRR50489 (Fig. 3a) and 80:20 in MA113 (Fig. 3b), a result that is consistent with the calcite 501 precipitation model presented above, whereby the addition of a carbon rich, oxidized meltwater, to a 502 reduced or intermediate CaCl2 brine, triggers calcite precipitation. As shown in extended data figure 8, the 503 calcite data imply formation when there is >30% of meltwater in the mixture. The array of calcite data can 504 also be fit by a mixing model that assumes an isotopically lighter carbon composition (δ 13 C = -15 ‰). While 505 feasible, this is a less appealing solution, as a δ 13 C of -15 ‰ does not match the composition of any specific 506 carbon source and would require a mixture of waters. The data presented here suggests that over the ~100ka 507 of aggregate sample precipitation, there are two consistent endmember waters: a reduced brine that is locally 508 derived (star in second quadrant, Fig. 3), and an oxidized meltwater that is from the polar plateau (star in 509 third quadrant, Fig. 3).