Reconstruction of Paleo-Circulation in the Western

Reconstruction of Quaternary Paleo-circulation in the Western Arctic Ocean Based on a

Neodymium Isotope Record from the Northwind Ridge

Thesis

Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in

the Graduate School of The Ohio State University

Rachael Elizabeth Gray, B.S.

Graduate Program in Geological Sciences

The Ohio State University

2012

Thesis Committee:

Lawrence Krissek, Advisor

Leonid Polyak

W. Berry Lyons

Rachael Elizabeth Gray

2012

Abstract

An understanding of past ocean circulation in the Arctic is critical for interpretations of past global ocean and atmospheric circulation, as well as predictions of future conditions.

The Arctic Ocean plays a major role in global climate, due to its contributions to both the

North Atlantic Deep Water (and subsequently the Atlantic Meridional Overturning

Circulation) and the planet's albedo (due to sea ice cover). A sediment core from the

Northwind Ridge in the western Arctic Ocean, ~800 km north of Alaska, has been sampled for measurement of radiogenic isotope ratios of neodymium and strontium.

Sediment grain coatings were leached from the bulk sediment and measured for 87/86Sr and εNd, a proxy for seawater source. Two leaching solutions, one using buffered acetic acid and the second using hydroxylamine hydrochloride, were applied to sediments.

Strontium data suggests that acetic acid best captures the seawater signal, while hydroxylamine hydrochloride leaching likely caused clay contamination of the hydrogenous data. εNd ratios were compared with independent lithologic proxies measured on the core and with results of earlier radiogenic-isotope studies in the Arctic

Ocean. Data obtained suggest that radiogenic waters dominated the western Arctic

Ocean during the estimated Early Pleistocene, probably due to increased Pacific water inputs and/or enhanced brine exclusion from sea ice formation on the Siberian shelves.

These conditions likely indicate relatively warm climatic environments with predominantly seasonal sea ice, and thus can be potentially used as a paleo-analog for the ii projected near-future state of the Arctic. Further upcore, in the estimated Middle to Late

Pleistocene, εNd values decrease overall, with high-amplitude fluctuations corresponding to glacial-interglacial cyclicity. Strongly non-radiogenic values in glacial intervals suggest the predominance of inputs from the Canadian Shield eroded by the Laurentide ice sheet. More radiogenic but gradually decreasing interglacial values indicate a change from Pacific to Atlantic water influence during the Middle to Late Pleistocene. Further isotope work on other cores may clarify the mechanisms and extent of this shift in circulation patterns.

iii

Acknowledgments

I would like to thank, first and foremost, my advisor, Dr. Leonid Polyak, without whom this thesis simply would not have been possible. I also owe my thanks to Dr.

Brian Haley for his scientific and technical contributions to this project, as well as his patience and sense of humor. Thanks also to my committee members, Dr. Larry Krissek and Dr. W. Berry Lyons for valuable feedback. I would also like to acknowledge my fellow students at Ohio State, particularly the past and present members of the paleoceanography group at Byrd Polar Research Center. A special thank-you goes to Dr.

Harunur Rashid for his advice and encouragement. Thanks to the faculty, staff and students of Oregon State’s College of Ocean and Atmospheric Sciences, who helped make my trips to Corvallis much more productive and enjoyable.

Tracie Fisher-Kline performed initial sediment sampling for this research. Allison

Kreinberg and Maya Wei-Haas generously assisted with laboratory equipment and space.

Edward Council and Chuang Xuan provided unpublished data cited in this thesis.

Finally, I would like to thank my friends, who have made the past few years some of the best of my life, and to my family, who have always believed in me.

This research was supported by NSF-OPP award ARC-1003777 to L. Polyak.

Vita

2009...... B.S. Geology, The Ohio State University

2009...... B.A. English, The Ohio State University

2009 to present ...... Graduate Research Associate, Byrd Polar

Research Center, The Ohio State University

Fields of Study

Major Field: Geological Sciences

Table of Contents

Abstract ...... ii

Acknowledgments...... iv

Vita ...... v

List of Tables ...... vii

List of Figures ...... viii

Chapter 1: Introduction ...... 1

Chapter 2: Arctic Ocean Physiography and Water Masses ...... 4

Chapter 3:Radiogenic Isotopes in the Arctic Ocean ...... 8

Chapter 4: Material ...... 14

Chapter 5: Methods ...... 21

Chapter 6: Results ...... 24

Chapter 7: Discussion ...... 28

Chapter 8: Summary ...... 36

References ...... 38

Appendix: Additional Data Tables ...... 46

List of Tables

Table 1. Major sources of Nd isotopes in the Arctic Ocean waters ...... 11

Table 2. Acetic acid leach data for P23 ...... 47

Table 1. Hydroxylamine hydrochloride leach data for P23 ...... 48

Table 1. Leach data from Winter et al. (1997) ...... 49

vii

List of Figures

Figure 1. Index map of the Arctic Ocean ...... 3

Figure 2. Downcore variability in sedimentary proxies in P23 ...... 16

Figure 3. Correlation of P23 with CESAR cores ...... 20

Figure 4. Flowchart of sample processing procedure ...... 22

Figure 5. P23 downcore distribution of εNd and 86/87Sr ...... 25

Figure 6. Comparison of acetic acid and hydroxylamine hydrochloride leaching results for core P23 ...... 27

Figure 7. Comparison of P23 acetic acid leaching data with oxalic acid leaching data from Winter et al. (1997) ...... 30

Figure 8. Acetic acid εNd results of P23 plotted with Mn and Ca contents ...... 32

Figure 9. Comparison of P23 acetic acid εNd values with results from ACEX ...... 35

viii

Chapter 1: Introduction

The Arctic Ocean is an important region for study of paleoclimate and paleoceanography due to its role in global oceanic and atmospheric circulation patterns combined with its sensitivity to climate change (e.g., Aagaard et al., 1985; Holland and

Bitz, 2003). The Arctic discharge affects the formation of the North Atlantic Deep Water and thus of the Atlantic Meridional Overturning Circulation (Peterson et al., 2006), one of the major modulators of the Earth’s climate. The sea ice coverage of the Arctic also plays an important role in global climate due to its impact on the planet's albedo (Serreze et al.,

2007). Current dramatic reduction in Arctic sea ice cover (Fig. 1) (Stroeve et al., 2012, and references therein) and related changes in oceanic and atmospheric circulation as well as biological and societal impacts (e.g., Wassmann, 2011; Allison et al., 2009) underline the urgency of comprehending this region and its behavior during past major shifts in climate.

Accurate records of climate and ocean conditions from the past are essential for understanding the region now and predicting future conditions. Due to logistical constraints, detailed high-quality sediment core records for the Arctic Ocean have been difficult to obtain until recent years, when icebreaker research expeditions have allowed the gathering of sediment cores across the Arctic Ocean including previously inaccessible regions (e.g., Darby et al., 2005; Polyak and Jakobsson, 2011). Measurements of various proxies in these cores contribute to a better understanding of the history of the Arctic 1

Ocean and thus the Earth’s climate as a whole. One important type of proxy is radiogenic isotopes (e.g., of neodymium, lead, strontium) in sediment grains and their hydrogenous coatings (Frank, 2002; Goldstein and Hemming, 2003). Downcore measurement of these radiogenic isotopes allows for interpretation of sediment provenance and water circulation changes through time.

The overall purpose of this study is to enhance the understanding of the circulation history in the western Arctic Ocean by information derived from radiogenic isotope compositions of hydrogenous marine sediment components (primarily neodymium) in a representative sedimentary record. This goal requires the following specific objectives:

1) measuring neodymium and strontium isotopes in coatings of sediment grains in a

selected down-core record;

2) comparing this new isotope data with other paleoenvironmental proxies

previously measured on the same sediment core, as well as with earlier radiogenic

isotope studies from the Arctic Ocean;

3) interpreting paleocirculation and sediment provenance in the western Arctic

Ocean as inferred from generated isotope values and related data.

We expect that a concerted change in neodymium isotope data and other proxies of the same core will characterize changes in paleocirculation, and a correspondence between the neodymium isotope record for this core and other Arctic cores will broaden the geographic and water-depth implication of the new data. The correspondence of the strontium record with the global seawater value will be used to verify that the leaching

2 process has successfully extracted the seawater signal.

Figure 1. Map of the Arctic Ocean with location of sediment cores: star – AR93-P23, circles – ACEX and CESAR11. Numbers show source εNd values. Lines show Pleistocene glaciations (dotted white) and summer sea-ice margins: 2007 (historical minimum, magenta) and late 20th century mean (light yellow). White arrows – surface Arctic circulation, orange and purple – Atlantic and Pacific waters, respectively (punctured – subsurface), light green – major rivers (with initial letters). NR, MR, AR, and LR – Northwind, Mendeleev, Alpha, and Lomonosov ridges, respectively. YP – Yermak Plateau, NGS – Norwegian-Greenland Sea. Bathymetry from IBCAO v.2 (Jakobsson et al., 2008).

Chapter 2: Arctic Ocean Physiography and Water Masses

The Arctic Ocean is both relatively small and closed to interaction with other oceans, with deep-water access only through the Fram Strait (Fig. 1). The deep central portion of the Arctic Ocean is divided into the Amerasian and Eurasian Basins by the

Lomonosov Ridge, and is surrounded by shallower shelf seas. The total volume of the

Arctic Ocean is estimated to be only ~1.4% of the world ocean volume, but ~4.3% of the area due to the extent of these shallow shelves, which make up 52.9% of the area of the

Arctic Ocean (Jakobsson 2002).

The Eurasian Basin is bounded by the Lomonosov Ridge, the Barents, Kara and

Laptev Seas, and Greenland. The Fram Strait allows deepwater access to the world ocean by way of the Greenland-Iceland-Norwegian (GIN) Seas, which connect to the North

Atlantic. The Lomonosov Ridge rises to a depth of 1000-1300 m below sea level for much of its length (Björk et al., 2007). The active-spreading Gakkel Ridge, roughly parallel to the Lomonosov Ridge, extends from Greenland to the Laptev Shelf north of

Siberia and divides the Eurasian Basin into two sub-basins.

The Amerasian (or Canada) Basin is separated from the Eurasian Basin by the

Lomonosov Ridge. Its other boundaries are formed by the Canadian Arctic Archipelago,

Alaska, and the Chukchi and East Siberian Seas. The abyssal Amerasian Basin is marked by several ridges extending from the surrounding continental shelves. The Mendeleev

Ridge extends from the East Siberian shelf, whereas the Alpha Ridge is connected to the 4

Canadian continental shelf. Together the Alpha and Mendeleev Ridges divide the

Amerasian Basin into two sub-basins, the Makarov Basin to the east and the Canada

Basin in the west.

The Chukchi Borderland occupies a space between the Chukchi Sea and Amerasia

Basin. It includes the Northwind Ridge to the east, extending north from the Chukchi shelf, and the Chukchi Rise and Cap to the west, with the Northwind Basin separating them. These features rise to relatively shallow water depths of <1000 m on the Northwind

Ridge and nearly 300 m on the Chukchi Cap and therefore affect water circulation in the western Arctic Ocean (Woodgate et al., 2007).

Major Arctic Ocean rivers include the Ob’, Yenisey, and Lena, and Kolyma rivers of Siberia and the Mackenzie River of North America (Fig. 1). These rivers, along with smaller ones, currently transport about 3300 km3 of freshwater annually (Aagaard and

Carmack, 1989). This very high river runoff, in addition to its major role in the Arctic

Ocean freshwater budget, contributes organic matter, nutrients, and other chemical constituents, thus affecting biological productivity and hydrochemistry in the Arctic

Ocean (e.g., Anderson et al., 1994).

The present-day hydrographic structure of the Arctic Ocean is complex, and the water column is strongly stratified (e.g., Rudels, 2009 and references therein). The upper

~50-m water mass, the Polar Mixed Layer, is characterized by a mixture of Atlantic and

Pacific waters, river discharge, and sea-ice meltwater. The resultant salinity of this layer is low, 30-34 psu (practical salinity units). This surface water is involved in circulation caused by two major wind-driven currents (Fig. 1). The clockwise Beaufort Gyre is

5 dominant in the western (Amerasian) Arctic, while the Transpolar Drift flows from the

Laptev Sea to the Fram Strait across the eastern Arctic and the Lomonosov Ridge.

Below the Polar Mixed Layer lies the halocline, which extends to ~200 m depth and exhibits strong vertical stratification due to variations in salinity with depth (e.g.,

Aagaard et al., 1981; Rudels et al., 1996). Various layers of the halocline are composed of

Pacific and/or Atlantic waters, which have been partially modified and mixed with local waters on the shelves. The relatively fresh upper halocline (33.2 psu) is thought to be mostly composed of Pacific water that enters via the Bering Strait and is modified on the

Chukchi shelf. The more saline lower halocline (34.2 psu) has a more complex spatial structure with predominantly Atlantic water in the eastern Arctic and winter Pacific water in the Amerasian Basin (Rudels et al., 1994; Woodgate et al., 2005).

Below the halocline, between ~200 and 900 m depth, are the intermediate waters formed by two branches of Atlantic water (Fig. 1), one which enters the Arctic Ocean through the Fram Strait and another that passes through the Barents Sea, where it is modified by brines and becomes colder (Rudels et al, 1994). The Fram Strait Branch

Water travels along the continental margin and is joined by the Barents Sea Branch Water as it moves eastward. Some of the Atlantic water travels back toward the Fram Strait along the Eurasian side of the Lomonosov Ridge, while a smaller portion, mostly Barents

Sea Branch Water, crosses the Lomonosov Ridge and enters the Amerasian Basin. This

Atlantic-derived water then flows along the Siberian margin, enters the Canada Basin near the Chukchi Plateau, and recirculates back to the Fram Strait between the

Lomonosov Ridge and Greenland (Rudels et al., 1994).

Deep water movement between the Amerasian and Eurasian Basins is limited by the presence of the Lomonosov Ridge, so that waters above its average depth of ~1500 m can flow freely, while deeper waters are restricted to the Eurasian and Amerasian basins.

The Eurasian bottom waters are overall slightly colder and less saline than those of the

Amerasian Basin (Rudels, 2009). The reason(s) for the occurrence of higher than predicted temperatures in deep/bottom waters is not well understood; geothermal heating has been suggested as a possible cause (Timmermanns et al., 2003; Björk and Winsor,

2006).

An important source of salinity for the halocline (and to some extent, the deeper layers) is the release of brines during sea-ice freezing on the broad Siberian shelves

(Aagaard et al., 1981; Rudels, 2009). If a sufﬁcient amount of ice forms, the released brines create dense (high salinity and near-freezing temperature) bottom waters, which cross the shelf break and descend downslope until their densities match those of the surrounding water. Less-dense plumes formed this way merge mostly with the upper layers of the halocline (e.g., summer Pacific/Chukchi Sea waters), while denser plumes descend to the lower halocline and even deeper waters, as observed north of the Barents

Sea (Schauer et al., 2002; Rudels, 2009).

Chapter 3: Radiogenic Isotopes in the Arctic Ocean

The distribution of radiogenic isotopes of dissolved elements varies throughout the world ocean based on residence time (e.g., Hodell et al., 1990; Frank, 2002). For example, the long oceanic residence time (~2.5 my) of strontium means that the 87/86Sr isotopic ratio of seawater is relatively uniform throughout the world ocean at any given time. In contrast, neodymium has a shorter residence time of ~1000 years. Because of this, neodymium does not become homogenized in the world ocean; therefore, neodymium isotope ratios can vary from one ocean basin to another, and water masses may have unique neodymium isotopic signatures.

Radiogenic isotopes in seawater are absorbed into the hydrogenous coating of sediment grains (typically ferromanganese hydroxides), from which they can be extracted by mild chemical leaching without incorporating the isotopic signal from the detrital portion of the sediment. While radiogenic isotopes of the detrital fraction reflect the sediment provenance, the grain coatings carry isotopic information about seawater at the time of deposition.

The neodymium isotope ratio used as a seawater/sediment provenance tracer is

143Nd/144Nd (Piepgras and Wasserburg, 1980). The common notation for this ratio, εNd, is defined as:

143푁푑 (144 ) 푁푑 휀푁푑 = 푠푎푚푝푙푒 − 1 ∗ 10000 143푁푑 ( ) 144푁푑 [ 퐶퐻푈푅 ]

143푁푑 where ( ) = 0.512638. Sampled neodymium ratios are normalized to CHUR, 144푁푑 퐶퐻푈푅 the Chondritic Uniform Reservoir. Variations in εNd are due to the decay of 147Sm

(samarium), with a half-life of 1.06x1011 yr, to 143Nd.

Nd isotopic composition in seawater is primarily controlled by continental inputs as rocks are weathered and neodymium solubilized into rivers and ultimately to the ocean basin (e.g., Frank, 2002). Generally speaking, older rock tends to correspond with a more negative (less radiogenic) εNd fingerprint (e.g., Goldstein and Hemming, 2003). While

143Nd/144Nd ratios increase for both the bulk earth and the continental crust over time, the crust ratio will increase more slowly as the original melt of continental rock includes less

147Sm than the melt residue (Stordal and Wasserburg, 1986; Goldstein and Hemming,

2003). However, the greater control on neodymium isotope ratios is elemental fractionation during petrogenesis (DePaolo and Wasserburg, 1976).

The distribution of εNd in the Arctic Ocean is controlled by the isotopic signature of incoming water masses and their circulation (Fig. 1; Table 1). The Atlantic Ocean is presently the chief source of seawater, and thus of seawater neodymium, for the Arctic as a whole. North Atlantic waters entering the Arctic Ocean tend to have εNd values around

-10.8 (Andersson et al., 2008). In contrast, North Pacific waters have a much more radiogenic signature overall, up to -2 (Porcelli et al., 2009), although those entering the 9

Arctic Ocean have somewhat lower values, around -4 to -5 (Dahlqvist et al., 2007). εNd ratios of the Siberian rivers range from -5 to -14 as controlled by variation in source terranes (Porcelli et al., 2009). The West Siberian Ob’ and Yenisey rivers have overall high εNd values of -5 to -6 consistent with extensive distribution of basaltic rocks southeast of the Kara Sea (Putorana Plateau), while East Siberian rivers show a more complex picture reflecting a diverse geological pattern. North American rivers, which come from very old Precambrian to Lower Paleozoic rocks of the Canadian Shield, have strongly negative εNd as demonstrated by the Mackenzie River εNd ratio of -14 (Porcelli et al., 2009). Isotopic signature of composites of rocks sampled from the Canadian Shield ranged from -32 to -7, with most εNd values being below -25 (McCulloch and

Wasserburg, 1978). Sediment in Baffin Bay, on the Atlantic side of the Canadian Shield, also shows εNd values as low as -26 (Stordal and Wasserburg, 1986).

Direct measurements of Nd in seawater in the Arctic Ocean (Andersson et al.,

2008; Porcelli et al., 2009) show a considerable variability in vertical distribution of εNd in the Polar Mixed Layer and the halocline, consistent with a strong stratification and variable sources in these waters. Deeper waters show less variability, with εNd values mostly between –10 to –11, indicative of the overall strong Atlantic influence.

Nevertheless, measurements from the Amundsen basin showed somewhat lower εNd values; whereas, a profile from the Canada basin near the Chukchi margin showed an elevated εNd value of –9 in water as deep as 2500 m. These data suggest inputs of water, possibly with suspended sediment from the shelves to deep water horizons. More measurements from water-column profiles are needed to better understand the processes

10 involved.

Source Signal εNd Reference

Fram Strait Atlantic -10.8 Andersson et al., 2008

Bering Strait Pacific -5 Dahlqvist et al., 2007

Mackenzie River North American -13 Porcelli et al., 2009

Lena River East Siberian -14 Porcelli et al., 2009

Kolyma River East Siberian -6 Porcelli et al., 2009

Yenisey River West Siberian -5 Porcelli et al., 2009

Ob River West Siberian -6 Porcelli et al., 2009

Table 1. Major sources of Nd isotopes in the Arctic Ocean waters.

Measurements of Nd isotopes in surface-sediment samples leached mostly using the same acetic acid technique as used for this study and geographically distributed across the Arctic Ocean show a good correspondence with the isotopic signature of bottom waters at those sites, and thus verify the validity of using εNd as a water tracer in the Arctic (Haley et al., 2008a; Haley and Polyak, in preparation). Throughout the Arctic

Ocean interior, sediment εNd values stay in a fairly narrow range between -11 to -10, reflecting the dominance of Atlantic water signal with some admixture from more or less radiogenic local sources. Considerably elevated values appear close to the Chukchi

11 margin, apparently due to the spread of radiogenic waters and/or sediments from the

Chukchi shelf, where εNd can be as high as around -2. These values are even higher than the Pacific water signature and possibly indicate leaching from more radiogenic suspended sediment; this pattern has yet to be investigated. Elevated εNd values also characterize other Siberian shelves, with predictably higher numbers of above -5 in the

Kara Sea affected by water and sediment inputs from the Putorana basalts. Application of the identified εNd sediment provenance proxy to down-core data allows for a record of changes in seawater sources and pathways over time.

While there have been several downcore studies of detrital εNd along with

87Sr/86Sr in Arctic sediments as a means of tracing sediment provenance through time

(Winter et al., 1997, Eisenhauer et al., 1999, Tütken et al., 2002; Haley et al., 2008a), only two studies have addressed Arctic paleo-circulation history from εNd in sediment grain coatings (Winter et al., 1997; Haley et al., 2008b). Winter et al. (1997) measured

εNd in manganese-oxide coatings of Fe-Mn micronodules and foraminifers, as well as in the silicate residues which remained post-leaching, in samples from a number of sediment cores from the Alpha Ridge, western Arctic Ocean, spliced lithostratigraphically in a composite record. These cores were taken from a range of water depths between

1380 to 3050 m, with most samples taken from core CESAR 11 from 1380 mwd.

Although the age model used at that time was later invalidated (e.g., Jakobsson et al.,

2000; Polyak et al., 2009; Stein et al., 2010), downcore changes identified in hydrogenous εNd have given first insights into the evolution of deep water masses in the

Arctic. A consistent change in εNd and detrital radiogenic isotopes 87/86Sr and 206/204Pb in

12 the upper part of the record towards less radiogenic values was interpreted as related to the increasing impact of Pleistocene glaciations in North America, which delivered sediment from ancient rocks of the Canadian Shield. In addition, a stratigraphic interval with elevated (more radiogenic) εNd values was identified in the pre-glacial part of the record, but this pattern remained unexplained.

A more recent and better age-constrained study was performed on samples from the ACEX (Arctic Coring Expedition) borehole from the Lomonosov Ridge (Fig. 1) covering the past ~15 Ma, combined with data from a nearby sediment core that provides more detail on the late Quaternary record (the past ~400 ka) (Haley et al., 2008b).

Interpretation of the hydrogenous εNd signal in this data set has suggested that the Arctic

Intermediate Water was affected by brine formation on the Eurasian Arctic margin between ca. 15 and 2 Ma, after which the circulation pattern switched to more input from the Atlantic water and less from brines. After ca. 2 Ma, the time of large Pleistocene glaciations at the Arctic Ocean periphery (Fig. 1), the εNd values fall into a pattern of higher variability with abrupt, high-amplitude switches between “interglacial” and

“glacial” modes. The radiogenic εNd signal during glacial times indicates that deep/intermediate waters in the Eurasian Basin were strongly influenced by brines formed on the shelves in front of ice sheets, while the entrance of the Atlantic water was presumably minimized by decreased glacial-time sea levels.

Chapter 4: Material

Sediment core P1-93AR-P23, hereafter referred to as P23, was chosen for this study. P23 is a piston core retrieved from the crest of the Northwind Ridge, ~800 km north of Alaska at 76°57.3'N, 155°03.9'W at a water depth of 951 m, and is 572 cm in length (Fig. 1) (Mullen and McNeil, 1995; Crawford, 2010). While low sedimentation rates in this area, as in the majority of the western Arctic Ocean (Polyak et al., 2009), mean that temporal resolution in P23 is lower than in many other Arctic Ocean cores, it also captures a longer period of time. This, together with the geographic position of P23 in the area of interaction of Pacific and Atlantic waters (Woodgate et al., 2007) and a rapid modern sea-ice retreat (Fig. 1), defines the core choice for this study.

Core P23 has been sampled for various proxies with the aim of developing its stratigraphy and sedimentary/paleoceanographic history such as sand content (>63μm), representative of ice-rafted debris (IRD), elemental composition measured by means of

XRF, paleomagnetic properties, foraminiferal content, and more (Polyak et al., 2009;

Crawford 2010). Based on these proxies the core can be divided into two main lithostratigraphic units with a boundary at ~130-140 cm (Fig. 2). The upper, grey to brown unit (grey on Fig. 2) is characterized by highly fluctuating contents of IRD, Mn, and Ca, indicative of cyclic inputs of sediment from the Laurentide ice sheet separated by interglacial intervals (e.g., Polyak et al., 2009; Stein et al., 2010). The lower, brown unit 14 has overall elevated Mn and related proxies of low-ice environments such as magnetic grain size (Karm/K) (O’Regan et al., 2009; Maerz et al., 2011; Xuan et al., 2012), while low IRD and Ca indicate limited glacial inputs. Sediment below ~350-360 cm is different from the rest of the lower unit by having somewhat higher content of IRD, tapering off towards the core bottom, and somewhat lower content of Mn and magnetic grain size.

Figure 2. Downcore variability in sedimentary proxies in P23 (from Crawford, 2009).

Two major lithostratigraphic units are shown in grey and brown. Numbers show estimated Marine Isotope Stages in the upper unit. Paleomagnetic and XRF elemental data courtesy of C. Xuan and E. Council.

An earlier study suggested Pliocene age for the lower part of core P23 based on the presence of benthic foraminifers characteristic for Pliocene and older sediments in the

Beaufort-Mackenzie basin (Mullen and McNeil, 1995). However, potentially time-

16 transgressive extinction of a benthic species allows for a younger age in deep-sea environments (Northwind Ridge) than in shallow waters (Beaufort-Mackenzie area) (e.g.,

King, 1989). The new age model developed for core P23 is primarily based on lithostratigraphic cycles indicative of glacial/interglacial stages along the lines of recent studies of the Arctic Ocean sediments (Fig. 3) (Jakobsson et al., 2000; O’Regan et al.,

2008; Polyak et al., 2009; Stein et al., 2010). Paleomagnetic data are being evaluated as an independent age constraint. While recent results suggest that inclination cannot be used to identify true magnetic reversals in Arctic Ocean stratigraphy due to diagenetic alterations of sediment (Channell and Xuan, 2008; Xuan and Channell, 2009), major changes in inclination appear to be stratigraphically consistent throughout the Arctic basins (Spielhagen et al., 2004; Polyak et al., 2009). In particular, the first (from top) major inclination drop tentatively dated to MIS7, between 200 to 250 ka (Jakobsson et al., 2000; Spielhagen et al., 2004) is identified in P23 at ~40 cm (Figs. 2, 3). A major inclination rise co-occurring with a general increase in Mn content at ~140 cm is also a consistent stratigraphic feature in the western Arctic Ocean, cyclostratigraphically estimated to occur sometime prior to MIS 16 (ca. 600-650 ka), near the end of the Early

Pleistocene (Polyak et al., 2009; Polyak and Jakobsson, 2011).

Below this boundary the age model is much less clear. If constant sedimentation rates are assumed, the core dates back to the Pliocene, >3 Ma. However, the consistent downcore strontium isotope ratio (as shown below) suggests that the core cannot be much older than approximately 1.5 Ma (Sr residence time). This inference implies that sedimentation rates were greater in the Early Pleistocene, which is consistent with a

17 generally elevated content of detrital Mn and other indicators of reduced ice conditions in the lower lithostratigraphic unit. Although the age model for P23, as well as other sediment cores from the western Arctic Ocean, is tentative and crude, it is consistent with the proxy paleoceanic/paleoclimatic history indicating high glacial inputs and sea ice in the Middle to Late Pleistocene (“glacial Pleistocene”) and less glaciated, more open- water conditions in the Early Pleistocene.

P23 can be correlated to other cores from the western Arctic Ocean based on lithostratigraphy and, where available, magnetic and biostratigraphic proxies (Polyak et al., 2009; Crawford, 2010). This correlation allows for a comparison of radiogenic isotope data from P23 with earlier data from the western Arctic (Winter et al., 1997) in a unified stratigraphic perspective, regardless of differences in the age models. Figure 3 shows a comparison of IRD (grains >63 m) in P23 and core CESAR 11 that holds half of the samples analyzed in Winter et al. (1997). The compressed nature of the stratigraphy and potential changes in sedimentation rates complicate a detailed correlation, but a prominent sand peak marked with an increase in Ca content at ~120 cm in P23 clearly matches Clark et al’s peak designated as Standard Lithological Unit (SLU)

F. The IRD peak below the major inclination rise (~ 160 cm in P23) does not show up in

CESAR 11, but has been described for the upper part of SLU A in other cores throughout the western Arctic Ocean (Clark et al., 1990), and the low-amplitude, broad peak in the lower part of the stratigraphy (SLU A2) is well expressed in both cores. The paleomagnetic inclination pattern in P23 also corresponds well to earlier investigated records including core CESAR 14 adjacent to CESAR 11 (Aksu, 1985), and the

18 correlation is corroborated by the position of a prominent subpolar planktonic foraminiferal peak in SLU G (Clark et al., 1990), corresponding to MIS 11 in the new age model.

Figure 3. Correlation of P23 with CESAR cores from the Alpha Ridge (Fig. 1 for location; Aksu (1985), Clark et al. (1990), and Crawford (2010), for CESAR11,

CESAR14, and P23 data, respectively). CESAR 11 was used for εNd study of Winter et al. (1997). Blue lines show correlation of key IRD peaks in P23 and CESAR11; green lines – major inclination swings in P23 (see Fig. 2) and CESAR14 (via

CESAR11 correlated lithostratigraphically to CESAR14: Mudie and Blasco, 1985).

Clark’s units, A3 to M, are shown to the left of CEASR11 and CESAR14 curves.

Asterisks show position of MIS 11 planktonic foraminiferal peak in P23 and

CESAR11.

Chapter 5: Methods

Bulk sediment for εNd measurements was collected downcore P23 from 1-cm- thick sampling intervals spaced at ~10-15 cm (48 samples total). Selection of sampling depths was guided by lithostratigraphic proxy data to best investigate intervals of paleoceanographic interest. For a better representation of surficial sediment, a top-most sample was taken from a nearby multi-core, HLY0503-3MC.

Each step involved shaking the samples with MilliQ water for 30 minutes in 50- mL acid-cleaned polypropylene tubes, then centrifuging at 4K RPM for 15 minutes, followed by removal of the supernatant liquid by decanting. 20mL of buffered acetic acid solution (2.1 M glacial acetic acid/ 0.6 M sodium acetate; pH 3.5-4) was added to the samples, which were then shaken for 150 minutes and centrifuged for 30 minutes.

After this step, 20mL of the supernatant liquid was pipetted into acid-cleaned Teflon vials. These extracted leach solutions were designated as the “A” samples. The sediment samples were again rinsed three times with MilliQ water, and then treated with 10mL of a hydroxylamine hydrochloride solution (0.04 M hydroxylamine HCl/ 3.9M glacial acetic acid/ 0.7M sodium hydroxide). This mixture was again shaken for 60 minutes, then centrifuged for 25 minutes at 4K RPM. The supernatant was again decanted and 15mL saved and designated as the “B” sample. The remaining sediment was again rinsed three times with MilliQ water, dried, and stored. All reagents used were of highest purity commercially available 21

Figure 4. Flowchart of sample processing procedure for measuring radiogenic isotopes in leached sediment grain coatings.

The ”A” and “B” leachate solutions were evaporated; brought up with concentrated distilled nitric acid and evaporated twice; and then brought up in distilled

6M hydrochloric acid before final evaporation. 1M hydrochloric acid was finally added and refluxed and in Teflon vials, prior to ion chromatographic separation, to destroy the matrix of the leach solution (Haley et al., 2008b).

The following treatment and analyses were performed at the College of Oceanic and Atmospheric Sciences of the Oregon State University under the direction of Dr. Brian

Haley. Samples were processed in a cation exchange column using Eichrom AG50 8-X cation exchange resin with 100-200 μm mesh to isolate rare earth elements and strontium.

Strontium was further purified using Eichrom Sr-Spec resin with 50-100 μm mesh.

Neodymium was extracted from the REEs using Ln-Spec resin with 100-150 μm mesh.

Samples were analyzed on a Nu Plasma HR MC-ICP-MS (multi-collector inductively-coupled plasma mass spectrometer). For neodymium reported values were corrected for instrument bias using the 50 ppb JNdi-1 reference standard (JNdi-1 =

0.512115, with a 2σ internal precision of ± 0.1 εNd units after 84 measurements) and 50 ppb SpecPure standard (SpecPure=0.511205, with an external 2σ reproducibility of ± 0.1

εNd units after 77 measurements). Initially, concentrations of neodymium were measured, and each sample was appropriately diluted with 3% nitric acid for an accurate measurement of εNd. Similarly, strontium concentrations were measured and then diluted for measurement of 87/86Sr ratios, which were corrected to the NBS 987 standard

(NBS987 = 0.710245; 2σ error = 0.000019; n = 56).

Of the 45 “A” samples, four were subjected to duplicate analysis. All of these ratios varied by 0.1 εNd unit or less from the original run, within the long-term instrumental precision of ±0.3. Additionally, three intervals were re-sampled entirely to evaluate the consistency of the leaching process. These all replicated the initial results within error. Of the 42 “B” samples, duplicate neodymium measurements were run on seven intervals. Of these, five were inconsistent over multiple runs, with repeat measurements showing variations greater than the instrumental error.

Chapter 6: Results

Grain coatings leached by buffered acetic acid (“A” samples) show considerable downcore εNd variation in P23 (Fig. 5). Three major intervals with distinct εNd patterns can be recognized. Samples taken below ~350-360 cm show little variability in εNd values, which range between -9.1 and -8.7. Values between ~350 and 200 cm are more radiogenic, ranging between -8.5 and -7.9. The increase in εNd corresponds with maximum values for magnetic grain size and manganese, as well as a decrease in sand content. Above ~200 cm grain coatings again show less radiogenic values, decreasing upcore along with a general decrease in manganese and magnetic grain size. In the upper lithostratigraphic unit, above ~140 cm, εNd values show abrupt, high-amplitude fluctuations, with highs varying from -11 to -8 and lows going down to below -15. This pattern corresponds to large, cyclical fluctuations in other proxies including Ca content that rises considerably in comparison with the lower unit (Fig. 2). Highs in εNd co-occur with Mn and magnetic grain size maxima, whereas low εNd values are generally consistent with peaks in Ca and IRD.

Figure 5. P23 downcore distribution of εNd and 86/87Sr in acetic leaches along with paleoceanographic proxies: IRD (>63mm), Mn, and magnetic grain size (Karm/K).

Numbers on top of the IRD curve show estimated Marine Isotope Stages (glacial) in the upper unit. Pink line – modern seawater 86/87Sr ratio.

The hydroxylamine hydrochloride leachates (“B” samples) show more variation downcore and display more frequent switches from radiogenic to non-radiogenic εNd measurements, including some very positive values (as great as 7.2). Like the “A”

25 samples, these εNd measurements show two distinct down-core patterns, with a shift occurring around 120 cm. Below this level εNd ratios fluctuate between -7.1 and 1.1, with more frequent and higher-amplitude switches from -10.7 up to 7.2 occurring above

120 cm. Beyond this broad trend of a pattern shift between the upper and lower core segments, “B” sample results are not consistent with those generated on the acetic acid leachates and do not show a noticeable association with any other sedimentary proxies.

Strontium isotope ratios for the “A” samples mostly ranged from 0.70911 to

0.70933, consistent with the modern seawater ratio of 0.70916, with ~30% of all samples having much more elevated values between 0.70950 and 0.70993 (Fig. 6). As these samples are randomly distributed along the core and do not show any stratigraphic pattern, they are considered as outliers. Strontium isotope values for the “B” samples show more variance in a range from 0.70910 to 0.71112 (Fig. 6). Strontium duplicates were run on ten “B” leachates, of which four did not show consistent values after multiple analyses. Additionally, the strontium ratios of the “B” samples tend to be notably higher than the expected seawater value. These inconsistencies suggest the hydroxylamine leach may be strong enough to partially leach the detrital sediment fraction and thus contaminate the leachate with its isotopic signal.

Figure 6. Comparison of acetic acid (left) and hydroxylamine hydrochloride (right) leaching results for core P23.

Chapter 7: Discussion

The meaningful down-core distribution of εNd in acetic lecheates, corresponding with independent sedimentary proxies (Fig. 5), suggests that acetic acid leaching captures the most accurate seawater signal in sediment samples as opposed to less correlative results of the hydroxylamine leaching, which is consistent with earlier work (Haley et al.,

2008b). This conclusion is further supported by the Sr isotope values in acetic leachates that most closely resemble the global seawater ratio. The acetic leachates have been chosen to be more appropriate for the interpretation of bottom-water paleoenvironments.

A comparison of acetic εNd in P23 with a record generated earlier on sediment cores from the western Arctic Ocean, mostly from the Alpha Ridge (Winter et al., 1997), shows a good correspondence between downcore distributions of the two data sets (Fig.

7). This correspondence not only attests to data similarity, but also indicates that results generated on P23 are applicable to a wide range of intermediate to deep water depths across the Amerasia Basin. A systematic offset between P23 and the Alpha Ridge data is possibly related to analytical differences such as the use of oxalic vs. acetic acid. Also, the use of hand-picked Fe-Mn micronodules by Winter et al. as opposed to bulk sediment used in this study may have introduced a bias towards more mature vs. more labile, amorphous Fe-Mn hydroxides, which may better represent the water signal. This inference is supported by an offset of Winter et al’s data from theoxide fraction towards the composition of the residual fraction, which indicates an admixture of detrital εNd 28 signal. The offset appears to be more pronounced in the lower part of the record, below

~200 cm on the P23 depth scale. The oxic εNd signal in this interval in the Alpha Ridge samples is decoupled from the record of radiogenic isotopes (εNd, 86/87Sr, and 206/204Pb) in total sediment digests (Winter et al., 1997), which indicates that εNd in Fe-Mn coatings here predominantly carries an ambient water signal. In contrast, in younger sediments corresponding to glacial Pleistocene, the oxide εNd signal corresponds with that of total- digest radiogenic isotopes indicating that the signal recorded in the coatings has been strongly affected by leaching from detrital grains either in sediment (in pore waters) or during the laboratory treatment. Although the cause of this pattern (natural vs. laboratory) yet needs to be understood, it can explain a less pronounced offset between our and

Winter et al’s εNd values in the upper part of the record.

Figure 7. Comparison of P23 acetic εNd data with oxalic acid leaches of Winter et al.

(1997) (ox – oxide fraction, res- residual fraction) spliced on P23 scale using correlation shown on Fig. 3.

Consistently high εNd values in the lower unit of P23 (Unit II; Fig. 7), especially below ~200 cm, indicate a steady supply of radiogenic waters and/or suspended sediment to the Amerasian basin during the Early Pleistocene. This condition is very different from the modern setting, where most of the Arctic Ocean interior, including the Northwind

Ridge, has non-radiogenic εNd values between –11 and –10 indicative of the predominant influence of Atlantic Ocean water. Elevated εNd values are currently only found close to the Chukchi shelf due to the spread of Pacific and Chukchi Sea waters and/or sediment from the shelf margin (Porcelli et al., 2009; Haley and Polyak, in prep.).

The downcore record suggests a stronger input of Pacific Ocean water via the Bering

Strait and/or more intense formation of sea ice and brines on the Siberian shelves in the

Early Pleistocene, which is indicative of widespread seasonal, rather than perennial ice.

This inference is consistent with overall elevated levels of such proxies as Mn and

Karm/K in Unit II (Figs. 2 and 5), which have been inferred to be related to stronger ice melting in recent studies (O’Regan et al., 2009; März et al., 2011; Xuan et al., 2012).

The interval between ~200 and 350 cm in P23, as well as the correlative interval in the Alpha Ridge record, shows the highest εNd values (Fig. 7), with a step increase of

>0.5 at ~350-360 cm indicating a change in water/sediment sourcing. Values around –9 before this change are similar to εNd composition in grain coatings from surface sediments of the Laptev and Eastern Siberian shelves (Haley and Polyak, in prep.), which could have been a water source area under intense sea-ice/brine forming conditions.

Consistently higher εNd values between ~200 and 350 cm values require increased mixing with a more radiogenic source signal. One potential source is a mixture of Pacific

Ocean and Chukchi Sea waters, both having high εNd values. The other potential source of radiogenic εNd could be the Kara Sea shelf as a mixture of Atlantic Ocean, Barents

Sea, and Kara Sea waters can descend to the Atlantic Intermediate Water layer or even to deeper waters and flow throughout the Arctic basins (e.g., Schauer et al., 2002; Rudels,

2009). Although more research is needed for a conclusive source discrimination, the

Pacific/Chukchi source appears more likely based on other sedimentary proxies in P23.

Notably, both Mn and Karm/K reach their maxima between 200 and 360 cm (Figs. 5 and

7) indicating strong ice melt over the Chukchi Borderland, which is consistent with higher Pacific influence. This inference is also corroborated by a benthic foraminiferal

31 fauna characteristic of a marginal ice zone (Crawford, 2010).

Figure 8. Interpretation of acetic εNd results in P23, plotted along with Ca and Mn contents (note reversed scale for Ca). Inferred sources for εNd levels (orange lines) are indicated in italics.

Above ~200 cm εNd values start to fall markedly along with the appearance of high IRD peaks and an overall decrease in Mn and Karm/K (Figs. 5, 8). This trend in

εNd, consistent with the Alpha Ridge data, continues to the core top, with large swings superimposed on the overall decrease in Unit I. In these swings, εNd minima correspond to lows in Mn and Karm/K and highs in total IRD and Ca (Figs. 5, 8). This pattern clearly indicates that low εNd values are controlled by inputs from North America during glacial 32 periods. Especially telling is the distribution of Ca, which is indicative of carbonate rocks

(primarily dolomites) from the Canadian Shield eroded by repeatedly growing and disintegrating Laurentide ice sheet (e.g., Bischof et al., 1996; Polyak et al., 2009). A dramatic increase in Ca content in IRD peaks towards the core top (Late Pleistocene), indicating higher discharges of eroded material from the Laurentide interior, corresponds to especially low εNd values reaching below -15. Such values are consistent with a strongly non-radiogenic composition of the Canadian Shield. It appears likely that the

εNd signal measured in grain coatings in these intervals results from leaching from sediment detritus enriched in carbonates. This inference is corroborated by a correspondence of εNd in Fe-Mn oxides with radiogenic isotopes in total sediment digests in correlative deposits from the Alpha Ridge (Winter et al., 1997). One possibility is that leaching took place in sediment through pore-water exchange, facilitated by fast deposition of glacigenic sediments from icebergs during pulses of ice-sheet disintegration. Alternately, leaching from the detrital fraction occurred during laboratory preparation. More measurements of 86/87Sr, possibly combined with Ca content in leachates, will help resolve this question.

Highs in the εNd record in Unit I, corresponding to interglacials, also decrease upcore. The last value as high as in the middle of Unit II, -7.9, occurs in the interval estimated as MIS 11, whereas in younger interglacials values are between 10 and -10.5.

This change through the Middle to Late Pleistocene suggests a shift from a stronger

Pacific Ocean contribution to the predominantly Atlantic Ocean influence as characteristic of the modern Arctic Ocean. It is important to note that glacial-interglacial

εNd swings in P23 are opposite to those measured in the ACEX and adjacent core record from the central Lomonosov Ridge, where sedimentation is controlled by the Transpolar

Drift (Fig. 9). This is not surprising as glacial intervals on the Lomonosov Ridge (even

MIS numbers) received sediment and bottom water from the glaciated Eurasian margin containing extensive outcrops of strongly radiogenic rocks of west Siberia; whereas interglacial sedimentation was controlled by Atlantic Ocean water inflow (Haley et al.,

2008b). This contrast in glacial-Pleistocene εNd patterns in P23 and the ACEX area corroborates a large difference in Quaternary sedimentation between the Beaufort Gyre and the Transpolar Drift circulation systems as inferred from some other sedimentary proxies (Polyak and Jakobsson, 2011).

Figure 9. Comparison of P23 acetic εNd values and related lithostratigraphic proxies with counterpart data from the central Lomonosov Ridge (O’Regan et al., 2009; Fig. 1 for location). εNd scale is reversed in both panels.

Chapter 8: Summary

1) Sediment core 93-AR-P23, from the Northwind Ridge in the western Arctic

Ocean, was measured downcore for neodymium and strontium isotopic ratios in

sediment grain coatings to establish changes in paleo-seawater circulation. This

was accomplished by leaching sediment samples to isolate the grain coating

signal, followed by chromatographic separation of ions and analysis by mass

spectrometer.

2) Distribution of strontium and neodymium data suggests that acetic acid leaching

better captures the sediment grain coating, and thus the seawater signal, than

hydroxylamine hydrochloride leaching. This is consistent with previous studies

in the Arctic (Haley et al., 2008b).

3) Neodymium record in core P23 is generally consistent with those in previously

analyzed cores from the western Arctic Ocean (Winter et al., 1997), with an offset

in data sets possibly explained by differences in methodology. This similarity

indicates that the P23 record is representative of changes in circulation at a wide

range of depths throughout the western Arctic Ocean.

4) Based on overall high εNd values in the lower part of the record, the Early

Pleistocene western Arctic Ocean was likely influenced strongly by Pacific waters

and/or increased brine rejection due to seasonal sea ice formation on the Siberian

shelf. The highest εNd values probably mark the interval with strongest Pacific 36

influence. These conditions likely indicate relatively warm climatic environments

with predominantly seasonal sea ice, and thus can be potentially used as a paleo-

analog for the projected near-future state of the Arctic.

5) Oscillations in εNd in the Late Pleistocene correspond well with

glacial/interglacial lithological proxies, with very non-radiogenic values

consistent with glacial times. These low εNd values are probably due to leaching

of high-carbonate detritus originating from ancient rocks of the Canadian Shield

eroded by the Laurentide Ice Sheet. Additional investigation is needed to conclude

whether the detrital signal in leached sediment coatings is due to in situ chemical

leaching in pore waters, or by the laboratory process.

6) The more gradual trend toward non-radiogenic εNd in interglacial intervals is

consistent with a shift away from dominant Pacific waters and/or shelf brine

formation, and toward the present-day circulation, with Atlantic waters

controlling the εNd signal in the Arctic Ocean. Further studies of radiogenic

isotopes in Arctic sediment cores, especially with higher resolution and improved

age control, are needed to clarify the mechanisms involved in this shift and their

implications for the modern Arctic change.

References

Aagaard, K., Swift, J.H., Carmack, E.C., 1985. Thermohaline circulation in the Arctic

Mediterranean seas. Journal of Geophysical Research 90 (C3), 4833-4846.

Aagaard, K., Carmack, E.C., 1989. The role of sea ice and other fresh water in the Arctic

circulation. Journal of Geophysical Research 94 (C10), 14485-14498.

Aagaard, K, Coachman, L.,Carmack, E., 1981. On the halocline of the Arctic Ocean.

Deep Sea Research Part A. Oceanographic Research Papers 28 (6), 529-545.

Aksu, A.E., 1985. Paleomagnetic stratigraphy of CESAR cores, in Initial Geological Report

on CESAR-the Canadian Expedition to Study the Alpha Ridge, Arctic Ocean.

H.R. Jackson, P.J. Mudie, and S.M. Blasco (eds.), Geological Survey of Canada

Paper. 84-22, 101-114.

Allison, I., Beland, M., et al., 2009. The state of Polar research: a statement from the

International Council for Science / World Meteorological Organization Joint Committee

for the International Polar Year 2007-2008. World Meteorological Organization, 12 p.

Anderson, L. G., Olsson, K., Skoog, A., 1994. Distribution of dissolved inorganic and

organic carbon in the Eurasian Basin of the Arctic Ocean. Geophysical

Monograph 85, 255-262.

Andersson, P.S., Dahlqvist, R., Porcelli, D., Frank, M., et al., 2008. Neodymium isotopes

in seawater from the Barents Sea and Fram Strait Arctic-Atlantic gateways.

Geochimica Et Cosmochimica Acta. 72 (12), 2854-2867.

Björk, G., Winsor, P., 2006. The deep waters of the Eurasian Basin, Arctic Ocean:

Geothermal heat flow, mixing and renewal. Deep-Sea Research Part I,

Oceanographic Research Papers 53 (7), 1253-1271.

Björk, G., Jakobsson, M., Rudels, B., Swift, J.H. et al., 2007. Bathymetry and deep-water

exchange across the central Lomonosov Ridge at 88–89°N.. Deep-Sea Research

Part I, Oceanographic Research Papers. 54 (8), 1197-1208.

Channell, J.E.T., Xuan, C., 2009. Self-reversal and apparent magnetic excursions in

Arctic sediments. Earth & Planetary Science Letters 284 (124-131).

Clark, D.L., Chern, L.A., Hogler, J.A., Mennicke, C.M., Atkins, E.D., 1990. Late

Neogene climate evolution of the central Arctic Ocean. Marine Geology 93, 69-

94.

Crawford, K.A., 2010. The quaternary stratigraphy of the northwind ridge, Arctic Ocean.

M.S. thesis Ohio State University, 113p.

Dahlqvist, R., Andersson, P. S. and Porcelli, D., 2007. Nd isotopes in Bering Strait and

Chukchi Sea water. Geochim. Cosmochim. Acta 71, A196.

Darby, D.A, Polyak, L., Jakobsson, M., 2009. The 2005 HOTRAX Expedition to the

Arctic Ocean, Global and Planetary Change 68 (1), 1-4.

DePaolo, D. J., Wasserburg, G.J., 1976. Nd isotopic variations and petrogenetic models.

Geophysical Research Letters. 3 (5), 249-252.

Eisenhauer, A, H. Meyer, V. Rachold, T. Tutken, B. Wiegand, B.T. Hansen, R.F.

Spielhagen, F. Lindemann, and H. Kassens, 1999. Grain size separation and

sediment mixing in Arctic Ocean sediments: evidence from the strontium isotope

systematic. Chemical Geology 158 (3), 173-188.

Frank, M., 2002. Radiogenic isotopes: Tracers of past ocean circulation and erosional

input. Reviews of Geophysics 40 (1), 1-38.

Goldstein S.L., Hemming, S.R., 2003. Long-lived isotopic tracers in oceanography,

paleoceanography, and ice-sheet dynamics. In Treatise on Geochemistry (eds.

H.D. Holland and K.K. Turekian), p. 453-489

Haley, B A., Frank, M., Spielhagen, R., Eisenhauer, A., 2008. Influence of brine

formation on Arctic Ocean circulation over the past 15 million years. Nature

Geosci., 1, 68-72.

Haley, B.A., Frank, M., Spielhagen, R., Fietzke, J., 2008. Radiogenic isotope record of

Arctic Ocean circulation and weathering inputs of the past 15 million years.

Paleoceanogr., 23, PA1S13.

Hodell, D., Mead, G., and Mueller, P., 1990. Variation in the strontium isotopic

composition of seawater (8 Ma to present): Implications for chemical weathering

rates and dissolved fluxes to the oceans. Chemical Geology: Isotope Geoscience

Section. 80 (4), 291-307.

Holland, M. M., Bitz, C.M., 2003. Polar amplification of climate change in coupled

models. Climate Dynamics. 21, 221-232.

Jakobsson, M., 2002. Hypsometry and volume of the Arctic Ocean and its constituent

seas. Geochemistry, Geophysics, Geosystems 3, 5.

Jakobsson, M., Løvlie, R., Al-Hanbali, H., Arnold, E., Backman, J., Mörth, M., 2000.

Manganese color cycles in Arctic Ocean sediments constrain Pleistocene

chronology. Geology 28, 23-26.

Jakobsson, M., R. Macnab, L. Mayer, R. Anderson, M. Edwards, J. Hatzky, H. W.

Schenke, and P. Johnson, 2008. An improved bathymetric portrayal of the Arctic

Ocean: Implications for ocean modeling and geological, geophysical and

oceanographic analyses, Geophysical Research Letters, 35:L07602.

King, C., 1989. Cenozoic of the North Sea. In Stratigraphical Atlas of Fossil

Foraminifera. 2nd ed. D.G. Jenkins and J.W. Murray (eds.).

März, C., Stratmann, A., Matthiessen, J., Meinhardt, A.-K., Eckert, S., Schnetger,

B., Vogt, C., Stein, R., Brumsack, H-J., 2011. Manganese-rich brown layers in

Arctic Ocean sediments: Composition, formation mechanisms, and diagenetic

overprint. Geochim. Cosmochim. Acta 75, 7668-7687.

McCulloch, M.T., Wasserburg, G.J., 1978. Sm-nd and rb-sr chronology of continental

crust formation. Science, 200 (4345), 1003-1011.

Mullen, M. W., and McNeil, D. H., 1995. Biostratigraphic and paleoclimatic significance

of a new Pliocene foraminiferal fauna from the central Arctic Ocean. Marine

Micropaleontology 26, 273–280.

O‘Regan, M., King, J., Backman, J., Jakobsson, M., Pälike, H., Moran, K., Heil, C.,

Sakamoto, T., Cronin, M., Jordan, R., 2008. Constraints on the Pleistocene

chronology of sediments from the Lomonosov Ridge. Paleoceanography 23,

PA1S19.

O'Regan, M., St. John, K., Moran, K., Backman, K., King, J., Haley, B.A., J

akobsson, M., Frank, M., Röhl, U., 2010. Plio-Pleistocene trends in ice rafted

debris on the Lomonosov Ridge, Quaternary International 219, 168-176.

Peterson, B., McClelland, J., Holmes, M., Curry, R., Walsh, J., Aagaard, K., 2006.

Trajectory shifts in the Arctic and subarctic freshwater cycle. Science 313 (5790),

1061-1066.

Piepgras, D. J., Wasserburg, G.J, 1980. Neodymium isotopic variations in seawater, Earth

Planet. Sci. Lett. 50, 128 –138.

Polyak L., Jakobsson, M., 2011. Quaternary sedimentation in the Arctic Ocean.

Oceanography. 24 (3), 52-64.

Polyak, L., Bischof, J., Ortiz, J., Darby, D., Channell, J., Xuan, C., Kaufman, D., Lovlie,

R., Schneider, D., Adler, R., 2009. Late Quaternary stratigraphy and

sedimentation patterns in the western Arctic Ocean. Global Planet. Change 68, 5-

17.

Porcelli, D., Andersson, P.S., Baskaran, M., Frank, M., Björk, G., Semiletov, I., 2009. The

distribution of neodymium isotopes in Arctic Ocean basins. Geochim.

Cosmochim. Acta 73, 2645-2659.

Rudels, B., Jones, E.P., Anderson, L.G., Kattner, G., 1994. On the intermediate depth

waters of the Arctic Ocean. Geophysical Monograph 85, 33-46.

Rudels, B., Anderson, L.G., Jones, E.P, 1996. Formation and evolution of the surface

mixed layer and the halocline of the Arctic Ocean. Journal of Geophysical

Research, 101, 8807–8821.

Rudels, B., Stipa ,T., Kuzmina, N., Zhurbas, V., Schauer U., 2009. Double-diffusive

convection and interleaving in the Arctic Ocean - distribution and importance.

Geophysica. 45 (1-2), 199-213.

Schauer, U., Rudels, B., Jones, E.P., Anderson, L.G., Muench, R.D., Bjork, G., Swift,

J.H., Ivanov, V., Larsson, A.-M., 2002. Confluence and redistribution of Atlantic

water in the Nansen, Amundsen and Makarov basins. Annales Geophysicae.

Atmospheres, Hydrospheres, and Space Sciences 20, 257-274.

Serreze, M.C., Holland, M.M., Stroeve, J., 2007. Perspectives on the Arctic's shrinking

sea-ice cover, Science 315, 1533-1536.

Spielhagen, R., Baumann, K., Erlenkeuser, H., Nowaczyk, N., Nørgaard-Pedersen, N.,

Vogt, C., Weiel, D., 2004. Arctic Ocean deep-sea record of northern Eurasian ice

sheet history. Quaternary Science Reviews 23, 1455-1483.

Stein, R., Matthiessen, J., Niessen, F., Krylov, A., Nam, S., Bazhenova E., and Shipboard

Geology Group. 2010. Towards a better (litho-) stratigraphy and reconstruction of

Quaternary paleoenvironment in the Amerasian Basin (Arctic Ocean).

Polarforschung 79 (2), 97-121.

Stordal, M.C., Wasserburg, G.J., 1986. Neodymium isotopic study of Baffin Bay water:

sources of REE from very old terranes. Earth and Planetary Science Letters. 77

(3-4), 259-272.

Stroeve, J. C., Serreze, M.C., Holland, M.M, Kay, J.E., Barrett, A.P., 2012. The Arctic's

rapidly shrinking sea ice cover: a research synthesis. Climatic Change 110, 3-4.

Timmermans, M.-L., Garrett, C., and Carmack, E., 2003. The thermohaline structure and

evolution of the deep waters in the Canada basin, Arctic ocean. Deep-Sea

Research. Part I, Oceanographic Research Papers. 50, 10-11.

Tütken, T., Eisenhauer, A., Wiegand, B., Hansen, B.T., 2002. Glacial-interglacial cycles

in Sr and Nd isotopic composition of Arctic marine sediments triggered by the

Svalbard/Barents Sea ice sheet. Marine Geology. 182 (3), 351-372.

Wassmann, P. 2011. Arctic marine ecosystems in an era of rapid climate change. Progress

in Oceanography 90 (1-4), 1-17.

Winter, B., Johnson, C.M., Clark, D.L., 1997. Strontium, neodymium and lead isotope

variations of authigenic silicate sediment components from the late Cenozoic

Arctic Ocean: Implications for sediment provenance and the source of trace

metals in sea water. Geochim. Cosmochim. Acta 61, 4181– 4200.

Woodgate, R.A., 2005. Pacific ventilation of the Arctic Ocean's lower halocline by

upwelling and diapycnal mixing over the continental margin. Geophysical

Research Letters. 32 (18), doi:10.1029/2005GL023999

Woodgate, R.A., K. Aagaard, J.H. Swift, W.M. Smethie, and K.K. Falkner. 2007. Atlantic

water circulation over the Mendeleev Ridge and Chukchi Borderland from

thermohaline intrusions and water mass properties. Journal of Geophysical

Research. 112 (C2), doi:10.1029/2005JC003416.

Xuan, C., Channell, J.E.T., 2008. Rock magnetic study of Arctic deep-sea sediments.

IRM Quarterly, 18 (3), 3-4.

Xuan, C., Channell, J.E.T., Polyak, L., Darby, D.A., 2012. Paleomagnetism of Quaternary

sediments from Lomonosov Ridge and Yermak Plateau: implications for age

models in the Arctic Ocean. Quaternary Science Reviews 32, 48-63.

Appendix: Additional Data Tables

εNd ± 0.3 87/86S r Depth (cm) run 1 run 2 run 3 average run 1 run 2 average comment 2-3 -15.4 -15.4 4-5 -17.2 -17.2 8-9 -11.6 -11.6 14-15 -10.9 -10.9 20-21 -10.4 -10.4 30-31 -10.5 -10.5 34-35 -10.5 -10.5 36-37 -11.7 -11.7 44-45 -9.0 -9.0 52.5-53.5 -10.6 -10.6 58.5-59.5 -9.4 -9.4 67-68 -8.3 -8.0 -8.2 0.70993 0.70993 73.5-74.5 -9.2 -9.2 0.70912 0.70910 0.70911 84.5-85.5 2 .6 -0.6 0.7 0.0 0.70916 0.70914 0.70915 84.5-85.5 -9.0 -9.0 repeat Nd analysis 108-109 -9.7 -9.7 0.71016 0.70920 0.70920 discarded S r value 108-109 -10.1 -10.1 repeat Nd analysis 119.5-120.5 -10.1 -10.2 -10.2 0.70934 0.70932 0.70933 124-125 -9.1 -9.1 140-141 -9.4 -9.4 0.70981 0.70981 150-151 -9.2 -9.2 0.70914 0.70914 162.5-163.5 -8.8 -8.8 0.70927 0.70925 0.70926 177.5-178.5 -8.9 -8.9 0.70919 0.70918 0.70918 191-192 -8.8 -8.8 0.70914 0.70911 0.70912 191-192 -8.9 -9.0 -8.9 0.70919 0.70922 0.70921 repeat Nd analysis 207-208 -8.1 -8.1 0.70958 0.70958 223-224 -8.1 -8.1 0.70917 0.70915 0.70916 235-236 -8.1 -8.1 0.70915 0.70913 0.70914 243-244 -8.0 -8.0 0.70915 0.70913 0.70914 243-244 -7.9 -7.9 repeat Nd analysis 256-257 -8.0 -8.0 0.70914 0.70914 269-270 -8.5 -8.5 0.70912 0.70912 280-281 -8.2 -8.2 0.70912 0.70912 296-297 -8.3 -8.3 0.70912 0.70912 314-315 -8.2 -8.2 0.70951 0.70951 327-328 -8.3 -8.3 0.70950 0.70950 344-345 -8.2 -8.2 0.70913 0.70913 369-370 -8.8 -8.8 0.70984 0.70916 0.70916 discarded S r value 391-392 -9.0 -9.0 0.70964 0.70964 413-414 -9.1 -9.1 -9.1 0.70912 0.70912 458-459 -8.9 -8.9 0.70965 0.70965 476-477 -8.7 -8.7 0.70914 0.70914 510-511 -8.9 -8.9 0.70979 0.70979 510-511 -8.9 -8.9 0.70915 0.70915 repeat Nd analysis

Table 2. Corrected εNd and 86/87Sr data for acetic acid (“A”) leachates of core P23.

eND (+/-0.3) 87/86S r Depth (cm) run 1 run 2 run 3 run 4 average run 1 run 2 run 3 average comment 2-3 -5.4 -5.4 4-5 -11.20 -11.2 14-15 -9.27 -9.3 20-21 -6.8 -6.8 30-31 7.23 7.2 34-35 3.1 3.1 36-37 -10.7 -10.7 44-45 -1.8 -1.8 52.5-53.5 -10.3 -10.3 58.5-59.5 1.7 -0.52 0.6 67-68 -4.2 -5.8 -5.0 0.70971 0.70971 73.5-74.5 -1.6 -1.6 0.70986 0.70986 73.5-74.5 -8.08 -8.1 repeat Nd value 84.5-85.5 2.5 2.9 2.7 0.71001 0.71001 84.5-85.5 -7.5 -7.5 repeat Nd value 108-109 -6.9 -6.9 0.70912 0.70912 108-109 -10.6 -10.6 repeat Nd value 119.5-120.5 -2.7 -2.7 0.70914 0.70914 124-125 -7.1 -7.1 0.71030 0.71030 140-141 -5.7 -5.7 0.71050 0.71050 150-151 -4.8 -4.8 0.70915 0.70915 162.5-163.5 -2.1 -2.1 0.71112 0.71112 177.5-178.5 0.5 0.5 0.71012 0.71012 191-192 -2.3 -6.4 -5.4 -4.5 -4.7 0.71049 0.71051 0.71050 207-208 -6.3 -6.3 0.70967 0.71018 0.71018 223-224 -0.6 -0.6 0.70980 0.70980 235-236 -0.8 -0.8 0.71001 0.71001 243-244 -0.2 -0.2 0.70995 0.70995 0.70995 243-244 -6.3 -6.3 repeat Nd value 256-257 -5.4 -5.4 0.70915 0.70915 269-270 -6.8 -6.7 -6.8 0.70916 0.71031 0.70974 280-281 0.6 0.6 0.70996 0.70996 296-297 -2.1 -2.1 0.70962 0.71018 0.70990 314-315 -5.9 -5.9 0.70969 0.71001 0.70980 0.70983 327-328 -6.6 -6.6 0.70988 0.70999 0.70965 0.70984 344-345 -4.6 -7.2 -6.7 -6.2 0.70917 0.70917 369-370 -4.6 -4.6 0.70913 0.70913 391-392 -7.7 -7.7 0.70914 0.70914 413-414 1.1 1.1 0.70980 0.70977 0.70978 432-433 -0.5 -0.5 0.70977 0.70976 0.70977 458-459 -4.3 -3.0 -3.6 0.70910 0.70910 476-477 -0.4 -0.4 0.71007 0.71007 510-511 -7.8 -7.8 0.70912 0.70912 527-528 -2.2 -2.2 0.71004 0.71004

Table 3. Corrected εNd and 86/87Sr data for hydroxylamine hydrochloride leachates of core P23. 48

Material Core Water Core depth Clark's unit Clark's age P23 depth ɛNd dupl ɛNd residue dupl 87/86Sr oxide dupl depth (m) (cm) (Ma) (cm) oxide 1 pl. forams FL-200 3038 0-1 M top 0.05 1 -11.0 -12.7 0.70921 2 pl. forams FL-474 1647 0-1 M top 0.05 1 -11.7 -12.2 0.70920 3 nodules FL-286 2316 106-109 mid K 0.78 25 -12.2 -14.3 0.71101 4 nodules FL-286 2316 112-115 low K 0.8 30 -12.5 0.70994 5 nodules FL-286 2316 202-205 mid I 1 45 -12.7 0.71035 6 nodules FL-443 2436 220-221 mid G 1.3 70 -10.1 -10.2 0.70972 7 nodules CESAR 11 1380 207-209 E/F 1.5 125 -10.1 -9.4 -9.23 0.71094 8 nodules FL-443 2436 289-290 mid D 1.7 135 -10.5 -9.81 -11.1 0.71033 0.71 9 nodules FL-275 2884 206-207 low D 1.72 140 -9.4 -9.4 0.71018 10 nodules CESAR 11 1380 210-211 C/D 1.75 145 -9.0 -9.21 -9.8 0.70956 11 nodules CESAR 11 1380 225-228 A top 1.8 165 -9.2** 0.71264 12 nodules CESAR 11 1380 227-228 upper A 2 180 -9.8 -9.4 0.71057 13 nodules FL-380 2401 285-286 A2-A3 2.5* 250 -8.8 -8.47 -10.1 0.70953 0.71 14 nodules CESAR 11 1380 276-277 mid A1 2.9 300 -9.4 -10.8 0.70946 15 nodules CESAR 11 1380 333-334 upper A2 3.4 370 -10.5 -11.0 0.70922 16 nodules CESAR 11 1380 395-396 mid A2 4.1 420 -10.5 -11.4 0.70901 0.71 49 17 nodules CESAR 11 1380 446-447 low A2 4.6 470 -11.1 -11.2 0.70902 18 nodules CESAR 11 1380 447-448 upper A3 4.65 490 -10.6 -10.4 -10.9 0.70915

Table 4. Radiogenic isotope data from Winter et al. (1997). The corresponding depth in core P23 has been estimated

based on similarities in isotope curves.

*age is arbitrarily assigned, as Winter et al. did not estimate an age for this interval **data point incorporates both oxide and residue