Volume 24, Issue 1
Regular Articles
Free Access

Discrimination of sources of terrigenous sediment deposited in the central Arctic Ocean through the Cenozoic

Nahysa C. Martinez

Nahysa C. Martinez

Department of Earth Sciences, Boston University, Boston, Massachusetts, USA

Search for more papers by this author
Richard W. Murray

Richard W. Murray

Department of Earth Sciences, Boston University, Boston, Massachusetts, USA

Search for more papers by this author
Gerald R. Dickens

Gerald R. Dickens

Department of Earth Sciences, Rice University, Houston, Texas, USA

Search for more papers by this author
Martin Kölling

Martin Kölling

Department of Geosciences, University of Bremen, Bremen, Germany

Search for more papers by this author
First published: 18 February 2009
Citations: 13

Abstract

[1] We analyzed a suite of sediment samples recovered in the central Arctic Ocean for major, trace, and rare earth elements in order to assess changes in terrigenous source material throughout the Cenozoic. The terrigenous component consists of two end-members. Input from a shale-like composition dominates bulk sediments, especially those deposited during the Paleocene and since the Miocene, and may represent sediment supply from the eastern Laptev Sea. Therefore, even though the environment and transport mechanisms may have varied from ice free to ice dominated, sequences of the early Paleogene and later Neogene appear to have been influenced by a single major terrigenous source. This suggests similar transport capabilities and trajectories for both ocean and drift currents through significant parts of the Cenozoic. Influence from a more mafic source appears to be more important through the early Eocene to the middle Miocene and most likely represents material from the western Laptev Sea or Kara Sea. Thus, Eocene major changes in surface water productivity appear broadly synchronous with those in terrigenous provenance. A combination of regional sea level variations, local shelf processes, and transport mechanisms are among the more probable causes for the observed source changes. Although the assignment of sources using chemistry presently is constrained by a lack of data from certain regions (e.g., eastern Siberian Sea) our results generally agree with inferences based on mineralogy or radiogenic isotopes and shed further light on long-term reconstructions of the central Arctic Ocean.

1. Introduction

[2] The Arctic Ocean (Figure 1) likely drove and responded significantly to global climate throughout the Cenozoic [Aagaard and Carmack, 1994; Spielhagen et al., 1997; Darby et al., 2002; Backman et al., 2004]. Because the Arctic may sensitively respond to future global climate change as well [Walsh, 1991; Winter et al., 1997], an improved understanding of Arctic Ocean paleoceanography should lead to a better definition of the role of the Arctic in global climate and ocean dynamics.

Details are in the caption following the image
Map of the Arctic Ocean and surroundings. Also shown are the Beaufort Gyre (BG), the Transpolar Drift (TD) and the main potential terrigenous source areas. The square box shows the location of the ACEX site. Modified from Jakobsson et al. [2000].

[3] Sediments in the central Arctic Ocean, particularly those on top of structural highs (e.g., Lomonosov Ridge) provide records of change in the region. Many researchers have explored the Plio-Pleistocene Arctic history by studying shallow piston cores [e.g., Winter et al., 1997; Clark et al., 2000; Jakobsson et al., 2001; Polyak et al., 2004]. A significant portion of this paleoceanographic research has focused on assessing changes in the source and transport of terrigenous material [e.g., Bischof et al., 1996; Bryce et al., 1997; Clark et al., 2000; Darby et al., 2002; Phillips and Grantz, 2001; Rachold, 1998; Schoster et al., 2000; Spielhagen et al., 2004]. These chemical, mineralogical, sedimentological, and/or isotopic studies have identified northern Siberia, northeast Siberia (including Chukchi Sea), and northern Canada, as the three main source areas for terrigenous supply. Discrimination between these possible sources has been used to infer changes in circum-Arctic land and shelf processes (e.g., ice sheet history), Arctic circulation, and their potential forcing(s) [e.g., Bischof et al., 1996; Bryce et al., 1997; Clark et al., 2000; Darby et al., 2002; Phillips and Grantz, 2001; Rachold, 1998; Schoster et al., 2000; Spielhagen et al., 2004].

[4] Until recently, the pre-Quaternary history of the Arctic Ocean remained poorly constrained because of logistical and technological limitations that precluded drilling in moving sea ice. Integrated Ocean Drilling Program (IODP) Expedition 302, also known as the Arctic Coring Expedition (ACEX), cored multiple holes on the Lomonosov Ridge. In total, ACEX recovered significant portions of an ∼430 m Holocene to Late Cretaceous record, albeit discontinuously because of several unconformities and coring gaps. The Cenozoic record shows the central Arctic Ocean to have been warm, ice free, and productive during the Paleocene to middle Eocene; cool with some ice in the middle Eocene (∼ 45 Ma); and following a major hiatus, cold and ice covered from the middle Miocene to present day [Moran et al., 2006; Stein et al., 2006]. The isotopic, sedimentologic, and mineralogic composition of the ACEX record has also provided new insights into provenance changes that are being used to further explore the history of the Central Arctic [Haley et al., 2008; Krylov et al., 2008; St. John, 2008; Darby, 2008], as will be discussed further below.

[5] Preliminary results from the Lomonosov Ridge, obtained by us during the onshore portion of ACEX, show that the chemistry of the sediments closely reflects the major lithological changes (Figure 2) [Backman et al., 2006a]. The distribution of various elements identify an upper siliciclastic rich Unit 1 (U1), an intermediate biosiliceous organic rich Unit 2 (U2), an upper, organic rich part of Unit 3 (subunit 3/1 of U3), and the lower part of a siliciclastic rich unit 3 (subunit 3/2 of U3). Major changes in detritally associated elements through U2 and subunit 3/1 were interpreted as recording a potential change of terrigenous sources [Backman et al., 2006a]. Here we have measured additional trace and REE data in a subset of the Arctic Coring Expedition (ACEX) samples. Using both data sets, we apply a multielemental proxy and multivariate statistical approach in order to further characterize the inorganic geochemistry of the sediments, and to identify the possible source areas and/or transport pathways of terrigenous matter in the central Arctic through the Cenozoic.

Details are in the caption following the image
Depth profiles of major and trace elements. Gray triangles are Boston University data. Black dots are Bremen data. Vertical dashed lines represent PAAS values. Horizontal dashed lines show the limits of the major lithoestratigraphic units as well as subunits 1/4, 1/5, and 1/6, and the major hiatus between 1/5 and 1/6. Horizontal gray solid line shows the bottom of the authigenic silica unit, which is lithologically (but not chemically) different from Unit 2. Profiles for Si, S, and Cl correspond to the Bremen data set [Backman et al., 2006a]. (right) A schematic stratigraphic column, showing the four lithostratigraphic units is shown. Age estimates as in the paper by Backman et al. [2008]. (left) Depth of the stratigraphic sequence.

2. Background

[6] Presently, sea ice covers most of the Arctic Ocean throughout the year, except for surrounding shelves, which become ice free during the summer [Spielhagen et al., 2004]. Arctic rivers deliver terrigenous material to these shelves; sea ice entrains this sediment and transports it across the Arctic Ocean by two main drift patterns, the Beaufort Gyre and the Transpolar Drift (TPD) [Nürnberg et al., 1994; Dethleff et al., 2000; Schoster et al., 2000]. Various studies have identified three broad areas for the source of the terrigenous material deposited in the Arctic Ocean. These are North Siberia, Northeast Siberia and Chukchi Sea, and North Canada. Some of the diagnostic geochemical and mineralogical characteristics of these areas are broadly constrained (Table 1). However, there are very few (or no) published works based on multielemental data sets of particulates from areas such as the McKenzie River (and/or Canadian platform), the Chukchi and North Siberian Sea, and even for the Kara Sea (Ob or Yenisei rivers).

Table 1. Summary of Geochemical and Mineralogical Characteristics of Some Circum-Arctic Regionsa
Sea/River Area/Lithology Geochemistry Mineralogy Reference
North Siberia (Taymir Peninsula)
E. Laptev Lena/Yana Siberian Platform: Cambrian/Precambrian limestones, Jurassic/Cretaceous terrigenous sediments, Quaternary alluvial sediments. Baikal folded region: Proterozoic metamorphic rocks. Verhoyansk Mountains: Paleozoic terrigenous. Triasic and Mesozoic volcanics and granitoids Shale-like and UC composition. High illite and chlorite, low smectite and kaolinite (compared to W. laptev) Rachold [1998], Dethleff et al. [2000], Schoster et al. [2000], Viscosi-Shirley et al. [2003]
W. Laptev Khatanga Siberian trap flood basalts (Permian/Triassic). Aldan highland: Archean/Proterozoic igneous and metamorphic rocks. High Cr/Al, Ti/Al, Ca/Al, clynopiroxene, and smectite. Low illite and chlorite (compared to E. Laptev). Enriched in Ca, Co, Cu, Fe, Mg, Ni, Ti, and V. Rachold [1998], Dethleff et al. [2000], Schoster et al. [2000], Viscosi-Shirley et al. [2003]
Kara Sea Putorana Mountains (part of the Siberian traps) Similar to W. Laptev but more enriched in smectite, Ni/Al, Cr/Al, Ti/Al, and with lower values of K/Al Schoster et al. [2000]
East Siberian
Indigirka and Kolyma (East Siberian Sea) Kolyma-Omolon: accretionary continental and island arc fragments. Verhoyansk Mountains superterrrain: Paleozoic terrigenous. Composition similar to average shale Viscosi-Shirley et al. [2003]
Chukchi Sea Chukotsk volcanic belt: felsic to mafic volcanics Chukotka terrain: indistinguish sedimentary rocks. Alaska: sedimentary rocks and accreated volcanic terrains. Basaltic-like composition. High Mg, illite, and chlorite. Smectite content lower than W. laptev or Kara Seas Viscosi-Shirley et al. [2003]
Northern Canada
Mckenzie River Sverdrup Basin, Mckenzie River Valley, Coronation Gulf Noncarbonate clasts. Abundant kaolinite and less smectite than North Siberia areas Bischof et al. [1996], Spielhagen et al. [1997], Phillips and Grantz [2001], Viscosi-Shirley et al. [2003]
Banks, Victoria, Queen Elizabeth Islands: early Paleozoic terrain. Carbonate-rich sediments Bischof et al. [1996], Spielhagen et al. [1997], Phillips and Grantz [2001], Viscosi-Shirley et al. [2003]
  • a UC, upper cust.

[7] Recent publications by ACEX participants have investigated sediment source changes during the Neogene and the top part of the Paleogene. Haley et al. [2008], for the relatively young part of the record (∼15 Ma, subunits 1/1 to 1/4), found the Sr-Nd isotopic composition of the bulk sediments to be most similar to sediments from the Eurasian shelf, and thus proposed predominant supplies of terrigenous material from this source area for the last ∼15 Ma. Krylov et al. [2008] investigated the clay and heavy mineral composition of sediments for the last 50 Ma (U1 and U2) and found sediments deposited prior to ∼14 Ma rich in clinopyroxene and smectite, whereas those deposited after ∼13–14 Ma were rich in hornblende and illite. This was interpreted as a switch from a western Laptev-Kara-Barent Sea to eastern Laptev–Eastern Siberia sources during the middle Miocene. St. John [2008], in a study of the composition and physical properties of IRD over the last ∼47 Ma, agrees with an overall Russian source, without differentiating between Western and Eastern Siberian sources. Finally, Darby [2008], using the composition of Fe oxide grains on IRD deposited over the last 14 Ma, suggested that for this time period shelves of both Siberia and North America (especially, the Canadian Islands) were important sources of sediments to Lomonosov Ridge.

3. Sites and Samples

3.1. Arctic Coring Expedition

[8] ACEX recovered sediments from five holes at four sites on the central Lomonosov Ridge, approximately 250 km from the North Pole [Backman et al., 2006b]. Samples and data from the holes have been integrated into a single composite sequence [Backman et al., 2006b], which has been assigned a “revised composite depth scale” (rmcd) [O'Regan et al., 2008b]. The total composite section sequence comprises 428 m of Holocene to Upper Cretaceous sediments and has been divided into four lithostratigraphic units (Figure 2) [Backman et al., 2006b]. Briefly, Unit 1 (Holocene to middle Eocene; 0–223.56 rmcd) consists of siliciclastic material with abundant silty clays, silty muds and clayey silts. Numerous sandy lenses and isolated pebbles were interpreted as fallout from IRD [Moran et al., 2006]. Near the bottom of this unit (200 rmcd), there is a major hiatus, represented by very slow or no deposition, or, indeed, erosion [Moran et al., 2006; Stein et al., 2006]. Unit 2 (middle Eocene; 223.56–313.6 rmcd) consists of very dark gray and black mud composed primarily of mud-bearing biosiliceous ooze with abundant pyrite and elevated total organic carbon (TOC, up to 2–3 wt%, compared to TOC < 0.5 wt% for most of Unit 1). The first occurrence of dropstones, among other observations, suggests the initiation of the transition between the ice-free (at least seasonally) and ice-covered Arctic waters [Moran et al., 2006]. Unit 3 (late Paleocene to early Eocene; 313.6–404.79 rmcd) consists of clays and silty clays with variable amounts of pyrite, siliceous material, and TOC that generally decrease from top to bottom. The PETM (∼55 Ma) has been recognized at ∼385 rmcd on the basis of the presence of Apectodinium Augustum (dinocyst biomarker) and a prominent negative carbon isotope excursion [Sluijs et al., 2006]. Sea surface temperatures (at least during the summer) were as high as 24°C, supporting ice-free conditions [Sluijs et al., 2006]. Unit 4 (Late Cretaceous; 404.79–427.63 rmcd) comprises silty clays and silty sands and is considered to be transitional to bedrock. The top of this unit is an unconformity.

3.2. Samples, Methods, and Data

[9] Two data sets are considered in this work. We first analyzed 156 discrete 10 cm3 samples of bulk sediment at the Department of Geosciences, University of Bremen, Germany, during the “onshore” component of ACEX [Backman et al., 2006c]. For these samples 48 major and trace elements were measured by energy dispersive X-ray fluorescence (Text S1). This set of samples spans the entire sedimentary sequence at a nominal spacing of 3 m, with some intervals at higher resolution.

[10] We also processed and analyzed 64 discrete 10 cm3 samples of bulk sediment in the Analytical Geochemistry Laboratory at the Department of Earth Sciences, Boston University (Text S1). Forty elements were measured (Table 2a2b) by Inductively Coupled Plasma Emission Spectrometry (ICP-ES) and by Inductively Coupled Plasma Mass Spectrometry (ICP-MS). These samples also span the entire sedimentary sequence and 40 of them are exact splits of the samples previously prepared and analyzed at Bremen. The purpose of these analyses was to measure additional trace elements, especially the REEs, which are important for discriminating sedimentary provenance [e.g., Olivarez et al., 1991; Ziegler et al., 2007, and references therein].

Table 2a. Major and Trace Element Concentrations, Integrated Ocean Drilling Program From Arctic Coring Expedition 302
Sample Depth (rmcd) Al (wt %) Ti (wt %) Fe (wt %) Mn (wt %) Ca (wt %) Mg (wt %) Na (wt %) K (wt %) P (wt %) Ba (ppm) Sr (ppm) Cr (ppm) Ni (ppm) Sc (ppm) V (ppm) Zr (ppm)
4C-01H-1W 20–22 0.20 8.34 0.520 5.34 0.286 4.04 1.71 1.86 2.15 0.085 542 212 129 84.5 16.4 176 167
4C-01H-2W 84–86 2.36 7.56 0.539 5.75 0.307 0.50 1.48 1.85 2.07 0.084 545 146 91 107.0 14.9 194 178
4C-02H-2W 64–66 5.62 7.84 0.468 4.67 0.149 0.46 1.13 1.70 1.90 0.075 609 161 105 52.5 14.1 156 158
3A-01H-4W 24–26 8.52 8.58 0.503 5.06 0.288 0.50 1.17 1.85 2.08 0.093 638 173 93 48.8 15.3 179 181
4C-03H-3W 44–46 12.41 7.46 0.494 5.06 0.089 0.42 1.15 1.84 2.10 0.093 618 133 80 44.1 12.4 179 160
2A-05X-2W 5–7 20.83 8.49 0.520 6.21 0.097 0.40 1.13 1.74 1.99 0.127 599 154 111 49.8 15.5 166 169
4C-06X-2W 56–58 22.01 8.82 0.504 6.16 0.079 0.40 1.22 1.69 1.89 0.120 562 153 106 52.0 16.8 187 165
2A-06X-2W 52–54 22.77 8.60 0.540 5.59 0.458 0.52 1.19 1.89 1.92 0.107 665 185 137 51.2 15.7 230 177
2A-07X-1W 114–116 27.02 7.59 0.496 5.59 0.059 0.46 1.12 1.87 2.02 0.108 712 171 75 45.9 13.6 189 164
2A-08X-1W 44–46 30.82 7.89 0.507 5.38 0.140 0.46 1.13 1.86 2.05 0.099 678 171 79 43.2 14.2 207 178
2A-10X-2W 070–72 41.52 8.71 0.514 4.84 0.200 0.47 1.04 1.85 2.14 0.093 725 193 100 42.0 14.7 202 173
2A-11X-2W 60–62 45.29 8.47 0.499 5.35 0.690 0.46 1.00 1.84 2.03 0.115 676 185 121 42.8 14.3 170 181
2A-12X-3W 57–59 50.84 8.76 0.513 5.03 0.269 0.46 1.04 1.69 1.99 0.066 658 176 98 40.2 14.8 221 184
2A-12X-3W 57B–59 50.84 8.07 0.514 5.07 0.269 0.43 1.05 1.68 2.02 0.070 663 170 83 42.0 13.7 227 167
2A-14X-3W 44–46 57.70 6.32 0.519 5.54 0.096 0.33 0.92 1.68 1.87 0.104 558 114 79 47.6 11.5 234 185
2A-16X-2W 90–92 66.55 7.90 0.541 6.40 0.189 0.37 1.17 1.55 1.93 0.086 663 146 90 49.4 15.5 222 175
2A-17X-2W 8 70.34 8.74 0.519 6.22 0.072 0.37 0.97 1.65 2.00 0.149 645 156 128 42.2 16.0 193 189
2A-20X-1W 44–46 81.39 8.71 0.517 5.06 0.028 0.36 1.03 1.72 1.94 0.073 701 165 82 43.5 16.5 291 172
2A-20X-2W 25–27 82.71 8.84 0.507 5.16 1.097 0.43 1.05 1.44 1.77 0.064 664 149 80 40.7 16.9 251 172
2A-20X-3W 114–146 85.10 7.57 0.532 4.95 0.029 0.32 1.01 1.56 1.97 0.075 649 140 81 48.3 12.9 263 186
2A-22X-1W 2–4 90.97 8.68 0.522 5.30 0.180 0.38 1.03 1.75 2.05 0.098 646 153 80 41.9 15.1 234 182
2A-25X-2W 25–27 107.21 9.34 0.517 4.63 0.071 0.33 1.00 1.68 2.05 0.068 617 152 100 42.0 16.4 248 188
2A-26X-2W 44–46 111.89 9.04 0.535 4.58 0.169 0.31 0.90 1.71 2.01 0.074 629 151 109 43.9 15.3 148 186
2A-29X-2W 25–27 124.70 8.20 0.479 4.39 0.051 0.35 0.81 1.59 1.74 0.078 690 162 89 37.9 13.8 231 164
2A-30X-1W 44–46 128.39 8.11 0.527 4.53 0.029 0.29 0.95 1.58 1.84 0.064 693 136 195 44.5 14.3 232 172
2A-30X-2W 25–27 129.71 6.34 0.522 4.82 0.026 0.26 0.81 1.53 1.73 0.082 636 99 78 40.7 10.7 231 186
2A-30X-3W 44–46 131.40 8.21 0.498 5.51 0.051 0.30 0.95 1.59 1.85 0.094 693 133 76 42.5 15.3 258 171
2A-32X-2W 25–27 136.20 9.47 0.529 3.86 0.026 0.27 1.00 1.64 2.02 0.043 794 142 132 44.4 16.8 235 201
2A-32X-2W(25B) 136.20 9.37 0.539 3.71 0.023 0.27 0.98 1.66 2.08 0.045 771 142 84 46.1 15.4 242 175
2A-32X-4W 44–46 139.41 8.91 0.543 4.76 0.037 0.27 0.95 1.74 2.02 0.078 681 140 80 40.9 16.5 249 203
2A-33X-2 143–150 SQC 142.70 9.28 0.511 6.28 0.083 0.27 0.99 1.48 1.97 0.107 731 141 114 46.7 17.4 189 193
2A-34X-2W 44–46 145.69 8.35 0.488 7.43 1.090 0.77 1.07 1.45 1.86 0.239 630 157 70 39.1 17.0 166 184
2A-35X-2W 25–27 149.27 9.52 0.537 6.20 0.240 0.29 0.97 1.55 2.08 0.121 678 142 115 45.0 19.6 225 203
2A-35X-5W 44–46 153.98 9.01 0.497 7.26 0.086 0.32 0.88 1.58 1.92 0.193 603 141 70 40.4 17.2 157 186
2A-38X-4W 44–46 164.78 9.58 0.511 6.19 0.682 0.30 0.98 1.55 1.85 0.117 782 183 71 69.1 17.3 266 181
2A-42X-2W 62–64 179.36 9.25 0.495 6.03 1.011 0.32 0.95 1.64 2.10 0.119 648 192 93 111.5 19.4 273 181
2A-44X-2W 44–46 188.68 9.48 0.505 2.82 0.027 0.30 0.81 1.57 2.22 0.045 2330 217 77 191.9 38.3 280 229
2A-48X-3 140–151 SQC 205.86 6.38 0.378 8.62 0.031 0.19 0.82 0.69 1.07 0.033 962 90 78 91.0 16.7 159 132
2A-49X-5W 44–46 212.24 7.31 0.424 8.14 0.029 0.25 0.93 0.96 1.31 0.034 815 110 101 69.4 20.2 236 157
4B-03X-2W 64–66 215.14 6.04 0.355 8.10 0.025 0.22 0.84 0.98 1.07 0.025 1294 84 78 64.9 15.7 179 129
4B-03X-2W 64A–66 215.14 7.08 0.384 8.06 0.025 0.20 0.87 1.06 1.17 0.027 1293 92 99 65.7 19.4 189 136
2A-52X-2W 4–6 221.32 3.78 0.219 4.80 0.093 0.56 0.57 1.48 0.98 0.090 171 86 52 44.8 11.2 147 101
2A-53X-2W 80–82 223.95 4.51 0.244 5.77 0.092 0.38 0.68 1.58 1.07 0.032 163 76 72 39.7 13.4 201 116
2A-55X-2W 80–82 231.78 2.75 0.149 4.80 0.088 0.36 0.47 1.52 0.61 0.026 124 62 41 33.1 8.4 134 68
2A-56X-2W 80–82 237.66 3.96 0.221 5.08 0.078 0.43 0.58 1.38 0.81 0.057 162 75 54 36.4 12.2 186 98
2A-57X-2W 80–82 242.16 3.39 0.178 4.76 0.073 0.53 0.53 1.70 0.89 0.085 136 81 48 37.2 11.1 169 90
2A-61X-2 80–82 261.60 3.25 0.153 5.35 0.078 0.36 0.54 1.38 0.66 0.035 122 62 42 38.0 9.7 162 71
2A-62X-2 80–82 266.54 3.29 0.157 5.22 0.075 0.39 0.59 1.42 0.76 0.049 109 65 43 33.5 10.2 164 67
4A-08X-1W 4–6 273.80 3.28 0.155 5.30 0.071 0.44 0.66 1.39 0.58 0.054 96 60 45 43.5 9.3 154 62
4A-09X-1W 4–6 278.80 3.66 0.158 6.22 0.082 0.61 0.71 1.40 0.62 0.132 103 68 49 42.0 10.7 155 65
4A-15X-CC (Fine) 313.35 2.54 0.116 4.87 0.050 0.55 0.52 0.37 0.45 0.109 148 48 34 32.3 7.9 129 51
4A-19X-1 144–151 SQC 313.43 2.83 0.124 5.75 0.062 0.31 0.52 0.80 0.60 0.030 100 51 26 29.2 6.6 77 55
4A-19X-2W 66–68 314.15 4.21 0.187 6.94 0.074 0.36 0.89 1.11 0.85 0.030 133 65 45 38.5 10.5 121 83
4A-21X-2W 66–68 322.93 2.36 0.099 4.18 0.055 0.30 0.41 0.97 0.55 0.036 84 48 18 23.8 4.7 56 43
4A-23X-1 SQC 334.50 3.74 0.164 5.36 0.063 0.39 0.69 0.92 0.85 0.043 124 66 34 42.6 8.9 108 76
4A-23X-2W 66–68 335.23 2.26 0.098 3.54 0.042 0.28 0.40 0.89 0.52 0.021 78 46 19 27.3 4.8 63 44
4A-27X-2W 66–68 361.02 8.35 0.417 7.09 0.012 0.51 0.92 1.35 2.47 0.077 377 231 110 107.5 23.3 287 170
4A-28X-2W 64–66 365.40 9.30 0.616 4.80 0.019 0.41 1.06 1.21 2.31 0.038 686 159 84 58.3 18.5 244 184
4A-29X-1W 97–99 369.27 8.72 0.485 2.39 0.018 0.41 0.95 1.36 2.34 0.031 2506 171 83 33.2 16.0 197 176
4A-30X-2 143–152 SQC 374.70 8.80 0.378 9.05 0.137 0.53 1.03 1.02 2.48 0.086 579 115 77 56.0 17.0 182 145
4A-34X-2 144–150 SQC 390.41 10.45 0.569 3.76 0.018 0.48 0.85 0.93 2.91 0.099 521 122 98 45.7 18.1 262 199
4A-35X-2W 75–77 394.91 9.89 0.585 3.53 0.022 0.41 0.86 1.08 2.57 0.074 536 118 104 49.2 17.9 269 207
4A-42X-1W 42–44 418.48 8.41 0.712 4.41 0.069 0.55 0.68 0.82 1.66 0.103 1567 113 101 55.7 17.7 156 262
4A-42X-1W 42–44 418.48 8.33 0.699 4.21 0.067 0.55 0.67 0.94 2.10 0.096 1468 110 109 54.0 17.7 156 264
Analytical reproducibility (%) 4 2 1 1 2 1 2 3 9 3 5
Table 2b. Major and Trace Element Concentrations, Integrated Ocean Drilling Program From Arctic Coring Expedition 302a
Sample Depth (rmcd) Li (ppm) Co (ppm) Cu (ppm) Zn (ppm) Rb (ppm) Y (ppm) Cs (ppm) La (ppm) Ce (ppm) Pr (ppm) Nd (ppm) Sm (ppm) Eu (ppm) Tb (ppm) Gd (ppm) Dy (ppm) Ho (ppm) Er (ppm) Tm (ppm) Yb (ppm) Lu (ppm) Hf (ppm) Pb (ppm) Th (ppm) U (ppm)
4C-01H-1W 20–22 65.9 70.6 62.9 138 124 30.31 8.08 42.31 97.46 10.81 37.53 7.61 1.70 0.99 6.13 5.48 1.10 3.40 0.51 3.09 0.48 4.87 30.13 17.65 2.54 65.9
4C-01H-2W 84–86 70.1 92.7 66.7 129 123 26.96 7.57 39.49 115.42 8.74 35.14 7.24 1.65 0.99 5.79 5.32 1.01 3.20 0.48 3.03 0.48 5.10 38.56 13.65 2.66 70.1
4C-02H-2W 64–66 53.8 35.0 38.0 105 110 24.80 6.33 35.22 74.38 7.69 31.36 6.39 1.45 0.83 5.06 4.57 0.92 2.81 0.42 2.67 0.41 4.55 19.78 10.56 2.19 53.8
3A-01H-4W 24–26 64.5 52.1 34.5 109 120 26.59 7.32 37.00 79.85 8.21 33.03 6.76 1.57 0.90 5.41 4.96 0.93 3.05 0.46 2.88 0.45 4.90 19.96 11.48 2.56 64.5
4C-03H-3W 44–46 64.9 16.8 29.4 120 116 20.03 7.44 30.32 66.31 6.54 28.51 5.73 1.29 0.72 4.51 3.84 0.76 2.36 0.35 2.19 0.35 4.45 23.24 9.33 2.17 64.9
2A-05X-2W 5–7 78.7 31.6 39.3 113 125 27.74 7.46 39.61 89.05 9.65 35.43 7.20 1.62 0.95 5.74 5.16 0.98 3.19 0.47 2.99 0.46 4.77 20.98 11.44 2.99 78.7
4C-06X-2W 56–58 75.0 31.6 22.3 119 126 25.23 7.41 37.62 83.44 9.89 34.28 7.03 1.62 0.90 5.50 4.91 0.98 3.03 0.46 2.82 0.44 4.69 13.57 14.91 2.59 75.0
2A-06X-2W 52–54 67.2 28.1 22.5 113 114 25.90 6.19 37.75 83.15 9.78 34.15 6.94 1.60 0.90 5.55 5.00 0.99 3.07 0.46 2.88 0.45 4.93 23.28 11.27 2.50 67.2
2A-07X-1W 114–116 75.9 25.2 17.6 107 126 23.73 6.87 40.11 88.60 9.86 36.31 7.33 1.72 0.92 5.63 4.83 0.89 2.91 0.43 2.72 0.42 4.63 9.84 12.07 2.40 75.9
2A-08X-1W 44–46 79.9 24.4 23.5 110 125 24.68 6.74 39.28 85.73 8.60 35.20 7.02 1.66 0.89 5.43 4.74 0.88 2.88 0.43 2.72 0.42 5.08 23.17 11.92 2.50 79.9
2A-10X-2W 070–72 69.3 23.6 26.1 145 121 23.42 6.35 41.19 89.28 10.55 36.77 7.29 1.66 0.90 5.58 4.72 0.92 2.86 0.43 2.64 0.41 4.86 61.84 12.33 3.04 69.3
2A-11X-2W 60–62 68.6 25.6 22.3 104 123 27.00 6.34 40.30 87.08 10.34 36.25 7.15 1.64 0.94 5.66 5.11 1.02 3.20 0.48 3.03 0.47 5.24 14.40 11.71 2.88 68.6
2A-12X-3W 57–59 69.3 20.6 25.0 100 118 22.69 6.32 37.90 82.91 8.72 33.92 6.83 1.63 0.87 5.22 4.60 0.85 2.78 0.42 2.63 0.40 5.15 22.58 11.86 2.22 69.3
2A-12X-3W 57B–59 69.3 18.7 25.0 106 119 21.53 6.40 39.24 84.35 10.13 35.14 7.01 1.56 0.84 5.21 4.40 0.85 2.64 0.40 2.48 0.38 4.77 14.62 12.72 2.13 69.3
2A-14X-3W 44–46 68.1 28.3 20.5 123 84 18.71 6.63 23.96 59.44 5.36 23.49 4.97 1.17 0.67 4.06 3.74 0.75 2.32 0.36 2.21 0.35 5.13 8.86 6.46 1.84 68.1
2A-16X-2W 90–92 72.3 26.8 25.9 112 116 27.28 6.60 40.25 92.36 10.60 37.17 7.58 1.76 0.98 6.00 5.34 1.06 3.24 0.47 2.85 0.45 5.15 15.44 12.41 2.41 72.3
2A-17X-2W 8 69.4 23.4 29.2 105 125 27.37 7.26 38.25 82.85 8.00 34.64 7.15 1.68 0.96 5.76 5.27 1.01 3.30 0.50 3.15 0.49 5.26 93.88 12.43 4.07 69.4
2A-20X-1W 44–46 75.9 21.6 27.4 119 114 25.40 6.39 41.87 101.26 11.36 40.01 8.13 1.88 1.02 6.33 5.36 1.04 3.14 0.47 2.92 0.44 4.98 31.14 13.93 4.17 75.9
2A-20X-2W 25–27 74.5 18.3 26.5 111 106 31.61 6.41 38.01 90.22 10.20 35.75 7.47 1.78 1.06 6.29 6.01 1.22 3.74 0.57 3.45 0.54 4.75 26.04 12.54 5.59 74.5
2A-20X-3W 114–146 96.6 26.7 24.4 107 121 23.03 6.99 39.37 90.70 9.87 36.05 7.32 1.67 0.92 5.60 4.82 0.90 2.86 0.43 2.73 0.41 5.19 37.55 12.67 2.98 96.6
2A-22X-1W 2–4 85.9 21.7 24.2 120 117 25.76 6.83 40.05 90.45 10.50 36.80 7.51 1.73 0.95 5.86 5.14 1.01 3.18 0.47 2.96 0.46 5.13 21.48 12.90 3.18 85.9
2A-25X-2W 25–27 90.0 20.0 25.8 175 117 23.34 6.88 42.22 98.84 11.14 38.84 7.81 1.75 0.94 5.84 4.87 0.94 2.93 0.44 2.74 0.42 5.38 19.87 13.58 3.34 90.0
2A-26X-2W 44–46 91.3 21.9 16.0 102 114 26.24 6.77 40.83 85.49 9.85 35.86 7.18 1.62 0.92 5.48 4.81 0.92 3.04 0.46 2.87 0.45 5.17 8.15 11.74 2.80 91.3
2A-29X-2W 25–27 74.5 18.0 24.9 93 98 23.93 5.15 40.15 96.73 10.19 35.81 7.22 1.66 0.90 5.61 4.81 0.94 2.88 0.43 2.67 0.41 4.80 17.06 10.73 2.45 74.5
2A-30X-1W 44–46 98.2 19.6 36.7 102 111 22.68 6.56 39.32 94.82 9.95 36.15 7.30 1.72 0.91 5.50 4.66 0.87 2.77 0.42 2.63 0.40 5.00 9.98 12.35 2.99 98.2
2A-30X-2W 25–27 85.3 18.4 23.7 113 80 17.42 6.32 22.45 54.06 5.65 22.87 5.01 1.18 0.66 4.04 3.61 0.70 2.14 0.32 1.99 0.31 5.12 11.02 5.62 1.70 85.3
2A-30X-3W 44–46 85.7 16.1 27.2 109 91 20.85 6.03 34.28 85.02 8.02 33.77 7.01 1.62 0.86 5.38 4.54 0.87 2.62 0.38 2.37 0.37 4.85 8.22 10.91 2.20 85.7
2A-32X-2W 25–27 118.9 18.4 29.1 167 106 23.41 7.16 38.93 94.92 9.45 36.93 7.49 1.77 0.93 5.64 4.88 0.90 2.87 0.43 2.71 0.42 5.57 21.66 12.59 6.23 118.9
2A-32X-2W(25B) 93.0 20.2 29.3 161 119 21.22 7.37 39.48 92.65 10.45 36.81 7.40 1.65 0.87 5.45 4.47 0.85 2.62 0.39 2.47 0.38 5.13 24.14 13.08 6.37 93.0
2A-32X-4W 44–46 118.7 18.9 27.2 107 116 24.85 6.96 39.52 89.77 8.75 36.19 7.25 1.69 0.92 5.59 4.94 0.93 3.06 0.46 2.95 0.46 5.60 30.08 12.25 2.95 118.7
2A-33X-2 143–150 SQC 106.4 21.7 18.9 113 118 24.90 6.90 40.27 93.17 10.19 37.09 7.61 1.77 0.94 5.72 4.96 0.93 3.03 0.45 2.83 0.44 5.35 12.00 12.70 2.43 106.4
2A-34X-2W 44–46 79.2 26.3 25.1 100 103 39.75 5.99 38.92 89.68 10.26 36.14 7.49 1.78 1.06 6.46 6.13 1.28 4.00 0.60 3.61 0.57 4.97 9.16 11.62 5.89 79.2
2A-35X-2W 25–27 103.3 26.2 27.4 125 117 30.28 7.08 40.51 94.54 11.00 38.84 8.04 1.82 1.05 6.29 5.78 1.16 3.56 0.55 3.35 0.53 5.68 13.05 17.63 3.01 103.3
2A-35X-5W 44–46 92.3 26.1 60.4 114 109 27.98 6.63 40.18 97.68 10.86 38.45 7.92 1.78 1.01 6.19 5.47 1.09 3.38 0.51 3.18 0.49 5.10 13.55 13.70 3.76 92.3
2A-38X-4W 44–46 143.0 48.1 39.4 116 113 29.92 7.43 45.91 129.25 11.81 44.51 9.28 2.14 1.19 7.17 6.19 1.13 3.63 0.54 3.25 0.49 5.19 30.34 14.98 2.95 143.0
2A-42X-2W 62–64 162.0 69.6 61.8 135 122 34.73 7.36 46.64 130.56 12.83 47.58 9.84 2.31 1.31 8.00 7.00 1.39 4.18 0.62 3.71 0.57 5.35 28.42 18.12 2.82 162.0
2A-44X-2W 44–46 125.3 170.7 118.6 297 128 52.14 8.40 63.39 266.35 18.03 72.56 16.16 4.12 2.27 13.81 12.08 2.24 6.43 0.92 5.48 0.83 7.52 29.07 19.35 8.99 125.3
2A-48X-3 140–151 SQC 48.9 128.4 111.6 234 99 24.17 6.41 31.23 66.11 6.90 29.64 6.47 1.75 0.85 5.12 4.69 0.85 2.70 0.40 2.52 0.39 3.83 32.78 13.72 11.34 48.9
2A-49X-5W 44–46 58.4 78.6 139.5 429 111 34.68 7.77 36.33 78.57 10.60 38.60 8.84 2.13 1.22 7.19 6.85 1.34 3.89 0.57 3.47 0.54 4.63 37.48 20.84 27.38 58.4
4B-03X-2W 64–66 49.4 86.4 90.5 168 102 22.57 6.53 30.17 62.97 7.67 29.31 6.31 1.45 0.81 4.98 4.46 0.87 2.61 0.39 2.44 0.37 3.96 45.99 13.65 8.94 49.4
4B-03X-2W 64A–66 47.0 84.6 106.4 159 100 21.88 6.52 29.38 60.87 6.31 28.78 6.25 1.46 0.80 4.92 4.40 0.86 2.58 0.39 2.39 0.37 3.86 29.45 16.65 8.95 47.0
2A-52X-2W 4–6 21.4 25.9 57.6 111 52 30.73 3.98 33.98 67.28 7.75 32.70 7.01 1.65 1.02 6.18 5.18 1.02 3.12 0.49 2.68 0.42 2.66 15.47 7.59 6.67 21.4
2A-53X-2W 80–82 26.7 28.6 61.1 100 58 15.71 4.51 28.14 55.54 6.16 24.42 4.95 1.14 0.62 3.91 2.99 0.56 1.72 0.27 1.54 0.24 2.99 18.12 8.79 5.66 26.7
2A-55X-2W 80–82 15.4 26.8 51.1 106 31 10.95 2.58 15.63 33.41 3.63 14.57 3.01 0.73 0.40 2.47 2.02 0.38 1.18 0.19 1.05 0.18 1.90 10.80 5.08 4.65 15.4
2A-56X-2W 80–82 23.7 26.6 66.8 107 50 20.77 4.12 33.10 64.93 7.09 29.03 5.99 1.40 0.79 4.94 3.84 0.74 2.23 0.35 1.93 0.29 2.84 13.70 7.58 8.22 23.7
2A-57X-2W 80–82 18.5 24.2 69.7 122 41 27.36 3.63 31.89 63.42 7.06 30.12 6.65 1.61 0.95 5.81 4.78 0.93 2.84 0.44 2.40 0.37 2.54 12.29 7.64 10.76 18.5
2A-61X-2 80–82 18.6 23.9 64.5 109 37 14.05 3.81 19.11 40.62 4.51 18.29 3.86 0.94 0.53 3.26 2.64 0.51 1.58 0.24 1.37 0.22 2.07 14.42 5.63 6.71 18.6
2A-62X-2 80–82 19.2 22.6 53.0 102 35 18.12 3.58 22.85 44.63 4.97 20.54 4.35 1.06 0.63 3.88 3.17 0.62 1.91 0.30 1.67 0.25 2.04 11.57 5.84 8.69 19.2
4A-08X-1W 4–6 18.2 26.9 61.0 132 36 16.46 3.10 19.45 41.47 4.73 19.50 4.23 1.06 0.61 3.64 3.01 0.58 1.74 0.27 1.51 0.24 1.83 11.64 6.28 8.53 18.2
4A-09X-1W 4–6 21.2 26.1 61.3 127 38 29.50 3.36 29.54 60.01 6.74 29.40 6.65 1.63 1.01 6.09 5.20 1.02 3.05 0.49 2.62 0.41 1.92 13.43 5.58 12.11 21.2
4A-15X-CC (Fine) 8.6 20.4 69.5 119 24 26.43 2.07 23.79 48.16 5.57 23.79 5.33 1.37 0.82 5.07 4.27 0.85 2.57 0.39 2.03 0.32 1.48 11.16 4.72 11.13 8.6
4A-19X-1 144–151 SQC 9.0 19.3 43.9 91 31 15.06 2.84 15.70 35.52 4.02 16.24 3.60 0.87 0.52 3.12 2.68 0.53 1.60 0.25 1.39 0.22 1.65 14.08 5.28 5.44 9.0
4A-19X-2W 66–68 16.1 23.6 50.5 112 47 17.18 3.99 22.27 45.85 5.11 20.66 4.40 1.05 0.62 3.78 3.11 0.62 1.88 0.30 1.68 0.26 2.35 17.66 6.77 6.66 16.1
4A-21X-2W 66–68 5.1 15.4 33.2 68 26 12.80 2.28 12.20 27.73 3.10 12.51 2.75 0.66 0.40 2.46 2.13 0.43 1.32 0.20 1.14 0.18 1.32 10.24 3.92 3.50 5.1
4A-23X-1 SQC 10.1 21.7 46.3 110 39 18.47 3.44 17.60 38.15 4.22 17.42 3.67 0.90 0.55 3.24 2.86 0.57 1.80 0.29 1.60 0.25 2.13 16.84 6.27 5.97 10.1
4A-23X-2W 66–68 5.2 13.6 34.1 68 23 9.91 1.99 10.57 22.21 2.61 10.54 2.16 0.53 0.31 1.88 1.64 0.32 0.99 0.15 0.85 0.14 1.26 9.84 3.66 3.96 5.2
4A-27X-2W 66–68 40.2 43.7 119.0 181 91 64.66 6.71 37.71 104.49 10.15 43.52 10.75 2.74 1.89 10.77 10.40 2.11 6.87 1.08 5.84 0.93 4.54 24.13 12.93 12.12 40.2
4A-28X-2W 64–66 53.2 18.3 78.3 99 101 30.39 6.96 40.26 107.76 9.92 39.31 8.18 2.01 1.09 6.72 5.46 1.05 3.27 0.53 3.01 0.47 5.07 24.97 12.16 4.79 53.2
4A-29X-1W 97–99 45.3 8.6 33.1 78 102 26.99 6.32 39.14 88.13 9.09 35.68 7.16 1.73 0.95 5.73 4.84 0.94 3.08 0.49 2.73 0.44 4.96 21.83 10.94 4.34 45.3
4A-30X-2 143–152 SQC 47.1 32.5 61.2 139 92 33.83 6.77 44.23 92.39 9.79 38.99 8.04 1.89 1.11 6.57 5.44 1.07 3.46 0.54 3.06 0.48 4.03 26.57 13.04 12.07 47.1
4A-34X-2 144–150 SQC 122.1 13.5 37.8 103 111 37.83 8.10 47.40 103.80 10.77 42.84 8.83 2.10 1.22 7.29 6.09 1.24 4.07 0.65 3.60 0.56 5.76 26.45 12.73 4.44 122.1
4A-35X-2W 75–77 120.1 14.4 38.8 123 105 32.57 7.44 43.56 100.27 9.84 40.09 8.13 1.93 1.09 6.63 5.42 1.09 3.53 0.59 3.35 0.52 5.67 25.36 11.76 4.54 120.1
4A-42X-1W 42–44 130.8 13.0 34.7 83 93 25.67 6.36 35.38 76.11 8.08 32.00 6.55 1.45 0.88 5.21 4.55 0.90 3.04 0.51 2.97 0.47 6.97 16.11 10.52 3.93 130.8
4A-42X-1W 42–44 132.5 13.0 36.8 88 95 26.12 6.42 36.10 77.12 8.11 32.24 6.60 1.45 0.88 5.19 4.56 0.91 2.99 0.52 2.94 0.46 6.96 16.57 10.61 3.94 132.5
Analytical reproducibility (%) 2.5 1.5 2.3 1.8 0.9 2.1 0.9 1.8 0.8 1.1 0.7 0.9 0.9 1.3 0.9 1.8 1.4 1.7 1.3 1.7 1.4 3.7 3.7 1.4 1.7
  • a S and Si concentrations are those measured at Bremen University [Backman et al., 2006a]. All other elements reported in this table correspond to analysis done at Boston University. Data overspecified for calculation purposes.

[11] The first data set (Bremen) is used in this work mostly when exploring chemical changes through the overall sedimentary sequence. Data tables, figures, and further discussion regarding this data can be found in the postcruise report [Backman et al., 2006a]. The second data set (Boston University) is used here specifically for terrigenous source discrimination.

4. Multivariate Statistics: Q-Mode Factor Analysis

[12] The combined geochemical data set was analyzed using multivariate Q-mode factor analysis. We took two approaches. First, the elements Al, Ca, Cl, Co, Mn, Ni, P, Rb, S, Si, and Ti, which collectively represent a wide variety of sediment components, were used to statistically define the sedimentary components. Second, a different suite of elements, including Al, Hf, K, La, Rb, Sc, Th, Ti, and Zr, all of which are commonly associated with the terrigenous component(s), were selected to investigate potential terrigenous source changes. Factor analysis techniques are summarized in Text S1 and are very similar to those used by Martinez et al. [2007]. Our interpretations are based on the resultant factor scores (the weight of each element on the discrimination of a single factor), compositional factor scores (the elemental composition of each factor), and square factor loading (the contribution of each factor to a single sample). In many cases care needs to be taken when interpreting compositional factor scores, since their actual values cannot always be matched directly to known sources. The most important strategy when using these composition scores is to pay attention to the relative “high” or “low” values of the elements and interelemental ratios between the factors (e.g., knowing that Factor X has the highest Ti/Al value of all factors is as (or, perhaps, more) meaningful than knowing the actual value).

5. Results and Discussion

[13] A first examination of the Boston University data confirms the main results obtained at Bremen University. Comparing those elements analyzed at both laboratories, the same patterns are found, although absolute values differ slightly (Figures 2, 3, and S1). A major difference was found when comparing some elemental ratios (e.g., Ti/Al, K/Al, Mg/Al, Rb/Al, and Th/Al). The preliminary results obtained at Bremen University show the highest contents of these detrital elemental ratios from ∼220–350 rmcd. The new data, however, do not record the same relative increase (see Section 6.1 and Figures 3 and S1) with the exception of La/Al, and Sc/Al (not analyzed in Bremen). We work with the newer data set when exploring changes in source material because the ICP-generated results are more analytically constrained. However, for understanding the entire sequence, both data sets are considered, as this gives a better stratigraphic resolution. We have not mixed the data sets for statistical analysis, and for each case we have specified which data set is used and why.

Details are in the caption following the image
Depth profiles of major and trace elemental ratios. Gray triangles are Boston University data. Black dots are Bremen data. Vertical dashed lines represent PAAS values. Horizontal dashed lines show the limits of subunits 1/4, 1/5, and 1/6, and the major hiatus between 1/5 and 1/6. Horizontal gray solid line shows the bottom of the authigenic silica unit, which is lithologically (but not chemically) different from Unit 2. Profiles for Si/Al, correspond to the Bremen data set [Backman et al., 2006a]. (right) A schematic stratigraphic column, showing the four lithostratigraphic units is shown. Age estimates as in the paper by Backman et al. [2008]. (left) Depth of the stratigraphic sequence.

5.1. Element Profiles: Overall Sequence

[14] Unit 1 is dominated by terrigenous material (>80 wt%) and shows a shale-like composition. Elements such as Si, Al, Ti, K, Zr, Rb, Sc, La (and other REEs), Th, and Hf, which are commonly associated with the detrital component, display values similar to Post Archean Australian Shale (PAAS) [Taylor and McLennan, 1985] (Figure 2). This is consistent with the initial observations [Backman et al., 2006a].

[15] Unit 2 and the top of Unit 3 are generally dominated by biosiliceous material (silica and organic matter), although the proportions and composition vary. The terrigenous material decreases considerably (averaging ∼30 wt%), because of dilution, and most detrital elements deviate significantly from PAAS. Si has its highest content because of the abundance of biosiliceous material (223.56–313.6 rmcd) and authigenic silica (313.6–350 rmcd). Similarly, Cl and Br values are elevated [Backman et al., 2006a]. The Cl increase is consistent with the relative high halite in Unit 2 (as observed by XRD [Backman et al., 2006a]). Between ∼220–350 rmcd, despite the abundance of pyrite [Backman et al., 2006a], a relative decrease of Fe and S, with respect to the top 20 m of Unit 2, again reflects dilution by the biogenic and organic material. Because ratios are unaffected by dilution, the high Fe/Al for the entire section between 200 and 350 rmcd reflects the high abundance of pyrite in this section. The increase of P/Al in Unit 2 reflects the common association of P and the biogenic component of sediments. Similarly, high Ca/Al reflects the relative abundance of gypsum and to a lesser extent the rare calcite [Backman et al., 2006a]. The major break in elemental contents occurs at ∼350 rmcd and not at the boundary between Unit 2 and Unit 3 (Figure 2).

[16] The bottom of Unit 3 (350–404.79 rmcd) is again dominated by terrigenous material. Most elements and elemental ratios show contents or values similar to those of Unit 1 (Figures 2 and 3), which is consistent with the similarities of terrigenous abundance (∼ 80 wt%) and lithology.

5.2. Q-Mode Factor Analysis: Overall Sequence

[17] To identify the broad occurrence of sedimentary components, we chose a set of elements that proved (by the correlation matri (Table S1) to be critical in representing the various components present in the sediments. Thus, the elements Al, Rb, Ti, S, Cl, Mn, P, Co, Ni, Si, and Ca were used for the analysis. Q-mode factor analysis yielded five factors that explain 97% of the data variability.

[18] Varimax factor scores (Figure 4b) and compositional factor scores (i.e., factor concentration) (Table 3) identify the various components, or end-members. Factor 1 explains 75 % of the data variability and possesses the highest concentrations of Rb, Ti, Si, and Ca, as well as high scores for these elements, representing the detrital component. The assignment of Ca to the detrital component reflects the low abundance of carbonate. Factor 2 explains 13% of the data variability and represents the sulfide component (pyrite) as S shows its highest concentration and score. Factor 3 explains 4% of the variability and is characterized by high Mn and Ni scores, identifying a hydrogenous/authigenic component. Factor 4 explains 3% of the data variability, has the highest Cl contents and scores, and thus we interpret this as the “evaporitic” component. Silicon is also important in defining this factor. The last factor (Factor 5; not shown on Figure 4a) accounts for only 2% of the data set variability and does not seem to represent any major component (see square loadings discussion, below). However, it has the highest Co and Ni contents and shows scores with strong Ca and P associations, so it seems to also represent a second, minor authigenic component.

Details are in the caption following the image
Results of factor analysis (sedimentary component discrimination). (a) Depth profiles of factor analysis square loadings showing the relative contribution of each factor (interpreted here as the detrital, sulfide, hydrogenous, and evaporitic components of the sediments). Also shown are the estimated terrigenous abundance, pyrite content, Mn/Al, and Cl profiles. (b) The varimax factor scores show the weight of each element on the definition of each factor. Dashed horizontal lines as in Figure 2.
Table 3. Factor Analysis Composition Scoresa
Element F1 (Terrigenous) F2 (Pyrite) F3 (Hydrogenous) F4 (Evaporitic) F5 (Authigenic)
Rb 129 18 84 5 97
Ti 5339 709 3567 71 3947
Mn 1847 b b 2331 4778
Cl 4737 3945 5821 14,186 6353
P 1180 370 1173 232 b
Co 81 87 71 22 103
Ni 56 47 41 b 71
S b 94,894 24,516 3968 21,853
Si 284,773 200,828 262,954 279,330 263,501
Ca 3985 1229 3902 1983 1424
Percent Explained 75 13 4 3 2
  • a Sedimentary components discrimination. All composition scores are reported as ppm. See Backman et al. [2006a].
  • b Information lost because of the extra positive rotation of the analysis.

[19] The bio-siliceous components are not directly captured by this statistical treatment. One problem is that silicon is found in opal and authigenic silica but also many common detrital minerals, so this element has little statistical power as a quantitative discriminator. Our chemical suite does not include enough elements that would be associated with organic material, and therefore the organic component was not resolved here statistically.

[20] To quantify the abundance of the various components, square loadings for each factor were plotted versus depth (Figure 4a). The interpreted detrital component (Factor 1) shows a depth profile very similar to those of Ti, Al, and the other refractory elements, and suggests average detrital contribution of ∼70 wt% for Unit 1, a decrease to 20% or less for Unit 2 and top of Unit 3, and higher average contributions of 45 wt% again for the bottom of Unit 3. These detrital contributions strongly agree with those calculated by normative methods (e.g., using PAAS), yet constitute an alternative and independent method of quantifying this component. Importantly, it uses more than one element and does not rely on any initial assumptions about the composition of the chemical detrital matter. The “sulfide” component (Factor 2) has a depth distribution that closely resembles the S and pyrite profiles. However, the estimated average 55 wt% of “sulfide” by the factor analysis in Unit 2 and top of Unit 3 is too high because of the failure of the analysis to resolve the siliceous biogenic component. The authigenic/hydrogenous component (Factor 3) shows low and homogenous contributions with the exception of a few isolated samples in the upper 200 m. The depth profile of this component is similar to the Mn profile and subsequently appears to represent authigenic Mn oxides. The fourth component (“evaporite”) contributes very little to units 1 and 3 but has average concentrations of 25 % in Unit 2. As expected, it resembles the depth profiles of Cl and Br. The relatively high scores of Si suggest that Factor 4 indirectly captures the biosiliceous component. Because opal, especially diatoms, has a high porosity, this component may be linked to high amounts of pore water and, therefore, Cl. In this way, Factor 4 might be not just a direct proxy of the evaporitic component but also an indirect proxy of the biosiliceous component. The fact that silica contents are high but F4 and Cl contents are relatively low from 300 to 350 rmcd can be explained by the conversion of opal to authigenic silica, which would decrease porosity. Factor 5 seems to be partially related to calcium phosphates since samples with the highest contributions of this component also show significant concentrations of fluorapatite.

6. Assessment of Terrigenous Sources to the Lomonosov Ridge

6.1. Elemental Correlations and Depth Profiles: Terrigenous Sources

[21] Using the 64 samples analyzed at Boston University, we assess changes in the source of the detrital material to the Lomonosov Ridge. A correlation matrix (Table S2) confirms that the elements Al, Ti, K, Zr, Rb, La, and Hf have the highest correlations among each other (r2 > 0.7). Scandium correlates well with Al, La, and the REEs (r2 between 0.7 and 0.9) but less so with other detrital elements such as Ti, K, and Rb (r2 ∼ 0.4). Thorium correlations are high with Sc, Rb Cs, La, Pr, and Nd, (r2 ∼ 0.7) but low with Ti (r2 ∼ 0.4), Zr (r2 ∼ 0.5), and Al (r2 ∼ 0.6). As expected, all the REEs are strongly correlated between each other (r2 between 0.7 and 1) and the middle and heavy REEs are highly correlated with Y (0.7 < r2 < 1).

[22] These general associations are evident in depth profiles (Figure 2). Because the biogenic component dilutes the nonbiogenic material, we have normalized all elements to Al such that we can study variations within the terrigenous component. This assumes that all Al resides in the detrital component. With such high terrigenous inventories, any scavenged Al [Murray and Leinen, 1996; Kryc et al., 2003] should be overwhelmed. Most detrital elemental ratios do not record significant variation through the sediment column (as compared to the Bremen results), and are close to those of PAAS (Figure 3). The exceptions are the profiles of La/Al, Th/Al, and Sc/Al, which increase through the bottom of Unit 1, Unit 2 in its entirety, and the top of Unit 3 (only Sc/Al), and are considerably greater than PAAS values. This suggests that terrigenous material in these sediments may differ slightly from that in most of Units 1 and 3 (see additional discussion in Text S2). Changes in the type of terrigenous material are also suggested by the relationship between Ti/Al and Sc/Rb (Figure 5), which discriminates the three units and especially separates Unit 2 from the rest.

Details are in the caption following the image
Ti/Al (g/g) versus Sc/Rb (g/g) scatterplot showing chemical discrimination between the three lithostratigraphic units. Lena (Central Laptev Sea) and Khatanga (Western Laptev Sea) river [Holemann et al., 1999], PAAS [Taylor and McLennan, 1985], lower-crust materials [Rudnick and Presper, 1990] and Archean shales [Condie, 1993] are also shown as comparison.

6.2. Q-Mode Factor Analysis: Terrigenous Sources

[23] Various combinations of detritally associated elements were used in the factor analysis in order to investigate terrigenous sources. Assessment of the elemental suite Al, Ti, Sc, Rb, La, and Th suggests the presence of three compositional end-members (Table 4). Factor 1 (F1) explains 66% of the total variability and is dominated by Al, Ti, and Rb (as seen in the varimax rotated factor scores; Figure 6b). Factor 2 (F2) explains 29% of the variability, with La and Th being unique in its discrimination, yet also including Sc. Factor 3 (F3) only explains 3% of variability, with Sc controlling its variance. The square loadings of these factors (Figure 6a) suggest F1 is the main contributor to most of Unit 1 (∼78% F1) and the bottom of Unit 3 (∼71% F1), while F2 contributes significantly to the bottom of Unit 1, Unit 2, and the top of Unit 3 (∼45% F2). The contribution of the third factor is smaller throughout all units but increases in distinct intervals through Unit 2 and Unit 3 (with minimal values in Unit 1, ∼9% in the intermediate section, and ∼2% in the bottom of Unit 3). When Cr is added to the elemental suite, a new factor is produced, with Cr being the sole element dominating its variance. However, the square loading depth profiles maintain the same overall pattern. Thus, regardless of the exact elemental groupings, it is clear that Al, Ti, and Rb group together and dominate Units 1 and bottom of Unit 3, and that La, Th, Sc, with varying amounts of Cr, and Ti are distributed between two other factors that identify Unit 2 and variability within Units 2 and 3.

Details are in the caption following the image
Results of factor analysis (source discrimination). (a) Depth profiles of factor analysis square loadings showing the relative contribution of each factor. The lithostratigraphic units are shown on the gray column. Dashed horizontal lines as in Figure 2. (b) Varimax factor scores showing the weight of each element on the definition of each factor. (c) Th-La-Sc Ternary diagram showing the resultant three factors (F1, F2, and F3), samples, and various rocks, sediments, and reference materials. PAAS is from Taylor and McLennan [1985]; Asian Loess and Kurile basalts are from Weber et al. [1996]; Archean shales are from Condie [1993]; Canadian upper crust is from Shaw et al. [1986]; Lena, Yana, and Khatanga River particulates are from Holemann et al. [1999]; and MORB is from Salters and Stracke [2004]. Notice that all samples plot on a mixing field between the three factors. Lower crust is from Rudnick and Presper [1990] and Rudnick and Fountain [1995].
Table 4. Factor Analysis Results Aa
Element F1 F2 F3
Al 71,659 76,269 70,089
Ti 4902 b 6452
Sc 5 51 91
Rb 95 111 35
La 8 211 b
Th 8 34 9
Percent Explained 66 29 3
  • a Terrigenous source discrimination. See Boston University Data Set. All composition scores are reported as ppm.
  • b Information lost because of the extra positive rotation of the analysis.

[24] We have compared the factor compositions to natural rocks, representative reference materials, and sediments that have been suggested as potential contributors of material to the Arctic Ocean (Figure 6c). Data and factors have been plotted on a La-Th-Sc ternary diagram [Olivarez et al., 1991; Weber et al., 1996; Ziegler et al., 2007]. Not all samples plot on a simple mixing line between a felsic and a mafic end-member. However, all samples plot on a mixing field between the three factors and especially close to F1 and F2. When considering only La, Th, and Sc, there is no clear discrimination between samples from one unit to another as they all plot close to each other. The samples are also similar in composition to PAAS and to particulates carried by rivers draining the Laptev Sea (e.g., the Lena and Yana Rivers).

[25] A comparison including the rest of the elements, and paying special attention to those that are stronger in defining each factor, show similarities between the factor compositions and some reference materials. For Factor 1, concentrations of Al, Rb and Rb/Al are similar to those of Lena and Yana Rivers. Some of these concentrations are also comparable to NASC, PAAS [Gromet et al., 1984; Taylor and McLennan, 1985] and Precambrian Upper crust material from Canada [Shaw et al., 1986]. The Ti and Ti/Al values, although slightly higher than PAAS and Yana River sediments, are too low with respect to basalts [Rudnick and Fountain, 1995]. The Sc/Al values of Factor 1 are the lowest of all factors and therefore more similar to those of Lena and Yana sediments than to basalts, Western Laptev Sea, and Kara Sea sediments. The high concentrations of 200 ppm La and 34 ppm Th for Factor 2, and 91 ppm Sc for Factor 3 are difficult to explain. Lanthanum values greater than 100 ppm are commonly found in slowly accumulating sediment, usually associated with authigenic phases [Turekian and Wedepohl, 1961; Plank and Langmuir, 1998]. The contents of Rb and Rb/Al for Factor 2 are comparable to average shale and materials from Lena and Yana Rivers, however Sc/Al contents suggest a more mafic source. Factor 3 shows the highest values of Ti, Sc, Ti/Al and Sc/Al and therefore we suggest that either Factor 2 or Factor 3 (or a combination of both) might indicate inputs from a more mafic source.

[26] Although it does not seem very likely that in these sediments La, Th, and Sc have a diagenetic/authigenic association (Text S2), an additional set of statistical analyses was performed using Al, Ti, K, Zr, Rb, Hf, and Cr but avoiding La, Th, and Sc. Depending on the specific combination and number of elements used, various results were obtained, yet a consistent pattern is observed regardless of the precise element menu used. There are two main factors that explain most of the data variability (Table 5 and Figures 7 and S2). One factor is dominated by Al, K, and Rb, while the other is dominated by Ti, Zr, and Hf (Figure 7). The square loadings show that the amount of the Ti-Zr-Hf factor increases slightly toward the bottom of Unit 1, peaks over Unit 2, and decreases to a minimum in the middle of Unit 3 (∼350–400 rmcd, depending on the specific analysis (Figure S2)). The square loading profile of the Al-K-Rb factor exhibits a pattern opposite to that of the Ti-Zr-Hf factor. If Cr, Th or both are included in the statistical treatment, they generate new factors (i.e., factors with very high Cr and/or Th) with high contributions over Unit 2. The Factor 1 compositional scores (Table 5) show that the contents of Al, K, and Rb are similar to those from Yana and Lena rivers, as well as to some Archean and Precambrian shales. Titanium abundances in Factor 2 are instead comparable to those of arc basalts and MORB. The high composition scores of Zr and Hf are more difficult to explain. Some of the highest Zr/Al values reported from the Circum Arctic region are from the Yana River and the Kara Sea. High Zr/Al are also common on some lower-crust granulites.

Details are in the caption following the image
Varimax factor scores that result from the second set of factor analysis (not including La, Sc, and/or Th on the variable set). The scores represent the weight of each element on the definition of each factor.
Table 5. Factor Analysis Results Ba
Element F1 F2
Al 68,746 82,105
Ti 1614 10,837
K 23,107 b
Zr b 554
Rb 105 61
Hf b 15
Percent Explained 52 47
  • a Terrigenous source discrimination. See Boston University Data Set. All composition scores are reported as ppm.
  • b Information lost because of the extra positive rotation of the analysis.

6.3. Sources of Detrital Material to Lomonosov Ridge

[27] On the basis of our multielemental approach and the few relevant chemical data sets available, two general statements can be made with regard to potential terrigenous sources regions. First, there is no strong and/or clear chemical distinction between most of Unit 1 and Unit 3 (especially the lower part of Unit 3), while the middle portions of the recovered sequence, namely, Unit 2, the bottom of Unit 1, and perhaps the top of Unit 3, are subtly different. Second, the overall chemistry of the sedimentary column mostly resembles the composition of dominant sources to the Laptev Sea, namely the Lena, Yana, and Khatanga rivers and associated materials.

[28] Although most of the samples from all units plot close to each other on various ternary diagrams (e.g., Figures 6 and S2), there are subtle differences. The Ti/Al versus Sc/Rb scatterplot (Figure 5) shows significant discrimination, with samples from Units 1 and 3 closer to the average composition of Lena particulates, and Unit 2 closer to the average composition of Khatanga material. The Sc/Al, La/Al, and Th/Al profiles also suggest that sediments between 170 and 184 rmcd and 313–350 rmcd (i.e., the bottom of Unit 1 and through Unit 2) have a slightly different composition, and hence, inferred origin. Moreover, the composition of this interval is less similar to representative average shales, such as PAAS, compared to the adjacent sediments. For Ti/Al and Sc/Al this different composition is also observed in the top of Unit 3.

[29] Factor analysis results suggest the presence of two or three detrital sources. The first set of results (Figure 6) indicate a factor with compositions similar to average shale and sediments from Lena and Yana rivers that contributes its greatest amount over Unit 1 and through part of Unit 3. Factor 2 presents characteristic elevated La, Th, and Sc, and contributes significantly over Unit 2. Although difficult to match to a specific source, the relationships between the elements suggest this second factor represent materials from Lena or Yana Rivers but that a small mafic component can also be present. Factor 3, which contributes the most in the bottom of Unit 1, Unit 2, and top of Unit 3, shows elevated Ti, Sc, and therefore suggests a relative increase in the input of a more mafic source over this interval. The second suite of statistical results (Table 5 and Figures 7 and S2) supports the presence of 2 compositional end-members. The compositional scores and loadings of this analysis suggest that sediments of Unit 1 and Unit 3 are influenced by materials with a shale-like composition such as those from the Lena and Yana rivers, and that the input from sources with a more basaltic affinity relatively increased over Unit 2, and at least the bottom of Unit 1. The high Ti and Zr values of this Factor 2 are consistent with increased input from a more basaltic source such as the Kara Sea.

[30] The second set of statistical results is also noteworthy in that it agrees well with a factor analysis of suspended particulate material in Siberian Rivers [Rachold, 1998]. Both our results and those of Rachold [1998] present a Factor 1 that includes Al, K, and Rb as discriminatory elements. In the paper by Rachold [1998], this factor represented Lena and Yana River sediments. Moreover, both studies also present a Factor 2 dominated by Ti, Zr, and Hf, among other elements. Factor analysis results from surface sediments on the Siberian-Arctic shelf [Viscosi-Shirley et al., 2003] also yielded a primary factor with high Al and K factor scores (their Figure 5), which they interpreted as a shale end-member and was abundant for samples from the eastern Laptev Sea. The fact that our second set of results collectively propose a “clay” dominated factor (Al, K, Rb), and a “high field strength element” factor (Ti, Zr, Hf) may alternatively suggest that this second factor represents material derived from a different source whose composition has also been modified by grain size sorting effects, as was suggested by Rachold [1998]. Studying such physical processes is beyond the scope of this paper because we are dealing with bulk chemical analysis, but future studies may wish to explore this further.

[31] Finally, the interelemental ratios V/Al, Cr/Al, Sc/Al, and Ca/Al, found to be high on the Western Laptev or Kara Sea, are also relatively high over Unit 2. Although Ti/Al is not elevated, which is somewhat surprising given the observed Ti enrichment in many basalts [Taylor and McLennan, 1985], Ti/Al similar to those from Unit 2 have been reported for the Kara Sea [Gordeev et al., 2004]. Nonetheless, the preponderance of the chemical data (e.g., chemical ratios) and the multivariate statistics suggest that the interval defined by Unit 2, the bottom of Unit 1, and possible the top of Unit 3 as well, is compositionally different than the rest and could represent increased inputs from materials with a more mafic-derived source.

7. Comparison to Other ACEX Provenance Studies

[32] Our study supports the idea that the Eurasian margin contributed most of the sediment to Lomonosov Ridge during the late Neogene (most of Unit 1) and the middle Eocene (subunit 1/6 and Unit 2) [Haley et al., 2008; Krylov et al., 2008; St. John, 2008]. Our results also support the suggestion of Krylov et al. [2008] regarding a source change, within Eurasia, during the middle Miocene (at the bottom of Unit 1). Significant changes in the chemical composition of terrigenous material are found where their study shows a shift of heavy and clay minerals. Although the chemistry of the sediments does not uniquely identify the sediment source for the middle Eocene (Unit 2), our study is consistent with a change in the proportion of sediment derived from sources with more mafic-derived materials (Western Laptev or Kara Sea) to more shale-like (Eastern Laptev Sea) materials. In contrast, our results do not show evidence of a Canadian source to the Lomonosov Ridge during deposition of Unit 1 [Darby, 2008].

[33] Focusing on the Neogene section, the observed uniformity of our chemical proxies through most of Unit 1 is consistent with recent interpretations that the present ice drift pattern and dominant ice-based sediment transport mechanism have operated for at least 14 Ma [Haley et al., 2008; Krylov et al., 2008]. Because the source of the sediments deposited between 50 and 13–14 Ma (i.e., Unit 2 and bottom of Unit 1) is different, they may have deposited under different paleoceanographic conditions. As suggested by Krylov et al. [2008], this may reflect a change from seasonal sea ice to perennial sea ice at ∼13–14 Ma, so that ice-borne material from the west became relatively less important than material from the Eastern Laptev Sea. A working hypothesis is that abundant sediment from distal Eastern sources would not reach the Central Arctic until sea ice could survive a melting season [Darby, 2008; Krylov et al., 2008].

[34] Both the geochemistry (this study) and mineralogy [Krylov et al., 2008] are consistent with an increase in the input of western Laptev-Kara Sea sources of terrigenous material for the recovered period from 50 Ma to 13–14 Ma (comprising all of Unit 2 and the bottom of Unit 1). There is, of course, large unknowns with regard to the depositional processes that occurred during the hiatus found between ∼18.2–44.4 Ma, and thus we cannot specifically document that the proposed western source was continuously active for the whole period from 50 Ma to 14 Ma. Moreover, our work suggests that not only sediments from the top of Unit 1 but also those deposited before the middle Eocene (Unit 3, or at least the bottom part thereof) came from the eastern Laptev Sea. This suggests that (1) a similar transport path (sea ice) was present in the Arctic during the early Paleogene, (2) another source in the past furnished material with a similar chemistry to that entering the modern eastern Laptev Sea, and/or (3) a different transport mechanism (e.g., ice free) supplied fine-grained sediment from the eastern Laptev Sea to Lomonosov Ridge. Each of these three options is discussed sequentially below.

[35] With respect to the Option 1, it is well established that Earth in the early Paleogene was very warm, especially at high latitudes, and probably ice free. This view is consistent with studies of the ACEX cores, which contain lithological, micropaleonotological and geochemical evidence for warm temperatures and ice-free conditions for Unit 3 and much of Unit 2 [Brinkhuis et al., 2006; Moran et al., 2006; Sluijs et al., 2006, 2008]. Although limited sea ice may have occurred in the middle Eocene [Moran et al., 2006], it is highly unlikely that sea ice transported terrigenous material from the Laptev Sea to Lomonosov Ridge in the early Paleogene.

[36] Regarding Option 2, the geography of the Arctic Basin was significantly different during the Paleocene and Eocene, with the Lomonosov Ridge being located much closer to the Eurasian margin and in particular to the Barent, Kara, and Western Laptev Seas [O'Regan et al., 2008a; Sluijs et al., 2008]. Sediments deposited at that time, however, do not show chemical characteristics similar to modern sediments recovered from those western regions. There is a slight chance that the extent of chemical weathering during the “greenhouse” Arctic conditions of the early Paleogene could have modified the fingerprints of the more basaltic western sources. In spite of this, there is evidence that erosion and weathering has not significantly altered the composition of ACEX sediment, at other times. For example, the Pb isotopic composition of Neogene sediments do not reveal major weathering changes over the last 15 Ma, a time spans which includes the initiation of major Northern Hemisphere glaciation at 2.7 Ma [Haley et al., 2008]. Hence, major changes in erosion and weathering appear to not have significantly overprinted the inorganic chemical signatures of at least the Neogene and Pleistocene sediments. By analogy, the contrasting composition between Unit 3 and Unit 2 is unlikely to be due to changes in weathering of a single source (namely Barent-Kara-Western Laptev Sea). Future studies extending the Sr-Nd and Pb isotopic studies to the Paleogene sediments would help address this issue. Also, the fact that we are specifically and intentionally studying refractory chemical elements well known to be more resistant than others to alteration during weathering and diagenesis [e.g., Taylor and McLennan, 1985], reduces the chances of sources' compositional modification masking our provenance assessment.

[37] Assuming that proximal Paleogene sources were not as important as the more distal the Eastern Laptev Sea source, this may indicate a weakened or nonexistent Siberian Branch of the TPD, or reflect different continental and/or shelf processes that prevented these proximal sources from providing sediments to the Central Arctic. For example, Haley et al. [2008] proposed that glacier erosion could favor the transport of sediments originally eroded form the Putorana Basalts and temporarily deposited during interglacials on the Western Laptev Sea shelf. By analogy, sea level changes between the early Eocene–late Paleocene and the mid-Eocene could affect the entrapment and/or mobilization of sediment from the Kara, Barent, or Western Laptev. For example, a lowering of sea level during glacials may increase marine hemipelagic deposition as rivers can more effectively cut subaerially exposed shelf and deposit sediments further from the coast. Therefore, a scenario of sea level rise for the time of deposition of Unit 3 would favor sediment entrapment on the Western Laptev shelf and a sea level decrease during deposition of Unit 2, together with the start of seasonal sea ice, could be consistent with the geochemical source changes. Although the global sea level reconstructions of Miller et al. [2005] show slight increases in sea level between 50 and 55 Ma (part of Unit 3) compared to that from 45.5 to 50 Ma (Unit 2), we are not certain that these are enough to produce such results, and local variability is also likely to be important. Changes in coastal geomorphology, which may not leave any geological record over such a long timeframe, may also play an important role in determining transport pathways from shore to sea.

[38] With respect to Option 3, tectonic reconstructions of the Cenozoic evolution of the Arctic show that the distance between the ACEX site and the Eastern Laptev Sea has not varied significantly between the late Paleocene and the present [O'Regan et al., 2008b, Figure 8]. Also, as derived from Option 1, transport most likely occurred by ocean currents instead of sea ice or iceberg rafting. Thus, we suggest that ocean currents trajectories could have been similar to those of the modern ice rafting (TPD) and thus still be capable of transporting fine sediments to such distal locations as to the Lomonosov Ridge. In the modern, both icebergs and sea ice concentrate fine and coarse particles (IRD), with the fines being preferentially incorporated [St. John, 2008]. Therefore, we hypothesize that sediments from the Eastern Laptev Sea deposited under ice rafting and river input could carry fine grains and hence provide the same chemical signatures, provided their loads follow the same transport trajectories. Even if perennial sea ice is not the only transport mechanism from the Eastern Laptev Sea, this mode of transport could significantly enhance its contribution and therefore our interpretation is not inconsistent with those of the studies of mineralogy [Krylov et al., 2008] and Fe grain [Darby, 2008]. Processes related to the hiatus comprising ∼ 26 Ma could have affected circulation patterns for the time period of the depositional transition between Units 1 and 2, however on the basis of our observations we cannot resolve details at this respect.

8. Summary and Conclusions

[39] 1. Elemental contents of bulk sediments clearly define major boundaries between lithostratigraphic units. Moreover, as we initially observed with preliminary work [Backman et al., 2006a], the top 40 m of Unit 3 is chemically similar to Unit 2 such that many chemical profiles show a major change at ∼350 rmcd and not at the boundary between Units 2 and 3 (313 rmcd). This is because the top of Unit 3 contains abundant chert, which was presumably biogenic silica before alteration, and also serves as a significant dilutant. The paleoceanographic changes responsible for a switch from siliciclastic-dominated deposition to biogenic-dominated deposition occurred around 350 rmcd and therefore earlier than if one assumed it happened at the Unit 2/Unit 3 boundary.

[40] 2. Factor analysis and ratios of “detrital” elements suggest subtle differences in terrigenous material between lithostratigraphic units. Unit 1 (especially subunits 1/1 to 1/4) and Unit 3 (especially the bottom) have a shale-like composition, as shown by elemental profiles with values close to PAAS (and other upper crustal materials) and by the composition of Factor 1 in both of our detrital factor analysis statistical treatments. Additionally, there are multiple pieces of evidence that suggest that Unit 2 has a subtly different composition that could result of increased input from materials with a more basaltic composition.

[41] 3. The provenance of the terrigenous components cannot be definitively established because data is lacking from some important potential source regions, notably the East Siberian Sea, Chukchi Sea, and the Canada. Nonetheless, the overall chemistry of the terrigenous component sedimentary sequence appears to resemble the composition of modern surficial sediments from the Siberian shelf (e.g., Laptev and Kara Sea). Furthermore, comparisons of majors and trace elements as well as factor analysis suggest that the eastern and central Laptev Sea (Lena and Yana Rivers drainage area) can be a significant source of the shale-like sediments observed over the whole ACEX sequence, but in particular for those deposited over Unit 1 (subunits 1/1 to 1/4) and at least the bottom portion of Unit 3. Sediment from the Western Laptev or Kara Sea (areas drained by the Khatanga, Ob and Yenisei River) appears to be a good candidate for the potential “mafic” source that became relatively more important during the deposition of the top of Unit 3, all of Unit 2, and the bottom of Unit 1 (namely, from the early Eocene to the middle Miocene, 350 rmcd to 170 rmcd). Moreover, our results do not show evidence of a Canadian source to the Lomonosov Ridge during deposition of Unit 1 [Darby, 2008].

[42] 4. The similarity of terrigenous sediment in Unit 1, when there was major sea ice, and Unit 3, when there is no evidence for sea ice, suggests different transport mechanisms from a single terrigenous source. Clearly, ice rafting along the TPD transports fine-grained sediments from the Eastern Laptev Sea to the Lomonosov Ridge at present. This implies that ocean currents under ice-free conditions could be equally capable of this type of transport. This similarity is also unexpected since the two sedimentary sections must have been deposited under very different erosion and weathering regimes (precontinental and postcontinental glaciation). However, on the basis of the chemistry, the suite of refractory elements used, and the confirmation of no significant isotopic changes due to other major climatic transitions, we argue that the similar composition of the top of Unit 1 and the bottom of Unit 3 are not caused by weathering.

[43] 5. That terrigenous material deposited through Unit 2 is compositionally different than the rest of the sequence, supports the possibility of paired major changes in both the biogenic and terrigenous system during the middle Eocene. We suggest that regional sea level variations can be one of the main reasons causing this paired change and that variations in coastal geomorphology could play an important role on the entrapment and/or availability of sediments from the Western Laptev–Kara Sea during this time period.

Acknowledgments

[44] We thank the geochemistry group at Bremen University for their help with sample preparation and analysis for the Bremen data set. We also thank L. Bolge and J. Sparks for their assistance in the Analytical Geochemistry Laboratories at Boston University. N. Pisias provided the MATLAB script used in the factor analysis. This research used samples and data from the Integrated Ocean Drilling Program (IODP). This research was funded by postcruise grant support by the U.S. Science Support Program (USSSP) to R. W. Murray and N. C. Martinez at Boston University and separately to G. R. Dickens at Rice University.