Introduction
Integrated Ocean Drilling Program (IODP) “Wilkes Land” Expedition 318 collected seven marine sediment cores from sites off the Terre Adélie and George V Coast during the summer of 2010 (Fig. 1). The objective was to reconstruct East Antarctic Ice Sheet (EAIS) behaviour during climate fluctuations from the Eocene epoch to present-day as an analogue for ice sheet behaviour in response to future global climate changes (Escutia et al. Reference Escutia, Brinkhuis and Klaus2011).
Fig. 1 Four of the seven Integrated Ocean Drilling Program (IODP) Expedition 318 site locations on the George V continental shelf and rise. Site U1358 on the continental shelf is the focus of this paper (modified from Beaman et al. Reference Beaman, O'Brien, Post and De Santis2011). Inset A: simplified bed topography (after Fretwell et al. Reference Fretwell, Pritchard and Vaughan2013), WSB = Wilkes Subglacial Basin. The study area is indicated by a red square. Inset B: line drawing of Oligocene and younger sequences above unconformity WL-U3 in seismic line IFP 107 (cf. Escutia et al. Reference Escutia, De Santis, Donda, Dunbar, Cooper, Brancolini and Eittreim2005).
Located offshore from the Wilkes Subglacial Basin, the margin was deemed an ideal location for investigations of ice sheet dynamics because the ice sheet is grounded below sea level and therefore possibly sensitive to ocean and surface warming. On the continental shelf west of Mertz Bank, two sediment cores from hole U1358 were retrieved using the rotary core barrel system (RCB). Core U1358B was the more successful of the two that retrieved marine and glacial sediment down to 35.6 m below sea floor (b.s.f.) (Expedition 318 Scientists 2011). A multichannel seismic reflection profile revealed a significant unconformity (WL-U8) at c. 165 m b.s.f. (Fig. 1), believed to indicate a change in glacial thermal regime in the mid- to late Miocene (Escutia et al. Reference Escutia, De Santis, Donda, Dunbar, Cooper, Brancolini and Eittreim2005), however, WL-U8 was not reached during drilling (Expedition 318 Scientists 2011).
Site U1358 is located in the Mertz trough, between the Adélie and the Mertz banks, on the continental shelf off the Adélie Coast at 499 m below sea level (m b.s.l.) (Fig. 1). The sea floor in the Mertz Trough displays a megaflute morphology indicative of ice advance by the Mertz Glacier to an outer shelf position just landward of site U1358 most likely during the Last Glacial Maximum (LGM) (Domack Reference Domack1982, Beaman & Harris Reference Beaman and Harris2003). Today and in the past, shelf currents and iceberg scouring in water shallower than 500 m have modified the diamictons outcropping at the sea floor on the George V shelf (Beaman & Harris Reference Beaman and Harris2003). Seismic profiles show that sometime between the mid–late Miocene and the LGM the ice sheet oscillated with glacial advances to the shelf break and extended beyond the LGM position (Escutia et al. Reference Escutia, De Santis, Donda, Dunbar, Cooper, Brancolini and Eittreim2005).
IODP site U1358B provides a unique opportunity to assess a portion of this older record, specifically, to reconstruct early Pliocene–Late Pleistocene ice sheet dynamics at an EAIS-proximal site. The objective of this paper is to: i) observe and describe variations in diamicton depositional processes using particle size distribution analysis, ii) investigate diamicton provenance and source terrains by conducting heavy mineral analyses, and iii) summarize changes in chemical weathering, sediment recycling, and biogenic productivity from the early Pliocene–Late Pleistocene. The compiled data from the three experiments were used to address the fundamental questions of ice sheet stability and drainage patterns on the Wilkes Land margin during this critical time of warming in the Southern Ocean (Whitehead & Bohaty Reference Whitehead and Bohaty2003, Escutia et al. Reference Escutia, Bárcena, Lucchi, Romero, Ballegeer, Gonzalez and Harwood2009).
Materials and methods
Cores from hole U1358B were described on-board ship through visual description of the cut face of the archive half (Expedition 318 Scientists 2011; Fig. S1, which will be found at http://dx.doi.org/10.1017/S0954102013000527). Core 1R consists of unconsolidated light brownish grey massive clast-rich muddy diamicton with trace abundances of diatoms. Cores 2R–4R consist of consolidated greenish grey to grey massive to crudely stratified clast-rich sandy and muddy diamicton (Fig. 2). A distinct colour change from greenish grey to grey was observed between cores 2R and 3R. Below 20.45 m b.s.f. the diamicton is sparsely stratified with planar horizontal to inclined beds and laminations. The diamictons contain between 5% and 7.5% clasts up to 24 cm in diameter (Fig. 2). Clast lithologies include facetted basalt and fine-grained metasediments, and polished granitic gneiss and quartzite, all with subangular to subrounded shape. Hole U1358B yielded 23 diamicton samples that were processed for laboratory analyses at Montclair State University (Montclair, NJ, USA).
Fig. 2 Core photographs of representative core sections. Arrows indicate sample locations. Depth in meters below sea floor (m b.s.f.) for the top of the core intervals are shown. Core numbers and offset in cm can be found at the bottom of each photograph. Of note are the large rock clast in interval 2R-1, 33–83 cm, the massive and homogenous nature of the diamictons in interval 3R-1, 5–55 cm, the inclined stratification in 3R-3, 50–100 cm, and the horizontal stratification in interval 4R-1, 52–102 cm.
Particle size analysis
A Malvern Mastersizer 2000 laser particle sizer was used to analyse grain size distributions and to determine the dominant diamicton sedimentation process on the Wilkes Land margin continental shelf. Based on the amount of available sample material, 21 of the 23 samples from hole U1358B were prepared for sediment particle size analysis (Konert & Vandenberghe Reference Konert and Vandenberghe1997, Sperazza et al. Reference Sperazza, Moore and Hendrix2004). Samples were prepared four at a time using 30% hydrogen peroxide to disaggregate the sediment and then boiled with deionized (DI) water and 10% HCl on a hot plate to remove carbonate and organic material. The disaggregated samples were centrifuged at 1500 rotations per minute (rpm) for 30 min and the supernatant liquid was removed from the centrifuge tubes. Sodium pyrophosphate was added as a dispersant and the samples were heated for complete dissolution of the dispersant. Once the samples were cooled, they were analysed on the laser particle sizer using a laser obscuration of 20–40%. Particle size distribution histograms were generated and raw data was utilized in the calculation of fractal dimensions with the purpose of determining changes in down core diamicton formation processes (Hooke & Iverson Reference Hooke and Iverson1995, Licht et al. Reference Licht, Dunbar, Andrews and Jennings1999, Benn & Gemmell Reference Benn and Gemmell2002, Principato et al. Reference Principato, Jennings, Kristjansdottir and Andrews2005).
Geochemical analysis
Major and trace element geochemical analyses of the matrix of 20 samples were carried out using a Jobin-Yvon ULTIMA C inductively coupled plasma optical emission spectrometer (ICP-OES; HORIBA Jobin Yvon Inc, Edison, NJ, USA). The fine fraction (< 63 μm) was isolated from bulk samples through standard wet-sieving with a rubber spatula and deionized water to break up the matrix. The fine fraction (< 63 μm) and coarse fraction (63–250 μm) were placed in ceramic bowls, dried at 75°C overnight, and transferred into vials. Sample preparation for ICP-OES analysis followed Murray et al. (Reference Murray, Miller and Kryc2000) and was carried out on the fine fraction (< 63 μm). Approximately 0.1 g of sample and 0.4 g of Lithium Metaborate flux (LiBO2) were mixed together and carefully transferred to graphite crucibles, which were placed in a furnace at 1050°C for 30–40 min. When removed from the furnace, samples were transferred into Teflon beakers with magnetic stir bars and 50 ml of 7% nitric acid and placed on a magnetic stir plate until the sample was dissolved. Once dissolved, it was poured over a filter into a sample bottle and placed in a refrigerator. Immediately before measurement on the ICP-OES, the samples, along with 12 USGS standards for calibration, were diluted using 2% nitric acid.
The ICP-OES geochemical data (Tables S1 & S2, which will be found at http://dx.doi.org/10.1017/S0954102013000527) is represented as element ratios (Fig. 3b). To infer the degree of chemical weathering, the chemical index of alteration (CIA) was calculated with a correction for the presence of biogenic and terrigenous carbonate (Nesbitt & Young Reference Nesbitt and Young1982). To assess sediment recycling we used Zr/Sc ratios, which are high for mature marine sediments on passive continental margins (McLennan et al. Reference McLennan, Hemming, McDaniel and Hanson1993). TiO2/Al2O3 ratios are primarily controlled by rock provenance and vary widely in soils developed on mafic vs felsic igneous source rocks (Nesbitt & Young Reference Nesbitt and Young1996). We calculated Ba-excess values to assess marine palaeoproductivity (Schenau et al. Reference Schenau, Prins, Delange and Monnin2001). For the calculation of the Ba-excess values the lithogenic Ba component was derived from the average whole-rock composition of clasts collected in dredges near the Mertz Glacier tongue (Goodge & Fanning Reference Goodge and Fanning2010). The clasts are probably derived from Mertz Glacier and are assumed to be a representative sampling of the local bedrock, which is largely inaccessible due to the presence of the ice.
Fig. 3 Downcore distributions of a. lithologies, b. major and trace element geochemistry, c. particle size distributions, and d. heavy mineral petrology for Integrated Ocean Drilling Program (IODP) hole U1358B. In d. numbers in brackets refer to mineral assemblages and source terrains reported in Table S3, which will be found at http://dx.doi.org/10.1017/S0954102013000527.
Heavy mineral analysis
A Hitachi scanning electron microscope (SEM) S-3400N with Bruker electron dispersive spectroscopy (EDS) system was used to determine the mineralogical variety in the sand fraction of 18 samples that were chosen based on the amount of available coarse fraction. Heavy minerals were separated by pouring the entire fine sand fraction (63–250 μm sieve residue) into 50 ml centrifuge tubes filled with heavy liquid (sodium polytungstate; 2.89 g c m-3). The centrifuge tubes were rotated for 15 min at 3000 rpm and when completed, dry ice was used to freeze the tips of the tubes to isolate the heavy minerals while pouring the lighter minerals through a filter. The remaining heavy mineral grains were filtered out, cleaned with deionized water, dried, weighed, and placed in 18 separate wells drilled in two lucite discs (nine wells per disc) using tape for the grains to adhere and prevent extraneous material from entering the specified wells. The wells were then filled with a 5:1 ratio of epoxicure resin and hardener. The analytical sides of the sample discs were diamond polished, then carbon coated and analysed at 15.0 keV with a c. 10 mm working distance. Thirty-five grains per sample were randomly selected for standardless EDS analysis. Electron dispersion spectra were compared to characteristic mineral spectra published by Reed (Reference Reed2005). Optical microscope analysis assisted in the identification of garnet and gedrite grains (Deer et al. Reference Deer, Howie and Zussman1992). Major elemental oxide weight percentages (wt%) with an accuracy of 10% for oxides with > 2 wt% abundances were determined using Esprit 1.9 software (Bruker). Compositions of pyroxenes and amphiboles were recalculated back to cation proportions using mineral formulae recalculation spread sheets from http://serc.carleton.edu/ and plotted on quadrilateral diagrams.
Results
Particle size distributions
Particle size distributions are variable and both bi-modal and uni-modal for cores 4R, 2R, and 1R, whereas the distributions for core 3R are more uniform and uni-modal (Fig. 3c). Fractal distribution values for all samples were between 3.0 and 3.5, with r 2 between 0.97 and 0.99 (Fig. S2, which will be found at http://dx.doi.org/10.1017/S0954102013000527). Despite this dominant trend, two samples from core 4R at 27.4 m b.s.f. and 27.28 m b.s.f. within stratified diamictons interbedded with mudstones displayed distinct coarse sand enrichments expressed as humps in both the fractal plots and the grain-size histograms.
Geochemical analysis
Al2O3/TiO2 ratios vary between 20.43 and 21.96 and support a uniform intermediate to felsic provenance for the fine sediment fraction. Carbonate percentages were measured shipboard on sediment samples in cores 1R, 3R and 4R and yielded 2.8, 2.6 and 3.4 wt%. The CIA ranged between 56 and 62 with one outlier sample in the core catcher at the base of hole U1358B yielding a CIA of 52 associated with exceptionally high CaO concentration (3.84 wt%). An outlier in Zr/Sc values was observed for one sample in core 2R at 9.03 m b.s.f., which also had an elevated Ba-excess ratio and a slightly elevated CIA of 62. Ba-excess values were the highest, however, in two samples in core 4R (Fig. 3b).
Heavy mineral analysis
Single mineral grains and minerals of rock fragments were assigned to consistent petrological assemblages that had been identified down-core. Pie diagrams were generated to provide a detailed representation of down-core variations in the relative abundance of rock types and protoliths in U1358B (Fig. 3d). Based on their co-occurrence in rock fragments, nine distinct petrological assemblages are distinguished, ranging from various grades of metamorphism with mafic/intermediate and mafic protoliths as well as a mafic dyke signature and a metamorphosed carbonate signature (Table S3, which will be found at http://dx.doi.org/10.1017/S0954102013000527). Other minerals not classified include monazite and pyritized biogenic clasts (Fig. 4). The quadrilateral diagrams reveal uniform down-core distribution of amphiboles such as kaersutite, actinolite, hornblende, and cummingtonite/grunerite and a few Ca-rich and Fe-rich pyroxenes (Fig. 5).
Fig. 4 SEM images of polished grains of pyritized biogenic fragments in sample U1358B-2R1-83-85 cm at c. 9.03 m b.s.f. This sample also displayed slightly elevated Ba-excess and Zr/Sc ratios for the bulk mud fraction and a sand mode in the particle size distributions (Fig. 3b).
Fig. 5 Recalculated cation proportions of Ca, Mg, and Fe of amphibole and pyroxenes plotted on quadrilateral diagrams. Symbol labels refer to core numbers (1R–4R) in hole U1358B.
Discussion
Depositional model
The grain-size distributions (Fig. 3c) and fractal dimensions (Fig. S2, which will be found at http://dx.doi.org/10.1017/S0954102013000527) of all 21 diamicton samples are characteristic of a source region dominated by glacial processes with an increase in the production of finer material through grain slippage and abrasion, and/or subglacial reworking of pre-existing (glacio) marine sediments (Hooke & Iverson Reference Hooke and Iverson1995, Benn & Gemmel 2002). The glacigenic origin for the diamictons recovered in hole U1358B is confirmed by the bulk major element analysis, which shows relatively low CIAs (56–62) indicative of parent rocks that are chemically unweathered (Nesbitt & Young Reference Nesbitt and Young1982).
Preliminary shipboard investigations concluded that the diamictons in the upper ∼9 m b.s.f. (core 1R and section 2R-1) and those, which are crudely stratified and interbedded with muds (below 20.45 m b.s.f.), were deposited from floating ice (Licht et al. Reference Licht, Dunbar, Andrews and Jennings1999). The stratification in sections 3R-3 and 4R-1 (Fig. 2) did not show evidence of boudinage or other characteristics typical of glaciotectonic laminations and were hence attributed to current activity. In agreement with this, the diamictons in cores 1R, and the upper portions of cores 2R and 4R show some enrichment of sand and some sorting (Fig. 3c). The coincidence of (non-glaciotectonic) planar laminations, mud interbeds, and unconsolidated diamictons in these intervals suggest that the primary sand modes are depositional or from winnowing, and not a reworked signature derived from the source sediment. Stratification and heterogeneous particle size distributions with sand enrichments characterize glaciomarine diamicts elsewhere on high latitude continental shelves (Licht et al. Reference Licht, Dunbar, Andrews and Jennings1999, Principato et al. Reference Principato, Jennings, Kristjansdottir and Andrews2005) and current activity and iceberg scouring are processes that have affected diamictons outcropping at the sea floor in this area during the current interglacial (Beaman & Harris Reference Beaman and Harris2003). In support of our interpretation, the glaciomarine diamictons in section 2R-1 and 4R-1 display elevated Ba-excess values (Fig. 2b) indicative of increased primary productivity (Schenau et al. Reference Schenau, Prins, Delange and Monnin2001). The inclined bedding of the stratified diamicton in section 3R-3 (Fig. 2) may signal iceberg scouring or a transition to more ice-proximal conditions, as discussed below.
The massive diamictons between ∼9 m b.s.f. and 20.45 m b.s.f. (bottom of core 2R and most of core 3R) were initially interpreted as subglacial deposits with possible remobilization by debris flows (Expedition 318 Scientists 2011). The uniform particle size distributions for the deposits in section 2R-CC and core 3R confirm the shipboard interpretations of subglacial deformation and debris flow as the dominant diamicton forming process (Licht et al. Reference Licht, Dunbar, Andrews and Jennings1999, Passchier et al. Reference Passchier, O'Brien, Damuth, Januszczack, Handwerger and Whitehead2003, Principato et al. Reference Principato, Jennings, Kristjansdottir and Andrews2005). Contemporaneous, sparse, diatom assemblages reworked into the tills in core 3R, however, are indicative of open-marine conditions (Expedition 318 Scientists 2011). Whereas the massive diamictons demonstrate that ice advanced to the outer shelf or shelf break during glacials of some part of the early Pliocene, the sparse diatoms reworked into these diamictons may represent retreat and open marine conditions during interglacials.
Sediment provenance
In shelf drill-cores, sand provenance studies can reveal valuable information on major changes in the loci of glacial erosion. From protolith/rock type pie diagrams (Fig. 3d), we observe a shift in provenance from a contribution from metapelitic rocks of the staurolite-sillimanite facies, and high mafic granulite-amphibolite rocks in samples from cores 4R and 3R, whereas samples from cores 2R and1R show an increase in the contribution from mafic greenschist rocks of prehnite-pumpellyite facies. A mafic igneous component remains constant throughout all sample intervals. The amphiboles (kaersutite, actinolite, hornblende, and cummingtonite/grunerite) point to a large contribution from high/intermediate grade metamorphic rocks (Fig. 5). The presence of cummingtonite and grunerite suggests that the amphiboles carry a Wilkes Land basement signature (Oliver & Fanning Reference Oliver and Fanning2002, Goodge & Fanning Reference Goodge and Fanning2010). Pyroxenes, possibly derived from basement granodiorites and/or the Ferrar Group, were identified throughout the core but comprise a negligible overall contribution to the heavy mineral assemblages in the fine sand fraction.
Specific U1358 mineral assemblages (Table S3, which will be found at http://dx.doi.org/10.1017/S0954102013000527) are indicative of the Archean metapelites and metagranitoids found in the region bounded by the Port Martin and Mertz Shear Zones (Oliver & Fanning Reference Oliver and Fanning2002, Goodge & Fanning Reference Goodge and Fanning2010). This cratonic terrane (Fig. 6) also contains mafic dyke intrusions, mafic granulite/amphibolite metamorphosed rocks, and marbles. Mineral assemblages, however, also identify a low-grade metamorphic rock (prehnite-pumpellyite facies), which resembles rock types found in outcrop far to the east in the Bowers Terrane of northern Victoria Land and Oates Coast (Wodzicki & Robert Reference Wodzicki and Robert1986). Wet-based ice discharging into Rennick Glacier covered this area until the late Miocene, followed by exposure and soil production (Mayewski et al. Reference Mayewski, Attig and Drewry1979, Van de Wateren et al. Reference Van der Wateren, Dunai, van Balen, Klas, Verbers, Passchier and Herpers1999). A more recent ice re-advance across this soil mantle led to the Rennick Glacier grounding line extending c. 43 km northward, followed by another, still ongoing, thinning and retreat (Mayewski et al. Reference Mayewski, Attig and Drewry1979, Pritchard et al. Reference Pritchard, Arthern, Vaughan and Edwards2009). Whereas material from this source area is a subordinate component in samples of cores 3R and 4R, its relative abundance is increased in the samples of core 2R. Hence, we postulate a change in source rock for the fine sand fraction within core 2R with the loci of sediment delivery through glacial erosion migrating eastward.
Fig. 6 a. Geological sketch map of East Antarctica with location of Integrated Ocean Drilling Program (IODP) site U1358 on the George V continental shelf overlain on bed topography and palaeobathymetry. Geology based on Gair et al. (Reference Gair, Sturm, Carryer and Grindley1969), Wodzicki & Robert (Reference Wodzicki and Robert1986), Oliver & Fanning (Reference Oliver and Fanning2002), and Goodge & Fanning (Reference Goodge and Fanning2010). NVL = northern Victoria Land, BT = Bowers Terrane, WT = Wilson Terrane, CG = Cape Gray, MSZ = Mertz Shear Zone, PMSZ = Port Martin Shear Zone. Bed topography and palaeobathymetry was plotted using GeoMap App (http://www.geomapapp.org) based on a data compilation of Ryan et al. (Reference Ryan, Carbotte and Coplan2009). Green areas are bed topography above sea level, the continental shelf is tan to pink, the continental slope orange, the continental rise is yellow, and the abyssal plain is blue. Iceberg tracks are based on satellite observations of Stuart & Long (Reference Stuart and Long2011). b. Scenario for sediment delivery during glacial minima during the early Pliocene with more icebergs melting near their source due to warm ocean temperatures. c. Scenario for sediment delivery during glacial maxima of the early Pliocene with ice extending to the shelf break supplying sand with local provenance.
Palaeoclimatological implications
Diatom assemblages provide tentative age control for hole U1358B (Expedition 318 Scientists 2011; Table S4, which will be found at http://dx.doi.org/10.1017/S0954102013000527). Based on the presence of Thalassiosira antarctica Comber, core 1R, the upper 0.42 m b.s.f., is assigned to the latest Pleistocene–Holocene (< 0.61 Ma), with the absence of Actinocyclus ingens Ratray with last occurrence (LO) at 0.54 Ma possibly constraining this interval to within the last ∼540 ky. A condensed interval or a hiatus was inferred in core 2R at ∼9 m b.s.f. Based on the presence of Thalassiosira inura Gersonde (2.54–4.74 Ma) and Thalassiosira torokina Brady (2.24–7.23 Ma), and the absence of Thalassiosira lentiginosa (Janisch) Fryxell with first occurrence (FO) at 3.99 Ma and Fragilariopsis curta (van Heurck) Hustedt (FO at 3.56 Ma), the strata recovered in the interval below the hiatus (∼9 m b.s.f in section 2R-1) are tentatively assigned an early Pliocene age.
The interval of no deposition or hiatus at ∼9 m b.s.f. (> 3.99 Ma to < 0.61/0.54 Ma) in U1358B overlaps with a widespread erosion surface (pp-12) with tentative age of 3.9–3.6 Ma on the Prydz Bay continental shelf (O'Brien et al. Reference O'Brien, Goodwin, Forsberg, Cooper and Whitehead2007) and a hiatus (> 4.2 Ma to < 3.4 Ma) within a diatomite in the AND-1B core marking an erosion or non-deposition event during full deglaciation of the Ross Sea (Naish et al. Reference Naish, Powell and Levy2009). The age control for the drill-holes mentioned here, however, is not sufficient to evaluate whether the erosion surfaces are coeval and the erosion surfaces are not directly correlated in seismic data.
Interestingly, we observe a significant change in provenance across the condensed interval or hiatus. Moreover, the sediment matrix of the diamicton sampled at 9.03 m b.s.f. in U1358B, which coincides with the hiatus, has an anomalous major and trace element composition and a bimodal grain-size distribution with a primary fine-sand mode consistent with sediment sorting and winnowing. SEM analysis revealed the presence of abundant biogenic pyritized sediment clasts suggestive of local scouring and winnowing of a substrate of biogenic sediments (Fig. 5).
Two different scenarios, or a combination of the two, explain the observed changes in diamicton sedimentology and sand provenance. First, variation in sand provenance may reflect a change in ice-flow direction, with sediment delivery directly by ice from more easterly sources during the Pleistocene compared to the Pliocene. Second, the data may reflect changes in sand delivery via iceberg rafting during interglacial conditions, with this sand fraction then being reworked into the diamicton upon glacial advance of the Mertz Glacier system.
Today, most icebergs follow the counter-clockwise path of the Antarctic coastal current (Stuart & Long Reference Stuart and Long2011). For the lower Pliocene glaciomarine strata (core 4R), the local signature of the sand provenance suggests low iceberg supply or survival rates from eastern sources, consistent with a somewhat retreated ice margin (Fig. 6b) and higher early Pliocene sea surface temperatures in the Southern Ocean (Whitehead & Bohaty Reference Whitehead and Bohaty2003, Escutia et al. Reference Escutia, Bárcena, Lucchi, Romero, Ballegeer, Gonzalez and Harwood2009). Diatom assemblages suggest a periodic high-nutrient, open-water environmental setting, similar to that of the modern-day Southern Ocean north of the winter sea ice extent (Expedition 318 Scientists 2011). In contrast, the local provenance of the ice-contact diamictons in core 3R is indicative of shelf-wide ice advances by the Mertz Glacier during the early Pliocene (Fig. 6c). Together, the sedimentological, diatom, and geochemical evidence suggests that the ice sheet may have lifted off the bed and stayed in a more stable retreated position during some portion of the early Pliocene (core 4R) with advances to shelf break at a later time in the early Pliocene (core 3R).
For the Pleistocene (cores 2R-1R), the increase in subglacial transport of sand from as far away as the Bowers Terrane would require unrealistically long glacial flow lines parallel to the coast and originating from Rennick Glacier (Fig. 6a). It is also noteworthy that the Al2O3/TiO2 ratios for the matrix of the diamictons remain constant throughout U1358 indicating little change in the provenance for the mud fraction (Fig. 3b). Given the distant location, a transport path for the sand fraction via iceberg rafting during interglacials seems more plausible. The increase in far-travelled material can be explained by the Late Pliocene decrease in sea surface temperatures in the Southern Ocean and an increase in iceberg survival rates (Whitehead & Bohaty Reference Whitehead and Bohaty2003, Escutia et al. Reference Escutia, Bárcena, Lucchi, Romero, Ballegeer, Gonzalez and Harwood2009). A Pleistocene configuration (cores 2R and 1R) with sand transport from the east is roughly in agreement with the previously predicted path of the last ice advance across the continental shelf in this area (Domack Reference Domack1982). However, based on a further increase in far-travelled sand-sized detritus originating from the Bowers Terrane, we also infer interglacial ice-rafting as a probable sediment transport mechanism for a portion of the sand fraction in the Late Pleistocene sediments.
In summary, during the early Pliocene a dynamic marine-based ice sheet retreated from the Wilkes Land continental shelf with periodic ice advances to the outer shelf beyond the LGM position. A change in sand provenance is indicative of a more stable Mertz Glacier system during the Late Pleistocene (< 0.54 Ma) with a larger proportion of sand delivered by ice rafting.
Conclusions
-
1) Pliocene–Pleistocene Wilkes Land margin diamictons are chemically unweathered with subglacial deformation and debris flow being the dominant forming process for the lower Pliocene diamictons in core 3R, whereas deposition from floating ice produced the lower Pliocene and Upper Pleistocene diamictons in cores 4R, 2R and 1R.
-
2) The sand provenance of the diamictons had a primarily local signature in the early Pliocene with an increase of a more distant source in northern Victoria Land during the Pleistocene suggestive of a more complex sediment delivery pathway.
-
3) The sedimentological and stratigraphic evidence is consistent with a dynamic ice margin with some retreat during the early Pliocene (> 3.99 Ma) with ice advances delivering local debris from the Mertz Glacier system to the outer shelf or shelf break. In contrast, a more stable ice margin is envisioned for the Late Pleistocene (< 0.54 Ma) with a decrease in the delivery of locally derived glacial debris and an increase in far-travelled ice-rafted material.
Acknowledgements
This research was supported by a shipboard salary and post-expedition activities award (# IUSSP410-T318A72) administered by the Consortium for Ocean Leadership, the National Science Foundation (award # OCE 1060080), and an undergraduate research award through the SHIP program at Montclair State University (sponsored by the Merck and Roche Foundation). Samples were provided by the Integrated Ocean Drilling Program. We thank the reviewers of this paper, Claus-Dieter Hillenbrand and Julie Brigham-Grette, for their valuable and insightful contributions.
