Palaeoenvironmental interpretation of core ANTA99-23

After careful lithological, micropalaeontological and geochemical observations, and taking dating problems in this area into account, we consider that the continental slope core ANTA99-23 represented the last climatic phases (marine isotope stage MIS 1 and MIS 2 at least), in a glacial-deglacial environment, corresponding to the peak of the LGM (Fig. 6).

Fig. 6

Proposed palaeoenvironmental evolution model. a. Glacial ice shelf prograding phase with glacio-marine deposition. b. Advanced glacial phase (LGM) with total covering from ice shelf, grounded on the basin flanks. The shaded arrows represent gravity flows and turbidity currents. c. Transitional phase from glacial to interglacial environment. d. A Holocene phase of seasonal open marine conditions with high productivity of seasonal diatom blooms and stable summer stratification of the water column, typical of the seasonal ice zone (SIZ).

The basal unit 23a showed low productivity and a mixed diatom assemblage. Paralia sulcata was the dominant taxon in the assemblage, which also included, in almost equal percentages, reworked Miocene taxa such as Denticulopsis spp., and the Pleistocene diagnostic markers A. ingens, and R. diploneides (reworked assemblage similar to those recorded for example by Sjunneskog & Scherer (Reference Sjunneskog and Scherer2005) and references therein) in diamictons, muds and diatomaceous muds from the central Ross Sea). These observations, together with our lithological ones, led us to conclude that this basal part of the core was composed of a considerable amount of reworked material transported by ice streams and/or eroded from older strata cropping out along the trough flanks in the inner parts of the Ross basin (Mosola & Anderson Reference Mosola and Anderson2006, Bart & Cone Reference Bart and Cone2011). The TOC had low variability (mean values < 0.4%) as a reflection of the polar climate. The N values were constant for the interval considered (mean value < 0.5%), and the C/N ratio showed the lowest values of the whole core (unit 23a and partially 23b). Moreover, even Thalassiosira lentiginosa, making up to 14% of the assemblage, was found to be poorly preserved (broken or with strong dissolution traces). This species, associated with F. kerguelensis, is generally considered as an open-ocean form (Crosta et al. Reference Crosta, Romero, Armand and Pichon2005), listed in the ‘permanent open ocean zone’ group (POOZ assemblage). Nowadays, the relative abundance of T. lentiginosa has an inverse relationship with sea ice-covered marine environments (with sea ice duration up to four months) (Armand et al. Reference Armand, Crosta, Romero and Pichon2005). On the basis of laboratory experiments, it has been shown that when dissolution increases, F. kerguelensis is preferentially depleted rather than T. lentiginosa (Crosta & Koç Reference Crosta and Koç2007). Due to the poor state of preservation of T. lentiginosa and its presence together with sea ice-related species, in low percentages but well preserved (T. antarctica, F. sublinearis and F. obliquecostata), we believe the T. lentiginosa in unit 23a originally accumulated in an open-water environment but was transported here and stressed by a proximal glacier/ice or by current transport. These results agree with Finocchiaro et al. (Reference Finocchiaro, Baroni, Colizza and Ivaldi2007) and Bart & Cone (Reference Bart and Cone2011), who assumed that the central sector of the Ross Sea and part of its coast (south of the Drygalski Ice Tongue) enjoyed open marine conditions in the pre-LGM era (27 500–24 000 yr bp). This assumption is based on evidence from marine sediments, terrestrial geomorphology and ¹⁴C dating analyses on raised marine beach deposits of penguin bones. Bart & Cone (Reference Bart and Cone2011) have worked to offer a detailed 3-D bathymetric multibeam survey of the middle Ross Sea continental shelf, coupled with new radiocarbon dates of foraminifera tests and the AIO component, and compared it with previously published radiocarbon dates (Fig. 6a).

Unit 23b records a glacial prograding phase with a shift of the sedimentary depocentre towards the slope and a decrease/dilution of the biogenic supply. Poor or barren samples were represented with residual diatom assemblages (the same as recorded in the lower unit). The end of this phase is marked by the cross-laminated episode (gravity current effect?) that testified to a reactivation of currents with hydrodynamic changes, also suggested by the geochemical data (decrease in the N supply at the end of this unit and a slight increase in the C/N ratio), while a very limited diatom supply (only reworked taxa) continued. The geochemical data clearly indicated a marine environment with probably continental shelf organic input. This record has been interpreted as linked to a more intense glacial cover phase (LGM) and containing the first deglacial phase with the detachment of the ice sheet from the continental shelf and current reactivation. A proximal ice shelf edge could be proposed, as highlighted by McKay et al. (Reference McKay, Naish, Powell, Barrett, Scherer, Talarico, Kyle, Monien, Kuhn, Jackolski and William2012), with an increase in benthonic activity (bioturbation in the sediments as evidenced in logs, see Figs 2a & 3a) supported by primary production (not recorded in the sediment but resulting in an increase in bioactivity probably advected beneath the ice shelf) (Fig. 6 a & b).

The following unit, unit 23c, was related to the first deglacial evolution phase and was characterized by a gradual increase in the palaeoproductivity signal (C/N ratio value > 4) and a decrease in the supply of glacial material. In fact, a rapid drop in the number of reworked forms was recorded at c. 180 cm, corresponding to a marked iceberg-calving IRD level. Instead, the diatom assemblages showed a clear shift in favour of highly silicified and seasonal sea ice-related species (Chaetoceros resting spores, F. obliquecostata, E. antarctica, T. antarctica and F. curta) (Armand et al. Reference Armand, Crosta, Romero and Pichon2005). Eucampia antarctica var. recta, was interpreted in this work, sensu Burkle 1984 (see Whitehead et al. Reference Whitehead, Wotherspoon and Bohaty2005 and references therein). From the micropalaeontological proxies, an ice-dominated environment with a seasonally stratified water column, cold, salty autumnal water production and multi-year sea ice was presumed. Chaetoceros resting spores (composing up to 37.5% of the assemblage in this unit) were linked to seasonal stratified meltwater, with Chaetoceros spp. blooms and nutrient depletion in the final seasonal stages, while the low/null presence of vegetative cells attested to silica-selection dissolution. The present temperature/abundance range of F. obliquecostata shows a summer temperature preference between -1.3 and 2°C (with the maximum abundance between -1 and 0°C) and suggested the presence of summer sea ice (Armand et al. Reference Armand, Crosta, Romero and Pichon2005, Crosta & Koç Reference Crosta and Koç2007). Even E. antarctica var. recta is considered an ice form linked to sea ice or ice shelf-edge bioproduction (Armand et al. Reference Armand, Crosta, Romero and Pichon2005, Whitehead et al. Reference Whitehead, Wotherspoon and Bohaty2005). Thalassiosira antarctica has the highest abundances in the present-day surface sediments just south of the maximum winter sea ice zone (Armand et al. Reference Armand, Crosta, Romero and Pichon2005). Denis et al. (Reference Denis, Crosta, Zaragosi, Romero, Martin and Mas2006) interpreted T. antarctica as an open-sea form that cannot bloom where there is sea ice. On the contrary Maddison et al. (Reference Maddison, Pike, Leventer, Dunbar, Brachfeld, Domack, Manley and McClennen2006), supported by the surface sediment data from the Ross Sea obtained by Cunningham & Leventer (Reference Cunningham and Leventer1998), considered this taxon related to autumnal seasonal bloom events preceding seasonal sea ice formation. Due to its presence in our unit 23c assemblage, together with Porosira spp. (probably in low percentages as selectively dissolved) and its major presence just after the IRD event (see logs in Figs 2a & 3a), T. antarctica is here considered, in agreement with Maddison et al. (Reference Maddison, Pike, Leventer, Dunbar, Brachfeld, Domack, Manley and McClennen2006), to be in its autumn-spore form and linked to the seasonal formation of sea ice and its corresponding production of very salty surface water.

It should be noted that cold, salty water production is actually characteristic of this Ross Sea area. This production is typical of continental shelf hydrological-oceanic features leading to the production of HSSW and is somehow linked to MCDW input on the continental slope and to energetic gravity currents (Gordon et al. Reference Gordon, Zanbianchi, Orsi, Visbeck, Giulivi, Whitworth III and Spezie2004) (Fig. 6c). As a matter of fact, a second more marked deglacial phase is recorded in unit 23d (the upper part of the core) (see Fig. 6d). It shows a higher productivity signal and a diatom assemblage shifted towards seasonal sea ice diatom taxa. Fragilariopsis curta and Chaetoceros (Hyalochetae) (both spores and vegetative cells) testified to a better preservation of the sediment and a more defined sea ice environment involving spring sea ice melting, a stratified water column, seasonal diatom blooms and higher summer sea-surface temperatures (from -0.5 up to 3.5°C). The presence of E. antarctica var. recta and F. obliquecostata in the assemblages could suggest recurrent yearly episodes of summer sea ice and a lower seasonal sea-surface temperature (SST). The presence of F. kerguelensis and F. rhombica was also recorded in the unit 23d assemblage. As we have seen before, F. kerguelensis is an endemic Antarctic diatom presently more abundant in the POOZ (with an SST ranging between 1° and 8°C) and abundant in conditions of summer ice-free sediments. Due to these characteristics it is considered a good tool for tracing the oceanic and nutrient conditions south of the Polar Front. Recent surface sediment analyses generally suggest that F. kerguelensis and F. curta are at their ecological distribution limit as F. curta begins to bloom when F. kerguelensis tends to decrease (Armand et al. Reference Armand, Crosta, Romero and Pichon2005, Crosta et al. Reference Crosta, Romero, Armand and Pichon2005, Cortese & Gersonde Reference Cortese and Gersonde2007, Crosta & Koç Reference Crosta and Koç2007). This record indicates the beginning of ‘Holocene style’ sedimentation in a developing seasonal sea ice-cycle environment (dated 14 379 cal yr bp) and promoted by warmer water inflows (with supposed surface water temperatures up to 8°C). These conditions were probably due to progressively higher influxes of MCDW on the slope. Influxes on the continental slope have been suggested by Presti et al. (Reference Presti, De Santis, Busetti and Harris2003), Denis et al. (Reference Denis, Crosta, Zaragosi, Romero, Martin and Mas2006) and by Cortese & Gersonde (Reference Cortese and Gersonde2007) on the George V Land shelf and have been observed by Fusco et al. (Reference Fusco, Budillon and Spezie2009), Barbara et al. (Reference Barbara, Crosta, Massé and Ther2010), and Dinniman et al. (Reference Dinniman, Klinck and Smith2011) in the Ross Sea and other East Antarctic and Antarctic Peninsula zones. We noted the beginning of a constant seasonal MCDW intrusion of the slope in this core just after the peak of the LGM (at c. 14 380 cal yr bp). These water masses, related to the Ross gyre, could have been warm and dense enough to change the slope front position and reach the grounding line, but not able to reach the inner part of the continental shelf. This changed hydrodynamic situation is thought to have had important consequences for long-term deglacial timing, in water-sea ice dynamics (meltwater pulse events, see McKay et al. Reference McKay, Naish, Powell, Barrett, Scherer, Talarico, Kyle, Monien, Kuhn, Jackolski and William2012) and heat-flux exchanges, all key topics for destabilization of continental ice and polynya formation (Budillon et al. Reference Budillon, Tucci, Artegiani and Spezie2000, Fusco et al. Reference Fusco, Budillon and Spezie2009, Fielder Reference Fielder2010, Barbara et al. Reference Barbara, Crosta, Massé and Ther2010). The geochemical data, overall, support our micropalaeontological assumptions. In fact, in unit 23a and unit 23b low C/N ratio values were probably due to the contribution of inorganic nitrogen from reworking of the material transported by ice streams and/or eroded from older strata cropping out along the trough flanks in the inner parts of the Ross basin. The C/N ratios of the marine organisms typically range from 4–10, as observed from unit 23c to the top of the core, emphasizing the beginning of stable open-marine conditions (Meyers Reference Meyers1997, Hillenbrand et al. Reference Hillenbrand, Larter, Dowdeswell, Ehrmann, Ó Cofaigh, Benetti, Graham and Grobe2010).

In unit 23c and unit 23d, the TOC values remained relatively low compared to the mid-Holocene values and may reflect dilution of this signal due to increased silty-clastic sedimentation (as would occur in close proximity to a receding glacier) (Taylor et al. Reference Taylor, Whitehead and Domack2001).

Palaeoenvironmental interpretation of core ANTA96-5bis

Core ANTA96-5bis fell into the same time context as core ANTA99-23, but was collected in the eastern part of the Ross Sea sector examined. The massive diamictites identified were difficult to interpret and may represent more than one specific subglacial or grounding-zone environment (McKay et al. Reference McKay, Naish, Powell, Barrett, Scherer, Talarico, Kyle, Monien, Kuhn, Jackolski and William2012). Licht et al. (Reference Licht, Jennings, Andrews and Williams1996) and Salvi et al. (Reference Salvi, Busetti, Marinoni and Brambati2006) identified diamicton units north of 74°S on the western continental shelf of the Ross Sea and in the Joides Basin. We classified unit 5a as a glacio-marine diamicton, comparing our lithological and geochemical data with those of the other diamicton units studied in the Ross Sea. Substantial barren samples or low diatom frustule values, together with the uniformity of the diatom assemblages (mainly reworked taxa) were linked to a glacial progradation phase (Fig. 6a). The diatom taxa found in the assemblage were similar to those found by Sjunneskog & Scherer (Reference Sjunneskog and Scherer2005), and came from Tertiary-Miocene strata, while the continuous presence of P. sulcata (highly silicified, and glacial transport- and dissolution-resistant) was recorded.

Units 5b and 5c consisted of gravity-flow sediments derived from the flank of the Glomar Challenger Basin and deposited over glacio-marine diamicton. The fact that the TOC content was lower than that of the diamicton and that the C/N ratio was higher testified to a depositional environment proximal to the grounding line, lacking biosiliceous content and with probable terrigenous material input. This slumping episode could testify to the proximity of grounded ice and its subsequent instability (Mosola & Anderson Reference Mosola and Anderson2006). Unit 5d, interpreted by Salvi et al. (Reference Salvi, Busetti, Marinoni and Brambati2006) as another gravity flow (on the basis of the sedimentological and geochemical data) demonstrated a different biosiliceous supply (Fig. 6b & c). The gradual increase in the palaeoproductivity and a change in the diatom assemblage could be attributed to diatom remains advected beneath the ice shelf from open waters and diluted in the mass flow (lower diatom abundance than other Holocene sub-ice shelf mudstones deposited proximally to the ice shelf calving line - see McKay et al. (Reference McKay, Naish, Powell, Barrett, Scherer, Talarico, Kyle, Monien, Kuhn, Jackolski and William2012)).

This entire core section represents a subglacial environment, resulting from mass-flow deposition and caused by a glacial progradation phase or by an overriding ice sheet only anchored to the basin flanks (as topographic ‘pinning points’). A progressive decrease in water circulation and a null palaeoproductivity-proxy signal are typical of proximal grounding line processes (McKay et al. Reference McKay, Naish, Powell, Barrett, Scherer, Talarico, Kyle, Monien, Kuhn, Jackolski and William2012). Geochemical data suggest poor organic and biosiliceous preservation, the remaining diatom assemblages being linked to the establishment of pseudoneritic environments with highly silicified taxa and specific cold-water or multi-year-ice diatom species, such as Eucampia antarctica var. recta and F. obliquecostata. The last gravity flow episode in unit 5d could testify to the beginning of the deglaciation phase with the detachment of the ice shelf and its subsequent retreat (Fig. 6c).

The transition from diamicton to biogenic mud/ooze corresponds not only to a lithological change but even to a change in diatom assemblages (see the literature related to the eastern Ross Sea, and Licht et al. Reference Licht, Jennings, Andrews and Williams1996, Sjunneskog & Scherer Reference Sjunneskog and Scherer2005, Mosola & Anderson Reference Mosola and Anderson2006, Barbara et al. Reference Barbara, Crosta, Massé and Ther2010). Moreover Licht et al. (Reference Licht, Jennings, Andrews and Williams1996) referred to a hiatus from c. 20 000–7000 yr bp observed in some parts of the eastern Ross Sea. Even in our core, the transition to marine muds, represented in unit 5e, was marked by a condensed sediment or, more probably, by a hiatus (also confirmed by the time dataset) (Fig. 6d). This lack of sedimentation could be due to the onset of modern environmental conditions with the strong gravity currents observed today over the continental slope and derived from the export of HSSW. In effect, HSSW, together with the less saline shelf water and the super-cooled low-salinity Ice Shelf Water (that forms beneath the Ross Ice Shelf), today flows off the shelf through the Drygalski trough (unofficial name), the Joides Basin (immediately to the east) and the Glomar Challenger Basin (Gordon et al. Reference Gordon, Zanbianchi, Orsi, Visbeck, Giulivi, Whitworth III and Spezie2004, Fusco et al. Reference Fusco, Budillon and Spezie2009 and references therein).

In unit 5e micropalaeontological analyses (a peak in diatom abundances up to 6.62 x 10⁵ valves g^-1 ds together with the presence of autochthonous calcareous foraminifera), sedimentological observations (an increase in the muddy fraction) and geochemical proxies (increases in TOC and N with lower and constant C/N ratio) attested to a clear change towards more open marine environments, dynamic but not enough to remove material and with higher water column productivity, probably resulting from increased diatom remains. These observations are all consistent with sedimentation models of modern and early Holocene ice shelves (Sjunneskog & Scherer Reference Sjunneskog and Scherer2005, McKay et al. Reference McKay, Naish, Powell, Barrett, Scherer, Talarico, Kyle, Monien, Kuhn, Jackolski and William2012 and references therein).

Notes on the reworked diatom forms

In the sedimentological environments of the Ross Sea, the biosiliceous record is considered to be the mixed signal of local biological production and reworked material linked to glacial-current transport under and near the ice shelf. Miocene taxa were similar to those observed by Sjunneskog & Scherer (Reference Sjunneskog and Scherer2005) in this area while the Rouxia spp., found in core ANTA99-23, include those studied by Zielinski & Gersonde (Reference Zielinski and Gersonde2002) and R. diploneides (Fig. S2 C and H, which will be found at http://dx.doi.org/10.1017/S0954102013000199). This last species, which has its last appearance datum at 2.51 m.y.a. (Sjunneskog, personal communication 2011), has been founded in unit 23a in a statistically valid percentage but in a mixed assemblage and has been here considered as reworked. Miocene and Plio-Pleistocene sediments cropping out on the inner part of the continental shelf and presenting signs of several grounding events (Scherer et al. 2007 and references therein, McKay et al. Reference McKay, Naish, Powell, Barrett, Scherer, Talarico, Kyle, Monien, Kuhn, Jackolski and William2012, Winter et al. Reference Winter, Sjunneskog, Scherer, Maffioli, Reisselman and Harwood2012), could explain the presence of the Miocene and Plio-Pleistocene reworked taxa in more recent continental slope sediments. Mosola & Anderson (Reference Mosola and Anderson2006) highlighted the glacial dynamics with the consequent transport of reworked taxa from the Ross embayment to the slope by several ice streams advancing over Miocene sediments and highly deformable, unconsolidated Plio-Pleistocene strata. Each ice stream had a different ice-collection basin (EAIS or WAIS) and the chronology of its advance and retreat was recorded by Licht et al. (Reference Licht, Lederer and Swope2005), who demonstrated that East and West Antarctic till samples had distinct compositions. McKay et al. (Reference McKay, Naish, Powell, Barrett, Scherer, Talarico, Kyle, Monien, Kuhn, Jackolski and William2012) and others (see references in McKay et al. Reference McKay, Naish, Powell, Barrett, Scherer, Talarico, Kyle, Monien, Kuhn, Jackolski and William2012) also found these taxa in sediments related to glacial, subglacial or glacial-prograding phases and in silty-clayey units overlying the diamictons in deep-basin sediment cores around Ross Island. Sediments containing Paralia spp. and Stephanopyxis spp. (abundant during the early Pliocene and found even in surface sediments by Cunningham & Leventer (Reference Cunningham and Leventer1998)) characterized sedimentation on the continental shelf of the Ross Sea but these forms are considered as reworked indexes and non-age-diagnostic taxa (Bart et al. Reference Bart, Sjunneskog and Chow2011). Moreover, reworked taxa could also be exclusive to glacial sediments and tended to gradually but definitely decrease in deglacial phases and in the seasonal opening phases (diatomaceous muds). We noted that Actinocyclus ingens, Denticulopsis spp., Trinacria spp. and Xanthiopyxis were present in the basal units of both cores, but R. diploneides was only present in the basal part of ANTA99-23 (Fig. S2 C, D, F & H, which will be found at http://dx.doi.org/10.1017/S0954102013000199). Paralia sulcata (Fig. S1 F, which will be found at http://dx.doi.org/10.1017/S0954102013000199) is a tychopelagic diatom taxon, ecologically tolerant but at present not found in Antarctic environments (Sjunneskog, personal communication 2011). Sjunneskog & Scherer (Reference Sjunneskog and Scherer2005) interpreted this taxon as reworked in the eastern part of the Ross Sea as we do in this work, but, due to its constant presence and abundance in both cores, we considered it separately. Paralia sulcata was found in the sediments of both cores in both glacial and glacial-prograding phases, with trends comparable with those of the reworked forms. It tended instead to differ from the other reworked species in all the deglacial units (unit 23c and unit 23d in ANTA99-23, and unit 5e in ANTA96-5bis). This species (with generally more than or c. 2% in the assemblage) was always present in the deglacial units, but with a gradual decrease over time, mainly in core ANTA96-5bis, while the other reworked species were sporadic and with a lower percentage (generally < 2%).

We presumed that these records of reworked taxa were linked to the variability in both deglacial ice-stream contributions and in the current sedimentary distribution. Ice streams flowing over a deformable bed transfer large quantities of sediment on the continental slope, as revealed by settings observed in the prograding and glacial phase in ANTA99-23 (units 23a and 23b). Furthermore, reworked material typology and distribution may have changed during the retreat of the ice shelf and Holocene events, due to hydrodynamic deglacial evolution on the continental shelf and to changing continental and ice-stream supply.

Western Ross embayment palaeoenvironmental reconstruction

Our results highlighted some changes in the environment, sedimentation rates and responses of productivity proxies in the different core locations. Without ignoring problems relating to the dating in this area, especially for reworked sediments and gravity flows in ANTA96-5bis, and assuming the validity of the corrected dating of ANTA99-23, we propose four deglaciation phases closely related to Ross Ice Shelf dynamics and oceanographic and ice affected environments:

1. A) A glacial ice shelf prograding phase with glacio-marine deposition (Fig. 6a). This phase is characterized by low productivity in the water column and deposition of Tertiary-Miocene reworked taxa (Trinacria spp. with others), highly silicified reworked diatom taxa (such as Paralia sulcata) and inputs of IRD. The oldest record is found at the base of ANTA96-5bis (diamictite unit 5a), which could refer to the final part of MIS 3 with environmental conditions mainly influenced by an overflowing ice sheet and debris flows (turbidity flows or conturites) from the Joides Basin flanks (unit 5b). We presume that, at the end of MIS 3/beginning of MIS 2, unit 23a (in the ANTA99-23 core from the Joides Basin slope) was deposited under conditions of high sedimentation rates and a marked incorporation of reworked Miocene and Plio-Pleistocene diatom taxa, coming from Miocene and Plio-Pleistocene sedimentary bodies cropping out on the inner continental shelf. Analyses of the diatom assemblages suggest a bioproductive supply from open water that was then transported and stressed by proximal glacial or current transport. These results agree with Finocchiaro et al. (Reference Finocchiaro, Baroni, Colizza and Ivaldi2007) and Bart & Cone (Reference Bart and Cone2011), who assumed that the centre of the Ross Sea and part of its coast (south of the Drygalski Ice Tongue) enjoyed open-marine conditions (during a pause in the advance of the West Antarctic Ice Sheet) in the pre-LGM era (27 500–24 000 bp).

2. B) An advanced glacial phase (LGM) with total coverage by the ice shelf, grounded on the basin flanks. Palaeoproductivity is null and both gravity flows and turbidity currents are present (shaded arrows in Fig. 6b). During this following phase a shift of the sedimentary depocentre towards the slope and a greater sedimentary supply at the ANTA99-23 core location are recorded (unit 23b shows an intensified bioturbation). The edge of a proximal ice shelf with an increase in benthonic activity, supported by primary productivity (which probably resulted in an increase in diatom remains advected beneath the ice shelf), could be assumed, as indicated by McKay et al. (Reference McKay, Naish, Powell, Barrett, Scherer, Talarico, Kyle, Monien, Kuhn, Jackolski and William2012). Moreover, the proximity of the edge of an ice shelf agrees with the hypothesis that the WAIS deflated in the central part of the Ross Sea and was well south of the middle shelf during the peak of the LGM, whereas thick ice was still grounded in the western Ross Sea (Bart & Cone Reference Bart and Cone2011) (probably overflowing from the ANTA96-5bis core location).

3. C) A deglacial phase. During this transitional phase the hydrodynamic situation re-awoke with the beginning of the influence of the MCDW, which gradually expanded over the continental slope (colourless arrows in Fig. 6c) and with the formation, on the continental shelf, of HSSW. In core ANTA96-5bis a sedimentary hiatus is recorded confirming the results of Licht et al. (Reference Licht, Jennings, Andrews and Williams1996) for this area. This hiatus (from c. 26 000 to c. 780 yr bp) spans from at least the beginning of the retreat to the sedimentation of the overlying diatomaceous muddy open-ocean mat and consequently this transitional record is missing. Nevertheless, considering that this dataset is surely altered by old organic inputs and therefore potentially unreliable, unit 5d is related to a gravity flow that brought proxies of more productive environments, probably during the first deglacial phase, with the detachment of the ice shelf and its subsequent retreat (Fig. 6c). The palaeoproductivity signal rise up with cold water diatom species and with summer-sea ice forms (Eucampia antarctica var. recta and F. obliquecostata), probably slumped from the Glomar Challenger Basin flanks. Paralia sulcata is present, its abundance in the assemblage is linked to ice streams’ terrigenous supplies and gradually tends to decrease. In core ANTA99-23 the end of the more intense glacial cover phase (LGM) was marked by a barren laminated level that corresponded to the first deglacial phase, with the ice sheet detaching itself from the continental shelf and oceanic current reactivation (laminated level in unit 23b). In unit 23c the palaeoproductivity, marked by highly silicified and seasonal or multi-year sea ice or cold-water diatom species, increased together with the calving of icebergs (IRD level) while the supply of glacial material from the continental shelf (marked by reworked taxa) sharply decreased. We presume that oceanic and continental shelf hydrological features evolved in this phase (since c. 14 300 yr bp), leading to the beginning of a ‘Holocene style’ environment and the establishment of seasonal sea ice cycles (with a probable winter sea ice-edge closer to the coast and summer open marine conditions), cold and salty waters production (HSSW) and consistent MCDW input on the continental slope.

4. D) A Holocene phase of seasonal open-marine conditions. In ANTA96-5bis (at c. 770 yr bp) a Holocene phase is recorded, with deposition of diatomaceous muds (unit 5e). This phase was completely missing in ANTA99-23 (Fig. 6c). Apart from a possible problem in core collection for core ANTA99-23, the hiatus found in ANTA96-5bis and the lack of Holocene sedimentation in ANTA99-23 could be due to the onset of modern environmental features with the strong gravity currents that can be observed over the continental slope today and that are derived from the export of HSSW. In fact, the HSSW, together with the less saline shelf waters and the super-cooled low-salinity Ice Shelf Water (forming beneath the Ross Ice Shelf), are actually characteristic of this Ross Sea area and flow off the shelf through the Drygalski trough, the Joides Basin (immediately to the east) and the Glomar Challenger Basin and into the lower CDW (Budillon et al. Reference Budillon, Tucci, Artegiani and Spezie2000, Gordon et al. Reference Gordon, Zanbianchi, Orsi, Visbeck, Giulivi, Whitworth III and Spezie2004, Fusco et al. Reference Fusco, Budillon and Spezie2009). The analyses of unit 5e showed a clear recent change towards a more open marine-shelf environment, dynamic but not enough to cause material removal and with higher water column productivity. This last is characteristic of a stable summer stratification of the water column environment (seasonal ice zone (SIZ)) with high productivity of seasonal diatom blooms (such as Chaetoceros spp. and F. curta) while a shift to the continent of the winter sea ice-edge is assumed. Periodical productions of cold waters (HSSW), linked to seasonal sea ice formation and/or to the probable presence of the polynya opening in this area, are supposed. Since it is assumed that the MCDW influences could have been increasingly marked during the Holocene, F. kerguelensis is plotted, in Fig. 6c, in an outer slope environment to remark its offshore oceanic productivity value while Holocene record in ANTA99-23 is lacking. These observations are all consistent with models of sedimentation of modern and Holocene ice shelves (Sjunneskog & Scherer Reference Sjunneskog and Scherer2005, McKay et al. Reference McKay, Naish, Powell, Barrett, Scherer, Talarico, Kyle, Monien, Kuhn, Jackolski and William2012, Livingstone et al. Reference Livingstone, Ó Cofaigh, Stokes, Hillenbrand, Vieli and Jamieson2012).

Believing the dating of the ANTA99-23 samples to be reliable, we can speculate that in this slope environment, early deglaciation signs began at c. 18 613 cal yr bp while the first evidence of a significant MCDW influence is recorded at c. 14 379 cal yr bp. This result fits well with that proposed by Bart & Cone (Reference Bart and Cone2011) for a grounded ice shelf followed by ice shelf retreat from the western Ross Sea between c. 16 000 ¹⁴C yr bp and 11 000 ¹⁴C yr bp and a lack of signs of deposition because of sea floor erosion in at least some parts of the eastern Ross Sea. Livingstone et al. (Reference Livingstone, Ó Cofaigh, Stokes, Hillenbrand, Vieli and Jamieson2012) noted that the Drygalski trough palaeo-ice stream seemed to have receded from its maximum position, just north of Coulman Island, on the mid-outer shelf, by c.14 000 cal yr bp. Moreover, Licht & Andrews (Reference Licht and Andrews2002) proposed a deglaciation date of c. 16 800 cal yr bp from the outer-shelf of the Pennell trough (unofficial name). We therefore deduced, from the sediment proxies in our cores, and in agreement with Barbara et al. (Reference Barbara, Crosta, Massé and Ther2010) and Bart & Cone (Reference Bart and Cone2011), that these oceanic sedimentological features suggest new interactions between atmosphere, cryosphere and ocean and could be related to a gradual southwards shift of the Ross gyre during the first Holocene phases (see Fig. 1 for the actual position of the Ross gyre southern limb). The current reorganization in the eastern Ross Sea could have removed Holocene pelagic sediments locally. Moreover, still inadequately studied interactions between atmosphere, cryosphere and ocean are thought to have had important consequences on long-term glacial-deglacial timing, in water-sea ice dynamics (meltwater pulse events, McKay et al. Reference McKay, Naish, Powell, Barrett, Scherer, Talarico, Kyle, Monien, Kuhn, Jackolski and William2012 and references therein) and heat flux exchanges between the slope and continental shelf. All these are key topics for destabilization of continental ice, for polynya formation and for current transport and paths (Fusco et al. Reference Fusco, Budillon and Spezie2009, Naish et al. Reference Naish, Carter, Wolff, Polland and Powell2009, Barbara et al. Reference Barbara, Crosta, Massé and Ther2010). As highlighted by Licht et al. (Reference Licht, Lederer and Swope2005), in the compositions of East Antarctic and West Antarctic till samples, we believe that the analysis of reworked diatom taxa and improved biostratigraphy will provide important constraints for the reconstruction of the LGM grounding line, ice-flow lines and ice-flow paths and an interesting tool for reconstructing palaeo-sub-bottom currents (Mosola & Anderson Reference Mosola and Anderson2006, Barbara et al. Reference Barbara, Crosta, Massé and Ther2010, Bart & Cone Reference Bart and Cone2011, Bart et al. Reference Bart, Sjunneskog and Chow2011 and references therein, Livingstone et al. Reference Livingstone, Ó Cofaigh, Stokes, Hillenbrand, Vieli and Jamieson2012).