Glacial Interglacial Cycles

Past influences of Antarctica on the Southern Ocean are best gauged from well-dated records of environmental proxies, which are interpreted against a history of Antarctic ice behaviour preserved in ice cores and sediment cores from high-latitude locations. While we focus on Antarctica, it is made with full acknowledgement of NH forcing especially as the dominant driver of eustatic sea-level oscillations and as a co-driver of thermohaline circulation.

11.4.1.1. Thermohaline circulation

Essentially, the sinking of upper ocean waters in northern high latitudes forms North Atlantic Deep Water (NADW) that migrates south to mix with southern-sourced waters comprised of (i) dense bottom waters generated around the Antarctic margin and (ii) deep waters recycled in the Indian and Pacific oceans especially within the Antarctic Circumpolar Current (see Carter et al., this volume Chapter 4). The resultant Circumpolar Deep Water is the most voluminous in the Southern Ocean and dominates the deep inflows into the Indian and Pacific basins (Fig. 11.5). Yet despite its hybrid character, it retains the signatures of its source waters in particular NADW, which is identified by the salinity maximum and, from a palaeoceanographic

0° H General regions

^ of bottom water formation

0° H General regions

^ of bottom water formation

\Main gateways for thermohaline circulation flow

Figure 11.5: Generalised outline of the modern Southern Ocean with the westerly wind-driven Antarctic Circumpolar Current defined by its zonal jets or fronts that include the Subantarctic Front (SAF), Polar Front (PF), Southern Front (SF) and Southern Boundary (SB) after Orsi et al. (1995). ODP Sites 1090 and 1123 containing >1myr records of ocean change are also located.

perspective, by its elevated 813C content ( 1992).

Changes in this circulation over G-I cycles are subject to a continuing debate, much of which focuses on the input of NADW to the Southern Ocean.

Charles and Fairbanks (1992), Rahmstorf (2002), Ninnemann and Charles (2002) amongst others favour a reduction in NADW input to the Southern Ocean in glacial times. Hall et al. (2001) support this decline and suggest that it was compensated by an increase in the supply of Antarctic-sourced deep water, at least in the New Zealand gateway for the Pacific Deep Western Boundary Current (DWBC). Glacial conditions potentially favoured bottom water production through increased windiness plus an expansion of sea ice and associated brine rejection. Certainly Hall et al. (2001) record an increase in the speed of the Pacific DWBC in glacial times as reflected by an increase in the size of 'sortable silt', a grain-size proxy for relative current strength (see also Venuti et al., 2007). For time frames longer than individual G-I cycles, the sortable silt data point to three phases of Pacific DWBC vigour during the last 1.2 myr. These three phases cover the transition from a world dominated by 41 kyr climatic cycles to one where 100 kyr cycles prevailed and include: Phase 1 (1.2-0.87 Ma) of moderately strong flow; Phase 2 or the MPT (0.87-0.40 Ma) of weaker flow, and Phase 3 (0.40 Ma to present) when strongest flows prevailed.

The alternative hypothesis that little or no decline in NADW inflow occurred in glacial times is supported by Moy et al. (2006), Rosenthal et al. (1997) and others on the basis of various palaeoceanographic proxies including the distribution of 813C between ocean basins over G-I cycles. As NADW flows to the Southern Ocean, its d13C content reduces due to remineralisation of organic carbon and mixing with southern-sourced waters with lower 813C. Depletion continues into the Indian and Pacific oceans, the latter having the lowest d13C (Kroopnick, 1985). By comparing 813C profiles measured on benthic foraminifers from the main ocean basins bathed by NADW-sourced waters, it is possible to identify any variability of those waters. Moy et al. (2006) compared benthic d13C for cores spanning the last 180 kyr from the equatorial Atlantic, the Indian-Pacific boundary at South Tasman Rise and the equatorial Pacific. Equatorial Atlantic 813C was higher than at South Tasman Rise, but followed the same G-I cyclicity and maintained a gradient suggesting maintenance of northern-source deep water for the last 160 kyr. In addition, d13C data for the Tasman Rise and equatorial Pacific were very similar highlighting a close tracking of Southern Ocean and central Pacific waters.

While the debate continues, most recent data (e.g. Moy et al., 2006; I. N. McCave, submitted and personal communication 2007) tend to favour a more subdued input of NADW and no major reorganisation of Southern Ocean water masses during glacial periods.

11.4.1.2. Intermediate waters

The dispersal of southern-sourced, intermediate depth waters is a key process in distributing heat, salt, nutrients and gases to equatorial and northern latitudes (e.g. Sarmiento et al., 2004). The northward transport of Subantarctic Mode Water (SAMW) and Antarctic Intermediate Water (AAIW) largely balances the southward export of NADW (Gordon, 1986). Both SAMW and AAIW form in the vicinity of the Subantarctic Front through buoyancy loss related to heat exchange, freshwater input and a northward Ekman transport although changes in wind stress and lateral advection may also play a role (e.g. McCartney, 1977; Rintoul et al., 2001; Toggweiler et al., 2006). While the mode(s) of intermediate water formation is not well understood, observations show that SAMW and underlying AAIW descend to ~ 500 and ~ 500-1,400 m depths, respectively. SAMW formation is fairly widespread, whereas AAIW generation is favoured in the SE Pacific and SW Atlantic.

Given their source at the wind-forced ocean surface, intermediate waters typically have elevated d13C contents as revealed by hydrographic data and intermediate depth benthic foraminifers, assuming other factors have not come into play (e.g. Lynch-Stieglitz et al., 1994; Pahnke and Zahn, 2005). In contrast, glacial benthic records show a marked reduction in 813C that is interpreted as a diminished supply of intermediate water. As outlined in more detail in the next section, the glacial Southern Ocean was (i) substantially cooler with sea-surface temperatures (SSTs) up to 6°C colder than now (e.g. Mashiotta et al., 1999; EPICA Community Members, 2004; Barrows et al., 2007), (ii) more extensive following an equatorward migration of southern waters by ~5-10° latitude (Howard and Prell, 1992; Gersonde et al., 2005), (iii) windier following an equatorward displacement and intensification of zonal westerly winds (Shulmeister et al., 2004; Toggweiler et al., 2006) and (iv) supported a more extensive cover of sea ice ~100% more than present (Gersonde et al., 2005). Pahnke and Zahn (2005) have argued that meltwater from the expanded sea ice, concomitant with its northward transport within an invigorated Ekman layer, reduced the surface density and the buoyancy-driven sinking of SAMW and AAIW. Yet despite this decline, intermediate waters remained a thermal conduit that linked Antarctica and the equatorial ocean judging by the strong coherence of their respective temperature records (Lamy et al., 2004; Kiefer et al., 2006). This linkage has been invoked by Weaver et al. (2003) as a potential control of meridional overturning in the North Atlantic. Pulses of meltwater from Antarctica could freshen AAIW, which upon arrival in the North Atlantic could potentially increase the density in contrast with surface waters to encourage sinking and the formation of NADW.

11.4.1.3. Surface waters

G-I cycles have a marked influence on the upper ocean by virtue of its direct interaction with the atmosphere. Reconstructions of SSTs for the last 25-400 kyr - the approximate range of records recovered by piston cores -highlight marked glacial cooling at orbital frequencies in the Atlantic sector of the Southern Ocean (e.g. Mortyn et al., 2002; Gersonde et al., 2003, 2005; Bianchi and Gersonde, 2004; Pahnke and Sachs, 2006), the Indian sector (e.g. Howard and Prell, 1992; Gersonde et al., 2005; Barrows et al., 2007), the SW Pacific (Fig. 11.5; Weaver et al., 1998; Barrows et al., 2000, 2007; Pahnke et al., 2003; Neil et al., 2004; Pahnke and Zahn, 2005) and the SE Pacific (e.g. Mashiotta et al., 1999; Lamy et al., 2004). A wide range of geochemical, isotopic and microfossil proxies is used to derive SSTs (see summary in Wefer et al., 1999), which can vary according to the proxy (Barrows et al.,

2007). The quest for reliable comparisons between ocean basins is further confounded by some palaeoceanographic observations that record localised effects such as upwelling, rather than regional signals (e.g. Pahnke and Sachs, 2006). With such limitations in mind, the aforementioned studies reveal that ocean cooling was most pronounced around 40-46°S during major glaciations (e.g. Marine Isotope Stages 2, 6, 8, 10, 12) when SSTs were 4-6°C lower than present (Fig. 11.6). This change was recorded in most sectors of the Southern Ocean with the possible exception of the central Pacific where a sparse database suggests that cooling was less severe (Gersonde et al., 2005). Certainly the western and eastern margins of the Pacific Basin were subject to 4-6°C coolings during the LGM (Lamy et al., 2004; Barrows et al., 2007). By comparison, the peak interglacials of Marine Isotope Stages 5e and 11 record SSTs up to 3°CS warmer than present. The amplitude of SST swings reduces southwards (Howard and Prell, 1992; Gersonde et al., 2003) and once south of the Antarctic Polar Front, LGM SSTs, for example were ~3°C cooler than present at ~50°S and <1°C cooler south of ~52°S, at least in the Atlantic-SE Indian sectors.

SST records extending over a million years or more from the Southern Ocean are restricted mainly to DSDP and ODP sites, which because of their sparse coverage prevent identification of regional patterns (Fig. 11.7). Nonetheless, they highlight temporal trends especially over the transition from 41 kyr-paced to 100kyr-paced G-I cycles. Of note are SST records from (i) ODP 1090 located between the modern Subtropical and Subantarctic fronts in the SE Atlantic (Becquey and Gersonde, 2002; Hodell et al., 2002), and (ii) ODP 1123 positioned at the northern limit of the Southern Ocean, within the Subtropical Front in the SW Pacific (Fig. 11.5; Crundwell et al.,

2008). They were chosen because they are both based on foraminifer transfer

Figure 11.6: (B) Stacked SST, normalised to a modern value of 0oC, for the Southern Ocean, compiled by Barrows et al. (2007). It highlights the marked cooling of the LGM, when temperatures were up to 50C cooler than now and the abrupt swings in SST that are near-synchronous with warm phases in Antarctica (C) but mainly leading prominent D-O events in the Northern

Hemisphere (A).

Figure 11.6: (B) Stacked SST, normalised to a modern value of 0oC, for the Southern Ocean, compiled by Barrows et al. (2007). It highlights the marked cooling of the LGM, when temperatures were up to 50C cooler than now and the abrupt swings in SST that are near-synchronous with warm phases in Antarctica (C) but mainly leading prominent D-O events in the Northern

Hemisphere (A).

functions and come from open-ocean sites beyond local complexities imposed by bathymetry. Both records exhibit the same three phases of SST variability (Fig. 11.7), but the type and degree of variability at times differ in response to the regional oceanography.

Phase 1: 41 kyr cycles (1.2 Ma and earlier to 0.87 Ma). SW Pacific temperatures exhibit high amplitude fluctuations of 4-70C within a G-I range of 9-180C. These fluctuations are superimposed on an overall cooling trend that is especially evident with interglacial optima, which decline progressively from 18 to 140C. Contemporaneous SE Atlantic SSTs have more restricted amplitudes of 1-30C and G-I range of 3-50C. A cooling trend is less clear,

EPICA Dome C

Mean sortable-silt record Site 1123 (Hall etal. 2001)

Southwest Pacific Site 1123 S. Atlantic Site 1090 Benthic oxygen isotope record (Crurxiwell etal., 2006) (Becquey & Gersonde 2002) Southwest Pacific site 1123 SST ra (Hall etal. 2001)*"0 (%=)

Figure 11.7: Composite plot of Antarctic and Southern Ocean climate proxies spanning the last 1.2myr. The data allow the coupled G-I variability within the atmospheric, oceanic and cryospheric systems to be evaluated in the context of longer-term changes extending from the 41 kyr climate system (Phase 1) through the MPT (Phase 2) and into the

Strengthened ACC, increased AABW production and intensify ction of deep Pacific inflow

Northward expansion of SAW

Northward migration of sea-ice margin

Large northen hemisphere ice sheets Expanded WAIS

Youngest evidence of open water

In Ross Sea MIS 31

Possible collapse of RIS and WAIS.

Figure 11.7: Composite plot of Antarctic and Southern Ocean climate proxies spanning the last 1.2myr. The data allow the coupled G-I variability within the atmospheric, oceanic and cryospheric systems to be evaluated in the context of longer-term changes extending from the 41 kyr climate system (Phase 1) through the MPT (Phase 2) and into the lOOkyr climate system (Phase 3).

AND-1B - Ross Sea Glacial Age (ka) NH

•g. sedimentary proximity insolation cycles cycles

Obliquity (degrees tilt)

Deep Pacific ventilation SW Pacific SST S Atlantic $$T Golobal ice volume Antarctic air temperature

EPICA Dome C

Mean sortable-silt record Site 1123 (Hall etal. 2001)

Southwest Pacific Site 1123 S. Atlantic Site 1090 Benthic oxygen isotope record (Crurxiwell etal., 2006) (Becquey & Gersonde 2002) Southwest Pacific site 1123 SST ra (Hall etal. 2001)*"0 (%=)

but an irregular decline is evident from ~ 1.06 to 0.87 Ma when interglacial optima went from 7 to 4°C. The general cooling at both sites reflects an expansion of the Antarctic cryosphere accompanied by a northward migration of the cold surface waters, while at depth there may have been a general increase in the thermohaline circulation at least into the Pacific Ocean (Hall et al., 2001). Thus, at ODP 1123, Subtropical Water was displaced by Subantarctic Water with the former returning at the last part of the terminations and early interglacials. In contrast, the more southerly sited ODP 1090 was mainly affected by circumpolar surface waters through much of the G-I cycles thus accounting for the colder temperatures with subdued amplitudes and G-I ranges. Phase 1 terminated at both sites at the prominent interglacial of Marine Isotope Stage 21.

Phase 2: MPT (~0.87-0.4Ma). This fundamental transition from a 41-100 kyr dominant world was marked at both sites by persistently warmer and more prolonged interglacial periods that were interrupted by some of the coldest glacial periods in the Pleistocene. G-I cyclicity was most marked in the SE Atlantic, which came under the influence of Subantarctic Water and even Subtropical Water during Marine Isotope Stage 15. Not surprisingly, that southward migration of Subtropical Water was particularly well shown at ODP 1123 where it dominated SSTs from Marine Isotope Stages 15-13 with the intervening glacial, Marine Isotope Stage 14, bringing only a modest cooling (<2°C). The general warming and southward migration of warmer waters during the MPT is consistent with a reduced Antarctic influence, which is supported by a less vigorous thermohaline circulation into the Pacific as suggested by Hall et al. (2001). The marked G-I variability is interpreted by Becquey and Gersonde (2002) to reflect the increasing influence of 100 kyr cycles, which in turn are driven primarily by changes in NH ice sheets. However, the impact of these drivers appears to be blunted by subtropical influences as attested by ODP 1123 record especially over Marine Isotope Stages 15-13.

Phase 3: 100kyr cycles (~0.4Ma to present). In the SE Atlantic the onset of the 100 kyr world was accompanied by highly variable SSTs with G-I contrasts of up to 8°C. These major fluctuations occurred against a background of mainly warmer temperatures at ODP 1090, which lay mainly within Subantarctic Water. While also exhibiting 100kyr-paced SSTs, the G-I differences at ODP 1123 are less marked with all but one (Marine Isotope Stages 6-5) less than 5°C. Such a muted response may reflect the influence of the south Pacific Subtropical Inflow (e.g. Carter et al., 2008). Despite differences in SST amplitudes, the SE Atlantic and SW Pacific data reveal a progressive reduction of ~3°C in interglacial optima over Phase 3.

11.4.1.4. Ocean fronts

From the preceding discussion it is clear that the variability of SSTs over G-I cycles is intimately linked to the meridional displacement of polar waters, which is affected by the state of the Antarctic cryosphere and zonal winds. Glaciations witnessed a major expansion of winter sea ice. In the LGM, sea ice grew by ~100% compared to now, extending 3-8° latitude in the Atlantic, 7-10° in the Indian and possibly ~2-5° in the Pacific, bearing in mind the last region has sparse data coverage (Gersonde et al., 2005). This expansion was accompanied by an intensification and equatorward displacement of zonal winds (Thiede, 1979; Shulmeister et al., 2004) along with a northward migration of cold water and ocean fronts (e.g. Howard and Prell, 1992; Becquey and Gersonde, 2002; Gersonde et al., 2003; Crundwell et al., 2008). The extent of the migrations depend upon the strength and duration of the climatic drivers, as well as the morphology of the Southern Ocean floor, for example any G-I displacement of fronts off eastern New Zealand, is limited by extensive submarine elevations such as the Chatham Rise where the STF is restricted to 1-2° latitude along the Rise crest (Weaver et al., 1998; Sikes et al., 2002). In contrast, frontal systems and associated water masses in the open Southern Ocean are free to migrate further afield (Fig. 11.5). However, it is an open question as to how SSTs represent shifts in ocean fronts. As pointed out by Gersonde et al. (2005), control sites are too scattered to delineate the temperature and salinity gradients that define frontal systems. Accordingly, the following outline of frontal migrations is indicative only. The data of Becquey and Gersonde (2002), Gersonde et al. (2005) and Howard and Prell (1992) indicate northward migrations of 4°, 510° and 2-3° for Antarctic Polar Front in the Atlantic, Indian and Pacific sectors, respectively, during the LGM. At the same time, the SAF shifted 4-5° and 5-10° in the Atlantic and Indian, respectively. SSTs from the easternmost (Lamy et al., 2004) and westernmost Pacific (Crundwell et al., 2008) suggest similar northward displacements. In contrast, migrations of the STF were restricted to only ca. 2-3° and 5° latitude in the Atlantic and Indian sectors suggesting an intensification of oceanographic gradients at the northern limit of the Southern Ocean (Gersonde et al., 2005). The reasons for such limited migration are not clear but it maybe reflect a counteraction by the subtropical gyres in the major basins.

In addition to the G-I cycles, long-term records from the ODP sites (Fig. 11.7) identify broad changes in the fronts. At ODP 1090, the PF prevailed, representing a possible shift of ~ 7° latitude north of its present position (Becquey and Gersonde, 2002). During the MPT the site was dominated by the SAF, but with interruptions by glacial excursions of the

PF. In Phase 3, the SAF again dominated but was interspersed by interglacial incursions of the STF. In the SW Pacific, the more northerly located ODP 1123 spent most of the last 1.2 myr within or north of the STF (Crundwell et al., 2008) except during peak glacial periods. Taking the mean annual SST of the northward edge of the modern SAF to be 11 °C (Uddstrom and Oien, 1999), then the mean annual palaeo-SST profiles reveal that (i) Phase 1 was a time of frequent glacial incursions of the SAF and associated waters representing a northward migration of ~ 5° latitude; (ii) Phase 2 MPT witnessed only infrequent SAF migrations, which ceased in the latter half when the STF prevailed and (iii) the SAF was located well south reaching Site 1123 only in the major glaciation of Marine Isotope Stage 6.

11.4.1.5. Surface ocean currents

As noted in the previous section, SST data indicate northward migrations of the PF and SAF during glacial periods. As these fronts accommodate most of the flow within the ACC (e.g. Whitworth, 1988; Rintoul et al., 2001) the inference is that this major current system also moved north, possibly by ~ 5° or more of latitude. However, in light of (i) limitations of SST data to define past frontal positions, (ii) the marked topographic steering of the modern ACC by mid-ocean ridges (e.g. Gordon et al., 1978; Orsi et al., 1995; Moore et al., 1999) and (iii) the strong latitudinal variability of modern ACC fronts such as the PF, which ranges over 5-7° of latitude (Moore et al., 1999), it is unclear how the ACC behaved. Uncertainty also arises as to whether the ACC strengthened under the stronger wind regimes of glacial periods. Certainly the ACC is mainly wind-driven, with the extent of this forcing dependent upon the position of mid-latitude westerly winds (Toggweiler et al., 2006). Wind strengthens the ACC by direct shear but also enhances the northward Ekman transport, which in turn is compensated by the southward transport and eventual upwelling of deep water at the Antarctic margin. These effects are most pronounced when winds are aligned with the ACC as is the case today. Although glacial period westerly winds (Moreno et al., 1999; Sigman and Boyle, 2000) and the main ACC fronts were located further north than today, it is not known if winds and current were aligned. However, where migration of the ACC is restricted by the seabed morphology as in the Scotia Sea (Pudsey and Howe, 1998), or off eastern New Zealand (Neil et al., 2004), the ACC appears to have strengthened, although it is unclear if this was a response to increased wind-forcing, compression of fronts against the bathymetry or both these processes.

11.4.1.6. Iceberg discharge and migration

A proxy for G-I behaviour of Antarctic ice is found in the IRD records of Southern Ocean sediments. ODP Site 1011, off the Antarctic Peninsula provides a generalised IRD history spanning 3 Ma (Cowan, 2002). Features are phases of pronounced IRD deposition around 2.8, 1.9 and 0.85 Ma, interleaved by more subdued deposition that followed G-I cycles with a frequency of 41 kyr from 2.2 to 1.0 Ma and 100 kyr frequency after 0.4 Ma. In the latter cycles, the late glacials and following interglacials were the main times of ice rafting. A more detailed record for the last 300 kyr in the Weddell Sea confirms the strong IRD signal at G-I transitions, but shows high deposition of IRD continuing into the early part of the following glacial period (Fig. 11.8; Grobe and Mackensen, 1992). The Weddell Sea record also reveals a consistent decline in IRD production since 200 kyr

Figure 11.8: Normalised stacked profiles for IRD generated at source in the Weddell Sea (Grobe and Mackensen, 1992) and deposited at a distant depocentre - the Campbell Plateau, New Zealand (Carter et al., 2002). Both profiles show similar broad outlines but the key difference is that the times of maximum IRD production are not faithfully reproduced off New Zealand especially during a G-I transition when a warming ocean and reducing windiness would be less favourable for iceberg preservation.

Normalised Stacks Campbell Plateau Weddell Sea

Normalised Stacks Campbell Plateau Weddell Sea

Figure 11.8: Normalised stacked profiles for IRD generated at source in the Weddell Sea (Grobe and Mackensen, 1992) and deposited at a distant depocentre - the Campbell Plateau, New Zealand (Carter et al., 2002). Both profiles show similar broad outlines but the key difference is that the times of maximum IRD production are not faithfully reproduced off New Zealand especially during a G-I transition when a warming ocean and reducing windiness would be less favourable for iceberg preservation.

(Marine Isotope Stage 7). Superimposed on the G-I cycles are IRD pulses of millennial-scale frequency which, given uncertainties of age models, are traceable from the SE Atlantic to SW Pacific, at least for the past 70 kyr (Labeyrie et al., 1986; Kanfoush et al., 2000; Carter et al., 2002). The Kanfoush et al. (2000) IRD events coincide with periods of NADW production and warm interstadials in the North Atlantic. This led to the proposition that the increased inflow of relatively warm NADW and/or a rise in sea level destabilised ice sheets to promote iceberg discharge.

Once calved, icebergs from the Weddell to Ross Sea regions were most likely captured within the Weddell, Ross and un-named coastal gyres judging by modern iceberg trajectories (Keys, 1990). Nevertheless, some were eventually entrained by the ACC to move east as identified from modern iceberg paths (Tchernia and Jeannin, 1983), last glacial IRD dispersal patterns (Cooke and Hays, 1982), and a meltwater signal generated about 35-17kyr (Labeyrie et al., 1986). Not surprisingly, the Weddell Sea IRD profiles (Grobe and Mackensen, 1992) broadly correlate with those off eastern New Zealand (Fig. 11.8; Carter et al., 2002). However, there are also differences in that the interglacial records are more subdued off New Zealand, a feature Carter et al. (2002) attributed to iceberg melting en route from Antarctica. In contrast, glacial records are better correlated indicating better preservation of travelling icebergs on account of colder glacial seas and possibly an intensified ACC (Pudsey and Howe, 1998; Neil et al., 2004).

0 0

Post a comment