Antarctic Oligocene Glaciation – Early Oligocene Glaciation Preceded by Export of Nutrient-Rich Northern Component Water

Australia: The Land Where Time Began

Antarctic Oligocene Glaciation – Early Oligocene Glaciation Preceded by Export of Nutrient-Rich Northern Component Water

The onset of the formation of North Atlantic Deep Water is believed to have coincided with the growth of the Antarctic ice sheet about 34 Ma. This timing is debated, however, partly due to questions about the geochemical signature of the ancient Northern Component Water (NCW) that formed in the deep North Atlantic. In this paper Coxall et al. present detailed geochemical records from sediment cores obtained from the North Atlantic close to sites where deep water is being formed. It was found that the northwestern Atlantic was stratified prior to 36 Ma, with bottom waters that were rich in nutrients and of low salinity. This restricted the transition of the basin into a conduit for Northern Component Water that began flowing to the south about 1 Myr prior to the initial glaciation of Antarctica. It was suggested by Coxall et al. that the probable trigger was tectonic adjustments in subarctic seas that allowed an increase in the exchange across the Greenland-Scotland Ridge. The Northern Component Water was strengthened by the increasing salinity and density of the surface water. The carbon isotope signatures of the Late Eocene deep water mass differed from modern values as a result of the leakage of fossil carbon from the Arctic Ocean. A transient pulse of CO₂ to the Earth system was provided by the export of nutrient-rich water, which possibly caused short-term warming, though the long-term effect of enhanced formation of Northern Component Water was an increased transport of heat to the north that cooled Antarctica.

In the North Atlantic Ocean the production of deep water plays a vital role in maintaining the global meridional overturning circulation (MOC). The North Atlantic Deep Water (NADW), which is the lower branch of the Atlantic part of the MOC (AMOC), forms as surface waters cool and densifies in the Labrador and Nordic Seas. Interplay of:

1) the stratification at convection sites, which is determined by the balance of warm salty water from low latitudes, cold fresh water from the Arctic Ocean and local heat and fluxes of freshwater, and

2) Upwelling in the Southern Ocean that is driven by wind, by which deep water is returned to the surface (de Boer, Toggweiler & Sigman, 2008; Toggweiler & Samuels, 1995). It is probable that both factors impacted the MOC state in the Early Cainozoic, when bathymetry and ocean gateways of the Atlantic were different and global temperatures were warmer than at the present (Abelson & Erez, 2017; Cramer et al., 2009).

It is, however, challenging to resolve the interplay at the onset of the production of North Atlantic Deep Water, which is referred to in this paper as its palaeo-precursor Northern Component Water (NCW), as the early history of the AMOC has remained poorly constrained.

The development of global deep water circulation is constrained by the benthic formation of δ¹⁸O and δ¹³C records by allowing insights into surface temperatures, salinity and nutrients (Cramer et al., 2009). It is widely believed that the Northern Component Water began filling the Atlantic Ocean close to the Eocene-Oligocene greenhouse-to-icehouse climate transition (EOT), at about 34 Ma (Abelson & Erez, 2017; Davies et al., 2001; Egloff & Johnson, 1975; Miller & Tucholke, 1983; Via & Thomas, 2006) or earlier (Borelli, Cramer & Katz, 2014; Boyle et al., 2017; Hohbein, Sexton & Cartwright, 2012; Langton et al., 2016). It has been argued by others that the emergence of a significant Northern Component Water was delayed until the Miocene (Wright & Miller, 1996). According to Coxall et al. modelling studies also diverge, with different studies suggesting that during the greenhouse-to-icehouse there was no Northern Component Water (Sijp, Miller & Huber, 2011), a strengthening/onset of the formation of bipolar deep water that was triggered by the deepening of Drake Passage (Elsworth et al., 2017; Tigchelaar, von der Heydt & Dijkstra, 2010; Zhang et al. 2011) or an ocean in which there was a Northern Component Water throughout (Huber & Sloan, 2001). It was assumed by previous data studies that argued for an onset of the production of Northern Component Water in the Late Eocene, that the Early Cainozoic Northern Component Water was nutrient poor with a high (‘young’) benthic δ¹³C signature that was similar to that of a modern North Atlantic Deep Water that was well ventilated (Abelson & Erez, 2017; Borelli, Cramer & Katz, 2014; Langton et al., 2016; Pusz, Thunell & Miller, 2011). Yet there is a lack of palaeodata that is required to characterise the Northern Component Water close to source regions.

Coxall et al. produced EOT benthic δ¹³C and δ¹⁸O records from deep sea sites at more than 1,000 m palaeodepths in the high latitudes of the North Atlantic in order to fill this gap. The most northerly EOT sequence that contained the calcareous microfossils that were necessary for δ¹⁸O and δ¹³C analysis of these is Site 647 in the southern Labrador Sea (SLS) (47^oN, palaeolatitude from 34 Ma, about 2,000-3,000 palaeodepth). For portions of the Late Eocene additional data were generated from the Deep Sea Drilling Project (DSDP) Sites 112 and 612, and integrated Ocean Drilling Project (IODP) Site U14112. According to Coxall et al. Northern Component Water export should be recorded by the latter 2 sites in the deep western boundary current (DWBC). The data were compared with an isotope compilation that incorporated 14 sites from the Atlantic that were recorded previously. Records of benthic foraminifera Mg/Ca, fish-tooth neodymium (Nd) that were reported in epsilon notation (εNd) and planktonic foraminifera δ¹⁸O and δ¹³C that had been obtained from Sites 647 and U1411 were also generated in order to provide constraints on the bottom-water temperature (BWT) and provenance, and the structure of the water column. Unique perspectives are added by these results on the Atlantic end-member deep-water properties and circulation changes in the circulation during EOT.

Late Eocene Southern Labrador Seas, high nutrient content of its deep waters

Coxall et al. recognised at site 647 the typical pattern of increase of δ¹⁸O and δ¹³C (>1.0‰ Vienna PeeDee Belemnite (VPDB) and ⁓0.5-1.0‰ VPDB, respectively) between 34 and 33.5 Ma, which is diagnostic for Antarctic glaciation in the Early Oligocene, and included in it the peak in δ¹⁸O that is seen at other sites (Coxall & Wilson, 2011), referred to in this paper as the Early Oligocene Glacial Maximum (EOGM). At site 647 the first novel observation is that prior to about 35.8 Ma δ¹³C of the SLS bottom water was, on average, 0.2-1‰ VPDB lower than at all the southerly sites. This is the opposite to the modern AMOC state, in which the northern deep waters have the highest δ¹³C as the result of the sinking of surface waters that are well-ventilated and nutrient poor (Kroopnick, 1985). Coxall et al suggest the low value of δ¹³C may be a reflection of the nutrient accumulation under stratified conditions, which is analogous to those of the modern Northern Pacific, i.e., the end of the circulation path. It could be implied by this that deep water that was southern sourced filled the southern Labrador Sea during the Late Eocene. At site 647, however, fish debris εNd, which is an isotopic tracer for the origin of deep water masses, bear the fingerprint of a source in the Northern Hemisphere (εNd = -11.4 to -9.4) throughout the stratified interval. Coxall et al. argued, consequently, that the low benthic δ¹³C reflects local bottom water that was sourced from surface waters that had a high concentration of nutrients within a narrow, restricted basinal deep water circulation in the North Atlantic. ‘Fossil’ carbon that leaks from the Arctic Ocean and subarctic Sea (Greenland Sea and Norwegian Sea) is a probable source of nutrients, which had high stocks of nutrients during the Eocene as a result of their semi-isolation, heavily vegetated margins (Golovneva, 2000) and high riverine inflow (Akhmetiev & Beniamovski, 2009; Gleason et al., 2009; O’Regan et al., 2011).

1) Regime 1. Based on the new Eocene-Oligocene proxy records of Coxall et al., and comparisons with published data, they identify 3 circulation regimes. Focusing first on δ¹³C, under Regime 1(>35.9 Ma), southern Labrador Sea bottom waters were isolated from the remainder of the Atlantic and a distinct Northern Component Water that had a low δ¹³C, bathed Site 647. After about 35.8 Ma there was a negative δ¹³C excursion (0.5-1‰ VPDB δ¹³C that continued for about 1.5-2 Ma, from about 35.8 to about 33.8 Ma, that was seen to varying degrees at sites 612, U1411 and 647, as well as some other sites in the Atlantic (1053, 1090 and 366).

2) Regime 2. The timing of offsets between sites is probably caused by age-model differences. The onset and peak of the δ¹³C excursion are encompassed in Regime 2. It is important that the δ¹³C excursions are largest, with a maximum of 1‰ VPDB, at sites 612 and U1411 which is in the Deep Western Boundary Current. The wider significance of this δ¹³C excursion has not been explored fully, though it has been noted previously (Pusz et al., 2009). The observation that Sites 612 and U1411, which are directly downstream of Site 647, gain benthic δ¹³C signals close to the Sothern Labrador Sea end member suggests the signal was propagated from the north. Therefore, it records the export to the south of Northern Component Water from the Arctic imprinted water, which was nutrient-rich. It is indicated by the increase towards the end of Regime 2 in Atlantic benthic δ¹³C that either the pulse of Northern Component Water ended, or surface water that was ventilated sufficiently with a higher δ¹³C was important to the convection sites.

3) Regime 3 Regime 3, which is described below, represents the phase during which a form of Northern Component Water that is more mature existed.

Deep Water cooling, salinification and stratification

The pattern of the Southern Labrador Sea benthic δ¹⁸O is a second prominent feature. During Regime 1in the Southern Labrador Sea benthic δ¹⁸O is 1-3‰ VPDB lower than that of the ensemble. The primary controls on the benthic δ¹⁸O are Bottom Water Temperature and the δ¹⁸O composition of seawater – the latter is a reflection of the global glacial ice volume and local salinity. It is indicated by the relatively low δ¹⁸O in the benthos in the Southern Labrador Sea that, assuming minimal ice prior to 34 Ma, a considerably warmer or fresher water mass in contact with the seafloor compared to the southern stations. From Sites 647 and 112 Benthic δ¹⁸O gradually increased from 36.0 to 35.4 Ma, and again from about 34.6 to 34.4 Ma, and had converged close to the dominant Atlantic trend by about 34.3 Ma, i.e., coincident with or just following the δ¹³C minimum that was Atlantic-wide. Diagenetic alteration of Site 647 benthic fossils (Arthur et al., 1989) is ruled out because of:

1) The excellent fossil calcite preservation,

2) The similarities of the new values of planktonic δ¹⁸O from Sites 647 and U1411.

Also, a similar pattern of benthic δ¹⁸O that is decreasing is present in the North Sea record (Nielsen et al., 2009), though at a shallower depth of about 500m.

Bottom Water Temperature Mg/Ca helps to deconvolve the temperature and salinity influences on δ¹⁸O. Across the Eocene-Oligocene transition from greenhouse-to-icehouse a Bottom Water Cooling of ⁓1^oC is suggested by Mg/Ca data at Site 647 Mg/Ca data combined with a ⁓0.6^oC VPDB increase in seawater δ¹⁸O (δ¹⁸O_sw), which is in agreement with previous studies (Lear et al., 2008). Coxall et al. estimate there is a gradual cooling of bottom water in the Southern Labrador Sea of ⁓3-4^oC between 37.5 and 35 Ma, which is similar to that of the northern high latitude cooling of the sea surface (Liu et al., 2009). By substituting the Mg/Ca Bottom Water Temperatures into a palaeotemperature equation yields ice-free δ¹⁸O_sw estimates of between -3 and -4‰ δ¹⁸O during this interval. It was estimated, based on these δ¹⁸O_sw constraints, that the Late Eocene Southern Labrador Sea bottom salinities were relevant modern δ¹⁸Osw-salinity of the ocean surface salinity (SSS) relationships (Waddell & Moore, 2008).

Applying a modern sea surface temperature-δ¹⁸O_sw relationship from eastern Greenland, which is at the present a conduit for low salinity water (32 practical salinity units (PSU)) Arctic flow implies the Bottom water salinity at Site 647 of 30-32 PSU prior to 36 Ma, which increase by 2-3 PSU between 36-34 Ma. The change in salinity is similar to what occurs when a Laptev Sea, which is at present fed by large Siberian Rivers, δ¹⁸O_sw-salinity regression is applied.

The values inferred by Coxall et al. are compatible with modern temperature-salinity fields, though Sea Surface Salinity-δ¹⁸O_sw relationships are widely variable spatially, and relationships of the present are only loose analogues for the Eocene. Coxall et al. therefore suggest that:

1) The Southern Labrador Sea bottom waters were relatively fresh, and

2) The Southern Labrador Sea salinity of the bottom waters increased from Regime 1 to regime 2.

Even if samples that are older than 35 Ma are biased to higher Mg this conclusion does not change as the salinity signal is buried in the benthic δ¹⁸O, which is independent of Mg/Ca. This interpretation is consistent with that for nutrients that were pre-formed as they both likely derived from the Arctic Ocean. It is agreed by proxies and models that during the Palaeogene there was a thick freshwater cap on the Arctic Ocean as a result of a strong hydrological system and inputs of fluvial water under greenhouse forcing which was combined with a salt input that was restricted (Gleason et al., 2009; Waddell & Moore, 2008; Stärz et al., 2017; Roberts, LeGrande & Tripati, 2009). As there was no opening to the Pacific Ocean at this time, major surface discharge occurred through the Nordic Seas (Stärz et al., 2017; Brinkhuis et al., 2006).

Planktonic foraminifera δ¹⁸O and δ¹³C at Site 647 add information on the upper part of the water column. The δ¹⁸O of species from the upper layer (surface) is 1-2‰ VPDB lower than species from deep water (subthermocline), which is consistent with an upper ocean that is stratified and calcification of species from the mixed layer high in the water column or during the warmest season. The δ¹⁸O of the planktonic species from deep water is indistinguishable from the benthic foraminifera, which is a reflection of deep water that is relatively fresh at subthermocline levels of the Southern Labrador Sea. Planktonic data from Site 647 are sparse prior to 34 Ma due to the low abundance of foraminifera as well as coring gaps. After 34.5 Ma, however, deep-dwelling planktonic and benthic δ¹⁸O the records separate coincident with the appearance of deep water that had temperature and salinity properties that were similar to values from the Atlantic Ocean. Also, it is documented that there was a progressive collapse of the planktonic-benthic δ¹³C gradient that captures the water column transition from being well-stratified with large vertical δ¹³C differences (1-1.5VPDB) during Regime 1, to a state with a smaller δ¹³C gradient (0.5‰ VPDB) comparable to the modern North Atlantic convection sites (Kroopnick, 1985) that were better-mixed by about 34.3 Ma. Both observations are consistent with increasing volume of the Northern Component Water. An abrupt shift in the benthic assemblages from Site 647 at 34.3 Ma from agglutinated species tolerant of carbonate-poor, nutrient-rich environments, to calcareous species that were suited to a stronger flow of the current (Kaminski & Ortiz, 2014), which was coincident with other changes, provides further evidence that convection increased. Circulation Regime 3 began at 34.3 Ma, when a form of Northern Component Water that was saltier and denser, with a higher δ¹³C is exported through the Southern Labrador Sea.

Deep-water sources and sinks

The fish debris εNd data from Site 647 that was obtained by Coxall et al. behave as a conservative tracer for northern sourced deep water and it is possible to compare it with ocean references that have been published (Burton et al., 1997; O’Nions et al., 1998) in order to identify the probable source of Northern Component Water. The sample obtained that dated to 39 Ma was similar to the remainder of the record, which implies there was no systematic change in εNd and therefore the bottom water provenance Northern Component Water evolved, though they did not reconstruct Nd directly from Regime 1. It is suggested by the comparison that the Southern Ocean, which has the highest end-member εNd signature in the compilation of Coxall et al. Also, they did not find evidence for the prominent shift to higher εNd values that were found in the records of the Southern Ocean (Scher & Martin, 2006). It has been suggested by previous studies that, prior to the deepening of the Greenland-Scotland Ridge, Northern Component Water had its source in the Labrador Sea (Borrelli, Cramer & Katz, 2014; Langton et al., 2016). Deep water in the Labrador Sea of the present, however, has εNd that is characteristically low, about -14), which reflects the erosional inputs from the cratonic hinterland (Lambelet et al., 2016). Site 647 Nd, in contrast, is significantly more radiogenic (εNd = -11.4 to -9.4), and is closer to the range of values that have been measured in the overflows of the Nordic Sea (εNd of about -12.0 to -8.4) (Lambelet et al., 2016; Lacan & Jeandel, 2005) and above 500 m in the proximal Arctic Basins (εNd = -11.7 to -8.8) (Porcelli et al., 2009). In the Palaeogene a presence of deep water at Site 647 that was sourced to the Tethys Ocean is another possibility, as the Tethyian εNd signature (εNd = -10.0 to -0.3) (Grandjean et al., 1987; Stille & Fischer, 1990) is indistinguishable from that of the North Atlantic water masses. Palaeogeographic reconstructions, however, suggest that the exchange of water mass between the European Tethys and the Nordic Seas was during in the Middle to Late Eocene (Akhmetiev & Beniamovski, 2009; Kharin & Lukashina, 2010), which makes this unlikely.

During the Palaeogene there were only shallow connections between the Arctic Ocean and the Nordic Seas (O’Regan et al., 2011; Kharin & Lukashina, 2010; Musatov & Pogrebitski, 20007). The transfer from the Arctic of freshened, nutrient-rich waters would have occurred by a proto Greenland Current. The similarity between the North Sea (Kysing-4 borehole) and Site 647 benthic δ¹⁸O, as well as independent evidence of low salinities in the Nordic Seas (Stärz et al., 2017; Porcelli et al., 2009), is, according to Coxall et al., consistent with this picture. Transport was also shallow from the subarctic seas to the Atlantic Ocean, and the sinking of Northern Component Water that was Arctic-imprinted must have occurred south of the Greenland-Scotland Ridge until it subsided. When there is sufficient cooling in the subarctic seas. The contrast in density between brackish Arctic waters with warmer, saltier surface waters from the North Atlantic permitted sinking, with the result that at Site 647 distinct bottom water is recorded.

It is important that Arctic-imprinted Northern Component Water formation was minimal prior to about 36 Ma, which implies there was a regular stratification and stagnation in the Southern Labrador Sea. According to Coxall et al. this is consistent with the considerable noise in δ¹³C and δ¹⁸O during Regime 1. It has remained uncertain how this deep water has remained isolated in the Southern Labrador Sea. According to Coxall et al. it is possible that in the North Atlantic the production and export of local deep waters were high compared with the influx of deep waters that were sourced from the south, subsequently increasing further as the cooler, saltier Northern Component Water began being produced. There is an alternative possibility that bathymetric highs that were associated with the now extinct Labrador Sea spreading ridges and the West Thulean Igneous Province to the south may have been the cause of the subsurface waters of the Southern Labrador Sea from the overall Atlantic during the early Palaeogene (Egloff & Johnson, 1975). In this case, it was probably important that the cessation of the Labrador Sea spreading close to the Eocene-Oligocene greenhouse-to-icehouse climate transition, as it allowed ridges to subside thereby enabling enhanced export of deep water.

In this paper Coxall et al. illustrate the isotopic evidence and sequence of EOT oceanic changes by use of a natural neighbour regridding of the isotopic data that has been compiled to produce the south-north depth transects in the Atlantic Ocean during time windows that were centred on circulation regimes 1-3. Prior to about 36 Ma, Regime 1, there was a strong isotopic δ¹⁸O and δ¹³C depletion that affected water masses down to about 2,000 m above 40^oN, which corresponded to small amounts of Arctic-imprinted Northern Component Water of low salinity, high-nutrient in the Southern Labrador Sea. Deep waters that have a more homogeneous δ¹⁸O that was sourced from southerly and possibly regions of low latitude (Borrelli, Cramer & Katz, 2014; Langton et al., 2016) filled the remainder of the Atlantic Ocean. A pulse over a period of 0.5-1.0 of Northern Component Water export during Regime 2 accompanied increasing subarctic δ¹⁸O, which reflects progressive salinification and densification of Nordic surface waters. The ‘fresh’ Southern Labrador Sea deep water did not exist by 33.3-34.3Ma, there was an increase in bottom water δ¹³C, and the acute phase of δ¹³C was over and a Northern Component Water was exported that was better ventilated (Regime 3). It is important that the initial pulse of export of Northern Component Water under Regime 2 is recorded by δ¹³C signals downstream of the Southern Labrador Sea that are decreasing in deep waters. In previous δ¹³C records (Borrelli, Cramer & Katz, 2014; Langton et al., 2016; Pusz, Thunell & Miller, 2011) the presence of Late Eocene Northern Component Water in the Atlantic Ocean is not identified as it was assumed that Northern Component Water had a high δ¹³C signature that was similar to that of the modern Atlantic Deep Water.

Causes and consequences of Northern Component Water

In the Late Eocene deepening of Greenland-Scotland Ridge (GSR) for which there is geological evidence (Abelson & Erez, 2017; Stärz et al., 2017), would have increased the Nordic overflows, and thereby strengthened the production of Northern Component Water. It is suggested by modelling that deepening of the Greenland-Scotland Ridge below a depth of 50 m initiates a threshold switch from lagoonal to estuarine circulation that salinifies the Nordic Seas enough to cause intensification of the production of northern deep water (Stärz et al., 2017). The bathymetric history of the GSR is currently too crude to accurately date such a change, though this idea is consistent with the findings of Coxall et al. Also, it is proposed by Coxall et al. that contemporary restrictions to the Arctic-Nordic Sea exchange also played a role. It is suggested by geological evidence that the Barents Sea-Arctic passageway shoaled in the latest Eocene (Kharin & Lukashina, 2010; Musatov & Pogrebitskij, 2000) and that variations in relative sea-level in the Arctic were decoupled from global trends from the Late Eocene to the Early Miocene (Hegewald & Jokat, 2013). The salinification of the Nordic Seas as brackish overflows from the Arctic were gradually cut off enhanced the palaeogeographic isolation of the Arctic.

Changes in the production of NCW had effects that were varied and competing. It is assumed that its onset affected polewards transport of heat in both hemispheres (Tigchelaar, von der Heydt & Dijkstra, 2010; Zhang et al., 2011). Coxall et al. suggest that the initial export of nutrient-rich, Arctic-imprinted NCW may have generated a pulse of CO₂ that was short-lived on the order of 100-200 ppm, which is consistent with the proxy compilations that display a temporary reversal in the dropping CO₂ trend between about 34 Ma and about 35 Ma (Anagnostou et al., 2016). A strengthening of the production of NCW and enhanced ocean heat transport to the north, on the other hand, could have played a role in the drawdown of CO₂ in the long term as a result of an accompanying rainfall increase over land and associated feedbacks with CO₂-weathering (Elsworth et al., 2017). The change in timing of the circulation change, 1-2 Myr before the glaciation of Antarctica, reinforces the idea that a role was played by the onset of NCW in preconditioning the Earth system during the Late Eocene for the transition from greenhouse to icehouse.

Sources & Further reading