Late Miocene Onset of Tasman Leakage and Southern Hemisphere Supergyre Ushers in Near‐Modern Circulation

This study provides a Miocene‐to‐recent history of Tasman Leakage (TL), driving surface‐to‐intermediate waters from the Pacific into the Indian Ocean. TL, in addition to Indonesian ThroughFlow (ITF), constitutes an important part of the Southern Hemisphere Supergyre. Here, we employ deep‐sea benthic δ13C timeseries from the southwestern Pacific and eastern Indian Oceans to identify the history of Tasman Leakage. The δ13C results combined with sedimentary evidence show that an inter‐ocean connection south of Australia existed from 7 Ma onward. A southward shift in Westerlies combined with a northward movement of Australia created the oceanic corridor necessary for Tasman Leakage (between Australia and the sub‐Antarctic Front) at this time. Furthermore, changes in the northern limb of the Supergyre (ITF) are evident in the sedimentary record on Broken Ridge from ∼3 to 2 Ma when Banda Sea intermediate waters started originating from the North Pacific.


Modern Tasman Leakage and the Southern Hemisphere Supergyre
The SHS is an oceanographic concept introduced by Speich et al. (2007) and Ridgway and Dunn (2007), although earlier works describe ocean gyres linked through pathways north and south of Australia (Tilburg et al., 2001). From most to least intense, the contributing oceanic gyres linked in the SHS are the Indian, South Pacific, and Atlantic (Ridgway & Dunn, 2007). In the SHS concept, the Pacific and Indian surface waters function as one system via the Indonesian Throughflow (ITF), and the TL connects the South Pacific and Indian Oceans at intermediate depths (Ridgway & Dunn, 2007). ITF, TL, and Drake Passage waters converge in the Agulhas Current to complete the SHS (Durgadoo et al., 2017;Speich et al., 2002). Recent works hint at the significance of the SHS in the global ocean circulation and related impacts on regional climate patterns. For example, Duan et al. (2013) report a cooling and freshening of intermediate waters within the SHS over the last decades, in concert with a central-south Pacific-focused 2.5° southward shift in position. Behrens et al. (2019) detail the influence of enhanced TL on regional climate, including diminished meridional temperature gradients and westerlies in the Southern Ocean and resulting in a positive feedback (reduced AAIW formation and southward expansion of the SHS). Fan et al. (2020) suggested the SHS may control drought in South Australia, whereby TL constitutes the teleconnection between continental precipitation and the SHS. These new insights, coupled with the projection of enhanced TL flow in future decades (Oliver et al., 2015), highlight the need for a comprehensive understanding of the TL and its role in the SHS.
Present-day TL is focused at upper intermediate depths as determined from observations, models, and float studies (Ridgway & Dunn, 2007;Rintoul & Bullister, 1999;Speich et al., 2002Speich et al., , 2007van Sebille et al., 2012van Sebille et al., , 2014; Figure 1). The East Australian Current (EAC) transports heat and high-salinity waters poleward to the mid-latitudes and into the Tasman Sea (Hu et al., 2015;Ridgway & Dunn, 2003;Sloyan & Rintoul, 2001; Figure 1). The TL originates when a portion of the EAC waters move westward around Tasmania within the upper 1,000 m of the water column (van Sebille et al., 2012(van Sebille et al., , 2014 to form TL (Speich et al., 2007;Figure 1). When flowing between the Australian continent to the north and the Antarctic Circumpolar Current (ACC) to the south (Ridgway & Dunn, 2007), the TL core sits between 400 and 900 m water depth (van Sebille et al., 2014). Within this oceanic corridor, TL is influenced by the position of the Subtropical Front (STF; Rintoul & Sokolov, 2001) and Australia itself. Once past the Australian continent, TL flows in a northwesterly direction between 110 and 95°E (Speich et al., 2002) over the Broken Ridge (Figure 1). TL gradually blends in with the subtropical gyre (Durgadoo et al., 2017;Speich et al., 2002) Speich et al., 2002), and partly directly feeding the Agulhas Current (Durgadoo et al., 2017). Retroflection from Southern Ocean AAIW and Agulhas Current somewhat freshen the high-salinity TL intermediate waters (Rintoul & Sokolov, 2001), but its salinity does not change much until reaching the Agulhas Current region (Ridgway & Dunn, 2007). Agulhas waters crossing the Cape of Good Hope are predominantly Pacific source water, with 50% of the water originating from the ITF and 42% from TL (Durgadoo et al., 2017).
The TL can be readily identified in salinity profiles around Australia (Figures 1 and S2; Table S1). TL flow is ∼4 Sv in both modeling (van Sebille et al., 2012) and float (Rosell-Fieschi et al., 2013) studies, compared to 14.3 Sv through the ITF (van Sebille et al., 2014), whereas more recent modeling suggests a somewhat greater role for TL (Rousselet et al., 2020). TL flow varies seasonally (Rosell-Fieschi et al., 2013;van Sebille et al., 2012) and largely depends on large-scale wind patterns over the Pacific (van Sebille et al., 2014). Indeed, the surface waters that feed TL display the greatest seasonal variability, whereas the upper intermediate TL waters are more perennial (Rosell-Fieschi et al., 2013). At 115°E flow increases to 5.0 ± 1.8 Sv (Rosell-Fieschi et al., 2013), as the TL is strengthened with waters from an anticyclonic loop from 139 to 146°E located between the Great Australian Bight (GAB) and the ACC (Rintoul & Sokolov, 2001).

Tasman Leakage Importance in the Global Thermohaline Circulation
Global thermohaline circulation was long thought to be a relatively simple system: deep water formed in the North Atlantic was distributed globally via the ACC, with most return surface flow occurring from the North Pacific to the South Atlantic via the ITF and the Agulhas Current. The reality, however, is more complex (e.g., , with the Indian Ocean increasingly seen as playing a more critical role in thermohaline circulation (Schott et al., 2009). It is now clear that both the ITF and TL are important inter-basinal connections that ultimately provide return waters for the North Atlantic through the Agulhas Current (Durgadoo et al., 2017;Rousselet et al., 2020). Flow through both the Tasman and Agulhas Leakages has been increasing over the past decades (Qu et al., 2019; Figure 1). Heat flux from the Pacific to the Indian through the TL is ∼1/3 of the ITF flux (van Sebille et al., 2012), whereby TL may even compensate for decreasing ITF volumes during El Niño times (van Sebille et al., 2014). Modern studies are shedding light on regional impacts and feedbacks (Behrens et al., 2019;Duan et al., 2013;Fan et al., 2020) of the TL and its role in the SHS. Unraveling the mysteries of the TL in the past will help define the complexity of Indian Ocean circulation and its relationship to the Atlantic and Pacific Oceans and may illuminate the global controls displayed by the SHS.

Detecting Tasman Leakage in the Geologic Record
This paper seeks to advance the understanding of TL and the SHS by determining its onset as recorded in geologic archives. Thereby, our objective is to provide first-order answers to three open questions. (a) When did the TL originate? (b) Did TL impact ocean circulation and global climate? (c) Did ITF restriction influence Indian Ocean circulation at intermediate-water depths (AAIW, TL, and Banda Intermediate Water, BIW)? We explore regional circulation patterns from ∼12 Ma onward, using benthic carbon-isotope and sedimentologic data from five sites ( Figure 1). We employ δ 13 C to identify TL and differentiate it from other water masses as previously demonstrated (e.g., Hodell & Venz-Curtis, 2006;Poore et al., 2006;Tian et al., 2018). We complement the δ 13 C proxy with regional sedimentologic observations (hiatuses, winnowing, and slumping) to reconstruct past ocean circulation change. This combined approach allows pinpointing of TL onset in geologic time and suggests a combination of paleoceanographic and tectonic controls: TL onset was only able to occur when there was room for the current to flow west, south of Australia, and north of the sub-Antarctic Front.

Materials and Methods
This project draws on the legacy of scientific ocean drilling sediments and data. It utilizes previously published δ 13 C benthic records from sediments collected through scientific ocean coring (Deep Sea Drilling Project  Table 1). These lower resolution records of mostly single genera of benthic foraminifera are readily comparable, providing valuable insight into long-term, regional trends. Since high-resolution age-depth models are not applicable for these sites, we created simple age-depth models by linear interpolation in-between calcareous nannoplankton biostratigraphic datums (Table S2), updated to the GTS2012 (Gradstein et al., 2012). As a first-order assessment of the individual age-depth models, benthic foraminiferal δ 18 O records of each study site were plotted against the δ 18 O benthic megasplice (De Vleeschouwer et al., 2017; Figure S5). Their congruency supports our linear age-depth conversion.

Results: Onset of the Tasman Outflow
Based on the modern system, we anticipate that the studied δ 13 C benthic records are largely dominated by two distinct signals: a South Pacific end-member (Tasman Sea/TL) and a Southern Ocean end-member (AAIW; Figure 1). Sites currently bearing a Pacific Ocean signature (DSDP Site 590 and ODP Sites 752 and 754) contrast with sites representing an AAIW signature (ODP Sites 751 and 1172; Figure 2). As expected, the AAIW sites largely track the Southern Ocean deep-water δ 13 C signal (e.g., Poore et al., 2006) throughout the entire studied interval. The Broken Ridge Sites 752 and 754, on the other hand, only track the AAIW records until ∼7 Ma, after which they diverge (Figure 2), suggesting the appearance of a more modern configuration. Prior to 7 Ma, Broken Ridge δ 13 C values are ∼0.3‰ heavier than AAIW sites δ 13 C, yet fall within the range for upper component deep waters reported by Hodell and Venz-Curtis (2006). DSDP Site 590 benthic δ 13 C values exhibit strong similarity with Pacific deepwater δ 13 C records of Cramer et al. (2009)  The simultaneous convergence of TL benthic δ 13 C and divergence between the Broken Ridge and AAIW δ 13 C records support a ∼7 Ma onset of waters flowing from the Tasman Sea into the Indian Ocean, supplying intermediate-depth bottom waters to Broken Ridge Sites 752 and 754. TL onset is thus synchronous with a late Miocene carbon isotope shift (LMCIS) in δ 13 C in intermediate and deeper (Hodell & Venz-Curtis, 2006;Poore et al., 2006;Tian et al., 2018) waters, but the LMCIS alone does not explain our results. Before 7 Ma, all three locations closely follow global patterns in deepwater δ 13 C, reflecting the contributions of those water masses (Figure 2). However, at 7 Ma, only AAIW continues to follow the same patterns as the other oceanic records (e.g., Hodell & Venz-Curtis, 2006) as TL upper intermediate water comes into existence and determines the δ 13 C signatures of Sites 590, 752, and 754. The ∼7 Ma onset of the southern limb of the SHS corresponds with the change to more modern δ 13 C gradients, particularly evident in South Atlantic and Pacific records (Hodell & Venz-Curtis, 2006 Dickens and Owen (1994). Winnowing data from House et al. (1992). SAF position from Groeneveld et al. (2017). Monsoon intensification from Betzler et al. (2016). Global sedimentary changes data from Eberli et al. (2019). Great Australian Bight data from Li et al. (2004) and International Ocean Discovery Program (ODP) Leg 182 Initial Reports (Feary et al., 2000). NE African and Gulf of Aden data from Liddy et al. (2016). Benthic oxygen isotope megasplice from De Vleeschouwer et al. (2017). Abundant radiolaria at Site 1172 from Diester-Haass et al. (2006) supplement.

Discussion: TL and Indian Ocean Circulation
Australia's northward motion led to the progressive uplift of the maritime continent and drove climate change in Australia Groeneveld et al., 2017;Sniderman et al., 2016), and altered surface (Auer et al., 2019;De Vleeschouwer et al., 2018Smith et al., 2020) and deeper (Karas et al., 2009(Karas et al., , 2011a(Karas et al., , 2011b water circulation. Australia's northward motion progressively constricted the ITF, defining the northern limb of the SHS. Australia's tectonic motion also opened an oceanic gateway to its south, setting up the southern limb of the SHS by creating physical space for westward intermediate water flow north of the ACC. The late Miocene position of Australia and Tasmania created a southwestern pathway for waters leaving the Tasman Sea to flow around the Tasman Plateau. Critically, at this time, Tasmania sat north of the Subantarctic Front (SAF; Nelson & Cooke, 2010), outside of the main ACC circulation. Southern Australia's migration out of this polar front was mapped using faunal and isotopic values in the Southern Ocean south of New Zealand (Nelson & Cooke, 2010) and is expressed in our study area by a rapid drop in radiolarian abundance at ∼7 Ma at ODP Site 1172 (located at 43°S today; Diester-Haass et al., 2006; Figure 2). It is also supported by the Groeneveld et al. (2017) study that documents a concurrent southerly shift in the subantarctic front and westerlies. Once Australia sat north of the SAF, the ACC's oceanographic barrier to westward flow was removed, facilitating TL and SHS onset. While it is possible that waters may have flowed west along this pathway before 7 Ma, aided by shifting westerlies , modeling suggests the position of the STF here is controlled less by wind belts and more by topography . In other words, there is no evidence for sustained TL before 7 Ma. Two intriguing intervals of convergence between the Broken Ridge and TL records at ∼10.5 and 9.5 Ma might be evidence for short windows permitted by a southward shift in both westerlies and the STF . However, these windows likely failed as attempts at TL onset as the more southerly position of Australia would have inhibited extensive flow by Pacific-sourced intermediate waters at those times.
We postulate that the early TL water mass was similar to the modern: waters sourced dominantly from Tasman or Coral Sea thermocline (EAC) waters that persist at upper intermediate depths beyond Tasmania ( Figure 1) (Sites 751 and 1172). This interpretation arises from the similarity between the Broken Ridge records and the deeper water AAIW records (e.g., Site 1088, Billups, 2002;Poore et al., 2006). The fact that the isotopic convergence between the Broken Ridge sites and the Tasman Sea Site 590 is coeval with the isotopic divergence of the Broken Ridge sites from the AAIW sites ( Figure 2) advocates a South Pacific gyre connection through the Tasman Sea and the onset of TL at 7 Ma. This interpretation requires the intermediate water pathway south of Australia to be open at 7 Ma. It largely accords with changes in benthic foraminiferal biofacies at ODP Site 757 (90E Ridge) from a well-oxygenated water mass with strong lateral advection to a sustained high flux of organic matter Singh and Gupta (2004) and illustrates the improved connectivity between the Indian and Pacific Oceans (Tasman Leakage).
Sedimentation patterns along the TL pathway (Coral Sea, GAB, and Broken Ridge) also hints at a ∼7 Ma TL onset (Figure 2 winnowing associated with TL onset at 7 Ma (Figure 2; House et al., 1992). In addition, the alignment of Broken Ridge winnowing (House et al., 1992) with evidence for a southerly shift in westerlies  provides additional support for an atmospheric enhancement to the TL pathway. Interpretation of seismic reflection profiles from ODP Leg 194 Marion Plateau indicates a shift in contourite drift sedimentation at 7.1 Ma (Betzler & Eberli, 2019), supporting enhanced SHS flow in the Coral Sea via the Pacific gyre associated with TL onset. Major changes in the Maldives in the northern Indian Ocean also support an invigorated circulation system ∼7 Ma ( Figure 2) and a shift toward a more complex Indian Ocean circulation. The Maldives underwent two major changes around the time of the onset of TL: a reduction in upwelling and a change in drift sedimentation by virtue of a N-NE directed bottom current (Betzler et al., 2016), likely driven by enhanced ventilation of the Indian Ocean. These events are coincident with the global peak in biogenic carbonate deposition (or preservation; Diester-Haass et al., 2006;Hermoyian & Owen, 2001).
The global change in depositional character of carbonate banks and platforms supports enhanced global thermohaline circulation (e.g., Hodell & Venz-Curtis, 2006;Keating-Bitonti & Peters, 2019) as predicted by the onset of SHS, and more North Atlantic water reaching the South Atlantic (e.g., Gruetzner et al., 2019). The AAIW ODP sites (751, 757; Figure 2) capture the globally observed LMCIS ∼0.8‰ δ 13 C shift toward more negative values (e.g., Hodell & Venz-Curtis, 2006;Keigwin, 1979;Keigwin & Shackleton, 1980), which is synchronous in surface and deep waters (Drury et al., 2017(Drury et al., , 2018. Tasman Sea Site 590 does not follow the trend of more negative values in the major oceans and instead converges with Sites 752 and 754 by increasing +1‰ as TL turns on, reflecting the influence of EAC-supplied, δ 13 C-heavier TL waters. The TL records (Sites 590 and 752/754) track one another until the Plio-Pleistocene boundary, when values at the Broken Ridge increase between 2.6 and 1.8 Ma. These dates are associated with the onset of the Arid Interval in NW Australia  and enhanced early Pleistocene cooling at Site U1463 (Smith et al., 2020), respectively. The 2.4 Ma onset of the Australian Arid Interval is a manifestation of progressive restriction of the ITF. Concurrence with increasing δ 13 C at ODP Sites 752/754 suggests enhanced contributions of ITF-sourced BIW at Broken Ridge as ITF restriction increased inflow of North Pacific waters (Karas et al., 2009). We infer a stronger earliest Pleistocene influence of the northern limb of the SHS at Broken Ridge and greater contributions of more positive δ 13 C BIW (Rochford, 1966;Talley & Sprintall, 2005) resulting from continued restriction within the ITF.
Dickens and Owen (1994) interpret a steep decline in Mn and Mn/Sc around 7 Ma at the Broken Ridge as evidence that the Indian Ocean Oxygen Minimum Zone (OMZ) extended that far south. Remarkably, the isotopic divergence between the AAIW and Broken Ridge sites is greatest at ∼5 Ma (Figure 2), which corresponds with what (Dickens & Owen, 1994) interpreted as further OMZ intensification at Broken Ridge driven by exceptionally high productivity in the Atlantic, Indian and Pacific Oceans during the latest Miocene and early Pliocene (i.e., the so-called biogenic bloom; Dickens & Owen, 1999 (Peterson et al., 1992) supports high productivity in the surface waters and explanation [1]. Dickens and Owen (1999), on the contrary, argue that the elevated MARs are the result of a reduction in winnowing and ocean circulation (in line with explanation [2]). Benthic foraminiferal analyses at Site 752 indeed evidence low-oxygen availability throughout this interval (Singh et al., 2012), but high variability in these records (Ridha et al., 2019;Singh et al., 2012) suggests that productivity levels were not constant, or that low-oxygen waters were moving in and out of the region. Furthermore, contrary to expectations under model [2], the OMZ event is less intense at the more northerly 90 E Ridge Site 757 (Dickens & Owen, 1999;Singh et al., 2012). We favor explanation [3] and emphasize that the opening of the TL gateway requires a reinterpretation of the benthic foraminiferal ecological variability as a reflection of changing balances between Indian deep waters, the advection of TL waters, and surface water productivity over Broken Ridge.

Conclusions
The onset of TL can be discerned in the sedimentary record from 7 Ma onwards. The late Miocene onset of the SHS is coeval with many important climatic and oceanographic events, including the LMCIS, establishment of "modern" intermediate waters (Hodell & Venz-Curtis, 2006) and simultaneous peaks in carbonate production between the eastern Pacific and the South Atlantic (Drury et al., 2021). Our study demonstrates that TL turned on at 7 Ma, and therewith provides a new piece of the cause-and-effect chain shaping the major transformation of oceans and climate at that time. However, detailed answers await better high-resolution paleoceanographic records, which in turn await better sediment core and future Ocean drilling. Indeed, numerous questions remain unanswered with the current paleoceanographic data sets. For example, how did the TL onset impact global heat flow? Did it play a direct role in the global cooling at 7 Ma (Herbert et al., 2016), or was its chief contribution that it served as a sentinel for the opening of the TL gateway and the associated oceanic circulation changes? Was this the start of enhanced global thermohaline circulation? Carbonate platform drowning at ∼7 Ma on the Marion Plateau (Eberli et al., 2010; Figure 2) further implicates TL and the SHS, suggesting enhanced circulation evident in the Indian Ocean is linked to invigorated global circulation. What controls the variability of surface water productivity and TL at the Broken Ridge in the late Miocene, and did a shift in westerlies aid in the onset of TL? Might this shift in westerlies be the atmospheric link that aids in regional (e.g., Feakins et al., 2020; and global patterns in cooling and drying at 7 Ma? These questions cannot be answered with the available sediments and research cannot progress without high quality core material that will allow for detailed analyses across these critical intervals at locations that define the limits of a heretofore unrecognized limb of global circulation in the sedimentary record. Re-drilling the Broken Ridge to provide suitable material to generate high-resolution proxy records is critical to answering these questions. Likewise, obtaining Miocene-age paleoceanographic records in the Tasman Sea is also paramount to defining the role of TL and the SHS in global circulation.

Acknowledgments
This work was supported by a German Academic Exchange Service (DAAD) fellowship (2019) to Christensen and the MARUM Cluster of Excellence: The Ocean Floor-Earth's Uncharted Interface to De Vleeschouwer. This study builds upon the over 50 years legacy of scientific ocean drilling, and so we are grateful to the crew and staff of the R/V Glomar Challenger and the R/V JOIDES Resolution for collecting the core, often in challenging conditions. Likewise we are indebted to the many scientists who first asked the questions that allowed for drilling to occur, and who then analyzed the sediments and published the data. Their efforts both advanced our knowledge at the time, and built the critical framework that allows us to continue to deepen our understanding of earth and its systems.