Evolution of the southwestern Pacific surface waters during the early Pleistocene

ABSTRACT

We present alkenone-derived sea surface temperature (SST) records from the New Zealand/Tasmania sector of the Southern Ocean (ODP Sites 1171 and 1125) during the Early Pleistocene (c. 1.7 to 0.7 Ma). Our data show that the subantarctic surface waters (ODP Site 1171) cooled by c. 4 °C and reached near modern conditions after 0.88 Ma, indicating a long-term northward migration of the Southern Ocean fronts. This long-term cooling occurred in two major steps between 1.25–1.1 Ma and 0.95–0.88 Ma, which correlate with significant increases in Antarctic ice volume and global circulation changes. The comparison between the subantarctic and subtropical SST trends tends to indicate that the Subtropical Front was unstable during the Early Pleistocene with significant southward incursions of subtropical waters during warm interglacials (1.07 Ma and 0.95 Ma identified as MIS31, MIS25).

KEYWORDS:

• Alkenone-derived sea surface temperatures
• Early Pleistocene
• Mid-Pleistocene transition
• New Zealand/Tasmania sector of the Southern Ocean
• Subtropical Front
• subantarctic zone

Introduction

The Southern Ocean connects the Atlantic, Indian and Pacific Oceans through the Antarctic Circumpolar Current and plays a prominent role in both sequestering CO₂ (Caldeira & Duffy 2000) and redistributing heat between hemispheres and from the surface to the deep ocean.

The climatic evolution of the Southern Ocean is closely linked to the migration of three prominent oceanic fronts (Figure 1): the Subtropical Front (STF) between the subtropical waters (STW) and subantarctic waters (SAW), the Subantarctic Front (SAF) at the boundary of the Subantarctic Zone (SAZ) and the Antarctic waters, and the Antarctic Polar Front at 62°S in the western Pacific sector of the Southern Ocean. It is well established that during the Pleistocene, the Southern Ocean frontal system migrated northward during cold periods and south during warm periods (Morley 1989; Howard & Prell 1992; Findlay & Giraudeau 2002; Pelejero et al. 2006; Sikes et al. 2009; Hayward et al. 2012; Bostock et al. 2015). However, in the southwest Pacific, the STF latitudinal displacements during glacial–interglacial (G–IG) climate cycles have been complicated by the shallow bathymetry around New Zealand and remained locked on the Chatham Rise over the last few glacial cycles (Weaver et al. 1998; Sikes et al. 2002).

Figure 1. Modern SST (March 2013; Naval Research Laboratory–Navy Coastal Ocean Model [Internet]. Available from: http://www7320.nrlssc.navy.mil/) and positions of the main water bodies, fronts and surface currents in the western sector of the Southern Ocean (after Orsi et al. 1995). S-Tasman Rise, South Tasman Rise; Ch. Rise, Chatham Rise; SEC, South Equatorial Current; EAC, East Australia Current; EAUC, East Auckland Current; ECC,  East Cape Current; TF, Tasman Front; N-STF, North Subtropical Front, S-STF, South Subtropical Front; STFZ, Subtropical Frontal Zone; SAF, Subantarctic Front.

In our study, we focus on the interval between c. 1.7 and 0.7 Ma, encompassing the Mid-Pleistocene Transition (MPT, between 1.25 and 0.7 Ma, Clark et al. 2006). This critical period in the Earth’s climate history features a long-term cooling trend affecting both low and high latitudes that started at 1.2 Ma [Marine Isotype Stage (MIS)35–34] and the transition from obliquity (41 kyr) paced ice volume variations to high-amplitude eccentricity (100 kyr) paced G–IG oscillations (Mudelsee & Stattegger 1997; McClymont et al. 2013).

The long-term Pleistocene cooling is well documented in the Atlantic sector of the Southern Ocean but less is known about the Pacific sector (Rodriguez-Sanz et al. 2012). The primary objective of this study was to document the long-term evolution of the subantarctic surface waters during the Early Pleistocene and to track the position of the STF in the southwestern sector of the Pacific through the MPT. To address this challenge, we reconstructed sea surface temperatures (SST) using alkenone paleothermometry () across the STF in the New Zealand/Tasmania sector of the Southern Ocean. We focus our analyses on two ODP sites: Site 1171 positioned south of Tasmania and south of the modern STF in the SAW; and Site 1125 located east of New Zealand, in the southwest Pacific STW directly north of the STF.

Modern oceanographic setting

The surface circulation in the southwest sector of the Pacific Ocean is marked by three components (Figure 1), the subtropical inflow from the north, the subantarctic inflow from the south and the occurrence of the two major frontal systems, the STF and the SAF. Subtropical inflow comprises warm (over 15 °C in summer), high-salinity, macronutrient-poor, micronutrient-rich STW.

The subtropical inflow in the southwest Pacific is connected to the warm east Australian current that flows from the north along the east coast of the Australian continent (Figure 1) and is deviated eastward around 32°S (Ridgway & Godfrey 1994; Ridgway et al. 2002) as a response to wind stress (Tilburg et al. 2001). It enters by the South Pacific subtropical gyres as the Tasman Front and flows along the eastern side of New Zealand, forming the East Cape Current before turning eastward along the northern slope of the Chatham Rise.

The subantarctic inflow comprises cool (< 15 °C in summer), low-salinity, macronutrient-rich and micronutrient-poor waters that flow northward along the eastern side of South Island of New Zealand before reaching the southern slope of the Chatham Rise.

The convergence between these two water masses (the STF) in the Southern Ocean is complex (Graham & De Boer 2013). In the Tasman Sea, the STF spans a 500 km frontal zone (STFZ) that is delimited by the North Subtropical Front (N-STF) and the South Subtropical Front (S-STF) (Figure 1; Smith et al. 2013; Chiswell et al. 2015 and reference herein). After flowing around the south of the New Zealand’s South Island, the STF bends north along the east coast, after which it becomes bathymetrically locked along the continental margin and the Chatham Rise (see Chiswell et al. 2015 for review). East of the Chatham Islands, the front becomes dynamic.

The S-STF, which is the poleward extent of the subtropical waters (STW) and is considered the main STF (Hamilton 2006), is characterised by a SST gradient of 4 °C over 0.5° of latitude (Rintoul et al. 1997). Southern Ocean fronts today migrate north and south seasonally across wide areas of the ocean (Belkin & Gordon 1996) and are generally controlled by wind (McCartney 1977), except in some locations where subsurface topography influences their location. For example, south of Tasmania, where ODP Site 1171 is located, the STF lies between the South Tasman Rise and Tasmania (Orsi et al. 1995; Belkin & Gordon 1996) and east of New Zealand, where ODP Site 1125 was drilled, the STF is steered by the Chatham Rise (Heath 1981, 1985; Belkin & Gordon 1996; McCave et al. 2004).

Material and methods

Core location and chronostratigraphy

ODP Site 1171 (48°30′S, 149°6′E) is located on the southwestern slope of the South Tasman Rise, c. 550 km south of Tasmania (Figure 1). This site lies in modern SAW. Pleistocene sediment mainly consists of foraminiferal nannofossil ooze. The modern average annual SST in this area is 10 °C, with the coolest temperatures (9 °C) recorded in July–September and the warmest (c. 11 °C) between January and March (Locarnini et al. 2010).

ODP Site 1125 (42°33′S, 178°10′W) is located 610 km east of New Zealand’s South Island (Figure 1) on the northern slope of the Chatham Rise at 1360 m depth (Carter et al. 1999). Sediments recovered at Site 1125 comprise pelagic and hemipelagic clayey nannofossil ooze interbedded with nannofossil-bearing silty clay that was deposited continuously with little disturbance over the last glacial cycles (Carter et al. 2000). The mean annual SST in this area is 16 °C, with winter temperatures averaging 13 °C and the warmest summer temperatures being 18 °C measured in January–March (Locarnini et al. 2010).

We used chronologies for Sites 1171 and 1125 from shipboard paleomagnetic data, as guided by the core catcher first (FAD) and last (LAD) appearance nannofossil datum (Figure 2), and published age model above 20.3 m at Site 1125 (Schaefer et al. 2005). Core catcher ages have a large depth error because FADs may occur up to 9 m deeper, and LADs 9 m shallower. At Site 1171, shipboard magnetic studies indicate that sediment demagnetises in low field strength alternating fields and that the dominant magnetic mineral in sediment is likely magnetite (Exon et al. 2001). Therefore, we used the 20 mT demagnetisation data measured at 5 cm intervals from holes B and C to construct the magnetostratigraphy. The magnetostratigraphy is relatively unambiguous, but demagnetisation data from hole B are better behaved than those from hole C, which contains anomalous shallow inclination intervals from the top of the core to c. 12 m. The anomalous shallow inclinations likely resulted from experiments performed to test the effect of different core liners and barrels (Exon et al. 2001). The correlation with the magnetic polarity timescale is unambiguous, as guided by the LADs of Pseudoemiliania lacunosa at 6.995 ± 0.05 m and Calcidiscus macintyrei at 24.06 ± 0.375 m, which have ages of 0.46 and 1.67 Ma, respectively (Stickley et al. 2004), and the LAD of Helicosphaera sellii (1.26 Ma) at 21.435 ± 0.38 m (Stant et al. 2004). Accordingly, we identify the base of C1n at c. 13.1 m, and the top and base of C1r.1n at c. 15.6 and 17.2 m, respectively. Overall, our age model agrees well with the most recent published age model of Stant et al. (2004), who used additional FAD and LADs. At Site 1125 we adopt the chronostratigraphies of Hayward (2002) and Schaefer et al. (2005) to a depth of 21.53 m (0.978 Ma), after which we use the magnetic polarity record from shipboard analyses. Shipboard analyses comprised natural remanent magnetization (NRM) magnetic moment measurements followed by 20 mT alternating field demagnetisation, which reduced the magnetisation to 10⁻⁶ A/m. The resulting demagnetisation data are poor and likely contaminated by a vertical drilling over print, which was pervasive on cores recovered during this leg (Carter et al. 1999). However, a recognisable reversal record is present and, as guided by calcareous nannofossil datum, provides age control for our records. The correlation of the magnetic polarity reversals with the geomagnetic polarity time scale is guided by the FAD of Gephyrocapsa parallela at 23.37 m and LAD of Reticulofenestra asanoi, which have ages of c. 0.95 and 0.85 Ma, respectively. The LAD of H. sellii at 31.31 m indicates an age of c. 1.26 Ma. Therefore, we identify the base of C1n at c. 20.2 m, and the top and base of C1r.1n at c. 21.3 and 23.3 m, respectively. The base of the record is constrained by the FAD of Gephyrocapsa (medium) at 37.88 m, which has an age of 1.69 Ma (Lourens et al. 2004), and the C2r.2r–C2An.1n reversal, which has an age of 2.581 Ma (Sabaa et al. 2004). Data were converted to age by linear interpolation between geomagnetic reversals.

Figure 2. Magnetic polarity stratigraphies for ODP Sites 1171 and 1125 constructed from shipboard, 20 mT demagnetisation data measured on working half cores. Nannofossil FADs and LADs, and existing age models of Hayward (2002) and Schaefer et al. (2005) guide the correlation with the magnetic polarity timescale (Gradstein et al. 2012).

All depths refer to the metres composite depth scale using the splice by Exon et al. (2001) for Site 1171 and Carter et al. (1999) for Site 1125.

Sea surface temperature reconstructions

For Site 1171A, alkenone unsaturation ratios were measured on samples from 1171A–2H–4W to 1171A–3H–5W at a resolution of one sample every 20 cm (between 12.94 and 22.65 m). For Site 1125A, we measured alkenones on samples from the interval between 1125A–3H–2W and 1125A–4H–6W. Because of drilling disturbances (Figure 2) at Site 1125, our sampling was not continuous. For the intervals of good sediment recovery, the average resolution is of one sample every 30 cm (between 18.78 and 33.2 m).

Alkenones were extracted from freeze-dried sediments (c. 6 g) using a Dionex 300 Accelerated Solvent Extractor with dichloromethane and methanol (9:1, v/v) at 1500 psi and 100 °C. Alkenones were isolated by silica gel chromatography using solvents of increasing polarity following the procedure of Sicre et al. (2001). The fractions containing alkenones were concentrated, transferred to clean glass vials, evaporated under nitrogen and reconstituted in dichloromethane. Gas chromatographic analyses were performed on a Hewlett-Packard 6890 gas chromatograph with a flame ionisation detector using a fused silica capillary column (Rxi-1 ms, 50 m long, 0.32 mm internal diameter, 0.25 μm film thickness). Helium was used as a carrier gas. The index was calculated from the C₃₇ alkenones to derive SSTs using the global calibration of Conte et al. (2006): , which is adapted for both subantarctic and subtropical Southern Ocean water masses and has a mean standard error of ± 1.2 °C (Conte et al. 2006). Alkenone-derived SSTs in the Southern Ocean are known to mainly reflect summer temperatures (King & Howard 2001, 2003; Sikes et al. 2005, 2009).

Results: sea surface temperature trends

Alkenone-derived SSTs measured at Site 1171 (SST₁₁₇₁) vary between 9.2 ± 1.2 °C and 18.2 ± 1.2 °C, whereas those at Site 1125 (SST₁₁₂₅) vary between 11.7 ± 1.2 °C and 19.3 ± 1.2 °C and are on average 3 °C warmer than SST₁₁₇₁ (Figure 3, Table S1). For the studied interval, SST₁₁₇₁ values are on average 3–4 °C warmer than the modern summer temperature until 0.85 Ma when they approach modern values. Although the SST record for Site 1125 is incomplete, our results show that for the studied interval, SST₁₁₇₁ and SST₁₁₂₅ record a long-term 3 °C cooling between c. 1.35 Ma and c. 0.88 Ma. From 0.85 Ma, SST₁₁₇₁ approaches the modern values and SST₁₁₂₅ averages 15 °C until the end of our record.

Figure 3. Summary of the evolution of the Subantarctic and Subtropical zones in the southwest Pacific sector of the Southern Ocean during the Early Pleistocene. The SAW are depicted by the alkenone-derived SSTs from ODP Site 1171. For the Subtropical Zone, we base our interpretations on SSTs from ODP Site 1125 (alkenone-derived SSTs from this study, modern analogue technique (MAT)-derived SSTs from Schaefer et al. 2005), ODP Site 1123 (Crundwell et al. 2008) and MD06-2989 (Hayward et al. 2012). For MAT-derived SSTs, the thick lines represent the 40 kyr smoothed records. For alkenone-derived SSTs, the error bars are shown. Each warm marine isotope stage (MIS) is shaded in grey and the extension of the Mid-Pleistocene Transition (MPT) is indicated by the vertical thick black bar. Modelled 40 kyr smoothed record of the Antarctic ice volume is from Pollard and Deconto (2009).

Discussion

Our long-term SST record at Site 1171 (Figure 3) is very similar to the alkenone-derived SST reconstruction of Martínez-Garcia et al. (2010) from the SAW of the Atlantic Ocean (ODP Site 1090). The agreement between these data sets indicates that our SST record from Site 1171 reflects changes that were Southern Ocean wide. To track the long-term interaction between the SAW and STW bodies in the southwest Pacific, we compare our new SST records with published ones. Available data for the STW (Schaefer et al. 2005; Crundwell et al. 2008; Hayward et al. 2012) were derived from the modern analogue technique and artificial neural network method on planktic foraminifera and overlap with our SST₁₁₇₁ during the interval between 1.1 and 0.7 Ma (Figure 3).

The pre-MPT: before 1.25 Ma

Prior to 1.25 Ma, subantarctic SSTs were on average 4 °C warmer than modern summer temperatures. Evidence of warm conditions is supported by the calcareous nannofossil assemblage composition at Site 1171, which is characteristic of the SAZ with abundant low latitude taxa (Stant et al. 2004). Alkenone-derived SSTs at subtropical Site 1125 are similar, with conditions that are on average 3–4 °C warmer than SST₁₁₇₁. The apparent consistency of the temperature gradient between SAW and STW, similar to the average modern latitudinal gradient across the STF, tends to indicate moderate Southern Ocean frontal migrations. The absence of significant ecological shifts to cool water taxa at Site 1171 (Exon et al. 2001; Stant et al. 2004), low average Antarctic ice volume (27 × 10⁶ km³; Figure 3) and moderate G–IG variations characterising the pre-MPT interval (41 kyr state; Clark et al. 2006) all lend support to the conclusion that there was relative warmth and only moderate migration of ocean fronts during this time. However, the short-lived appearance of subtropical calcareous nannofossil Umbilicosphaera jafari (Stant et al. 2004) at Site 1171 around 1.13 Ma seems to indicate a brief incursion of warm STW all the way to the latitude of Site 1171, which is not clearly featured in our SST reconstructions.

Onset of the MPT: 1.25 Ma to 1.1 Ma

We identify the first significant transition between 1.25 and 1.1 Ma (Figure 3) by a progressive SST cooling of c. 5 °C, which is also reflected in the calcareous nannofossil assemblage (Stant et al. 2004) with the increasing abundance of cold water SAZ species Coccolithus pelagicus and Calcidiscus leptoporus (e.g. McIntyre & Bé 1967; Hiramatsu & De Deckker 1997; Eynaud et al. 1999; Baumann et al. 2000).

A similar SST decrease was observed in the Atlantic sector (ODP Site 1090) of the Southern Ocean (Martínez-Garcia et al. 2010) between 1.25 and 1.15 Ma at the onset of the MPT (McClymont et al. 2013). At Site 1090 (42.9°S–8.9°W), this trend appears to be driven by a 6 °C cooling during MIS34, which we do not fully capture at Site 1171 because of the lower sampling resolution in this interval.

The SST₁₁₇₁ cooling, the temporary disappearance of the low latitude taxa Gephyrocapsa caribbeanica, and the shift to subantarctic calcareous nannofossil populations (Stant et al. 2004) attest to major changes in the subantarctic surface waters. Rodriguez-Sanz et al. (2012) suggest that during this interval, the Atlantic Ocean surface waters were under the influence of a northward extension of the polar frontal system. At Site 1123 (41°47.2′S–171°29.9′W), situated on the northern edge of the STF east of New Zealand, Crundwell et al. (2008) reported a similar long-term cooling of the mean annual SST from 1.185 Ma, which indicates a northward expansion of SAW (Figure 3). We consider that during the onset of the MPT, at least the STF migrated northward in the southwest Pacific region and possibly the entire Southern Ocean frontal system in response to Antarctic ice growth (Figure 3).

The MPT part 1: between 1.1 Ma and 0.88 Ma

This interval corresponds to a long-term SST cooling of c. 3.5 °C at the STZ sites (Site 1123, MD06–2989 and Site 1125 alkenone-derived SSTs), which has been attributed to an increased inflow of SAW (Crundwell et al. 2008). The northward displacement of the STF in the southwest Pacific echoes the general northward migration of SAF in the Atlantic interpreted by Marino et al. (2009) from calcareous nannofossil assemblages and is in agreement with a dynamic Pacific Deep Western Boundary Current (Hall et al. 2001; Venuti et al. 2007).

The subantarctic SSTs (Site 1171) during that interval display low variability and an overall stable long-term trend interrupted by two short warm events at 1.07 and 0.95 Ma, which may be MIS31 and MIS25 (Figure 3). Stant et al. (2004) noted the dominance of Helicosphaera carteri and the temporary reappearance of the low latitude eutrophic specie G. caribbeanica at around 1.07 Ma at Site 1171, indicating surface-ocean warming. The ecology of H. carteri is controversial and their increased abundance may indicate a warming of waters (McIntyre 1967; Winter et al. 1979) or iceberg melting (Sierro et al. 2005). Flores and Sierro (2007) reported an increase in H. carteri at Site 1094 south of the Polar Front in the Atlantic sector of the Southern Ocean during the MIS31, arguing for the southward migration of the front and surface ocean warming. Similarly, the SST₁₁₇₁ increase at 0.95 Ma is correlated with a temporary increase in the abundance of G. caribbeanica (Stant et al. 2004). At Site 1123 (Figure 3), MIS31 and MIS25 are also clearly identified with an SST increase averaging 5 °C, indicating a temporary southward migration of the STF. At Site 1125, our SST reconstruction is patchy and we do not observe a clear temperature increase during MIS31 or MIS25. However, Schaefer et al. (2005) also did not identify unusual SST warming during MIS25, which may indicate that SSTs were generally more stable in this location. The apparent SST stability at Site 1125 compared with Site 1123 may be due to complex interactions of the STF with the topography in this region (Smith et al. 2013). At the location of Site 1125, sitting half way along the Chatham Rise, the subsurface topography limits the STF shifts, whereas Site 1123 is located east of the Chatham Rise in a region where STF can migrate latitudinally (Crundwell et al. 2008; Hayward et al. 2012).

Although Hayward et al. (2012) consider that during all the MPT interglacials (including MIS31 and MIS25), the STF was diffuse and located close to its modern position in the East Tasman Sea, we argue that, at least during MIS31 and MIS25, the STF migrated southward, bringing warm waters close to the latitude of ODP Site 1171.

The MPT part 2: from 0.88 Ma

At c. 0.88 Ma (MIS22), the subtropics and the subantarctic surface waters warm simultaneously. This abrupt event marks the beginning of a clear separation between the subantarctic and subtropical zones associated with the major perturbation of the Pacific Deep Western Boundary Current (Hall et al. 2001; Venuti et al. 2007) and global thermohaline circulation.

From 0.87 Ma, at Site 1171 we observe a c. 4 °C cooling to near modern SSTs with persistent average SSTs of 11 °C until 0.74 Ma, whereas in the subtropical zone, the SSTs vary between long intervals of temperate interglacials and short G–IG with increasing subtropical inflow north of the STF (Crundwell et al. 2008; Figure 3). At Site 1171, calcareous nannofossils also reflect the cooling trend with episodic decreased abundance of C. pelagicus, C. leptoporus and reticulofenestrids (Stant et al. 2004). This is in agreement with the deposition of laminated diatom mats in the Atlantic SAZ indicating a northward displacement of the Polar Front between 0.89 and 0.74 Ma (Kemp et al. 2010).

Conclusions

Our low-resolution SST record at Site 1171 demonstrates that through the Early Pleistocene, subantarctic surface waters cooled by c. 4°C and reached near modern conditions after 0.88 Ma. This long-term cooling has also been recognised in the Atlantic SAZ (Martínez-Garcia et al. 2010) and indicates a northward migration of the Southern Ocean fronts. The cooling occurred in two major steps between 1.25–1.1 Ma and 0.95–0.88 Ma, and correlates with significant increases in Antarctic ice volume (Naish et al. 2009; Pollard & DeConto 2009) and changes in global circulation (Hall et al. 2001; Venuti et al. 2007; Marino et al. 2009).

Based on a comparison between the long-term SST evolution of the subtropical and SAZ, we argue that the position of the STF in the southwest Pacific was unstable and that, at least during two episodes, dated at 1.07 and 0.95 Ma (identified as MIS31 and MIS25), the STF migrated south significantly. Further effort should be dedicated to investigating this later frontal displacement, which has not been identified in the southwest Pacific. Investigations should include alkenone-derived SST reconstructions in the subtropical zone to determine the latitudinal temperature gradient across the STFZ, avoiding any proxy-related bias (modern analogue technique versus ).

Supplementary data

Table S1. Alkenone-derived SST estimates from ODP Sites 1171 and 1125.

Acknowledgements

We thank Nick Mortimer and the reviewers for their constructive and valuable reviews. Special thanks go to Helen Bostock who greatly improved the manuscript with her comments and suggestions. We thank the curators from the USA and Japan IODP core repositories for their efficiency in providing the samples that were needed for this study.

Associate Editor: Associate Professor Brent Alloway.

Disclosure statement

No potential conflict of interest was reported by the authors.