Temporal evolution of mechanisms controlling ocean carbon uptake during the last glacial cycle

Earth and Planetary Science Letters 472 (2017) 206–215

Contents lists available at ScienceDirect

Earth and Planetary Science Letters www.elsevier.com/locate/epsl

Temporal evolution of mechanisms controlling ocean carbon uptake during the last glacial cycle Karen E. Kohfeld a,∗ , Zanna Chase b a b

School of Resource and Environmental Management, Simon Fraser University, 888 University Drive, Burnaby, BC, V5A 1S6, Canada Institute for Marine and Antarctic Studies, University of Tasmania, Private Bag 129, Hobart, Tasmania, 7001, Australia

a r t i c l e

i n f o

Article history: Received 11 January 2017 Received in revised form 3 May 2017 Accepted 12 May 2017 Available online xxxx Editor: M. Frank Keywords: carbon cycle ice age ocean circulation paleoceanography glacial–interglacial

a b s t r a c t Many mechanisms have been proposed to explain the ∼85–90 ppm decrease in atmospheric carbon dioxide (CO2 ) during the last glacial cycle, between 127,000 and 18,000 yrs ago. When taken together, these mechanisms can, in some models, account for the full glacial–interglacial CO2 drawdown. Most proxy-based evaluations focus on the peak of the Last Glacial Maximum, 24,000–18,000 yrs ago, and little has been done to determine the sequential timing of processes affecting CO2 during the last glacial cycle. Here we use a new compilation of sea-surface temperature records together with time-sequenced records of carbon and Nd isotopes, and other proxies to determine when the most commonly proposed mechanisms could have been important for CO2 drawdown. We find that the initial major drawdown of 35 ppm 115,000 yrs ago was most likely a result of Antarctic sea ice expansion. Importantly, changes in deep ocean circulation and mixing did not play a major role until at least 30,000 yrs after the first CO2 drawdown. The second phase of CO2 drawdown occurred ∼70,000 yrs ago and was also coincident with the first significant influences of enhanced ocean productivity due to dust. Finally, minimum concentrations of atmospheric CO2 during the Last Glacial Maximum resulted from the combination of physical and biological factors, including the barrier effect of expanded Southern Ocean sea ice, slower ventilation of the deep sea, and ocean biological feedbacks. © 2017 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction Many hypotheses have been put forth to explain the 80– 100 ppm changes in atmospheric carbon dioxide (CO2 ) concentrations that occurred during glacial–interglacial cycles over the past 800,000 yrs, including physical and biological changes affecting the partitioning of carbon between the ocean and the atmosphere (Sigman et al., 2010) and changes in volcanic emissions of CO2 (Huybers and Langmuir, 2009). Many hypotheses put forth to explain glacial–interglacial CO2 change focus on the total amplitude of CO2 change that can be attributed to individual mechanisms, and then consider the total amount of CO2 that could be sequestered at equilibrium by summing up various mechanisms until they account for the total change observed (Kohfeld and Ridgwell, 2009). However, the mechanisms of CO2 removal need not have occurred simultaneously (Gildor and Tziperman, 2001; Peacock et al., 2006; Sigman et al., 2010). While CO2 release to the atmosphere during the last

*

Corresponding author. E-mail addresses: [email protected] (K.E. Kohfeld), [email protected] (Z. Chase).

deglaciation occurred rather abruptly over the ∼10,000 yrs following the Last Glacial Maximum (hereafter LGM, ∼18–24 ka), CO2 removal from the atmosphere occurred through multiple steps between the last Interglacial period ∼127,000 yrs ago and the LGM. We focus on three intervals of CO2 drawdown: the first occurred 115–100 ka, the second from 72–65 ka, and the final step, after a brief return to slightly higher CO2 levels, occurred 40–18 ka (Fig. 1). The steps are roughly equivalent to the transitions between Marine Isotope Stages (MIS) 5e to 5d, MIS 5a to MIS 4, and the end of MIS 3 to MIS 2, respectively. How these steps of atmospheric CO2 reductions are linked with specific physical and biological mechanisms remains an open question. Both process-based modeling (Kohfeld and Ridgwell, 2009; Sigman et al., 2010; Hain et al., 2014) and paleo-environmental data reconstructions (Kohfeld et al., 2005) suggest that marine biology feedbacks are important but cannot account for the full 80–100 ppm magnitude of change, and cannot explain CO2 drawdown during the early part of the glacial cycle. These studies implicate physical processes as important drivers of ocean carbon uptake. Some proposed physical mechanisms include a) reduced air-sea gas exchange in response to sea ice expansion (Stephens and Keeling, 2000; Ferrari et al., 2014), b) a decrease

http://dx.doi.org/10.1016/j.epsl.2017.05.015 0012-821X/© 2017 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

K.E. Kohfeld, Z. Chase / Earth and Planetary Science Letters 472 (2017) 206–215

207

most recent glacial cycle. We use SST because this variable provides a critical link between the atmosphere and ocean, influencing processes such as buoyancy forcing (Watson et al., 2015) and sea ice formation (Gordon, 1981). Furthermore, SST is the focus of many paleoceanographic reconstructions, and therefore has near-global coverage, high temporal resolution, and broad agreement between multiple proxy types (see Supplementary Information). We use this compilation, together with observational constraints from sea ice and ocean circulation proxies, to evaluate which mechanisms were acting during glaciation, and to develop a plausible sequence of events that enhanced CO2 uptake by the world’s oceans during the last glacial cycle. Changes in ocean circulation are constrained using a global compilation of the carbon isotopic composition of benthic foraminifera (Oliver et al., 2010) and supported by more limited reconstructions using the Nd isotope proxy (Piotrowski et al., 2005; Böhm et al., 2014; Jonkers et al., 2015; Wilson et al., 2015). To track changes in Southern Ocean sea ice, we use the ice core proxy of sea salt sodium (ssNa) fluxes (Wolff et al., 2010) and diatom-based proxies from marine sediment cores (Gersonde and Zielinski, 2000; Crosta et al., 2004). 2. Materials and methods

Fig. 1. (A) June insolation at 60◦ N (red line) and December insolation at 60◦ S (blue dashed line) (Berger and Loutre, 1991), (B) relative sea level (green line) reconstructed from benthic foraminiferal oxygen isotope records (Waelbroeck et al., 2002), (C) atmospheric carbon dioxide concentration (blue) (Jouzel et al., 2007) and (D) ice core temperatures reconstructed from Antarctica (EPICA Dome C, grey) (Jouzel et al., 2007). Marine Isotope Stages (MIS) 5e, 5d, 5a, 4, 3, 2, and 1 are indicated at the top and bottom of the figure; grey shading indicates the 10,000-yr time periods over which characteristics are averaged in Fig. 4.

in the rate of supply of nutrients and CO2 to polar surface waters, initially called ‘polar stratification’ (François et al., 1997; Sigman et al., 2004), and c) slower ventilation of the deep ocean, leading to increased isolation of deep waters from the atmosphere (Toggweiler, 1999; Watson and Naveiro Garabato, 2006; Adkins, 2013; De Boer and Hogg, 2014). Here we use a new compilation of sea-surface temperatures (SSTs) from 136 deep-sea core records for the past 130,000 yrs (Fig. 2; Supplementary Tables A.1–A.2) to constrain the physical mechanisms responsible for oceanic uptake of CO2 during the

We compiled SST reconstructions and high latitude North Atlantic faunal assemblages that extend from 130,000 yrs ago to today (Fig. 2). The compilation includes SST data reconstructed using a range of techniques, including alkenones (52 sites), Mg/Ca ratios (16 sites), and faunal assemblage reconstructions (planktonic foraminifera, diatoms, and radiolaria, 78 sites; Fig. A.1). These estimates come from a total of 136 deep-sea cores distributed between 72◦ N and 57◦ S (Tables A.1 and A.2). We examine conditions for several time slices of comparable length during the last glacial cycle: Marine Isotope Stages 5e (118–127 ka), 5d (105–114 ka), 5a (75–85 ka), 4 (59–68 ka), 3 (45–55 ka), 2 (18–28 ka), and 1 (0–9 ka) (Table A.3). The cores used to reconstruct changes in SST are found between 56◦ S and 57◦ N and thus do not provide adequate coverage of surface ocean changes in the highest polar latitudes. We supplement SST reconstructions using data on relative percentages of N. pachyderma (s.) from North Atlantic sites between 60 and 76◦ N (10 sites; Tables A.1–A.2 and Fig. A.2). Many previous studies have shown that the relative percentage of N. pachyderma (s.) in the planktonic foraminiferal assemblage provides a reasonable, first order estimate of mean annual surface temperatures between 3 and 12 ◦ C in the North Atlantic Ocean (R 2 = 0.83, Kohfeld et al., 1996). We have used the latest published age model associated with each deep-sea core with minor modification. The published age

Fig. 2. Core locations for sea surface temperature estimates made from alkenone (green triangles), Mg/Ca ratios from planktonic foraminiferal calcite (orange triangles), and faunal assemblages (circles) of planktonic foraminifera (blue), diatoms (light blue) and radiolaria (dark blue). Red diamonds show locations of % N. pachyderma (s.) sites, and black square denotes location of the EPICA Dome C ice core.

208

K.E. Kohfeld, Z. Chase / Earth and Planetary Science Letters 472 (2017) 206–215

Fig. 3. Stacked records of SST changes for the last 127,000 yrs using 136 deep-sea core records, presenting areal-weighted global averages (black lines) and zonally averaged estimates for seven latitudinal bands. (A) absolute SSTs (shading represents one standard deviation around the mean values); (B) changes relative to average SSTs estimated for 127 ka. Vertical lines indicate three stages of CO2 decrease discussed in this study (100–115 ka; 72–65 ka, 40–18 ka).

models incorporate a mix of AMS radiocarbon dates (calibrated to calendar years), oxygen isotope stratigraphies, and stratigraphic correlation. In all but five cases, the period between 59 and 130 ka is based on correlation with oxygen isotope chronologies (Imbrie et al., 1984; Martinson et al., 1987; Shackleton, 1995; Lisiecki and Raymo, 2005). To account for slight differences in the timing of marine isotopic stage boundaries between published time scales, we have converted all oxygen isotope age models so that they are consistent with the time scale of Lisiecki and Raymo (2005). This age conversion had minimal impact on patterns of SST change (Fig. A.3, Table A.4a). Some studies have suggested that offsets can occur between benthic oxygen isotope chronologies collected from different water masses (Govin et al., 2009). We attempt to minimize the effects of a possible, thousand-year lag on our results by examining average SST changes over the ten-thousandyear “time slices” established above. We have calculated areal-weighted global average SST by averaging SST estimates by latitude band and weighting the SST contribution to the global average by the ocean area covered by each latitude band (Table A.4b). In addition to calculating averages for each time-slice using original, non-interpolated SST estimates, we have also produced SST stacks by latitude, and globally. These were calculated by linearly interpolating all records having a temporal resolution of 3 ka or better to a common, 500-yr resolution time-scale before averaging by latitude bands as above (Table S5). Results were not sensitive to small changes in interpolation or resolution cut-off. As a first approximation of SST changes during different stages of the last glacial cycle, we combine all SST data compiled for this study. However, we recognize that vital effects, postdepositional modifications, and processing technique can influence inter-comparisons between these different methods of temperature reconstruction (e.g., Kucera et al., 2005). Thus we also include SST estimates made for each method individually (Table A.4b; Fig. A.4), with the caveat that the number of sites with SST estimates using any one method is small. In 12 cores, SST reconstructions were

made using more than one method (Table A.6). These paired reconstructions show an average temperature difference between methods of 0.85 ◦ C, with no evidence for a bias by latitude or marine isotope stage, although the comparison does suggest a tendency for SSTs based on planktonic foraminifera to be warmer than those based on other methods. 3. Results 3.1. 115–100 ka The first major drop in atmospheric CO2 of ∼35 ppmv below interglacial levels occurred 115–100 ka. Global average SSTs cooled by 1.3 ◦ C during this time, but not all regions cooled equally (Figs. 3 and A.5; Table A.4a). The tropics between 30◦ S and 30◦ N cooled by 0.8 ◦ C while the high latitudes between 50 and 60◦ cooled by 1.6 ◦ C in the south and 3 ◦ C in the north (Figs. 3b and 4). In the high-latitude Atlantic Ocean, early cooling is also observed between 61 and 76◦ N from increased abundances of the polar planktonic foraminiferal species N. pachyderma (s.) (Fig. A.2). Eight of 10 cores reached assemblages of nearly 100% N. pachyderma (s.) by 115–110 ka and remained saturated at 100% until after the LGM. The remaining two cores at 61◦ S show initial shifts towards polar assemblages by 110–105 ka, with polar species dominating the assemblages by ∼70 ka (during MIS 4). In the southern high-latitudes, surface temperatures in Antarctica dropped by 6.5 ◦ C (Fig. 4), and the ssNa sea ice proxy from the EPICA Dome C ice core reached 71% of its full glacial–interglacial amplitude, although we note that this change is likely an overestimate of early sea ice expansion due to this proxy’s underestimation of sea ice extents during the LGM (Wolff et al., 2010). Sediment core reconstructions from the Atlantic sector of the Southern Ocean (Bianchi and Gersonde, 2002) also show an increase in winter sea ice during MIS 5d (Fig. 5). Nitrogen isotope proxies show the first evidence of increased surface water nutrient utilization in the central Pacific Antarctic Zone (Fig. 5; Studer et al., 2015).

K.E. Kohfeld, Z. Chase / Earth and Planetary Science Letters 472 (2017) 206–215

209

Fig. 4. Average changes in regional conditions for Marine Isotope Stages 5d–5a, 4, 3, and 2, compared with the last Interglacial Stage 5e, for 50–60◦ N (red), 50–50◦ N (orange), 40–50◦ S (light blue), and 50–60◦ S (aqua blue), EPICA Dome C temperature (dark navy blue) (EPICA Community Members, 2004), and EPICA sea salt Sodium flux (ssNa, purple) (Wolff et al., 2010) which is taken as a proxy for sea ice extent. Changes in ssNa fluxes are expressed as the percent of the total change estimated for this proxy between MIS 5e (118–127 ka) and 2 (18–28 ka). Black line represents EPICA Dome C carbon dioxide concentrations (EPICA Community Members, 2004). Shading indicates time periods averaged for MIS 5e (118–127 ka), 5d (105–114 ka), 5a (75–85 ka), 4 (59–68 ka), 3 (45–55 ka), and 2 (18–28 ka) (with intervals for MIS 5b–c indicated in Table A.3).

Not all proxies changed between 115 and 100 ka. Diatom-based sea ice proxies from the Antarctic Polar Front Zone of the southwest Pacific and Atlantic sectors of the Southern Ocean show minimal increases (Fig. 5). Similarly, deep ocean Nd records (Fig. 5) and the δ 13 C of benthic foraminifera (Fig. 6a–c) show little evidence for widespread changes in deep ocean circulation during this first stage of CO2 decline. 3.2. 72–65 ka The second major 40 ppmv drop in CO2 of occurred between

∼72–65 ka (MIS 4), coincident with cooling of ∼7–8 ◦ C in northern high latitudes (Figs. 3b and 4). The minimum in CO2 at the end of this period, 190 ppmv, is only ∼5 ppmv higher than the lowest CO2 concentrations reached during the LGM (Fig. 1). Similarly, SST reductions nearly reached their full glacial–interglacial amplitude, with global SSTs that were 3.1 ◦ C below interglacial levels (Fig. 3b; Table A.4a). In the southern high latitudes, the diatom sea ice proxies also show evidence for an increase in the duration of sea ice cover, and the presence of sea ice at sites previously ice-free (Fig. 5). During this second step of CO2 decline, we see clear evidence for a change in deep ocean circulation in both the δ 13 C (Fig. 6d) and Nd isotope records (Fig. 5; Piotrowski et al., 2005; Böhm et al., 2014; Jonkers et al., 2015; Wilson et al., 2015). Low δ 13 C values were extensive in the South Atlantic below 2000 m and were found for the first time north of the equator between 2500 and 4000 m. Individual ε Nd records from the South Atlantic (Fig. 5; Piotrowski et al., 2005) and Indian Ocean (not shown; Wilson et al., 2015) clearly show the transition in deep ocean circulation at the start of MIS 4. Other records (not shown) suggest later changes, with steepest decreases during Stage 3 (45–55 ka) in the South Atlantic (Jonkers et al., 2015), and initiation of decreases at Bermuda Rise around 50 ka (Böhm et al., 2014). Finally, coincident with this second drop of CO2 during MIS 4, iron, dust, and lithogenic deposition rates show their first substantial increases in Antarctic ice cores (Fig. 5) and Subantarctic marine sediments, along with evidence of increased opal fluxes and nutrient utilization (Lamy et al., 2014; Martínez-García et al., 2014). 3.3. 40–18 ka A final 5–10 ppmv decrease in atmospheric CO2 occurred between 40–18 ka and is associated with maximum changes in all

environmental variables considered. Many variables continued to change beyond their MIS 4 levels to reach maximum values during the LGM. Specifically, surface Antarctic temperatures continued to cool, dropping to more than 10 ◦ C below interglacial levels. SSTs between 50–60◦ N also continued to drop during the last step of CO2 decline (Fig. 3b and 5). Furthermore, ocean circulation proxies (ε Nd and benthic δ 13 C) and diatom sea ice duration proxies also show evidence for some continued change across this time, although of a smaller magnitude compared to the changes observed during the first two steps of CO2 decline. Finally, starting at around 39 ka, dust delivery to Antarctica (Fig. 5) and iron and terrigenous deposition in Subantarctic cores increased for a second time to reach maximum values, together with maximum increases in biological productivity (Kohfeld et al., 2005; Martínez-García et al., 2014) In contrast, some proxies show very little change between the second and final CO2 drops. In particular, nitrate utilization proxies and SSTs in the Southern Ocean showed very little change during this period, having reached near-full glacial levels already during the second CO2 drop during MIS 4. 4. Discussion 4.1. Initial decrease in atmospheric CO2 during glacial inception (115–100 ka) Ocean cooling, increased global salinity, and decreases in terrestrial carbon would all affect atmospheric CO2 during inception, but their combined effects are small. Based on solubility alone, a whole-ocean SST reduction of 1.3 ◦ C could account for approximately 15 ppmv reduction in atmospheric CO2 . Using the sea level approximation of Waelbroeck et al. (2002), we estimate that the effect of ice volume on global salinity had only reached 30% of its glacial–interglacial impact, which would have added about 2 ppmv of CO2 to the atmosphere. Combining these two effects results in 13 ppmv reduction, or roughly two fifths of the observed 35 ppmv reduction in atmospheric CO2 . Release of terrestrial carbon would offset the 13 ppmv uptake by the oceans. Ciais et al. (2012) estimate the terrestrial biosphere was 330 ± 400 PgC smaller during the LGM compared to pre-industrial climate. Based on the sea-level reconstruction of Waelbroeck et al. (2002) and assuming 80% of the released CO2 would react with the inorganic carbon chemistry of seawater over

210

K.E. Kohfeld, Z. Chase / Earth and Planetary Science Letters 472 (2017) 206–215

Fig. 5. Summary of changes driving ocean CO2 uptake during glacial inception. (A) Stacked records of SST at 40–50◦ N (orange) and 50–60◦ N (red), showing relative change since 127 ka; (B) composite Nd isotope record from core SK129-CR2 (3.8 km) in the Indian Ocean, showing changes in deep ocean circulation (Wilson et al., 2015); (C) proxies of sea ice duration (mo/yr, light brown) and winter sea ice extent (% F. curta, purple) from the Antarctic Polar Front Zones (PFZ) in the SW Pacific (SO136-11; Crosta et al., 2004) and Atlantic (PS1768-8; Gersonde and Zielinski, 2000) sectors of Southern Ocean; (D) proxy of surface nitrate utilization using δ 15 N from diatom-bound organic matter, taken from south of the PFZ in the Pacific sector of the Southern Ocean (PS75/072-4) (Studer et al., 2015); (E) Stacked record of SST at 50–60◦ S showing relative change since 127 ka (light blue), and a proxy for sea ice extent using ssNa fluxes from the EPICA Dome C ice core (dark blue) (Wolff et al., 2010); (F) EPICA Dome C carbon dioxide concentrations (EPICA Community Members, 2004). Vertical shaded bars of (i) blue, (ii) pink, and (iii) grey highlight important periods of CO2 drawdown which we suggest were driven primarily by (i) Southern Ocean sea ice expansion and near-surface ocean stratification in response to polar cooling (ii) deep-ocean stratification and increased efficiency of the biological soft tissue pump associated with dust deposition, and (iii) further increases in Fe-enhanced biological soft tissue pump and deep-ocean stratification with associated carbon trapping.

several hundred years (Archer et al., 1997), and ignoring dissolution of sea-floor calcite, we estimate that the decrease in the terrestrial biosphere would have caused an increase in atmospheric CO2 of at most 10 ppmv during this time. Thus, the combined reduction in atmospheric CO2 attributable to changes in SST, salinity, and the terrestrial biosphere is on the order of ∼4 ppmv, leaving us to seek other mechanisms to explain the first drawdown of atmospheric CO2 during glacial inception. Here we focus on physical oceanographic changes, recognizing that (a) changes in biological productivity occur later (Kohfeld et al., 2005), and (b) the impact of volcanic processes requires further quantification, but was likely on the order of 10 ppmv (Huybers and Langmuir, 2009). Physical oceanographic changes can be char-

acterized as either a) ‘barrier’ mechanisms of Southern Ocean sea ice and near-surface stratification acting to prevent CO2 release to the atmosphere from upwelled Circumpolar Deep Water (CDW) or b) changes that lead to a slower ventilation of the deep ocean, and a more sluggish circulation (Sigman et al., 2010). We argue based on existing proxy data that the initial drawdown of atmospheric CO2 was driven primarily by barrier mechanisms, mostly through expanded sea ice cover. We rule out deep ocean ventilation as a mechanism for early CO2 drawdown because proxies show no evidence for widespread changes in deep ocean circulation until the MIS 5/4 transition ∼30 ka after the initial CO2 drop. The main proxy tools for reconstructing the deep ocean circulation are benthic temperature, benthic 13 C, and Nd isotopes. Some evidence for early cooling of deep waters between 100 and 115 ka has been documented in the South Atlantic (Adkins, 2013), along with some evidence for cooling and expansion of AABW in the Indian Ocean (Govin et al., 2009; Adkins, 2013). However, the benthic foraminifera δ 13 C data compilation from the Atlantic Ocean (Oliver et al., 2010) suggests that a basin-wide change in water mass geometry did not occur until MIS 4 (Fig. 6d). The handful of Nd isotope studies also suggests that the response of deep ocean circulation was delayed until ∼MIS 4. Taken together, these records suggest that the slow down of deep-ocean ventilation occurred well after MIS 5d and cannot explain the first 35 ppmv decrease in atmospheric CO2 at 115–100 ka. We have been deliberately non-committal in our description of ‘slower ventilation’. While there is consensus that reduced deep ocean ventilation is a means of lowering CO2 , proxy data have been used to support multiple interpretations of the altered ocean circulation, all of which are consistent with increased storage of carbon in the deep sea. For example, the deep-ocean δ 13 C distribution depicted in Fig. 6e is typically interpreted as an expansion of AABW and a shoaling and weakening NADW (e.g., Lynch-Stieglitz et al., 2007). However, other studies argue that low deep-water δ 13 C values reflect enhanced trapping of nutrients (and carbon), and that glacial NADW remained vigorous and similarly mixed with southern source waters (Böhm et al., 2014; Gebbie, 2014; Howe et al., 2016). Despite multiple possible scenarios, there appears to be consensus that an altered circulation would leave some imprint on proxy data. The lack of a change in ε Nd or δ 13 C distributions during the initial CO2 decrease therefore argues against decreased ventilation as the primary driving mechanism during early glaciation. Having ruled out deep ocean changes, we argue that increased Antarctic sea ice production played a large role in the initial CO2 drawdown. The ice core ssNa proxy (Wolff et al., 2010) shows Antarctic sea ice responded early during glacial inception between 121 and 117 ka (Fig. 5), roughly consistent with limited diatombased records (Bianchi and Gersonde, 2002). The rate of air-sea gas exchange is reduced by at least an order of magnitude in the presence of sea ice, and even areas partially covered by ice experience reduced gas exchange (Rutgers van der Loeff et al., 2014). In addition to the barrier effect of sea ice, waters seasonally impacted by ice may experience reduced gas exchange due to surface water stratification induced by melt-water (Moore et al., 2000; Stephens and Keeling, 2000). Such a melt-water effect is consistent with the δ 15 N observations from the Pacific Antarctic zone (Studer et al., 2015), which show a clear increase in nitrate utilization by 115 ka during MIS 5d. This increased nitrate utilization occurred without concomitant increases in export production or dust deposition and is consistent with a reduced supply of nutrients to the surface ocean, as might occur due to summer stratification induced by meltwater. We note that Studer et al. (2015), following earlier work (François et al., 1997; Sigman et al., 2004), use the term ‘strati-

K.E. Kohfeld, Z. Chase / Earth and Planetary Science Letters 472 (2017) 206–215

211

Fig. 6. The δ 13 C from benthic foraminifera from the Atlantic Ocean (between 70◦ W and 15◦ E) during Stages (A) 5e, (B) 5d, (C) 5a, (D) 4, (E) 2, and (F) the Holocene. Black line represents the 0.3h (PDB) contour. Data from Oliver et al. (2010).

fication’ not to mean density stratification, but rather to describe a reduction in the input of subsurface water to the surface layer, i.e., a reduction in upwelling of CDW. Given the lack of proxy evidence for deep-water changes during MIS 5d described above, we argue the existing proxy data during the first CO2 drop are more consistent with a local density stratification, likely driven by sea ice melt during summer, but decoupled from deep-water responses. Any reductions in upwelling, driven by either wind or buoyancy forcing, would likely result in shifts in water mass geometry (De Boer and Hogg, 2014), which are not observed until the MIS 5/4 transition. This separation between surface and deep-water processes during MIS 5d becomes quite important because it implicates sea ice expansion and the inhibition of gas exchange as the main driver of the initial drop in CO2 ; the overall rate of upwelling likely did not change during this time, but locally, summer density stratification in the upper ocean may have limited nutrient supply, thereby increasing relative nitrate uptake but decreasing export production (which was limited by the supply of iron from below). The early response of Antarctic sea ice is tied to early polar cooling observed in both hemispheres, in response to insolation forcing. High latitude summer insolation forcing at 60◦ N and 60◦ S reached the lowest levels observed in the past 130 ka during early glaciation (Fig. 1). Consequently, in the South by 105 ka, extreme cooling is observed in Antarctic surface air and water temperatures, and ice core-derived sea ice changes were already extensive (Table A.4b, Fig. 4). In the North, strong early cooling is directly coincident with minimum values in insolation forcing 115,000 yrs ago. While northern hemisphere insolation forcing leads the south by several thousand years, we suggest that cooling in the formation region of NADW contributed to early cooling of Southern Ocean surface waters and expansion of Antarctic sea ice when colder NADW upwelled south of the Antarctic Polar Front (Crowley and Parkinson, 1988; Gildor and Tziperman, 2001). Following the early glacial minima between 105–115 ka, summer insolation forcing increased in both hemispheres until the next local minima around 80–70 ka (Fig. 1). While SSTs and sea

ice records reflect some of this variability, the cooling trajectory towards full glacial conditions continued. This cooling was probably sustained by a combination of internal climate feedbacks, including snow-albedo changes, land and sea ice growth, and CO2 reductions (Imbrie et al., 1992; Khodri et al., 2001, 2005; Brovkin et al., 2012). 4.2. Second decrease in atmospheric CO2 , 80–65 ka Between MIS 5a and 4, atmospheric CO2 dropped a second time by 40 ppmv, just as polar temperatures cooled to near-Ice Age levels, North Atlantic SSTs plunged to 8 ◦ C below interglacial levels, and benthic δ 13 C and ε Nd proxies show their largest deep ocean changes (Fig. 5 and 6d). We argue that this second decrease in atmospheric CO2 was driven largely by a more sluggish overturning circulation, which trapped respired CO2 in the deep ocean and increased whole ocean alkalinity via the carbonate compensation mechanism (Watson and Naveiro Garabato, 2006; Lund et al., 2011; Adkins, 2013). As described above, the proxy evidence points to such a change in deep ocean circulation around the MIS 5/4 transition, coincident with this second large decrease in atmospheric CO2 . What initiated a more sluggish overturning circulation at this time? A number of drivers have been proposed, including: (1) weakening and equatorward migration of the southern westerly winds (Toggweiler, 1999); (2) decreased buoyancy forcing in the Southern Ocean as a result of colder atmospheric temperatures (Watson and Naveiro Garabato, 2006; Watson et al., 2015); (3) production of denser, more saline AABW through reduced melting of land ice as colder North Atlantic Deep Water (NADW) upwelled on the Antarctic shelf (Adkins, 2013); (4) production of denser, more saline AABW through increased sea ice production and brine rejection (Bouttes et al., 2010; Brovkin et al., 2012); and (5) a reorganization of deep ocean circulation driven by an expansion of summer sea ice extent in the Southern Ocean (Ferrari et al., 2014). Few studies have examined the timing of these potential drivers of deep ocean circulation

212

K.E. Kohfeld, Z. Chase / Earth and Planetary Science Letters 472 (2017) 206–215

changes, but the SST reconstruction presented here, and associated proxy records, can help clarify the timing of potential drivers. The first mechanism, involving the position and strength of southern hemisphere westerly winds, is very difficult to constrain with proxy data, even for the relatively well-studied LGM (Kohfeld et al., 2013), so we are unable to address this driver. The second mechanism involves colder polar air temperatures reducing the buoyancy flux from the atmosphere to the ocean, which in turn reduced upwelling and deep-ocean ventilation (Watson and Naveiro Garabato, 2006). Our SST compilation shows that SSTs had essentially reached full glacial values by MIS 4, 3.1–3.7 ◦ C below interglacial levels between 40 and 60◦ S. These SST reductions – both in the Southern Ocean and globally – are expected to modify atmospheric circulation sufficiently to lower buoyancy forcing south of ∼45◦ S (Boos, 2012). Thus, observed SST reductions moved in the right direction for temperature-driven buoyancy forcing to reach near peak-glacial levels during MIS4. However, the magnitude of change needed to initiate a shift in deep-ocean circulation remains unclear. During the first CO2 drop, SST reductions of 1.6–2.0 ◦ C, and Antarctic cooling of 6.5 ◦ C appear to have had little or no effect on deep-ocean stratification. Cooling of North Atlantic surface waters was even greater than in the Southern Ocean between the first and second CO2 steps (7.1–8 ◦ C below interglacial levels) and may have been sufficient to trigger the third mechanism, proposed by Adkins (2013). According to this mechanism, when pre-cooled NADW was upwelled along the coast of Antarctica, it reduced the melting of land ice and its associated freshwater contribution to AABW. As a result, a colder, more saline AABW was formed, lowering the density of NADW relative to the more saline AABW, and thus initiating deep water stratification. Adkins (2013) originally proposed this mechanism as a cause for early CO2 drawdown between 115 and 100 ka; we suggest that the larger, later North Atlantic cooling associated with MIS 4 and the second CO2 drop is better aligned with evidence for deep-ocean circulation changes and that the initial North Atlantic cooling may not have been sufficient to trigger deep-water re-organization. The remaining two drivers of deep ocean circulation change are linked to Southern Ocean sea ice production, either through brine rejection (Watson and Naveiro Garabato, 2006; Bouttes et al., 2010) or by a northward shift in the summer sea ice edge, representing the boundary separating the upper and lower cells of the meridional overturning circulation (Ferrari et al., 2014). The ice corederived sea ice proxy (Wolff et al., 2010) responded early during glacial inception, well before the change in deep-ocean circulation, and shows a relatively smaller change during the second step in CO2 decline (Fig. 4). This proxy clearly reflects ice conditions close to Antarctica, in the region relevant to the brine rejection hypothesis. While a small response of this proxy during the second CO2 decline seems inconsistent with brine rejection as the mechanism initiating deep ocean stratification, we note that the lack of response may also reflect saturation of this proxy at large ice extents (Röthlisberger et al., 2010). The critical element linking Southern Ocean sea ice with deepocean circulation in the final mechanism proposed by Ferrari et al. (2014) is the extent of summer sea ice. The sea ice ssNa proxy appears biased to spring-time ice extent (Wolff et al., 2010) and does not necessarily equate with summer sea ice expansion. While few records of summer sea ice cover the full glacial cycle, existing records suggest increases in summer sea ice extent did not occur until the start of MIS 4 (Fig. 5) (Gersonde and Zielinski, 2000; Esper and Gersonde, 2014), roughly coincident with the deepocean changes. As suggested for the first CO2 step, the strong additional North Atlantic cooling between MIS 5a and 4 (Fig. 5) could have resulted in cooler CDW upwelling south of the Antarc-

tic Polar Front, and thus contributed to the expansion of summer sea ice during this time. Although the timing of summer sea ice expansion supports its role in triggering a change in overturning circulation, the magnitude of the expansion does not. Most diatom-based sea ice reconstructions suggest that, even at the LGM, summer sea ice increased in extent by at most 1◦ of latitude (Crosta et al., 1998; Gersonde et al., 2003), not the 5◦ assumed by Ferrari et al. (2014). More records of summer sea ice extent, covering the last glacial– interglacial cycle, as well as further constraints on the sea ice properties captured by the ice core ssNa proxy, are needed to test the link between sea ice extent and deep ocean circulation, and their contribution to drawdown of atmospheric CO2 during glacial periods. While this analysis has focused on physical changes, we note that increased dust deposition and associated increases in carbon uptake by the Subantarctic productivity likely also contributed to CO2 drawdown during the second CO2 decrease in MIS 4 (Lamy et al., 2014; Martínez-García et al., 2014). Estimates of either global ocean productivity or global deposition of dust to the ocean are not available for MIS 4, but the best ‘guesses’ of the maximum contribution of carbon uptake by Fe-enhanced productivity during the LGM are in the range of 15–20 ppmv (Röthlisberger et al., 2004; Kohfeld and Ridgwell, 2009; Ciais et al., 2013). Dust deposition to Antarctic ice cores during MIS 4 was approximately 66–75% of the full LGM deposition rates (Fig. 5). These two pieces of information together suggest that the effects of iron fertilization, while most likely less than half of the 40 ppmv CO2 change observed during this time, still could have been substantial. 4.3. Final decrease in atmospheric CO2 , 40–18 ka The final 5–10 ppmv decrease in atmospheric CO2 occurred between 40 to 18 ka. The first, strong candidate for enhanced CO2 uptake by the ocean during this time is an increase in the strength of the biological pump, as dust deposition increased to the highest levels observed during the glacial cycle (Fig. 5). Therefore, the maximum 15–20 ppmv effect of this CO2 uptake mechanism is likely to have occurred during this time. A second, complementary possibility is that enhanced deep-ocean stratification allowed for additional trapping of carbon in the deepest layers of the ocean. Continued reductions in ε Nd (Fig. 5) along with δ 13 C evidence of further shoaling of NADW (Fig. 6e) both suggest that deep-ocean stratification continued to intensify to a maximum extent during this final CO2 drop. Continued small decreases in high latitude SSTs between 50 and 60◦ N as well as increased sea ice durations (Fig. 5) would both support the continued enhancement of the effect of polar cooling and expanded sea ice cover on deep-ocean stratification. 4.4. Future work This analysis uses existing data to establish a logical ordering of oceanic mechanisms that contributed to the sequential drawdown of atmospheric CO2 and provides a benchmark for testing earth system models. But it also points to some key areas where data and information are lacking. First, some latitudinal bands and large areas of the Pacific and Indian ocean basins still lack SST data for the full glacial cycle, and SST reconstructions in some regions are weighted toward one type of SST estimation method. While this work provides a first-order reconstruction of global SST changes, we recognize our strong focus on changes in the Atlantic Ocean and the need for further study of the other basins. Second, our assessment of the role of sea ice expansion, both in driving the initial reductions in CO2 and acting as a potential trigger for subsequent

K.E. Kohfeld, Z. Chase / Earth and Planetary Science Letters 472 (2017) 206–215

213

deep-ocean circulation changes, relies on a limited number of reconstructions of nitrate utilization and winter and summer sea ice changes. Thus, this work also accentuates the need for more temporally extensive reconstructions of polar surface conditions. Finally, our analysis suggests that extensive modification of polar surface ocean temperatures and sea ice extent may be required to trigger a switch to the glacial overturning circulation, but the exact nature of links between surface modifications and deep-ocean changes remains unclear. Recognizing the absence of proxy evidence to diagnose one of the five mechanisms (winds), we have found that SST and sea ice data are consistent with three mechanisms that enhance buoyancy loss over critical areas of the Southern Ocean during the second CO2 drop. The largest change appears to be North Atlantic cooling, which would have reduced land-based freshwater inputs and increased summer sea ice extent in the Southern Ocean. Temperature reductions and increased brine rejection from sea ice expansion are likely contributors to buoyancy losses, but questions remain about why earlier (MIS 5d) changes were not sufficient. A useful focus for future mechanistic studies is to understand what magnitude and/or combination of these three changes are necessary to trigger a shift in circulation. 5. Conclusions Atmospheric carbon dioxide removal from the atmosphere occurred through three major steps between the last Interglacial period ∼127,000 yrs ago and the LGM. We combine a new, global SST compilation with proxies of sea ice extent, ocean circulation, and polar nutrient utilization to propose a sequential timing of the dominant oceanic mechanisms contributing to the drawdown of CO2 during a glacial cycle (Fig. 7). Substantial surface ocean cooling, particularly in high-latitudes, had already begun when the first ∼35 ppmv drop in atmospheric CO2 occurred between 115 and 110 ka, during MIS 5d. Global cooling, when coupled with estimated changes in ocean salinity and the terrestrial biosphere, likely resulted in only a ∼3 ppmv reduction in CO2 during this time. Sea ice expansion in the Southern Ocean played an early role in reducing atmospheric CO2 . A combination of direct, orbitally forced cooling and the cooling of preformed NADW upwelled in the Southern Ocean likely provided the early cooling needed to enhance sea ice formation. Evidence from ocean circulation proxies (δ 13 C and Nd isotopes) suggest that deepocean changes did not play a dominant role until the second major 40 ppmv drop in atmospheric CO2 , which occurred 30,000 yrs later during MIS 4. During MIS4, we suggest that several mechanisms could have contributed deep-ocean ventilation changes. First, the strong, 7–8 ◦ C cooling of Northern Hemisphere SSTs would have cooled NADW, which could have acted as a trigger for deep-ocean stratification, either by (a) decreasing the contribution of freshwater from melting Antarctic land ice to AABW, or (b) promoting summer sea ice expansion, both of which would have increased deep ocean stratification and carbon sequestration. Second, as Southern Ocean SSTs approached their minimum values, temperature-dependent changes in buoyancy loss over this region also acted to reduce upwelling and influence deep-ocean circulation. Linking deep-ocean circulation to sea ice changes is somewhat complicated because existing sea ice reconstructions suggest only a modest (∼1 degree) poleward expansion of the summer sea ice zone. Further work is needed to confirm the extent of summer sea ice, and to assess the extent of the expansion needed to trigger reorganization of the deep ocean. Likewise, the atmospheric cooling required to initiate a buoyancy-forced reduction in ventilation is unclear. In general, it is not obvious why deep ocean changes occurred so late in the glacial cycle, after substantial polar cooling and expansion of sea ice had already occurred. Further modeling studies should examine

Fig. 7. Timing of physical oceanographic changes during Glacial Inception. (A) Interglacial conditions during Stage 5e. (B) During MIS 5d, high-latitude sea-surface cooling occurs in response to insolation forcing, cools NADW, and results in Southern Ocean sea ice expansion, which acts early to draw down CO2 . (C) Changes in deep ocean circulation that affect atmospheric CO2 appear ∼30,000 yrs later during MIS 4, aided by temperature-dependent buoyancy losses and sea ice induced brine rejection in the Southern Ocean. CO2 decreases are also enhanced by Fe-induced productivity changes.

the nature of the links between sea ice, polar temperatures, and deep-ocean circulation, to improve our mechanistic understanding of how changes in temperature and permanent sea ice extent exert control on Southern Ocean buoyancy fluxes, deep-ocean circulation, and carbon sequestration. MIS 4 also saw an increase in dust deposition to the Subantarctic ocean, which drove an increase in the soft tissue biological pump, acting in synergy with further increasing ocean stratification to draw down atmospheric CO2 (Jaccard et al., 2016). The final, 5–10 ppmv drop in atmospheric CO2 occurred between 40 and 18 ka, when most environmental variables had reached their full glacial–interglacial amplitudes of change. Proxies of dust deposition, sea ice duration, and deep ocean stratification show small but continued changes between the second and third steps in CO2 reductions. This suggests that some combination of enhancements of the soft tissue biological pump in response to iron fertilization and carbon trapping in the deep ocean may have been responsible for the final CO2 decreases. In conclusion, this proposed sequencing of mechanisms for ocean carbon uptake during a glacial cycle

214

K.E. Kohfeld, Z. Chase / Earth and Planetary Science Letters 472 (2017) 206–215

can serve as a starting point for testing transient simulations of glacial–interglacial carbon cycling and highlights where new observations are needed to further constrain the timing and mechanistic linkages between surface ocean processes and ocean carbon sequestration. Acknowledgements We gratefully acknowledge the many researchers who published the original data used in this study as well as the people, institutions, and funding agencies that support the PANGAEA and NOAA NGDC Paleoclimatology data archives, which make access to these data possible. B. Shandro assisted with graphics. We thank A. De Boer, P. Buchanan, and M. Nikurashin for valuable discussions. We also thank A. Piotrowski and an anonymous reviewer for their comments which greatly enriched the manuscript. KEK is supported by the National Sciences and Engineering Research Council of Canada (NSERC) Discovery Grants (#RGPIN2013-342251) and Canada Research Chairs (#950-228166) programs. Zanna Chase is supported by an Australian Research Council Future Fellowship (FT120100759). Appendix A. Supplementary material Supplementary material related to this article can be found online at http://dx.doi.org/10.1016/j.epsl.2017.05.015. These data include the Google map of the most important areas described in this article. References Adkins, J.F., 2013. The role of deep ocean circulation in setting glacial climates. Paleoceanography 28, 539–561. Archer, D., Kheshgi, H., Maier-Reimer, E., 1997. Multiple timescales for neutralization of fossil fuel CO2 . Geophys. Res. Lett. 24, 405–408. Berger, A., Loutre, M.F., 1991. Insolation values for the climate of the last 10 million years. Quat. Sci. Rev. 10, 297–317. Bianchi, C., Gersonde, R., 2002. The Southern Ocean surface between Marine Isotope Stages 6 and 5d: shape and timing of climate changes. Palaeogeogr. Palaeoclimatol. Palaeoecol. 187, 151–177. Böhm, E., Lippold, J., Gutjahr, M., Frank, M., Blaser, P., Antz, B., Fohlmeister, J., Frank, N., Andersen, M.B., Deininger, M., 2014. Strong and deep Atlantic meridional overturning circulation during the last glacial cycle. Nature 517, 73–76. Boos, W.R., 2012. Thermodynamic scaling of the hydrological cycle of the Last Glacial Maximum. J. Climate 25, 992–1006. Bouttes, N., Paillard, D., Roche, D.M., 2010. Impact of brine-induced stratification on the glacial carbon cycle. Clim. Past 6, 575–589. Brovkin, V., Ganopolski, A., Archer, D., Munhoven, G., 2012. Glacial CO2 cycle as a succession of key physical and biogeochemical processes. Clim. Past 8, 251–264. Ciais, P., Sabine, C.L., Bala, G., Bopp, L., Brovkin, V., Canadell, J., Chhabra, A., DeFries, R., Galloway, J., Heimann, M., Jones, C., Le Quéré, C., Myneni, R.B., Piao, S.L., Thornton, P., 2013. Carbon and other biogeochemical cycles. In: Stocker, T.F., Qin, D., Plattner, G.-K., Tignor, M., Allen, S.K., Boschung, J., Nauels, A., Xia, Y., Bex, V., Midgley, P.M. (Eds.), Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK, pp. 465–483. Ciais, P., Tagliabue, A., Cuntz, M., Bopp, L., Scholze, M., Hoffmann, G., Lourantou, A., Harrison, S.P., Prentice, I.C., Kelley, D.I., Koven, C., Piao, S.L., 2012. Large inert carbon pool in the terrestrial biosphere during the Last Glacial Maximum. Nat. Geosci. 5, 74–79. Crosta, X., Pichon, J.-J., Burckle, L.H., 1998. Application of modern analog technique to marine Antarctic diatoms: reconstruction of maximum sea-ice extent at the Last Glacial Maximum. Paleoceanography 13, 284–297. Crosta, X., Sturm, A., Armand, L., Pichon, J.-J., 2004. Late Quaternary sea ice history in the Indian sector of the Southern Ocean as recorded by diatom assemblages. Mar. Micropaleontol. 50, 209–223. Crowley, T.J., Parkinson, C.L., 1988. Late Pleistocene variations in Antarctic sea ice, II, Effecto of interhemispheric deep ocean heat exchange. Clim. Dyn. 3, 93–105. De Boer, A.M., Hogg, A.M., 2014. Control of the glacial carbon budget by topographically induced mixing. Geophys. Res. Lett. 41. http://dx.doi.org/10.1002/ 2014GL059963. EPICA Community Members, 2004. Eight glacial cycles from an Antarctic ice core. Nature 429, 623–628.

Esper, O., Gersonde, R., 2014. New tools for the reconstruction of Pleistocene Antarctic sea ice. Palaeogeogr. Palaeoclimatol. Palaeoecol. 399, 260–283. Ferrari, R., Jansen, M.F., Adkins, J.F., Burke, A., Stewart, A.L., Thompson, A.F., 2014. Antarctic sea ice control on ocean circulation in present and glacial times. Proc. Natl. Acad. Sci. 111, 8753–8758. François, R., Altabet, M.A., Yu, E.-F., Sigman, D.M., Bacon, M.P., Frank, M., Bohrmann, G., Bareille, G., Labeyrie, L.D., 1997. Contribution of Southern Ocean surfacewater stratification to low atmospheric CO2 concentrations during the last glacial period. Nature 389, 929–935. Gebbie, G., 2014. How much did glacial North Atlantic water shoal? Paleoceanography 29, 190–209. Gersonde, R., Abelmann, A., Brathauer, U., Becquey, S., Bianchi, C., Cortese, G., Grobe, H., Kuhn, G., Niebler, H.S., Segl, M., Sieger, R., Zielinski, U., Futterer, D.K., 2003. Last glacial sea surface temperatures and sea-ice extent in the Southern Ocean (Atlantic–Indian sector): a multiproxy approach. Paleoceanography 18. Gersonde, R., Zielinski, U., 2000. The reconstruction of late Quaternary Antarctic sea-ice distribution – the use of diatoms as a proxy for sea-ice. Palaeogeogr. Palaeoclimatol. Palaeoecol. 162, 263–286. Gildor, H., Tziperman, E., 2001. Physical mechanisms behind biogeochemical glacial– interglacial CO2 variations. Geophys. Res. Lett. 28, 2421–2424. Gordon, A.L., 1981. Seasonality of Southern Ocean sea ice. J. Geophys. Res. 86, 4193–4197. Govin, A., Michel, E., Labeyrie, L.D., Waelbroeck, C., Dewilde, F., Jansen, E., 2009. Evidence for northward expansion of Antarctic bottom water mass in the Southern Ocean during the last glacial inception. Paleoceanography 24. http://dx.doi.org/ 10.1029/2008PA001603. Hain, M., Sigman, D., Haug, G., 2014. 8.18 The Biological Pump in the Past, The Oceans and Marine Geochemistry. Elsevier Ltd, pp. 485–517. Howe, J.N.W., Piotrowski, A.M., Noble, T.L., Mulitza, S., Chiessi, C.M., Bayon, G., 2016. North Atlantic deep water production during the Last Glacial Maximum. Nat. Commun. 7. Huybers, P., Langmuir, C., 2009. Feedback between deglaciation, volcanism, and atmospheric CO2 . Earth Planet. Sci. Lett. 286, 479–491. Imbrie, J., Hays, J.D., Martinson, D.G., McIntyre, A., Mix, A.C., Morley, J.J., Pisias, N.G., Prell, W.L., Shackleton, N.J., 1984. The orbital theory of Pleistocene climate: support from a revised chronology of the marine δ 18 O record. In: Berger, A., Imbrie, J., Hays, J., Kukla, G., Saltzman, B. (Eds.), Milankovitch and Climate Part 1. D. Reidel Publishing Company, Dordrecht, pp. 269–305. Imbrie, J.D., Boyle, E.A., Clemens, S.C., Duffy, A., Howard, W.R., Kukla, G.J., Kutzbach, J., Martinson, D.G., McIntyre, A., Mix, A.C., Molfino, B., Morley, J.J., Peterson, L.S., Pisias, N.G., Prell, W.L., Raymo, M.E., Shackleton, N.J., Toggweiler, J.R., 1992. On the structure and origin of major glaciation cycles; 1. Linear responses to Milankovitch forcing. Paleoceanography 7, 701–738. Jaccard, S.L., Galbraith, E.D., Martínez-García, A., Anderson, R.F., 2016. Covariation of deep Southern Ocean oxygenation and atmospheric CO2 through the last ice age. Nature 530, 207–210. Jonkers, L., Zahn, R., Thomas, A., Henderson, G., Abouchami, W., François, R., Masquea, P., Hall, I.R., Bickert, T., 2015. Deep circulation changes in the central South Atlantic during the past 145 kyrs reflected in a combined 231 Pa/230 Th, neodymium isotope and benthic δ 13 C record. Earth Planet. Sci. Lett. 419, 14–21. Jouzel, J., Masson-Delmotte, V., Cattani, O., Dreyfus, G., Falourd, S., Hoffmann, G., Minster, B., Nouet, J., Barnola, J.M., Chappellaz, J., Fischer, H., Gallet, J.C., Johnsen, S., Leuenberger, M., Loulergue, L., Luethi, D., Oerter, H., Parrenin, F., Raisbeck, G., Raynaud, D., Schilt, A., Schwander, J., Selmo, E., Souchez, R., Spahni, R., Stauffer, B., Steffensen, J.P., Stenni, B., Stocker, T.F., Tison, J.L., Werner, M., Wolff, E.W., 2007. Orbital and millennial Antarctic climate variability over the past 800,000 years. Science 317, 793–796. Khodri, M., Cane, M.A., Kukla, G.J., Gavin, J., Braconnot, P., 2005. The impact of precession changes on the Arctic climate during the last interglacial–glacial transition. Earth Planet. Sci. Lett. 236, 285–304. Khodri, M., Leclainche, Y., Ramstein, G., Braconnot, P., Marti, O., Cortijo, E., 2001. Simulating the amplification of orbital forcing by ocean feedbacks in the last glaciation. Nature 410, 570–574. Kohfeld, K.E., Fairbanks, R.G., Smith, S.L., Walsh, I.D., 1996. Neogloboquadrina pachyderma (sinistral coiling) as paleoceanographic tracers in polar oceans: evidence from northeast water Polynya plankton tows, sediment traps, and surface sediments. Paleoceanography 11, 679–699. Kohfeld, K.E., Graham, R.M., De Boer, A.M., Sime, L., Wolff, E.W., Le Quéré, C., Bopp, L., 2013. Glacial–interglacial changes in Southern Hemisphere Westerly winds: paleo-data synthesis. Quat. Sci. Rev. 68, 76–95. Kohfeld, K.E., Le Quéré, C., Harrison, S.P., Anderson, R.F., 2005. Role of marine biology in glacial–interglacial CO2 cycles. Science 308, 74–78. Kohfeld, K.E., Ridgwell, A., 2009. Glacial–interglacial variability in atmospheric CO2 . In: Le Quere, C., Salzman, E. (Eds.), The Surface Ocean – Lower Atmosphere Processes. In: Geophysical Monograph Series. AGU, Washington, DC. Kucera, M., Weinelt, M., Kiefer, T., Pflaumann, U., Hayes, A., Weinelt, M., Chen, M.T., Mix, A.C., Barrows, T.T., Cortijo, E., Duprat, J., Juggins, S., Waelbroeck, C., 2005. Reconstruction of sea-surface temperatures from assemblages of planktonic foraminifera: multi-technique approach based on geographically constrained calibration data sets and its application to glacial Atlantic and Pacific oceans. Quat. Sci. Rev. 24, 951–998.

K.E. Kohfeld, Z. Chase / Earth and Planetary Science Letters 472 (2017) 206–215

Lamy, F., Gersonde, R., Winckler, G., Esper, O., Jaeschke, A., Kuhn, G., Ullermann, J., Martinez-Garcia, A., Lambert, F., Kilian, R., 2014. Increased dust deposition in the Pacific Southern Ocean during glacial periods. Science 343, 403–407. Lisiecki, L.E., Raymo, M.E., 2005. A Pliocene–Pleistocene stack of 57 globally distributed benthic δ 18 O records. Paleoceanography 20. http://dx.doi.org/10.1029/ 2004PA001071. Lund, D.C., Adkins, J.F., Ferrari, R., 2011. Abyssal Atlantic circulation during the Last Glacial Maximum: constraining the ratio between transport and vertical mixing. Paleoceanography 26. http://dx.doi.org/10.1029/2010PA001938. Lynch-Stieglitz, J., Adkins, J.F., Curry, W.B., Dokken, T., Hall, I.R., Herguera, J.C., Hirschi, J.J.M., Ivanova, E.V., Kissel, C., Marchal, O., Marchitto, T.M., McCave, I.N., McManus, J.F., Mulitza, S., Ninnemann, U., Peeters, F., Yu, E.F., Zahn, R., 2007. Atlantic meridional overturning circulation during the Last Glacial Maximum. Science 316, 66–69. Martínez-García, A., Sigman, D.M., Ren, H., Anderson, R.F., Straub, M., Hodell, D.A., Jaccard, S.L., Eglinton, T.I., Haug, G.H., 2014. Iron fertilization of the subantarctic ocean during the Last Ice Age. Science 343, 1347–1350. Martinson, D.G., Pisias, N.G., Hays, J.D., Imbrie, J., Moore, T.C., Shackleton, N.J., 1987. Age dating and the orbital theory of the ice ages – development of a highresolution—0 to 300,000-year chronostratigraphy. Quat. Res. 27, 1–29. Moore, J.K., Abbott, M.R., Richman, J.G., Nelson, D.M., 2000. The Southern Ocean at the last glacial maximum: a strong sink for atmospheric carbon dioxide. Glob. Biogeochem. Cycles 14, 455–475. Oliver, K.I.C., Hoogakker, B.A.A., Crowhurst, S., Henderson, G.M., Rickaby, R.E.M., Edwards, N.R., Elderfield, H., 2010. A synthesis of marine sediment core δ 13 C data over the last 150,000 years. Clim. Past 6, 645–673. Peacock, S., Lane, E., Restrepo, J.M., 2006. A possible sequence of events for the generalized glacial–interglacial cycle. Glob. Biogeochem. Cycles 20. Piotrowski, A.M., Goldstein, S.L., Hemming, S.R., Fairbanks, R.G., 2005. Temporal relationships of carbon cycling and ocean circulation at glacial boundaries. Science 307, 1933–1938. Röthlisberger, R., Bigler, M., Wolff, E.W., Joos, F., Monnin, E., Hutterli, M.A., 2004. Ice core evidence for the extent of past atmospheric CO2 change due to iron fertilisation. Geophys. Res. Lett. 31. http://dx.doi.org/10.1029/2004GL020338. Röthlisberger, R., Crosta, X., Abram, N.J., Armand, L., Wolff, E.W., 2010. Potential and limitations of marine and ice core sea ice proxies: an example from the Indian Ocean sector. Quat. Sci. Rev. 29, 296–302.

215

Rutgers van der Loeff, M.M., Cassar, N., Nicolaus, M., Rabe, B., Stimac, I., 2014. The influence of sea ice cover on air-sea gas exchange estimated with radon-222 profiles. J. Geophys. Res., Oceans 119, 2735–2751. Shackleton, N.J., 1995. New data on the evolution of Pliocene climate variability. In: Vrba, E.S., Denton, G.H., Partridge, T.C., Burckle, L.H. (Eds.), Paleoclimate and Evolution, with Emphasis on Human Origins. Yale University Press, New Haven, CT, pp. 242–248. Sigman, D., Hain, M., Haug, G., 2010. The polar ocean and glacial cycles in atmospheric CO(2) concentration. Nature 466, 47–55. Sigman, D.M., Jaccard, S.L., Haug, G.H., 2004. Polar ocean stratification in a cold climate. Nature 428, 59–63. Stephens, B.B., Keeling, R.F., 2000. The influence of Antarctic sea ice on glacial– interglacial CO2 variations. Nature 404, 171–174. Studer, A.S., Sigman, D.M., Martinez-Garcia, A., Benz, V., Winckler, G., Kuhn, G., Esper, O., Lamy, F., Jaccard, S.L., Wacker, L., Oleynik, S., Gersonde, R., Haug, G.H., 2015. Antarctic zone nutrient conditions during the last two glacial cycles. Paleoceanography 30, 845–862. Toggweiler, J.R., 1999. Variation of atmospheric CO2 by ventilation of the ocean’s deepest water. Paleoceanography 14, 571–588. Waelbroeck, C., Labeyrie, L., Michel, E., Duplessy, J.C., McManus, J.F., Lambeck, K., Balbon, E., Labracherie, M., 2002. Sea-level and deep water temperature changes derived from benthic foraminifera isotopic records. Quat. Sci. Rev. 21, 295–305. Watson, A.J., Naveiro Garabato, A.C., 2006. The role of Southern Ocean mixing and upwelling in glacial–interglacial atmospheric CO2 change. Tellus, Ser. B Chem. Phys. Meteorol. 58, 73–87. Watson, A.J., Vallis, G.K., Nikurashin, M., 2015. Southern Ocean buoyancy forcing of ocean ventilation and glacial atmospheric CO2 . Nat. Geosci. 8, 861–865. Wilson, D.J., Piotrowski, A.M., Galy, A., Banakar, V.K., 2015. Interhemispheric controls on deep ocean circulation and carbon chemistry during the last two glacial cycles. Paleoceanography 30, 621–641. Wolff, E.W., Barbante, C., Becagli, S., Bigler, M., Boutron, C.F., Castellano, E., de Angelis, M., Federer, U., Fischer, H., Fundel, F., Hansson, M., Hutterli, M., Jonsell, U., Karlin, T., Kaufmann, P., Lambert, F., Littot, G.C., Mulvaney, R., Roöthlisberger, R., Ruth, U., Severi, M., Siggaard-Andersen, M.L., Sime, L.C., Steffensen, J.P., Stocker, T.F., Traversi, R., Twarloh, B., Udisti, R., Wagenbach, D., Wegner, A., 2010. Changes in environment over the last 800,000 years from chemical analysis of the EPICA dome C ice core. Quat. Sci. Rev. 29, 285–295.