Persistent deep water anoxia in the eastern South Atlantic during the last ice age

Significance

Climate variations are closely related to changes in ocean circulation, and this has important implications for the availability of oxygen in deep waters and the ocean’s CO₂ sequestration capacity. Potentially anoxic conditions have previously been inferred for the deep waters of the Southern Ocean and the Pacific, and low oxygen has been proposed for the North Atlantic deep waters during glacial cycles. Here, we provide clear evidence that fully anoxic conditions occurred in the deep waters of the eastern South Atlantic during the last glacial interval. This is an important finding implying that the Atlantic Ocean circulation was more sluggish during the advanced stage of the last ice age than previously assumed.

Abstract

During the last glacial interval, marine sediments recorded reduced current ventilation within the ocean interior below water depths of approximately >1,500 m [B. A. Hoogakker et al., Nat. Geosci. 8, 40–43 (2015)]. The degree of the associated oxygen depletion in the different ocean basins, however, is still poorly constrained. Here, we present sedimentary records of redox-sensitive metals from the southwest African margin. These records show evidence of continuous bottom water anoxia in the eastern South Atlantic during the last glaciation that led to enhanced carbon burial over a prolonged period of time. Our geochemical data indicate that upwelling-related productivity and the associated oxygen minimum zone in the eastern South Atlantic shifted far seaward during the last glacial period and only slowly retreated during deglaciation times. While increased productivity during the last ice age may have contributed to oxygen depletion in bottom waters, especially on the upper slope, slow-down of the Late Quaternary deep water circulation pattern [Rutberg et al., Nature 405, 935–938 (2000)] appears to be the ultimate driver of anoxic conditions in deep waters.

The ocean plays an important role in the global carbon cycle by taking up carbon dioxide (CO₂) from the atmosphere. The oceanic CO₂ sequestration capacity is determined by the temperature and chemistry of the water masses and strongly influenced by the efficiency of the biological carbon pump (refer to ref. 1 and references therein). If more organic carbon is exported from the surface ocean and buried in the sediment or stored as respired CO₂ within deep water masses, this would increase the sink function of the ocean for atmospheric CO₂. Of essential importance for the carbon burial efficiency is the oxygen (O₂) content of the deep ocean (2). Under oxygen-deficient or even anoxic conditions, less organic carbon is respired via aerobic respiration and more organic carbon is buried in the sediment (3). Against the background of the anthropogenic CO₂ increase and current climate change, it is crucial to determine and evaluate processes and mechanisms that impact the oceanic bottom water oxygen content.

The ocean carbon cycle and the distribution of oxygen in the ocean are closely entangled with the meridional overturning circulation (4). In the Atlantic part of the meridional overturning circulation (AMOC), two water masses are dominant: the warm, oxygen-rich, and nutrient-depleted North Atlantic Deep Water (NADW) and the cold and nutrient-rich Antarctic Bottom Water (5, 6) (Fig. 1). The current strengths of these water masses have strongly varied between glacial and interglacial periods, and this is intrinsically tied to the marine carbon cycle (4, 9). The oxygen content of the bottom waters is of central importance in this context. It is determined by the supply of oxygenated waters via currents (10, 11) and the O₂ consumption through organic carbon respiration, both locally and transported downstream (12).

Fig. 1.

Modern oceanographic conditions along the study sites at the Namibian continental margin. (A) Cross-section of dissolved oxygen concentrations (O₂) displaying the present-day location of the OMZ and overview map. (B) Water column salinity with major currents and flow directions. The tropical surface waters are underlain by the South Atlantic Central Water (SACW). Originating from the Indian Ocean, the Antarctic Intermediate Water (AAIW) propagates northwards, similar to the Upper Circumpolar Deep Water (UCDW) which lies below the AAIW and originates from the Antarctic Circumpolar Current. Advecting southwards, the NADW is underlain by the northward propagating Lower Circumpolar Deep Water (LCDW) and Antarctic Bottom Water (AABW) (refer to ref. 6 and references therein). Plots were created using the software Ocean Data View (7). Red dots indicate study sites; dark gray dot shows the location of Site RC 13–229 (8).

Reliable information on the extent of bottom water oxygen deficiency during glacial intervals is therefore extremely important for our understanding of past global carbon cycle dynamics. Due to its position at the onset of the global thermohaline conveyor belt, the eastern South Atlantic is of particular relevance in this respect. The data presented here provide a detailed geochemical perspective on the bottom water redox conditions within the last 46 ky along a lateral transect in the Cape Basin at the Southwest African continental slope. Strong, redox-sensitive metal accumulation and enhanced organic carbon burial provide clear evidence that the redox conditions must have been much more reducing during the last glaciation than previously considered.

Organic Carbon and Trace Metal Accumulation

The investigated sites are located in an upwelling system on the continental margin off Namibia and are characterized by high amounts of total organic carbon (TOC) in the sediments (up to 11.4 wt%), generally increasing with depth (Fig. 2). Highest TOC mass accumulation rates (TOC_MAR) of up to 29.4 gC m⁻² y⁻¹ are observed in the glacial interval (Table 1) concurrent with elevated amounts of the redox-sensitive metals molybdenum (Mo) and uranium (U) ranging from 20.0 to 72.5 and 8.1 to 34.0 mg kg⁻¹, respectively (Fig. 2).

Fig. 2.

Reconstruction of the redox conditions for the last 46 ky along the Namibian Continental Margin. (A) TOC, molybdenum concentration (Mo), Mo isotope (δ⁹⁸Mo; 2SD), and Mo/U ratios for the lower slope Site GeoB 8470. (B and C) Geochemical data for the sites located on the upper slope, Sites GeoB 8455 and GeoB 8426, respectively. Red arrows depict ¹⁴C ages. The yellow bar indicates the Holocene (11.5 ky BP to present) and the gray bar the LGM (24 to 18.7 ky BP) interval within the respective MIS.

Table 1.

Average geochemical sediment compositions at the West African transect sites

Time interval*	SR	Mo	MoMAR	U	UMAR	TOC	TOCMAR
	cm ky−1	mg kg−1	nmol cm−2 y−1	mg kg−1	nmol cm−2 y−1	wt.%	gC m−2 y−1
GeoB 8426 (25°28.9 S, 13°21.1 E, 1,045 m WD)
Holocene	14.4	23.7 ± 2.0	2.62	20.9 ± 2.5	2.38	6.5	7.42
MIS 1	17.4	24.0 ± 3.0	3.37	19.6 ± 3.0	2.71	6.9	9.56
LGM	22.5	46.6 ± 4.5	10.00	28.6 ± 4.3	6.18	7.5	15.2
MIS 2†	22.5	35.8 ± 8.9	7.52	21.7 ± 6.0	4.59	7.6	15.1
MIS 3†	22.5	38.9 ± 6.2	7.50	17.0 ± 2.9	3.30	9.2	16.7
GeoB 8455 (25°30.4 S, 13°11.0 E, 1,502 m WD)
Holocene	5.0	21.3 ± 0.5	1.04	25.7 ± 2.2	1.25	4.8	2.31
MIS 1	6.3	22.1 ± 1.7	1.10	26.0 ± 2.1	1.30	5.1	2.45
LGM	12.8	26.4 ± 4.5	2.96	18.5 ± 4.9	2.06	6.1	6.55
MIS 2	58.0	40.1 ± 10.1	16.71	17.0 ± 4.9	6.07	7.2	29.4
MIS 3	23.5	29.9 ± 5.4	5.09	13.9 ± 2.7	2.34	7.6	12.4
GeoB 8470 (25°32.7 S, 12°51.6 E, 2,470 m WD)
Holocene	8.0	9.8 ± 5.8	0.59	6.1 ± 3.2	0.37	2.1	1.20
MIS 1	8.2	12.1 ± 7.2	0.74	7.3 ± 3.9	0.45	2.3	1.34
LGM	9.6	30.8 ± 6.2	1.82	18.2 ± 2.5	1.08	4.7	2.64
MIS 2	9.6	36.3 ± 8.7	2.41	18.2 ± 2.9	1.20	5.3	3.32
MIS 3	9.9	59.5 ± 8.2	4.32	24.2 ± 2.3	1.76	6.8	4.73

Average sedimentation rates (SR), molybdenum (Mo) and uranium (U) contents and SD (SD), and mass accumulation for molybdenum (Mo_MAR) and uranium (U_MAR) for specific time intervals at each investigated site.

*Data calculation for MIS 3 interval cut off at 46 ky.

^†Estimated age based on average SR determined for the upper ∼14.5 ky BP. WD: Water Depth. Event ages after ref. 13 and references therein: MIS 2 33 to 14.4 ky BP; LGM 24 to 18.7 ky BP; MIS 1 14.5 ky BP to present; and Holocene 11.5 ky BP to preset.

Under fully oxic water column conditions, Mo contents can be increased in the sediments due to reactive iron (Fe) and manganese (Mn) oxide phases on which Mo can adsorb (14, 15). For example, in surface sediments of Baja California deposited under oxic conditions, high Mo contents were observed (up to 117 mg kg⁻¹) correlating with high manganese concentrations (Mn >2 wt%) (16, 17). However, those high Mo values are limited to the sediment surface where oxides are present. Below this depth, Mn and Fe oxides are reduced, releasing any adsorbed Mo into the pore water. This process results in the restriction of Mo enrichments to the upper surface sediments. Mn oxides as the major Mo host phase would also be reflected by the Mo isotope composition as Mo adsorbed onto Mn oxides is isotopically light (δ⁹⁸Mo of −0.5‰ compared to +2.3‰ in seawater) (18–20). Similar to other high productivity environments such as the Mexican or Peruvian margin, where Mn contents stay below 280 mg kg⁻¹ (21), the Mn contents in the sediment investigated along the Namibia continental margin transect are low (<210 mg kg⁻¹) and could thus not explain the high Mo contents. Similarly, Fe oxide input on the lower slope is low; compared to the Peruvian margin in which total Fe averages 2.88 wt% (22), the average total iron is 2.44 wt%, 1.69 wt%, and 1.56 wt% at Sites GeoB 8426, GeoB 8455, and GeoB 8470, respectively. Furthermore, the analyzed high-Mo isotopic composition of the sediment (δ⁹⁸Mo > 0.7‰) in the glacial intervals at all three sites (Fig. 2) argues against Mo input coupled to Mn or Fe cycling.

To accumulate appreciable amounts of Mo (>30 mg kg⁻¹) in the sediments, as observed in our study area at all three sites during the glacial interval (Fig. 2 and Table 1), severe oxygen drawdown in the overlying water is required. In the absence of increased iron or manganese oxide deposition, it was previously noticed—based on a global data compilation—that for an enrichment of ∼25 mg kg⁻¹ Mo or higher in the sediment, dissolved sulfide must be present in the (bottom) water, either permanently or intermittently (23). In sulfide-containing environments, Mo is converted from molybdate into particle reactive thiomolybdate or reactive Mo-polysulfide (24–26). This leads to moderate Mo accumulation (<25 mg kg⁻¹) if sulfide is restricted to the sediments and results in high sedimentary Mo uptake and burial if sulfide is available in the overlying water column (21, 26, 27). Though, in shallow margin settings with high organic carbon input, mildly oxygenated (<15 µM O₂) to anoxic (nonsulfidic) conditions seem to be sufficient to lead to high Mo accumulations in the sediment (21, 22, 28), with Mo not exceeding 30 mg kg⁻¹ when O₂ is still available (28). These types of redox conditions are also recorded in the Mo isotopes in which δ⁹⁸Mo values can be up to 0.7‰ lower than average seawater δ⁹⁸Mo (2.3‰) (24, 29).

In contrast to Mo, the main delivery pathway of U to the sediments is diffusion from the bottom water into anoxic sediments with only slightly increased U uptake in the presence of dissolved sulfide (30–33). Thus, compared to Mo, the uptake of U into the sediment is higher under anoxic, nonsulfidic conditions, while under sulfidic conditions, this relation is reversed (21, 34). Uranium accumulation in sediments occurs under anoxic conditions as a result of the reduction of soluble U(VI) to insoluble U(IV) during the oxidation of organic carbon (30, 31, 35, 36). Increased organic carbon flux can also act as another input source of U to the sediment (31, 37). Consequently, U is often found to be enriched in organic-rich sediments (34, 38). The observed U mass accumulation rates (U_MAR) along the investigated transect (Table 1) are similar to or much higher in the older (>14.5 ky) sediments compared to values reported for continental margin systems with very low to undetectable O₂ concentration (<10 µM) in the overlying bottom waters (21). In those settings, the U_MAR rates are mainly around 0.4 nmol cm⁻² y⁻¹ with highest rates not exceeding 1.97 ± 0.37 nmol cm⁻² y⁻¹ (21). Therefore, based on the determined U contents and accumulation rates and the Mo/U ratios, we have to conclude that the bottom waters at the investigated sites were not just low but completely O₂-free during deposition.

Increased deposition of organic matter in an organic carbon depocenter along the upper slope off Namibia has been related to suspension loads with material from the shelf transported laterally downslope via nepheloid layers (39). While such a downslope transport process could explain the observed high TOC amounts, it would also lead to the oxidation of reduced Mo and U phases and remobilization into the water column unless anoxic bottom water conditions were present (40). At all three sites, bottom water anoxia during the glacial period is also mirrored by sedimentary Mo isotope values (Fig. 2) being similar to those observed in modern anoxic upwelling areas (22, 41). In the upper Holocene, lower δ⁹⁸Mo values, in conjunction with low Mo concentrations (<30 mg kg⁻¹), suggest an increase in bottom water ventilation with oxygenated to somewhat oxygen-depleted conditions similar to those found in the present-day bottom waters (Fig. 1A). Overall, our data show that the bottom water along the investigated transect was predominantly anoxic during the last glacial interval, even at the deepest site (GeoB 8470; 2,470 m water depth).

Deep Water Chemistry Response to Ocean Circulation

In the current oceanographic configuration, Site GeoB 8470 is situated within the NADW flow path. Sites GeoB 8426 and GeoB 8455 are located within the low-oxygen, nutrient-rich intermediate waters flowing northwards above the NADW—the Antarctic Intermediate Water (AAIW) and the lower Upper Circumpolar Deep Water, respectively (6, 42) (Fig. 1A). The calculated Mo_MAR and U_MAR (Table 1) for the glacial interval are comparable to rates reported for other modern, open-ocean sedimentary environments that are strongly depleted in or free of bottom water O₂ such as the California margin, the Gulf of California, and the high-productivity areas off Namibia (shelf), Mexico, and Peru (21, 22, 33, 43) (SI Appendix, Table S1). However, in all these cases, the reduced oxygen conditions occur at much shallower water depths, mostly on the upper slope and shelf (<600 m), whereas our investigated sites record bottom water redox conditions at much greater water depth (>1,000 m) in the open ocean.

The observed anoxic bottom water conditions on the slope in the glacial Cape Basin are likely linked to increased productivity and a seaward shift of the oxygen minimum zone (OMZ). Because of the lower sea level during the last glacial period, high productivity regions shifted from shelf to slope, increasing the export of organic material to the deeper ocean interior (44). In addition, erosion and downslope transport of organic material from the exposed or shallower shelf of the Cape Basin may have enhanced organic carbon delivery (39, 45). The elevated organic carbon delivery increased both the oxygen demand within the deep water and the burial flux of organic carbon within deep-sea sediment (45). Concurrently, cooler temperatures resulted in lower remineralization rates broadening the zone of maximum oxygen demand (10, 46). It seems unlikely, however, that the seaward-shifted OMZ broadened throughout the water column to depths >2,000 m (47). Therefore, a mechanism other than the local increase in productivity and associated OMZ expansion alone is required to explain the observed O₂-free bottom waters at our deepest study site during glacial times, which is indicated by highly elevated, redox-sensitive metal accumulation.

Importantly, similar metal accumulation profiles were reported for a deeper site (RC13-229; 4,191 m; Fig. 1) (8) along the same transact as our study sites. These basin sediments show elevated U values (up to 7.6 mg kg⁻¹) at ∼30 to 15 ky (marine isotope stage, MIS, 2) followed by a strong decrease during the last 14.5 ky (MIS 1; average 0.28 mg kg⁻¹), while the TOC changes are slightly less pronounced, from mean 0.85 to 0.2 wt%, respectively (SI Appendix, Fig. S1) (8). This observation is in good agreement with previously published results that suggest low oxygen in deep Southern Ocean bottom water during the Last Glacial Maximum (LGM) based on Mn/aluminum (Al) ratios and authigenic U (12, 37). Based on our findings along the Cape Basin margin and previously reported similar observations in the deeper basin, we propose that changes in deep water circulation were the driving factor leading to the observed anoxic bottom water conditions during glacial times in the deeper (>1,500 m) eastern South Atlantic. The observed U_MAR at our study site is higher compared to previously published values further offshore (SI Appendix, Fig. S1), which is likely related to a higher export production along the continental margin (31, 37).

Specifically at the two shallow(er) sites, bottom water redox conditions are most likely controlled by both current ventilation and the local organic carbon flux. We speculate that the glacial increase in ventilation in the intermediate waters related to enhanced AAIW production (48) is overprinted in the geochemical record at the shallower sites by a seaward shift of the upwelling cell and an increase in export production. It cannot be excluded, however, that even the intermediate waters in the eastern South Atlantic were far less ventilated during the last ice age than previously assumed (10, 48).

Implication for the Global Carbon Cycle

It has been suggested that during the last glacial period, the deeper water masses experienced reduced or possibly reversed flow (42, 49, 50) or shoaling (51). In addition, the strength of the AMOC was strongly influenced by a decrease of the supply of warm Indian Ocean surface water to the eastern South Atlantic during glacial stages (52). Either mechanism resulted in lower oxygen concentrations within the deep water during the last glaciation accompanied by a build-up of respired carbon (48, 53) and increased organic carbon burial in the sediment (54). A similar change in current strength and ventilation was suggested for the Pacific deep waters during the LGM, most likely causing strongly O₂-depleted bottom water conditions and enhanced carbon burial (11, 12, 55). We propose that the decreased ventilation of the deep ocean waters during the glacial interval of the last 46 ky was far more pronounced than previously expected. Low O₂ concentrations (about 20% lower compared to modern concentrations) have been suggested for the deep North Atlantic (10) and the South Atlantic (56) during the last glacial period related to weaker NADW formation. In contrast, our records indicate completely anoxic conditions in the deep waters of the eastern South Atlantic. With the onset of deglaciation, the NADW deep water became more ventilated (48), and carbon respiration increased again, leading to less carbon burial and accumulation of redox-sensitive metals. These changes in current strength and ventilation and the shift of organic carbon export to shallower depth is recorded in the sediment of the Cape Basin slope. Compared to the deeper basin or the Southern Ocean, the slope sediment records (SI Appendix, Fig. S1) show a slightly delayed shoreward response in adjusting to the new conditions similar to trends observed in carbon-rich glacial Equatorial Pacific sediments (SI Appendix, Fig. S1) (55). Based on our findings, we suggest that the deep waters of the eastern South Atlantic experienced not just oxygen deficiency but fully anoxic bottom water conditions due to strong changes in the AMOC. These conditions led to higher carbon burial (54) for a prolonged period of time during the ice age, which was amplified on the West African slope by increased productivity and an offshore relocation of the OMZ. While low oxygen bottom waters during glacial cycles have previously been suggested for the deep Pacific and Southern Ocean (10, 11, 37, 55), our data clearly indicate complete O₂-depletion in the deep waters of the eastern South Atlantic. Such deep water redox conditions lead to increased sequestration of CO₂ in the ocean via organic carbon burial or storage of respired CO₂ within deep waters (1, 3). The extent of deep ocean carbon storage and its spatial and temporal variability in the eastern South Atlantic over the last 46 ky provides important information for global carbon cycle models attempting to quantify the ocean’s impact on paleo-climate variability and the oceanic CO₂ sequestration capacity.

Materials and Methods

Sampling.

Gravity cores GeoB 8426–3 (25°28.9 S, 13°21.1 E) at a water depth of 1,045 m, GeoB 8455–2 (25°30.4 S, 13°11.0 E) at a water depth of 1,502 m, and GeoB 8470–4 (25°32.7 S, 12°51.6 E) at 2,470 m water depth were collected during the RV METEOR expedition M57/2 in the eastern South Atlantic on the continental margin off Namibia (57). Gravity cores were cut into 1-m segments on deck immediately after retrieval and placed in a cooling container at a temperature of about 4 °C. The cores were cut lengthwise into work and archive halves (within the first 2 d after recovery). All further processing occurred in an argon-purged glove box under anoxic conditions. Solid-phase samples were taken at 10-cm intervals and stored in gas-tight glass bottles under argon atmosphere at −20 °C. Detailed sample collection is described in refs. 57 and 58.

Analytical Methods.

For samples from Site GeoB 8455, the total carbon (TC) and TOC contents were determined using a LECO CS-300 carbon sulfur analyzer at the University of Bremen (57). Total inorganic carbon (TIC) was determined by subtracting TOC from TC in each sample. Additional samples from Site GeoB 8455 and all samples from Sites GeoB 8426 and GeoB 8470 were analyzed for TC and TIC content on an ELTRA 2000 Carbon Sulfur Determinator at Oklahoma State University (OSU) in Stillwater, OK. About 100 mg of sample was ashed in a furnace at 1,350 °C to determine TC. For TIC, about 100 mg of sample was treated with 20 mL of 20% HCl. The TOC was calculated by subtracting TIC from TC.

Trace metal analysis was performed at OSU. Approximately 100 mg of each sample was ashed at 550 °C for 10 h in a Thermo Fisher Scientific furnace to remove organic carbon. A multi-acid total-digestion procedure (3 mL ultra-pure nitric acid, 3 mL ultra-pure hydrochloric acid, and 2 mL ultra-pure hydrofluoric acid) was carried out on the ashed samples using a PicoTrace© Pressure Digestion System. The vials were gradually heated up to 180 °C and kept there for 5 h, followed by an evaporation step for ∼10 h. Dried samples were taken up in 10 mL 5% ultra-pure nitric acid solution. Diluted samples were analyzed on a Thermo Fisher Scientific iCAP Qc Inductively Coupled Plasma Mass Spectrometer (ICP-MS). For quality control, a standard reference material (SRM 2702; National Institute of Standards and Technology [NIST]) was included in every sample batch, and the analytical error was <6%. All metal data are reported as total content due to the overall small difference to calculated authigenic element contents (<3.3% for Mo and 5.9% for U).

To determine the isotopic composition of Mo, a chemical separation was performed in 0.5 M HCl on 2 mL of Bio-Rad AG50W-X8 cation resin to remove Fe. Following this step, the remaining matrix was separated in 4 M HCl on 1 mL Bio-Rad AG1-X8 anion exchange resin and elution in 2 M HNO₃. Total procedural blanks were <1 ng of Mo. Samples were analyzed on a Nu Instruments Multi Collector (MC) ICP-MS with DSN-100 Desolvation Nebulizer System (GEOMAR, Kiel) (for details, refer to ref. 59). A Mo isotope double spike (¹⁰⁰Mo, ⁹⁷Mo) was used to compensate for instrumental mass bias and possible mass fractionation during column chromatography (22). All Mo isotopic variations are presented in delta notation as the deviation of the ⁹⁸Mo/⁹⁵Mo ratio relative to a standard in parts per thousand (‰):

$${\text{δ}}^{98}\text{Mo}=\left[\left({}^{98}\text{Mo}{/}^{95}\text{Mo}\right)\text{sample}/\left({}^{98}\text{Mo}{/}^{95}\text{Mo}\right)\text{standard}–1\right]*1000.$$ [1]

Samples were measured relative to Alfa Aesar Mo plasma standard solution Specpure #38791 (lot no. 011895D) (refer to discussion in ref. 60). International standard NIST-SRM-3134 was analyzed constantly and has an offset from the Alfa Aesar standard of +0.15 ± 0.08‰ (2 SD, n = 46), which is consistent with published values of refs. 60 and 61. Following ref. 60, results are presented in the delta notation relative to the NIST-SRM-3134 scale with an offset of +0.25‰. This makes results comparable to earlier studies on Mo isotope fractionation in oceanic environments. Certified reference standard SDO-1 (Devonian Ohio Shale, US Geological Survey) was processed through chemistry and measured with each sample run. The long-term external reproducibility of SDO-1 is 0.09‰ (2 SD, average δ⁹⁸Mo is +0.99‰). The long-term external reproducibility of the Specpure standard solution is <0.1‰ (2 SD).

Age Model.

Radiocarbon (¹⁴C) analyses on planktonic foraminifera from 11 sediment samples (SI Appendix, Table S2) were carried out in the Carbon Dating Laboratory of the Alfred-Wegener-Institute (AWI) in Bremerhaven, Germany, by using an accelerator mass spectrometer (AMS) (AWI nos. 2321 through 2326, 2562 through 2564, 2565, and 2566). For calibration of the AMS ¹⁴C ages, the modeled ocean average curve (Marine13) (62) was used as well as a weighted mean ΔR of 146 ± 85 ¹⁴C y. The ages were calibrated (calendar age before present [BP]) with Calib 7.1 software (63). For Sites GeoB 8455 and GeoB 8470, sliding age models were adapted. In order to expand the age model at Station GeoB 8455 back to 50 ky BP, a comparison with the age model of the very nearby Ocean Drilling Program Site 1084 (Leg 175; 25° 31′S, 13° 2′E, water depth 1,992 m) was used (SI Appendix, Fig. S2). For Site GeoB 8426, a linear model was adapted. Final ages are reported in calendar years BP, in which present is 1950.

MAR.

MAR were calculated using sedimentation rates (SR) derived from the ¹⁴C age models, an average dry bulk density (DBD) of 2.7 g/cm⁻³ (58), and the porosity (φ) (57, 58): MAR = SR * DBD* (1-φ).

Data Availability

Numerical data have been deposited in the PANGAEA Data Publisher (https://doi.org/10.1594/PANGAEA.926028).