A new proxy method for [CO32−] applied to the SE Pacific
A new dissolution index (%Low-CT-number calcite volume) obtained from XMCT scanning, enables us to measure the planktic foraminiferal test density of G. bulloides excluding the effect of wall thickness, and provides a reliable method to evaluate the dissolution intensity of tests.
The internal structure of G. bulloides tests shows selective dissolution of inner calcite, and CT number histograms changed from mono-modal to bi-modal distribution with the progress of dissolution (Figure 2a), indicating that relative volume of low CT number calcite increases with dissolution, confirming earlier studies31,32. Calibration between dissolution IDX (%Low-CT-number calcite volume) and deep-water Δ[CO32−] at each core site shows that the % Low-CT-number calcite volume serves as a quantitative proxy of deep-water Δ[CO32−] (Figure 2b). Our regression equation is as follows:
(1) Δ[CO32−] = –1.45 × (%Low-CT-number calcite volume) + 46.3, R2 = 0.91
Assuming constant [Ca2+] on a time scale shorter than 100 ka34, deep-water Δ[CO32−] is controlled by [CO32−]. Based on this equation [CO32−] at each core site was calculated using the following equation under the assumption of stable temperature and salinity of deep-water after the last glacial period:
(2) [CO32−] = -1.45× (%Low-CT-number calcite volume) + α
The α is a constant depending on the sediment core water depth, bottom water temperature and salinity at each core site. Based on the regression between in-situ deep-water Δ[CO32−] and the %Low-CT-number calcite volume, the uncertainty associated with reconstructing deep-water Δ[CO32−] is 3.4 µmol kg−1, which makes it possible to detect ~10 µmol kg−1 variations of [CO32−] on the millennial time-scales.
Deglacial [CO 3 2− ] reconstruction: evolution of deep SE Pacific carbonate chemistry
Among four sediment cores, water depths in the SE Pacific with water depths ranging from ca. 1500 to 4100 m (Figure 3a-d) with our first shallowest mesopelagic site PC01 (46˚04 S, 75˚41 W, 1535 m), bathed in UCDW and AAIW, yields low [CO32−] values of 65 to 73 µmol kg−1 during the early deglaciation (15-19 ka BP). Thereafter, site PC01 values decrease to around 60 µmol kg−1 at the end of the Younger Dryas (11.5 ka BP) (Figure 3a). Our second, slightly deeper upper bathyal site PC02 (46˚04 S, 76˚32 W, 2793 m) is also bathed in UCDW, but under a more pronounced influence of aged NADW. During the LGM (19-21 ka BP), PC02 shows values between 77-90 µmol kg−1, followed by a short, foraminifera-barren interval, supposedly driven by a carbonate dissolution event due to low [CO32−] deep water intrusion at the end of the LGM (19-19.5 ka BP). Thereafter, the [CO32−] in PC02 increased to values of around 85 µmol kg−1 during the early deglaciation (17-19 ka BP), followed by a significant decrease to 67 µmol kg−1 towards end around 15 ka BP) (Figure 3). Our third, lower bathyal site PC03 (46˚24 S, 77˚19 W, 3072 m) is bathed in LCDW with influence of AABW. Notably, this site shows mostly similar variations as PC02 during the LGM (19-21 ka BP), albeit with lower values around 75 µmol kg−1 at the end of the LGM around 19 ka BP. Thereafter, values in both cores start to diverge during the early deglaciation (15 - 19 ka BP), with significant [CO32−] increases from 75 to 95 µmol kg−1 at the deeper site PC03 (Figure 3c). Our deepest, abyssal site PS75/054-1 (56 ˚S, 115 ˚W, 4085 m), is bathed in LCDW and admixture of AABW from below. In contrast to our three shallower sites, this latter record shows a continuous decrease from relatively high glacial values of 100 to Early Holocene values of ca. 84 µmol kg−1 (Figure 3d).
A short-term, but distinct carbonate preservation event lasting from 18 – 18.4 ka BP is evident in our bathyal cores PC02 and PC03. During this time interval, a significant maximum in sediment deposition occurs, as evidenced by an age model-based 50-100-fold sedimentation rate increase to ca. 200 cm ka−1. which likewise, XRF-scanning-derived Ti/K ratios yield simultaneous maxima, be caused by suggesting higher input of terrestrial materials from the volcanic area of the Andes Cordillera35, while Zr/Rb ratio increases indicate stronger bottom current flow36 (Figure 3f). We thus assume that a transient high-sedimentation event of terrestrial material likely promoted carbonate preservation at the seafloor37,38, with foraminiferal test dissolution being in temporary disequilibrium with ambient deep-water, while it was the fact that the amount of carbonate dissolution at deep sea floor have clearly decline during this period. Therefore, we consider that reconstructed [CO32−] values during this short 400-year period between 18 and 18.4 ka BP are unreliable, and thus do not consider these data points in the following discussion.
Factors influencing [CO 3 2− ] changes: potential effects of export production and vertical mixing
Biological productivity in the surface ocean and organic matter remineralization below is one of the potential factors that alters deep water [CO32−]. In the Southern Ocean, surface ocean biological productivity is considered to have been generally low in the LGM and high in the early deglaciation39. At a nearby upper Chilean margin, variations in the Br/Ca ratio as an indication of biological carbon pump efficiency in this area, were used to suggest a more efficient biological carbon pump at the end of LGM (19 ka BP), combined with lower efficiency during the early deglaciation (15 -19 ka BP)40. In our study, however, the Br/Ca ratio at deeper bathyal sites PC02 and PC03 (Figure 3e) are generally higher during the early deglaciation (15-19 ka BP) than at the end of LGM (19 ka BP). This differing result is likely caused by local differences in surface productivity and/or depth-dependent differences in the carbonate preservation at our bathyal PC02 and PC03 sites, which indicates that organic carbon was effectively buried in sediment during the early deglaciation (15-19 ka BP) (Figure 3e). Considering that carbonate was relatively better preserved during the early deglaciation (15-19 ka BP) than the LGM (19 ka BP), we suppose that the biological productivity at sea surface is not a principal controlling factor of deep water [CO32−] variation at our study sites.
On the other hand, variations in the geochemical characteristics of deep-water masses are an alternative factor that potentially alters deep water [CO32−]. In previous studies, ventilation age reconstruction (i.e., 14C age difference between planktic and benthic foraminifera) suggested that South Pacific deep-water mass were less ventilated and more isolated from the atmosphere, in line with a presumed strong stratification, during the LGM (19 ka BP), followed by enhanced mixing with well-ventilated surface water during the early deglaciation (15-19 ka BP)11,12. In addition, radiogenic isotope results, which indicate the contribution from AABW supplied from the Ross Sea, suggested the breakup of deep bathyal water stratification after the LGM and enhanced mixing of AABW into shallower water masses13. In our study, the results of difference in stable carbon isotope of planktic and benthic foraminifera (Δδ13C planktic -benthic) in bathyal PC03, which imply the strength of bottom-surface ventilation11, suggests the stratification at the LGM (19 ka BP), followed by enhanced ventilation at the end of early deglaciation (15 ka BP) (Figure 3g). Furthermore, the results of Zr/Rb ratios in bathyal sites (PC02 and PC03), which cab bu used to trace variations in current strength in ocean bottom currents36, suggests an increase at the end of early deglaciation (15-16.5 ka BP) (Figure 3f). These results are consistent with previous interpretations that South Pacific deep-water masses were well-stratified and isolated from the atmosphere during the LGM, followed by maximum mixing and ventilation at the end of early deglaciation (15 ka BP). Therefore, we consider the carbonate chemical condition in deep-water mass to be principally governed by the chemical dynamics of deep-water mass structure in our area.
Transient zonal and meridional dynamics of Pacific deep-water carbonate chemistry
Deep-water mass biogeochemical configurations contribute to changes in vertical [CO32−] profiles in two ways. One is the variation in deep-water structures due to change in deep-water circulation, and the other one is changes of DIC within each water mass due to release and uptake of atmospheric CO2. However, these factors interact with each other, and change in deep-water mass structures can be interpreted as changes in carbon storage at a specific water depth and area. In this study, we found significant post-LGM variations of 65 to 90 µmol kg−1 in the [CO32−] at bathyal sites PC02 and PC03. These variations fluctuate between the [CO32−] of the mesopelagic site PC01 and abyssal site PS75/054-1 (Figure 4b), implying that the deep-water masses at bathyal sites around the depth of 2000-3000 m are principally a mixing product of the surrounding water masses.
Among our four sediment cores, site PC01 is strongly influenced by upper water masses originating from the north. A previous ventilation age reconstruction on a core close to site PC01 (MD07-3088) suggested a strong stratification and effect of PDW at the end of LGM (19 ka BP), followed by enhanced vertical mixing and increasing influence of well-ventilated deep waters during the early deglacial (Figure 4d)11. Upstream of our core sites, the [CO32−] of PDW was reconstructed in the Equatorial East Pacific at Site ODP1240 (2921 m: shown in Figure 4b), using B/Ca of benthic foraminifera, suggesting aged, low [CO32−] PDW during the LGM22. On the other hand, our deepest, abyssal site PS75/054-1, reflecting AABW supplied from the Ross Sea remains largely invariant in its [CO32−], despite previous suggestions of increased mixing with bathyal water masses from above during the early deglaciation based on εNd analyses13. Therefore, we assume that the [CO32−] of CDW as recorded in our bathyal sites PC02 and PC03 is mainly controlled by varying mixing of northern-sourced low-[CO32−] PDW and southern-source higher-[CO32−] AABW.
During the LGM, our bathyal [CO32−] reconstructions from sites PC02 and PC03 significantly differ from values at shallower and deeper sites PC01 and PS75/054-1, respectively. Considering that the South Pacific deep-water mass structure was supposedly strongly stratified and isolated from the atmosphere during this period (Figure 4d, e)11,13, glacial CDW was likely less influenced by water masses supplied from north and south. Some proxy reconstructions suggest that the AMOC was relatively strong (Figure 5c) during the LGM41,42, and therefore, glacial CDW should have been affected by deep-water supply from NADW. In fact, reconstructions from the subantarctic Southern Atlantic revealed that the [CO32−] of NADW at the depth of ca. 3800 m was around 70 µmol kg−1 during the LGM43, similar to CDW measured here. During the early deglacial (15-19 ka BP), the AMOC was weakened (Figure 5c), with suppressed NADW supply. Furthermore, deep-water stratification in the Southern Pacific was gradually breaking up11–13 (Figure 4d, e), and CDW became variably controlled by both, inflow of northern-sourced shallower, and southern-sourced deeper water masses.
In the following, we discuss the varying contributions of surrounding water masses to CDW during the deglaciation (15 -19 ka BP). Firstly, at the beginning of the early deglaciation (17-19 ka BP), the bathyal [CO32−] in both PC02 and PC03 temporally increased to 85 µmol kg−1, converging to similar values with that of abyssal site PS75/054-1, suggesting that CDW was temporally homogenized with a more pronounced AABW. At that time, SE Pacific deep-water masses were still isolated from the atmosphere, as evidenced by ventilation age reconstruction in nearby core MD07-3088 and Δδ13C (planktic -benthic foraminifera) in PC03 (Figure 3g). Thereafter, the [CO32−] in our bathyal sites PC02 and PC03 gradually diverged, with the shallower core PC02 converging to northern-sourced PDW signatures towards the end of early deglaciation (15 ka BP). This implies that the influence of northern-sourced PDW gradually increased, and reached a maximum expansion at the end of early deglaciation (15 ka BP).
Our quantitative reconstruction of SE Pacific deep-water [CO32−] enables us to compare our location to other key regions and to assess potential connections and spatial distributions of deep-water carbon storage. As a first step, we compare the variations in deep-water carbon storage based on deep-water [CO32−] between the SW Pacific24,25 and the SE Pacific (this study). Under modern conditions, the main flow of abyssal AABW to the north is located in the SW Pacific in the Deep Western Boundary Current system. In contrast, the return flow of DIC-rich and low-[CO32−] PDW is mostly located in the SE Pacific basins at bathyal depths of 2000-3500 m44. In the subantarctic SW Pacific, the variations in [CO32−] were recently reconstructed along a depth transect of three sites from 1160 m, 1630 m and 2540 m located off New Zealand, using B/Ca ratios of benthic foraminifera24,25 (Figure 4c). In particular, the mesopelagic SW Pacific showed low and stable [CO32−] values of around 60 µmol kg−1 at the beginning of the early deglaciation (17-19 ka BP), followed by a mid-deglacial 15−20 µmol kg−1 increase beginning around 17 ka BP. These results provide evidence for a reduction in deep-water carbon storage accompanied by strengthened vertical mixing at the end of the early deglaciation12. Compared to our vertical profile of [CO32−] in the SE Pacific that without [CO32−] increase in mesopelagic and upper bathyal sites, this suggests that a reduction in deep-water carbon storage occurred in shallower depths in the SW than in the SE Pacific (Figure 4). This asymmetrical east-west imbalance in Southern Pacific carbon storage was most likely caused by the preferential low-[CO32−] PDW import from the north into the SE Pacific. Thus, spatial reconstructions of deep-water masses and their biogeochemical signatures in both meridional (north-south) and zonal (east-west) directions are indispensable for a quantitative and process-oriented understanding of variations in marine carbon storage and its transfer between different reservoirs.
Pacific–Atlantic deep-water export through ACC transport: Implications for the Southern Ocean carbon budget
In order to elucidate the mechanisms causing pre-anthropogenic atmospheric CO2 variations and marine -atmospheric carbon transfer, one has to understand the inter-basin transport within the global MOC through the Southern Ocean as central intersection. In fact, it was suggested that glacial PDW, supplied into the Atlantic Ocean via Drake Passage, played a positive feedback role in increased carbon storage in the South Atlantic during the LGM43. However, data of previous deep-water [CO32−] reconstructions are limited to the SW and equatorial Pacific22,24,25, both distal to the Drake Passage as entrance to the Atlantic Ocean. Hence, our cores from the SE Pacific Chilean margin, close to the Drake Passage, enable us to better assess the transport and carbonate chemistry of deep-water masses exchanged between the Pacific and Atlantic. The intensity of Drake Passage throughflow is governed by the Antarctic Circumpolar Current (ACC), and past changes in throughflow over the last glacial have been reconstructed by grain size and geochemical analyses36,45,46. These results suggest that the Drake Passage throughflow was weaker during the LGM, and gradually strengthening in the termination, with a maximum at the end of early deglaciation (Figure 5d). In the South Atlantic, variations in [CO32−] of CDW and AABW were reconstructed with B/Ca ratios of benthic foraminifera43, showing different values between CDW and AABW during the LGM and early deglaciation (16.5-19 ka BP), while the [CO32−] of CDW increased after 16.5 ka and showed similar values as AABW at 15 ka BP (Figure 5b). Based on our deep-water [CO32−] reconstruction, we estimated the DIC variations at each site in the SE Pacific Chilean margin, using modern bottom water parameters (Total Alkalinity, Temperature and Salinity) as boundary conditions (Figure 5a). These estimates indicate a temporal change in the Southern Ocean’s deep-water structure, with a more pronounced influence of higher deep-water transport from the Pacific to the Atlantic on the carbon cycle after the LGM. (Figure 6)
During the LGM (Figure 6a), Southern Ocean surface biological productivity is thought to have been generally low due to high surface stratification39. However, data from the Chilean margin suggest that the biological carbon pump was relatively stronger during the LGM than the deglaciation, due to lowered carbonate production40, which contributed to effective carbon uptake from the atmosphere. The deep Southern Ocean is considered to have been well-stratified12,13, and therefore, Pacific CDW was more isolated from the surrounding water masses, ultimately leading to DIC-rich, aged PDW flowing eastward through the Drake Passage. The DIC estimate implies that carbon storage was larger in bathyal Pacific CDW than in AABW (Figure 5a), and therefore, Atlantic southern-sourced deep-waters were likely affected by such carbon-rich PDW-influenced CDW export from the SE Pacific via Drake Passage.
At the beginning of the early deglaciation (17-19 ka BP), our results in SE Pacific suggest expansion and mixing of AABW into overlying CDW, shown by homogenous [CO32−] in Pacific CDW. At the same time, deep-water [CO32−] reconstructions in Atlantic CDW and AABW yield diverging values43, suggesting that low [CO32−] Atlantic CDW was relatively isolated from Atlantic AABW. Deep-water in the SE Pacific was still strongly stratified and carbon discharge to the atmosphere hence suppressed, implying that the [CO32−] of deep-water remained low. In addition, deep-water export from the Pacific to the Atlantic through Drake Passage was strengthened gradually, enabling a leakage of low [CO32−], carbon-rich Pacific UCDW, contributing to the low Atlantic CDW [CO32−] values (70 µmol kg−1). Thereafter, at the end of the early deglaciation (15-16.5 ka BP) (Figure 6b), deep-water [CO32−] in Atlantic CDW increased and showed values similar to AABW (90 µmol kg−1). This seems to have been caused by enhanced transport of Pacific LCDW into the South Atlantic, due to strengthened Drake Passage flow. In addition, re-juvenated carbon-poor waters originating from upwelled northern-sourced PDW were transported into Atlantic CDW. At this period, the DIC estimate of Pacific LCDW decreases (Figure 4a), suggesting that better ventilated, low DIC deep-water was exported to the Atlantic and mixed into Atlantic LCDW and AABW (Figure 6b).
Our quantitative reconstruction of deep-water [CO32−] based on the depth transect in the SE Pacific, a central junction point connecting the Pacific and the Atlantic Ocean, suggests that dynamic, transient variations in deep-water structure and carbon chemistry characteristics largely shape and determine the amount of carbon stored in the Southern Ocean. In particular, the volumetric expansion of Pacific CDW with high [CO32−], caused by the deglacial reconfiguration of abyssal to bathyal waters in the Southern Ocean likely contributed to the reduction of the carbon storage in CDW, and was followed by effective carbon discharge due to enhanced surface ocean ventilation of deep-water masses.
Our [CO32−] reconstruction also enables us to roughly estimate changes in the bathyal to abyssal carbon budget. During the LGM to early deglaciation (15-19 ka BP), atmospheric pCO2 increased from 180 to 210 ppm, which corresponds to carbon efflux of approximately 63 GtC to the atmosphere. The [CO32−] reconstructions for our bathyal site PC03 show a ca. 20 ± 9 µmol kg−1 increase in the SE Pacific, which corresponds to a 40 ± 15 µmol kg−1 reduction in DIC under the assumption that bottom water parameters (Total Alkalinity, Temperature and Salinity) were stable, analog to the modern condition. If deep-water DIC of LCDW in the South Pacific (assumption of water mass distribution range: 40°S-70°S, 135°W-150°W, 3000-4000 m water depth) was uniformly reduced by 40 µmol kg−1, the amount of carbon efflux from this water mass can be estimated as ca.12 GtC. This corresponds to 19±7% of the total carbon emission to the atmosphere during this period. This estimate is based on the assumption that deep-water parameters are stable after the LGM to modern. In fact, however, ocean alkalinity varies on glacial-interglacial time scale, and deep-water alkalinity is generally higher in glacials than interglacials45. Previous proxy studies suggested that deep Southern Ocean alkalinity was ∼25 µmolkg−1 higher in glacials than in interglacials46. In addition, box model studies estimated higher alkalinity values of ∼300 µmol kg−1 in the deep Southern Ocean during glacial periods47. Assuming glacial alkalinity of the deep ocean was higher than under modern conditions, deep-water DIC would also be also higher than under the assumption of stable alkalinity, suggesting that the amount of carbon efflux from Pacific LCDW after the LGM is likely higher than 12 GtC. Therefore, our results imply that the amount of carbon corresponds to 19% or more of the total amount of carbon discharge from ocean to atmosphere during the early deglaciation, was emitted from Pacific LCDW during this period. Lastly, we suggest that carbon storage in South Pacific LCDW played an indispensable role in the atmospheric pCO2 rise during the early deglaciation.