Deep ocean carbonate chemistry and glacial-interglacial atmospheric CO 2 changes

USAGE Permission is granted to copy this article for use in teaching and research. Republication, systematic reproduction, or collective redistribution of any portion of this article by photocopy machine, reposting, or other means is permitted only with the approval of The Oceanography Society. Send all correspondence to: info@tos.org or The Oceanography Society, PO Box 1931, Rockville, MD 20849-1931, USA. Oceanography THE OFFICIAL MAGAZINE OF THE OCEANOGRAPHY SOCIETY

Deepwater carbonate ion concentration, [CO 3 2-], is a critical parameter in investigations of carbon reorganization in the climate system.To a rst approximation, [CO 3 2-] ≈ ALK -DIC (Zeebe and Wolf-Gladrow, 2001), where ALK is alkalinity and DIC is dissolved inorganic carbon both of which a ect atmospheric CO 2 (Figure 2).ALK represents the excess base that a ects deprotonation of aqueous CO 2 into bicarbonate ion (HCO 3 -) and CO 3 2-.Everything else being equal, an increase in ALK causes more DIC to take the form of CO 3 2-and less of it to be aqueous CO 2 , lowering CO 2 partial pressure (Figure 2B) and e ectively increasing CO 2 solubility in seawater and decreasing atmospheric CO 2 .Deepwater

MODERN OCEAN CO 3 2  VARIABILIT Y INTRODUCTION
One of the most intriguing ndings in paleoclimate research concerns systematic glacial-interglacial atmospheric CO 2 variations, as revealed by measurements of air bubbles trapped in ice cores from Antarctica.During the last 800,000 years, atmospheric CO 2 uctuations have been highly correlated with climate on millennial time scales (Figure 1; Lisiecki and Raymo, 2005;Jouzel et al., 2007;Lüthi et al., 2008).Because CO 2 is a greenhouse gas, atmospheric CO 2 uctuations likely caused, or at least ampli ed, the climatic changes.us, resolving the reasons for glacial-interglacial atmospheric CO 2 uctuations is essential for improved understanding of the mechanisms that control the global climate system.Despite intensive research, the mechanisms for past atmospheric CO 2 changes remain elusive.Because the ocean is the largest carbon reservoir that equilibrates with the atmosphere on a ~ 1,000 year time scale, Broecker (1982) argued that the ocean must have played a critical role in modulating past atmospheric CO 2 variations.Changes in physical conditions, including temperature, salinity, and ocean volume (due to waxing and waning of ice sheets; Figure 1), have opposing e ects, resulting in a net decrease in atmospheric CO 2 by ~ 15 ppm during glacials (Sigman and Boyle, 2000;Brovkin et al., 2007). is decrease is barely enough to counter the atmospheric CO 2 rise that is due to the transport of ~ 500 Gt (1 Gt = 10 15 g) of carbon from the land biosphere to the ocean during interglacial to glacial transitions (Bird et al., 1994;Ciais et al., 2012).To explain up to ~ 100 ppm glacial-interglacial atmospheric CO 2 uctuations requires more complex ocean processes.Existing evidence suggests that biogeochemistry and physical Figure 1.Records of (A) atmospheric pCO 2 (Lüthi et al., 2008) compared with (B) Antarctic air temperature changes (Jouzel et al., 2007) and (C) benthic foraminiferal δ 18 O stack curve (Lisiecki and Raymo, 2005)  (C organic :C CaCO3 ) of 4:1 (Broecker and Peng, 1982), photosynthesis consumes seawater DIC and ALK at a ratio of 3.58:1 (Yu et al., 2008) and leads to an increase in seawater [CO 3 2-] (Figure 2A).
Di erent from other regions, the Southern Ocean exhibits reduced surface-deep water [CO 3 2-] contrast (Figure 3A), due to intensive upwelling of very deep waters and ine cient nutrient utilization (Broecker et al., 1998).
Sinking and subsequent decomposition of biological matter decreases [CO 3 2-] in the deep ocean (Figure 2A), but di erent water masses bear distinctive values that re ect their di erent histories (Figure 3).  2] (unit: µmol kg -1 ) and pCO 2 (unit: µatm) are calculated at temperature of 25°C, salinity of 35, pressure of 1 atm (water depth of 0 m).CaCO 3 dissolution increases seawater ALK and DIC in a ratio of 2:1, with a net effect to raise seawater [CO 3 2-] and decrease pCO 2 ; CaCO 3 formation has the opposite effects.Invasion and release of CO 2 into/from the ocean only affects seawater DIC.Photosynthesis mainly consumes seawater DIC and slightly increases seawater ALK due to the uptake of nitrate; respiration has the reverse effect.requirement for analyses (Hönisch et al., 2008;Yu et al., 2010b;Rae et al., 2011).
Some empirical approaches have also been developed to quantify deepwater [CO 3 2-].For example, the partitioning of Zn into benthic foraminiferal carbonates appears to be correlated with the degree of deepwater calcite satura- values below ~ 25 µmol kg -1 .However, the prerequisite to knowing past seawater [Zn], which is nontrivial, makes application of this method a challenge (Marchitto et al., 2000).
Here, we focus on a recently developed empirical proxy, which uses the benthic foraminiferal boron-to-calcium ratio (B/Ca).e method is supported by extensive core-top calibrations (Figure 4; Yu and Elder eld, 2007;Brown et al., 2011;Rae et al., 2011;Raitzsch et al., 2011;Yu et al., 2013a)  2-] and the saturation horizons in equatorial Atlantic and Pacific Oceans.e saturation horizon is defined as the depth where in situ [CO 3 2-] equals the saturation [CO 3 2-] (which is largely determined by pressure or water depth).Above the saturation horizon, seawater is saturated and CaCO 3 tends to be preserved, whereas below the saturation horizon, seawater is undersaturated and CaCO 3 has the potential to dissolve.Panel (A) is generated using Ocean Data View (Schlitzer, 2006).Data are from the GLobal Ocean Data Analysis Project (GLODAP) data set (Key et al., 2004).
During the LGM, the carbon lost from the atmosphere (~ 200 GtC) and the terrestrial biosphere (~ 500 GtC) must have been stored as DIC in the deep ocean (Broecker, 1982) et al., 2012) reduced the removal of ALK by reef aragonite on continental shelves (Opdyke and Walker, 1992) (Broecker and Peng, 1987).Reduced CaCO 3 burial introduces ALK and DIC in a 2:1 ratio into the ocean, causing oceanic [CO 3 2-] to rise (Figure 2A) until the ALK balance is restored on a time scale of ~ 5,000 years (Broecker and Peng, 1987).A net ALK gain occurred during glaciations.By increasing CO 2 solubility in seawater, increased ocean ALK helps to further lower atmospheric CO 2 during glaciations (Figure 2B; Broecker and Peng, 1987;Boyle, 1988).likely indicates an ALK decrease possibly associated with carbonate compensation (Broecker and Peng, 1987).e initial deepwater [CO 3 2-] increase due to CO 2 release during the early deglaciation would promote the preservation and burial of CaCO 3 in sediments, which is observed at many locations worldwide from 14,000-10,000 years ago (Berger, 1977;Farrell andPrell, 1989, 1991;Hodell et al., 2001;Anderson et al., 2008;Yu et al., 2010a).Improved CaCO 3 preservation during the last deglaciation in turn decreased the oceanic ALK inventory and, hence, lowered oceanic  1992).By reducing CO 2 solubility in the ocean, the decrease in the whole ocean ALK inventory drives up atmospheric CO 2 (Figure 2B) and contributes to the 20 ppm rise in atmospheric CO 2 since ~ 8,000 years ago (Figure 6G), as demonstrated by modeling (Ridgwell et al., 2003;Menviel and Joos, 2012).

The Last Deglaciation
S P EC I A L I S S U E O N C H A N G I N G O C E A N C H E M I S T R Y  PA S T R E C O R D S : PA L E O C E N E T O H O L O C E N E (inset photo) Benthic foram Cibicidoides wuellerstorfi.
water depth than at 5 km water depth in the equatorial Paci c (Figure 3C).Were this DIC increase solely due to remineralization of biological particles sinking from the surface, then ALK would be expected to rise by ~ 15 µmol kg -1 , which is less than the observed increase of ~ 40 µmol kg -1 . is discrepancy suggests an introduction of ALK into deep waters by CaCO 3 dissolution on the sea oor.Compared to the deep equatorial Atlantic Ocean, the observed ~ 25 µmol kg -1 decrease in deep equatorial Paci c [CO 3 2-] corresponds to shoaling of the calcite saturation horizon by ~ 1.3 km (Figure 3B), deterioration of the calcite saturation degree, and promotion of CaCO 3 dissolution in the Paci c Ocean.As a consequence, the calcite lysocline-the depth where CaCO 3 dissolution intensi es

Figure
Figure 2. Effects of various processes on (A) carbonate ion concentration, [CO 32-], and (B) pCO 2 in ALK-DIC space (ALK = alkalinity and DIC = dissolved inorganic carbon).Seawater [CO 32-] (unit: µmol kg -1 ) and pCO 2 (unit: µatm) are calculated at temperature of 25°C, salinity of 35, pressure of 1 atm (water depth of 0 m).CaCO 3 dissolution increases seawater ALK and DIC in a ratio of 2:1, with a net effect to raise seawater [CO 32-] and decrease pCO 2 ; CaCO 3 formation has the opposite effects.Invasion and release of CO 2 into/from the ocean only affects seawater DIC.Photosynthesis mainly consumes seawater DIC and slightly increases seawater ALK due to the uptake of nitrate; respiration has the reverse effect. Photosynthesis Figure 3. (A) Distribution of seawater [CO 3 2-] in the global modern ocean.Four water masses are indicated: NADW = North Atlantic Deep Water, AAIW = Antarctic Intermediate Water, LCDW = Lower Circumpolar Deep Water, and NPDW = North Pacific Deep Water.Inset shows the locations of hydrographic sites used for the [CO 3 2-] section.Circles and diamonds show, respectively, the locations of cores for deep and surface water [CO 3 2-] reconstructions displayed in Figure 6.Vertical dashed lines indicate locations of hydrographic sites for depth profiles shown in (B-D).(B) Bathymetric profile of [CO 3 2-] with salinity normalized to account for effects of mixing and evaporation-precipitation. (C) and (D) Bathymetric profiles, respectively, of DIC and ALK in the equatorial Atlantic and Pacific Oceans.(B) also shows calcite saturation [CO 32-] and the saturation horizons in equatorial Atlantic and Pacific Oceans.e saturation horizon is defined as the depth where in situ [CO 32-] equals the saturation [CO 3 2-] (which is largely determined by pressure or water depth).Above the saturation horizon, seawater is saturated and CaCO 3 tends to be preserved, whereas below the saturation horizon, seawater is undersaturated and CaCO 3 has the potential to dissolve.Panel (A) is generated using Ocean Data View(Schlitzer, 2006).Data are from the GLobal Ocean Data Analysis Project (GLODAP) data set(Key et al., 2004).
shows, B/Ca in the species Cibicidoides wuellerstor and Cibicidoides mundulus shows clear linear relationships with deepwater Δ[CO 3 2-], and there appears to be no threshold in the B/Ca:Δ[CO 3 2-] sensitivity at high Δ[CO 3 2-], allowing reconstructions using shallow sites.Both C. wuellerstor and C. mundulus are epifaunal species, with a habitat above the sediment-seawater interface (Lutze and iel, 1989), which minimizes complications in B/Ca due to pore water chemistry.Because B is conservative in seawater, changes in the B concentration or inventory over time scales of 14-20 million years (Lemarchand et al., 2000) have limited in uence.Existing data suggest an absence of temperature and dissolution e ects on benthic B/Ca (Yu and Elder eld, 2007).In addition, the method requires small sample sizes, and C. mundulus also suggest biological in uences on B incorporation into benthic foraminiferal carbonate.Despite incomplete understanding of the mechanisms, a down-core comparison between benthic B/Ca and δ 11 B (Yu et al., 2010b) strongly supports the feasibility of the core-top calibrations for past deepwater [CO 3 2-] reconstructions as shown in Figure 4. Currently, calibrations are only available for a few species, and B/Ca in other benthic species remains to be explored.
Figure 5. Bathymetric distribution of seawater [CO 32-] in the North Atlantic and Indo-Pacific Oceans during the Late Holocene (0-5,000 years ago) and the Last Glacial Maximum (LGM; 18,000-22,000 years ago)(Yu et al., 2013a).

Figure
Figure 5 shows bathymetric distributions of benthic B/Ca-derived deepwater [CO 3 2-] in the North Atlantic and Indo-Paci c Oceans during the Last Glacial Maximum (LGM; 18,000-22,000 years ago) and Late Holocene (0-5,000 years ago) (Yu et al., 2013a).e large LGMto-Holocene [CO 3 2-] changes from about +20 to -35 µmol kg -1 in the North Atlantic Ocean are consistent with substantial changes in the geometry of Atlantic Meridional Ocean Circulation.During the LGM, Glacial North Atlantic Intermediate Water (GNAIW) was characterized by low nutrient content, high δ 13 C, and high [CO 3 2-] values, and sank to ~ 2.5 km in the North Atlantic (Boyle and Keigwin, 1987; Curry and Oppo, 2005; Yu et al., 2008).Below GNAIW were Antarctic-derived waters with high-nutrient, low-δ 13 C, and low-[CO 3 2-] properties (Figures 3A and 5).During the LGM, elevated [CO 3 2-] at the mid-depth range likely resulted from high surface preformed [CO 3 2-] in the North Atlantic (Figure 6A,B) (Henehan et al., 2013; Yu et al., 2013b), while decreased [CO 3 2-] in the deep North Atlantic is consistent with a greater penetration of low-[CO 3 2-] deep waters from the Southern Ocean (Figure 3A; Curry and Oppo, 2005).Because (1) the Atlantic is relatively small in volume (~ 25% of the global ocean), and (2) opposite changes in [CO 3 2-] at shallow and deep water masses tend to cancel each other, changes in Atlantic [CO 3 2-] have negligible impacts on the global mean deep ocean [CO 3 2-].In contrast, LGM-to-Holocene [CO 3 2-] changes in the deep Indo-Paci c Oceans (accounting for ~75% of the global
values that were similar to those observed in the modern deep sea (Figure 2A).Increased respiratory CO 2 in the glacial deep sea is consistent with reduced dissolved O 2 concentrations in the glacial deep Paci c Ocean resulted in shoaling of the saturation horizon, which increased CaCO 3 dissolution on the sea oor.In addition, the replacement of NADW with GNAIW allowed more corrosive Southernsourced deep waters (Figure 3A) to occupy the deep Atlantic, further strengthening CaCO 3 dissolution there (Figure 6D).Moreover, reduced coral growth during glacial times due to substantial shrinkage of shelf areas caused by ~ 120 m sea level drop (e.g., Grant

Figure
Figure 6 shows benthic B/Ca-derived deepwater [CO 3 2-] at four locations from the global deep ocean (middle panel), together with some surface water [CO 3 2-] (top panel) and atmospheric CO 2 and Antarctic temperature changes (bottom panel) during the last 25,000 years.During the early deglaciation (17,500-14,500 years ago), [CO 3 2-] rose by about 10 µmol kg -1 in the deep Atlantic, Indian, and equatorial Paci c Oceans (Figure 6D-F). is cannot observed in the Atlantic Ocean(Figure 6C,D).Because benthic δ 13 C (which is sensitive to biological respiration) has remained roughly constant (e.g.,Yu et al., 2010a), the Holocene [CO 3 2-] decline in the deep Indo-Paci c Oceans likely resulted from a depletion in oceanic ALK. is inferred ALK decrease is consistent with continued carbonate compensation in response to the carbon reorganization that occurred during the last deglaciation.Additionally, coral reef buildup on shelves due to an increase in the area of shallow water environments at high sea level stand may have contributed to the removal of ALK and resulted in a decrease in whole ocean [CO 3 2-] (Opdyke and Walker, strati cation, and sea ice retreat) has suggested to many that Southern Ocean processes are critical to carbon release from the deep ocean during glacial terminations.e deep ocean .Development of highresolution records since the last glacial and of reconstructions that span several glacial-interglacial cycles will improve and quantify our understanding of the role of deep ocean carbonate chemistry in past atmospheric CO 2 changes.