ARTICLES https://doi.org/10.1038/s41561-019-0450-3 Reduced continental weathering and marine calcification linked to late Neogene decline in atmospheric CO2 Weimin Si 1,2* and Yair Rosenthal1,2 The globally averaged calcite compensation depth has deepened by several hundred metres in the past 15 Myr. This deepening has previously been interpreted to reflect increased alkalinity supply to the ocean driven by enhanced continental weathering due to the Himalayan orogeny during the late Neogene period. Here we examine mass accumulation rates of the main marine calcifying groups and show that global accumulation of pelagic carbonates has decreased from the late Miocene epoch to the late Pleistocene epoch even though CaCO3 preservation has improved, suggesting a decrease in weathering alkalinity input to the ocean, thus opposing expectations from the Himalayan uplift hypothesis. Instead, changes in relative contributions of coccoliths and planktonic foraminifera to the pelagic carbonates in relative shallow sites, where dissolution has not taken its toll, suggest that coccolith production in the euphotic zone decreased concomitantly with the reduction in weathering alkalinity inputs as registered by the decline in pelagic carbonate accumulation. Our work highlights a mechanism whereby, in addition to deep-sea dissolution, changes in marine calcification acted to modulate carbonate compensation in response to reduced weathering linked to the late Neogene cooling and decline in atmospheric partial pressure of carbon dioxide. arth’s climate in the Neogene (~23–2.58 million years ago and Berner’s hypothesis2 have different predictions as to the change (Ma)) was characterized by successive cooling steps that culmi- in global CaCO3 burial through the Neogene. Berner’s hypoth- Enated in Quaternary glaciations1. The origin of this long-term esis predicts high alkalinity and burial fluxes in the early Neogene cooling, however, remains debatable. On geological timescales, the decreasing to the present, whereas Raymo’s hypothesis predicts the atmospheric partial pressure of carbon dioxide ( pCO2 ), considered opposite. These trends should, in principle, be reflected in the CCD. the prime climate forcing, is regulated through the Ibalance between For the past 15 Myr, the globally averaged CCD has deepened by volcanic/metamorphic outgassing, silicate weathering and organic ~500 m (ref. 13). This deepening has been interpreted as increased matter burial2. In the Neogene, seafloor spreading rates and, by carbonate burial fluxes in response to higher riverine alkalinity inference, outgassing rates appear to have been relatively constant3, input as the result of enhanced chemical weathering from mountain 4,5 leading to the corollary that the Neogene decrease in pCO2 and cli- building . However, the CCD-based interpretation relies critically mate cooling may have been primarily driven by enhancedI weather- on the assumption of constant carbonate production. Instead, we ing; it has been argued that the Neogene uplift of the Himalayas may propose that environmentally driven changes in the production of have enhanced rock weathering due to increased exposure of weath- marine calcifiers have also played an important role in the long- erable rock surfaces4,5. The weathering hypothesis (Raymo’s hypoth- term calcite compensation, thereby modulating the pelagic carbon- esis), however, is controversial because it argues that weathering can ate burial rate. This supports recent model results suggesting that act as a forcing rather than a stabilizing feedback of Earth’s thermo- the CCD alone is insufficient to constrain either the dissolution or stat2,6,7. The alternative hypothesis (Berner’s hypothesis) relates high the accumulation of pelagic carbonate14. Therefore, instead of recon- weathering fluxes to the impact of high atmospheric pCO2 and warm structing the palaeo-CCD, as was done previously, we reconstruct surface temperatures, which provide a necessary feedbackI to stabi- late Neogene (0–15 Ma) pelagic carbonate production and dissolu- lize climate2. Accordingly, Neogene cooling should have resulted in tion by documenting mass accumulation rates of carbonate sediment reduced, rather than enhanced, continental weathering. (MARc) and its main producers, coccolithophores (MAR-coccolith) In a steady-state ocean, weathering fluxes of dissolved solutes and planktonic foraminifera (MAR-foram), from various ocean basins to the sea must be balanced by output fluxes. With regard to the and depths (Fig. 1), particularly along depth transects where the ocean’s carbonate alkalinity budget, the balance between riverine effects of production and dissolution can be distinguished. Because alkalinity input and CaCO3 burial is achieved through CaCO3 disso- coccolithophores and foraminifera are different in terms of their lution. Pelagic calcifying species, planktonic foraminifera and coc- ecology (autotrophic versus heterotrophic) and calcification pro- –1 colithophores produce today ~50 Tmol yr carbonates in the upper cesses, distinguishing between the histories of MAR-coccolith and 8–10 ocean , consuming approximately three times the alkalinity that MAR-foram is critical to an improved understanding of long-term is made available to the ocean from rivers (~33 Teq yr–1 (ref. 11)). To carbonate cycle changes. Application of this approach to five deep- restore the ocean’s alkalinity budget, dissolution will take place in sea cores selected above the lysocline (<2.6 km depth) suggests that the deep sea. The balancing process is arguably manifested in the decreasing pelagic carbonate burial rates over the late Neogene were 12 calcite compensation depth (CCD) , below which seawater is suf- probably driven by decreasing MAR-coccolith, possibly in response to 4 15 ficiently undersaturated to dissolve all CaCO3. Raymo’s hypothesis a decline in continental weathering . Here we extend the study to 1Department of Earth and Planetary Sciences, Rutgers University, Piscataway, NJ, USA. 2Department of Marine and Coastal Sciences, Rutgers University, New Brunswick, NJ, USA. *e-mail: [email protected] NATURE GEoscIENCE | www.nature.com/naturegeoscience ARTICLES NATURE GEOSCIENCE 90° N 100 60° N 982 80 607 1208 608 Seafloor %CaCO 558 30° N 999 60 758 807 803 925 711 U1337 0° 846 709 710 806 804 U1338 928 667 Latitude 707 708 40 3 588 1237 (wt) 754 1264 30° S 590 1266 593 1088 1 171 20 60° S 90° S 0 30° E 60° E 90° E 120° E 150° E 180° 150° W 120° W 90° W 60° W 30° W 0° 30° E Longitude 40 Fig. 1 | Deep-sea sites plotted with gridded seafloor %CaCO3 (ref. ). We focus on open ocean sites from mid and low latitudes that are ecologically favoured by pelagic calcifiers and where the seafloor is above the CCD. Pelagic settings that satisfy both criteria account for most pelagic carbonate accumulation9,40. The map was prepared using Ocean Data View41. more than 30 sites to test the globality of the suggested changes and CCD reconstruction of a deepening towards the Pleistocene. provide stronger constraints on its causes and implications to the Higher Pleistocene MARc at site 1266, however, may be partially carbon cycle. due to winnowing and redeposition (Supplementary Information). The equatorial Indian Ocean exhibits the lowest pelagic MARc –2 –1 –2 –1 Changes in MARc over the past 15 Myr (<1.5 g cm kyr ) and the smallest variations (<0.5 g cm kyr ) of Among the MARc records we investigated, many show a decreas- all basins throughout the late Neogene (Supplementary Fig. 1). ing trend following the late Miocene productivity pulse to the late Pleistocene, consistent with previous studies15,16. To summarize the Dissolution of deep-sea carbonate during the past 15 Myr trends, we average the MARc in three approximately evenly spaced Decreases in MARc in the Pacific and the Atlantic can be due to intervals, that is, Pleistocene, 0–2.5 Ma, late Miocene, 5.3–7.5 Ma, increased dissolution and/or decreased production. We examine and middle Miocene, 11.5–13.5 Ma (Fig. 2 and Supplementary the carbonate preservation along a depth transect on the Ontong- Table 3). Multimillion-year averages are used because they presum- Java Plateau in the western equatorial Pacific (sites 806–804) where ably reflect long-term steady states and minimize age model uncer- the deep water likely represents the mean ocean. The CaCO3 con- tainties. Because a range of processes can affect local carbonate tent at this transect is ~90% between ~15 and 6 Ma decreasing to accumulation, temporal variations of MARc from the middle–late ~80% during the past 6 Myr (Fig. 3a). Given relatively constant Miocene to the Pleistocene are expected to be spatially heteroge- non-carbonate MAR (~0.2–0.4 g cm–2 kyr–1, Supplementary Fig. 1), neous. Two global features, however, stand out. this decrease may be interpreted as increased dissolution in the past First, MARc have decreased significantly in relatively shallow 6 Myr, particularly at deep site 804 (3,800 mbsl). sites (<3,000 metres below sea level (mbsl)) but show little varia- The preservation of the planktonic foraminifera, however, tion in deeper sites. This is best illustrated by two depth transects indicates the opposite. We introduce a coccolith-free size dis- in the equatorial Pacific (Fig. 2a,b). In the western equatorial solution index (CF-size index: weight ratio of >60 µm/>20 µm) Pacific, MARc of the shallowest site (site 806) decreased from to qualitatively evaluate the preservation of planktonic foramin- >4 g cm–2 kyr–1 in the Miocene to ~1.7 g cm–2 kyr–1 in the Pleistocene. ifera (Supplementary Information). Compared with the conven- 19–22 Concomitantly, MARc in the shallow central equatorial Pacific tional dissolution index based on coarse fraction content , the decreased from ~2–3 g cm–2 kyr–1 to ~1 g cm2 kyr–1. By contrast, CF-size index examines only the foraminiferal fraction (>20 µm) MARc of both regions show little variation at ~4,000 mbsl, consis- by excluding the coccolith fraction (<20 µm) that may potentially tent with a previous reconstruction suggesting relatively constant bias dissolution estimates due to changes in coccolith produc- CCD in the Pacific since 15 Ma (ref.