Between about 1.25 and 0.7 million years ago, Earth's glacial cycles switched from a dominantly 41,000-year rhythm to the ~100,000-year cycles of the late Pleistocene, although the pacing of Earth's orbit did not change; the cause must be internal to the Earth system and remains unresolved. A leading hypothesis is that the glacial ocean stored more respired carbon, lowering atmospheric CO2 and enabling larger ice sheets. Benthic foraminiferal B/Ca, Cd/Ca and neodymium-isotope evidence shows that the deep Atlantic gained roughly 50 Gt of carbon during Marine Isotope Stages 24–22 (~950–870 thousand years ago), but this estimate assumes unchanged deep-ocean alkalinity; the partition between Southern Ocean nutrient utilisation and deep-ocean stratification is undetermined; the implied 20–40 ppm glacial CO2 decline cannot yet be tested against a direct, high-resolution atmospheric record; and the proposed Antarctic buoyancy-forcing trigger is only indirectly constrained. The problem is to determine the magnitude, mechanism and atmospheric expression of glacial ocean carbon storage across the Mid-Pleistocene Transition — the carbon-cycle feedback hypothesised to have produced the 100,000-year ice-age world.

Background

Earth's climate has oscillated between glacial and interglacial states for the past ~3 million years, paced by periodic variations in the geometry of Earth's orbit around the Sun (Milankovitch forcing). These oscillations are recorded in the oxygen-isotope composition (δ18O\delta^{18}\mathrm{O}) of benthic (seafloor-dwelling) foraminifera, which tracks global ice volume and deep-water temperature. From ~3 to 1.25 million years ago (Ma) the dominant pacing was a 41,000-year (kyr) cycle matching the obliquity (axial tilt) of Earth's orbit; during the Mid-Pleistocene Transition (MPT; ~1,250–700 thousand years ago, ka) the cycles lengthened to ~100 kyr, roughly doubled in amplitude, and acquired a sawtooth shape — with no corresponding change in orbital forcing. Because the trigger must therefore be internal to the Earth system, the origin of the "100-kyr world" is one of the longest-standing open problems of paleoclimate science. Successive climate states are numbered as Marine Isotope Stages (MIS); MIS 24–22 (~950–870 ka) spans the first extreme glacial cycle of the MPT.

A leading class of explanations invokes atmospheric carbon dioxide. Air trapped in Antarctic ice shows glacial CO2 of ~180 ppm and interglacial values of ~280 ppm over the past 800 kyr; a lower glacial CO2 level emerging at the MPT would cool the planet and permit larger ice sheets. The ocean holds ~60 times the atmosphere's carbon as dissolved inorganic carbon (DIC — the sum of dissolved CO2, bicarbonate and carbonate ions), so glacial CO2 decline must be accounted for mainly by ocean carbon storage. Three ocean processes dominate the hypotheses: (i) the biological pump — the photosynthetic export of organic matter to depth, whose efficiency in the Southern Ocean depends on nutrient utilisation, i.e. the fraction of upwelled major nutrients (phosphate, nitrate) consumed before the water returns to depth, which can be raised by longer surface residence times, by dust-borne iron relieving iron limitation, or by northward shifts of the oceanic fronts; (ii) deep-ocean stratification — density layering that isolates abyssal "respired" carbon (generated by remineralisation of exported organic matter) from the surface; and (iii) the meridional overturning circulation (MOC), whose Atlantic arm carries North Atlantic Deep Water (NADW) southward at depth against the northward spread of Southern-Ocean-sourced bottom water (often called Antarctic Bottom Water, AABW). The Southern Ocean is the principal window through which deep carbon re-equilibrates with the atmosphere, so its surface buoyancy forcing (freshwater input, sea ice, winds) and its biological pump together set the strength of the deep-ocean carbon trap.

Reconstructions rely on geochemical proxies measured on benthic foraminifera: B/Ca ratios track the carbonate-ion saturation state of deep water (Δ[CO32−]\Delta[\mathrm{CO_3^{2-}}], which falls as DIC rises at fixed alkalinity); Cd/Ca tracks dissolved phosphate (cadmium mimics nutrient cycling); and neodymium isotopes (εNd\varepsilon_{\mathrm{Nd}}) track water-mass provenance because NADW and AABW carry distinct isotope signatures. Converting these proxies into absolute DIC changes requires knowing total alkalinity (TA, the ocean's acid–base buffering capacity), which evolves through carbonate compensation — the dissolution or burial of seafloor calcium carbonate that restores carbonate equilibrium over several thousand years.

In this setting, Farmer, Hönisch, Haynes, Kroon, Jung, Ford, Raymo, Jaume-Seguí, Bell, Goldstein, Pena, Yehudai and Kim (2019; doi:10.1038/s41561-019-0334-6) combined paired B/Ca, Cd/Ca and εNd\varepsilon_{\mathrm{Nd}} records from deep South Atlantic Ocean Drilling Program Site 1267 with published North Atlantic Site 607 data and showed that between 950 and 900 ka the carbonate-ion saturation fell by ~30 µmol/kg, phosphate rose by ~0.5 µmol/kg, and the NADW contribution to the abyssal South Atlantic dropped by ~20%. Interpreted as DIC, these signals imply that the glacial deep Atlantic gained 42±1542\pm15 µmol/kg of carbon (about 50 Gt C below 3 km depth) during the transition to 100-kyr cycles, with a corresponding interglacial increase of 30±1630\pm16 µmol/kg. The authors proposed that East Antarctic Ice Sheet growth around ~1,000 ka altered Southern Ocean surface buoyancy forcing, increasing both nutrient utilisation and deep stratification, so that — combined with the weakened overturning — the deep ocean stored enough carbon to lower glacial atmospheric pCO2 by 20–40 ppm and thereby enable the larger ice sheets of the 100-kyr world. The same B/Ca-based approach had earlier identified at least 50 Gt of deep-Atlantic carbon sequestration at the onset of the last glaciation (Yu et al. 2016; doi:10.1038/ngeo2657), establishing the Last-Glacial-Maximum baseline for the method.

The quantitative claim rests on acknowledged weak points. The B/Ca-to-DIC conversion assumes deep-ocean alkalinity did not change (ΔTA=0\Delta\mathrm{TA}=0), even though the MPT visibly reorganised deep-sea carbonate sedimentation; the Atlantic estimate pools only two sites; and the predicted CO2 decline could be compared only with sparse evidence — a boron-isotope sea-surface pCO2 record with roughly two glacial and two interglacial points across MIS 24–22 (Hönisch et al. 2009; doi:10.1126/science.1171477) and a discontinuous ~1 Ma "blue ice" record from the Allan Hills, Antarctica, spanning only about half of the late-Pleistocene CO2 range (Higgins et al. 2015; doi:10.1073/pnas.1420232112). An orbitally resolved boron-isotope CO2 record across the MPT (Chalk et al. 2017; doi:10.1073/pnas.1702143114) attributes the growth of the glacial–interglacial CO2 amplitude mainly to dust-borne iron fertilization of the Southern Ocean — an attribution within the nutrient-utilisation pathway that differs from the stratification emphasis of later carbon-storage work. Subsequent studies have contested the magnitude of the circulation change (Hines et al. 2024; doi:10.1126/science.adn4154, finding no evidence for a substantial MPT deep-circulation "crisis" in Cape Basin paired Nd–C–O isotopes) and the reliability of empirical benthic-proxy-to-DIC conversions in general (Adloff et al. 2025; doi:10.1029/2025PA005125, finding large prediction errors in Earth-system-model tests). These contestations define the open problem stated below.

Problem Statement

The problem is to determine, quantitatively and mechanistically, the ocean carbon-storage change that accompanied the Mid-Pleistocene Transition — its magnitude, its Southern Ocean pathway partition, its expression in atmospheric CO2, and its trigger — and thereby to test whether deep-ocean carbon storage was the internal feedback that lowered glacial CO2 enough to enable the 100,000-year glacial world. The objective requires jointly resolving:

  1. Magnitude with a defensible alkalinity treatment. The timing and magnitude of the glacial (and, where records allow, interglacial) DIC increase in the deep Atlantic — and ideally the abyssal ocean generally — across MIS 24–22 and the wider MPT, with the carbonate-compensation and alkalinity evolution (regional versus global changes in carbonate burial and dissolution) constrained well enough that proxy-derived ΔDIC\Delta\mathrm{DIC} estimates no longer rest on the assumption of unchanged deep-water total alkalinity.

  2. Pathway partition. The relative contributions to that carbon accumulation of (i) enhanced Southern Ocean nutrient utilisation — biological-pump efficiency, surface residence time, frontal and upwelling shifts, iron fertilization — versus (ii) increased deep-ocean stratification, together with the actual magnitude of any reduction in North Atlantic Deep Water production, which recent paired isotope work disputes.

  3. Atmospheric CO2 test. A comparison of the predicted glacial CO2 decline (20–40 ppm) in both timing and magnitude against the best available atmospheric CO2 constraints for MIS 24–22 — the proxy and blue-ice records, with their uncertainties quantified — and a statement of what remains unresolved because a direct, well-resolved ice-core-quality record does not yet exist.

  4. Trigger constraint. Direct empirical constraints on the hypothesised link between Antarctic ice-sheet expansion (timing and geometry) and Southern Ocean surface buoyancy forcing that would set off the circulation and biogeochemical response — as distinct from the competing Northern-Hemisphere (regolith/erosion) trigger.

These four requirements are evidence-requirements of one scientific objective — attributing the MPT carbon-cycle change — not independent problems. The authors also raised a further, distinct question about how a future weakening of the Atlantic overturning circulation would interact with the biological pump and anthropogenic CO2 uptake; that future-projection question is not part of the present problem.

References

  1. Farmer, J. R., Hönisch, B., Haynes, L. L., Kroon, D., Jung, S., Ford, H. L., Raymo, M. E., Jaume-Seguí, M., Bell, D. B., Goldstein, S. L., Pena, L. D., Yehudai, M., Kim, J. (2019). Deep Atlantic Ocean carbon storage and the rise of 100,000-year glacial cycles. Nature Geoscience 12: 355–360. doi:10.1038/s41561-019-0334-6. https://doi.org/10.1038/s41561-019-0334-6
  2. Hönisch, B., Hemming, N. G., Archer, D., Siddall, M., McManus, J. F. (2009). Atmospheric carbon dioxide concentration across the mid-Pleistocene transition. Science 324(5934): 1551–1554. doi:10.1126/science.1171477. https://doi.org/10.1126/science.1171477
  3. Higgins, J. A., Kurbatov, A. V., Spaulding, N. E., Brook, E., Introne, D. S., Chimiak, L. M., Yan, Y., Mayewski, P. A., Bender, M. L. (2015). Atmospheric composition 1 million years ago from blue ice in the Allan Hills, Antarctica. Proceedings of the National Academy of Sciences 112(22): 6887–6891. doi:10.1073/pnas.1420232112. https://doi.org/10.1073/pnas.1420232112
  4. Chalk, T. B., Hain, M. P., Foster, G. L., Rohling, E. J., Sexton, P. F., Badger, M. P. S., Cherry, S. G., Hasenfratz, A. P., Haug, G. H., Jaccard, S. L., Martínez-García, A., Pälike, H., Pancost, R. D., Wilson, P. A. (2017). Causes of ice age intensification across the Mid-Pleistocene Transition. Proceedings of the National Academy of Sciences 114(50): 13114–13119. doi:10.1073/pnas.1702143114. https://doi.org/10.1073/pnas.1702143114
  5. Yu, J., Menviel, L., Jin, Z. D., Thornalley, D. J. R., Barker, S., Marino, G., Rohling, E. J., Cai, Y., Zhang, F., Wang, X., Dai, Y., Chen, P., Broecker, W. S. (2016). Sequestration of carbon in the deep Atlantic during the last glaciation. Nature Geoscience 9: 319–324. doi:10.1038/ngeo2657. https://doi.org/10.1038/ngeo2657
  6. Yehudai, M., Kim, J., Pena, L. D., Jaume-Seguí, M., Knudson, K. P., Bolge, L., Malinverno, A., Bickert, T., Goldstein, S. L. (2021). Evidence for a Northern Hemispheric trigger of the 100,000-y glacial cyclicity. Proceedings of the National Academy of Sciences 118: e2020260118. doi:10.1073/pnas.2020260118. https://doi.org/10.1073/pnas.2020260118
  7. Thomas, N. C., Bradbury, H. J., Hodell, D. A. (2022). Changes in North Atlantic deep-water oxygenation across the Middle Pleistocene Transition. Science 377(6606): 654–659. doi:10.1126/science.abj7761. https://doi.org/10.1126/science.abj7761
  8. Qin, B., Jia, Q., Xiong, Z., Li, T., Algeo, T. J., Dang, H. (2022). Sustained Deep Pacific Carbon Storage After the Mid-Pleistocene Transition Linked to Enhanced Southern Ocean Stratification. Geophysical Research Letters 49: e2021GL097121. doi:10.1029/2021GL097121. https://doi.org/10.1029/2021GL097121
  9. Yamamoto, M., Clemens, S. C., Seki, O., Tsuchiya, Y., Huang, Y., O'ishi, R., Abe-Ouchi, A. (2022). Increased interglacial atmospheric CO2 levels followed the mid-Pleistocene Transition. Nature Geoscience 15: 307–313. doi:10.1038/s41561-022-00918-1. https://doi.org/10.1038/s41561-022-00918-1
  10. Marcks, B. A., Dos Santos, T. P., Lessa, D. V. O., Cartagena-Sierra, A., Berke, M. A., Starr, A., Hall, I. R., Kelly, R. P., Robinson, R. S. (2023). Glacial Southern Ocean Expansion Recorded in Foraminifera-Bound Nitrogen Isotopes From the Agulhas Plateau During the Mid-Pleistocene Transition. Paleoceanography and Paleoclimatology 38: e2022PA004482. doi:10.1029/2022PA004482. https://doi.org/10.1029/2022PA004482
  11. Hines, S. K. V., Charles, C. D., Starr, A., Goldstein, S. L., Hemming, S. R., Hall, I. R., Lathika, N., Passacantando, M., Bolge, L. (2024). Revisiting the mid-Pleistocene transition ocean circulation crisis. Science 386(6719): 681–686. doi:10.1126/science.adn4154. https://doi.org/10.1126/science.adn4154
  12. Martin, J. R. W., Pedro, J. B., Vance, T. R. (2024). Predicting trends in atmospheric CO2 across the Mid-Pleistocene Transition using existing climate archives. Climate of the Past 20: 2487–2497. doi:10.5194/cp-20-2487-2024. https://doi.org/10.5194/cp-20-2487-2024
  13. Scherrenberg, M. D. W., Berends, C. J., van de Wal, R. S. W. (2025). CO2 and summer insolation as drivers for the Mid-Pleistocene Transition. Climate of the Past 21: 1061–1077. doi:10.5194/cp-21-1061-2025. https://doi.org/10.5194/cp-21-1061-2025
  14. Adloff, M., Jeltsch-Thömmes, A. J., Rae, J. W. B., Pöppelmeier, F., Trudgill, M., Stocker, T. F., Joos, F. (2025). Evaluating Glacial CO2 System Reconstructions From Benthic Proxies in Earth System Simulations. Paleoceanography and Paleoclimatology 40: e2025PA005125. doi:10.1029/2025PA005125. https://doi.org/10.1029/2025PA005125
  15. Garity, M., Lund, D., Jerris, H., McBride, J. (2025). Progressively greater biological carbon storage in the deep Atlantic during glacial inception. Proceedings of the National Academy of Sciences 122(32): e2510171122. doi:10.1073/pnas.2510171122. https://doi.org/10.1073/pnas.2510171122
  16. Basak, C., Kapuge, A. K. I. U., Farmer, J. R., Symes, E., Middleton, J. L., Gottschalk, J., Arz, H. W., Lamy, F., Winckler, G., Cruz, A. P. S. (2026). Southern Hemisphere initiation of the mid-Pleistocene transition. Science Advances 12: eaea6811. doi:10.1126/sciadv.aea6811. https://doi.org/10.1126/sciadv.aea6811
  17. Starr, A., Hall, I. R., Ninnemann, U. S., Barker, S., Charles, C. D., Nederbragt, A., Owen, L., Tangunan, D., Zhang, X. (2026). Southern Ocean Vertical Carbon Stratification Across the Mid-Pleistocene Transition. Paleoceanography and Paleoclimatology 41: e2024PA005090. doi:10.1029/2024PA005090. https://doi.org/10.1029/2024PA005090
  18. Thomas, N. C., Ford, H. L., Greaves, M., Hodell, D. A. (2026). Increased abyssal ocean density stratification across the Middle Pleistocene Transition. Climate of the Past 22: 1559–1584. doi:10.5194/cp-22-1559-2026. https://doi.org/10.5194/cp-22-1559-2026
  19. Yun, K.-S., Timmermann, A. (2026). Increased sensitivity of the Antarctic Ice Sheet to decreasing CO2 across the Mid-Pleistocene Transition. Nature Geoscience 19: 683–690. doi:10.1038/s41561-026-01979-2. https://doi.org/10.1038/s41561-026-01979-2