Analyses of boron isotopes in ancient marine carbonate sediments provide an enlightening perspective on the links between carbon dioxide and ice-cap cover at a climatically momentous time in Earth's history.
On page 1110 of this issue, Pearson et al.1 report how they have peered back to a time, around 33.5 million years ago, when Earth underwent a drastic interval of climatic change associated with the formation of the Antarctic ice cap. The authors' work provides a fund of data about the connection between climate and atmospheric levels of the greenhouse gas carbon dioxide.
Coupled records of past variations in temperature and atmospheric CO2 are precious: they provide fundamental information on climate dynamics that can help in predicting future change. These records are available for recent times from instrumental measurements and, going further back, from data provided by oxygen isotopes in ice cores and by CO2 trapped in air bubbles in such cores. The ice-core data take us back almost a million years2,3, but that still covers only the last few glacial–interglacial cycles, which were characterized by atmospheric CO2 concentrations of 170–300 parts per million by volume (p.p.m.v.) — lower than the present-day value of about 380 p.p.m.v.
To test climate theories in the present-day condition of massive carbon emissions to the atmosphere, what is needed is access to geological archives that provide clues about how past climate responded to CO2 levels higher than those of today. Analyses of oxygen isotopes in the shells of marine organisms that form carbonate sediments can provide proxy data on ancient global temperature and ice volume, in the same way that oxygen isotopes in ice samples provide such data for more recent times. But, so far, reconstructions of ancient levels of CO2 have been rare4, and they remain much needed.
This is where Pearson et al.1 come in. They have used boron isotopes in exceptionally well-preserved marine microfossils — the carbonate shells of organisms called foraminifera — as an indirect measurement of atmospheric CO2 during the Eocene–Oligocene climate transition that began about 34 million years ago. This was a time of dramatic climate perturbation, which saw the formation of the still-extant Antarctic ice cap, and which has been attributed to changes in the global carbon cycle5.
Boron has a physicochemical property that is particularly appropriate for reconstructing historical levels of atmospheric CO2. The relative abundance of its two stable isotopes (11B/10B, designated δ11B) in the shells of foraminifera is correlated with the pH of the uppermost ocean layers in which these organisms lived6,7. Because atmospheric CO2 is in chemical equilibrium with the pH of surface sea water, it is possible to infer past CO2 concentrations from boron isotopes in carbonate sediments8. By comparing their boron-isotope results with published oxygen data9 from samples of the same age, Pearson et al.1 reach two intriguing conclusions.
First, they conclude that the slow temperature decline recorded by oxygen isotopes was concomitant with a decline of atmospheric CO2 from about 1,100 p.p.m.v. to a threshold concentration of about 750 p.p.m.v., at which the main phase of Antarctic ice-cap growth was initiated. This finding confirms model predictions5 that — contrary to what might be expected — the initiation and the rapid expansion of the Antarctic ice sheet occurred about 33.5 million years ago at levels of atmospheric CO2 that were more than twice the present-day value. Pearson et al. propose that the Antarctic glaciation was preconditioned by the global cooling associated with the decline of atmospheric CO2. But the glaciation really started only when Earth's orbital parameters, which change periodically, favoured the process.
The authors' second conclusion is that, although the newly formed ice cap may have shrunk somewhat, it largely survived a subsequent and rapid recovery of atmospheric CO2 back to levels of 1,000 p.p.m.v. or more. Such a rise in CO2 after the main phase of ice-sheet growth is predicted by climate models10. But the boron isotopes indicate that it occurred within the following 50,000 years, which is faster than the model prediction. This disparity highlights a need for more refinement in modelling the carbon cycle and understanding its relationship with global climate.
The inferences drawn by Pearson et al.1 about relative variations in boron isotopes — and, hence, in the pH of 'palaeo-seawater' — are solid. But caveats must be mentioned about the extrapolation of boron-isotope data in determining the corresponding atmospheric CO2 concentration11. Reliable absolute values of seawater pH can be deduced from boron isotopes in shells of ancient marine organisms only once the δ11B value of sea water itself is known, which is not readily achieved for the situation 33.5 million years ago. Even when the pH of palaeo-seawater is correctly estimated, more information on seawater chemistry (in particular with respect to the dissolved carbonate species) is still required to deduce the corresponding atmospheric CO2 at equilibrium. Possible errors in doing so may arise from the method itself — which involves using fluid inclusions in salt deposits to reconstruct open-ocean chemistry12 — or from the determination of the ancient seawater saturation state with respect to carbonate13. The boron data might tell a slightly different story if the model used for palaeo-seawater chemistry turns out to have flaws.
Nonetheless, as they stand, the results validate climate models at CO2 concentrations not observable in the instrumental and ice-core archives. The unequivocal advance made by Pearson et al.1 is to demonstrate that marine organisms that existed when the Antarctic ice cap formed show much lower δ11B values than do such organisms today — probably indicating lower seawater pH and higher CO2 levels than today — and that they record sharp variations in CO2 associated with the main phase of ice-cap growth. Their high-quality data will further invigorate study of the coupling between global temperature, ice volume and atmospheric CO2 in ancient times — and that is no mean achievement.