Close mass balance of long-term carbon fluxes from ice-core CO2 and ocean chemistry records

Feedbacks controlling long-term fluxes in the carbon cycle and in particular atmospheric carbon dioxide are critical in stabilizing the Earth’s long-term climate. It has been hypothesized that atmospheric CO2 concentrations overmillions of years are controlled by a CO2-driven weathering feedback that maintains a mass balance between the CO2 input to the atmosphere from volcanism, metamorphism and net organic matter oxidation, and its removal by silicate rock weathering and subsequent carbonate mineral burial. However, this hypothesis is frequently challenged by alternative suggestions, many involving continental uplift and either avoiding the need for a mass balance or invoking fortuitously balanced fluxes in the organic carbon cycle. Here, we present observational evidence for a closemass balance of carbon cycle fluxes during the late Pleistocene epoch. Using atmospheric CO2 concentrations from ice cores, we show that the mean long-term trend of atmospheric CO2 levels is no more than 22 p.p.m.v. over the past 610,000 years. When these data are used in combination with indicators of ocean carbonate mineral saturation to force carbon cycle models, the maximum imbalance between the supply and uptake of CO2 is 1–2% during the late Pleistocene. This long-term balance holds despite glacial–interglacial variations on shorter timescales. Our results provide support for a weathering feedback driven by atmospheric CO2 concentrations that maintains the observed fine mass balance. Carbon dioxide is released from volcanism and metamorphism to the atmosphere at a rate that would double the amount of carbon in the combined ocean–atmosphere system within less than 600 kyr (refs 13,14). Chemical weathering of Ca and Mg silicate rocks followed by carbonate burial removes CO2 from the atmosphere through weathering reactions. The weathering rates of the silicate rocks increase with soil CO2 concentration and temperature. The long-term balance between carbon release and removal has been hypothesized to be maintained by a negative stabilizing feedback, which results from the influence of the partial pressure of atmospheric CO2 on the Earth’s surface temperature and silicate rock weathering rates. Note that there are different types of this feedback, for example, before and after the rise of land plants. If the rate of carbon input to the ocean–atmosphere system would exceed the rate of chemical weathering, atmospheric CO2 and temperature would rise, thereby elevating weathering rates until a new balance is achieved and vice versa. Thus, CO2 levels over millions of years would be controlled by a CO2-driven weathering feedback that maintains a mass balance between CO2 input to and removal from the atmosphere. One school of thought has promoted this feedback as a key player in the long-term stabilization of the Earth’s climate. However, this view is controversial and frequently challenged by conflicting hypotheses. These hypotheses often involve continental uplift and avoid the need for a mass balance or postulate organic carbon fluxes that coincidentally balance the cycle. Until present, the advocates of the weathering feedback have theoretically argued that CO2 would undergo large variations within a few million years, if the feedback was absent. Unfortunately, the mechanism is difficult to prove, partly because the involved fluxes cannot be measured directly over millions of years. As a result, the CO2-driven weathering feedback, which has probably prevented runaway greenhouse and icehouse conditions over timescales of millions to billions of years, remains speculative. To provide observational constraints on long-term carbon fluxes, we have turned to recently generated ice-core CO2 records, now covering 650 kyr of the late Pleistocene epoch. Previous studies have focused on analysing the causes of the glacial–interglacial variations. However, despite variations on the timescale of thousands of years, there is remarkably little trend on the timescale of hundreds of thousands of years. Here, we show that the lack of a strong trend in atmospheric CO2 or ocean chemistry over this time period implies a tight coupling between CO2 sources to and sinks from the atmosphere and oceans. We have analysed Antarctic ice-core CO2 records spanning the past 650 kyr (refs 10–12) (Fig. 1). The calculated linear, long-term trends in CO2 based on different fit models (see the Methods section) range from −22± 2 p.p.m.v to +10± 6 p.p.m.v. per 610 kyr (Table 1). In other words, themean, long-term atmospheric CO2 change during the past 610 kyr is at most 22 p.p.m.v. Superimposed on this trend are the well-known glacial–interglacial variations. In general, causes for a mean pCO2 change can be split into two categories (see Supplementary Information, Fig. S1). (i) ‘Surface recycling’: a redistribution of carbon between or within surficial reservoirs of ocean (inventory=M C ), atmosphere (M atm C ) and terrestrial biosphere (M trr C ), with total carbon inventory M S C (refs 13,14):