A study confirms that volcanism set off one of Earth's fastest global-warming events. But the release of greenhouse gases was slow enough for negative feedbacks to mitigate impacts such as ocean acidification. See Letter p.573
On page 573, Gutjahr et al.1 report a fundamental breakthrough in our understanding of a global-warming event that occurred 56 million years ago and was caused by an increase in concentrations of atmospheric greenhouse gases. The episode, known as the Palaeocene/Eocene Thermal Maximum (PETM), was one of the most rapid warming events in Earth's history. The authors find that volcanism was the main source of carbon dioxide, and that the rate of emission of this gas was much lower than it is today. Understanding the processes that operated during warming events such as the PETM is crucial to help predict future climate conditions — we cannot rely solely on climate models, because they potentially misrepresent key processes that occur in warm climates2. But CO2 concentrations have not been as high as modern values for at least the past 3 million years3, so we need to go further back in time to find potential analogues of current and future climate change.
Estimates of the magnitude of warming during the PETM are relatively well constrained4 (4–5 °C), but the amount and source of carbon that fuelled the event has been hotly debated. Possible culprits5 include compounds known as methane hydrates, found in deep-sea sediments; volcanism that coincided with the initial opening of the Norwegian–Greenland Sea; and thawing of permafrost. Quantifying the magnitude, origin and emission rates of carbon is important. A reliable estimate of the amount of carbon released would constrain estimates of Earth's climate sensitivity6 (the change in the planet's equilibrium temperature in response to a doubling of atmospheric CO2) and therefore refine projections of future temperature rise. Determining the sources of carbon is particularly crucial, because destabilization of 'fast-release' reservoirs of organic carbon (such as methane hydrates or permafrost) could cause more damage to terrestrial and marine organisms than 'slow-release' mechanisms such as volcanism (Fig. 1).
Gutjahr et al. used boron isotope data measured in a sediment core from the northeast Atlantic Ocean to constrain changes in atmospheric CO2 during the PETM. Boron isotope ratios are a proxy for sea-surface pH, which depends on the amount of CO2 dissolved in the ocean, and therefore on atmospheric CO2 concentration. This is not the first time that the boron proxy has been used to study the PETM7, but the authors report a new data-assimilation technique that yields the most accurate assessment yet of the total volume and sources of carbon.
Using an Earth-system model, the authors simulated sea-surface pH at approximately weekly intervals, and compared the results at each time step with the interpolated pH record reconstructed from the boron data. They then adjusted the atmospheric CO2 levels in their model to minimize any differences between the simulated and observed pH for the next time step. This allowed them to reconstruct time series of CO2 concentrations throughout the PETM, and to convert these into carbon emissions. The authors also measured the ratio of carbon-13 to carbon-12 in sediments from the same core, to provide information about the possible mix of carbon sources.
The researchers concluded that the PETM was caused by a release of 10,200–12,200 petagrams of carbon (PgC; 1 Pg is 1015 grams). They also calculated that carbon-emission rates peaked at 0.58 PgC per year, which is less than one-tenth of current fossil-fuel emissions8 (about 10 PgC per year), and in agreement with earlier studies9. Their results point to volcanism as the main source (up to 90%) of emissions, with the remainder coming from a reservoir of organic carbon. The proposed mixture makes geological sense, because volcanic activity in the Norwegian–Greenland Sea is known to have triggered the release of methane from organic-rich sediments and coal10.
The study does have several uncertainties. The deep-sea sediment analysed was deposited at rates of a few centimetres per thousand years. It therefore cannot reveal changes that might have occurred over decadal or centennial intervals at the start of the PETM, including catastrophic, abrupt releases of massive amounts of carbon. Higher-resolution sedimentary records are available in coastal areas, but these are complicated by regional effects. In addition, to faithfully simulate climate recovery from the PETM, the authors needed to modify their model to increase the amount of organic carbon that was removed from the ocean's surface — confirming that our understanding and parameterization of ocean biogeochemical processes, and in particular biological productivity, is still rudimentary.
Even with these uncertainties, Gutjahr and co-workers' study is simultaneously good and bad news for the future of anthropogenic climate change. The first piece of good news is that the results imply that warm climates have a climate sensitivity consistent with, and at the lower end of, estimates from climate-model predictions11. Second, the findings suggest it is unlikely that catastrophic release of carbon from methane hydrates or permafrost was the main trigger of the PETM, although such processes probably amplified the effects of climate change. Yet the risk of potential future destabilization of these fast-release reservoirs cannot be ruled out, because they might be more vulnerable in today's much more-quickly warming world than they were in the past.
The relatively slow release of carbon during the PETM would have allowed time for amplified chemical breakdown of silicate and carbonate minerals in continental rocks under warm conditions. The carbon release would also have led to an extended period of corrosive conditions at the deep sea floor, which would have dissolved the calcium carbonate shells of marine organisms (calcifiers) buried in sediments. Increased atmospheric CO2 levels would have led to depletion of carbonate ions in the ocean, but the mineral breakdown and carbonate dissolution would have restored ocean carbonate levels, making the surface waters less corrosive for living calcifiers, and increasing the amount of atmospheric CO2 taken up by the ocean through a chemical 'buffering' effect. The chemical breakdown of silicate minerals would also have directly removed CO2 from the atmosphere.
The effects described above constitute negative feedbacks that would have dampened the initial spike in atmospheric CO2 concentrations resulting from a fast-release scenario. In the surface ocean, they would also have decoupled the decrease in pH and carbonate ions from each other — so sea-surface species were mostly affected by warming during the PETM, rather than by acidification12. But much of the CO2 initially released by volcanism was sequestered in the deep sea, where it caused severe acidification and resulted in one of the largest known extinctions of deep-sea species13. The bad news is that any current or future rapid carbon release will result in much stronger surface acidification than occurred during the PETM, because there won't be time for negative feedbacks to take effect. The effects on sea-surface species, including corals and marine calcifiers at the bottom of the food chain, will therefore be much worse.
Gutjahr and colleagues' study provides a much clearer understanding of the drivers of climate and ecosystem change during the PETM. It shows that warming and increases in atmospheric CO2 unfolded much more slowly than they are today. We are therefore entering uncharted waters.