When it comes to carbon dioxide (CO2) and climate, the past is prologue. Barring radical change to humanity’s voracious consumption of fossil fuels, atmospheric CO2 is bound to go up and up, driving global warming. But it won’t be the first time that CO2 has surged. In Earth’s ancient atmosphere, scientists see the faint outlines of a CO2 roller coaster, climbing and dipping across deep time in repeated bouts of climate change. “Each little slice in Earth’s past is a replicated experiment,” says Dana Royer, a paleoclimatologist at Wesleyan University in Middletown, Connecticut. “It helps us think about where we may be headed in the near future.”
If only the past could be seen more clearly. Models of ancient atmospheres and tools for teasing out past CO2 levels from fossils and rocks all have limitations. Now, scientists have developed a new method for wringing CO2 estimates from fossilized leaves—one that can go deeper into the past, and with more certainty. “At the moment, it’s very promising and it’s probably the best tool that we’ve got,” says David Beerling, a biogeochemist at the University of Sheffield in the United Kingdom who helped develop the so-called fossil leaf gas exchange technique. Already, it is solving ancient climate puzzles and delivering some unsettling news about the future.
Last month, at a meeting of the American Geophysical Union in San Francisco, California, another pioneer of the technique, plant physiologist Peter Franks of the University of Sydney in Australia, trained it on one of those puzzles: the time shortly after an asteroid impact killed off the dinosaurs 66 million years ago. Tropical forests were growing at temperate latitudes, yet earlier studies suggested CO2 levels of about 350 parts per million (ppm)—less than levels today and seemingly too low to create a global hothouse. Based on a gas exchange analysis of fossil leaves in what was once a tropical forest at Castle Rock, near Denver, Franks and his colleagues now conclude that the atmosphere 1.5 million years after the impact contained CO2 at about 650 ppm—a far more plausible level.
But in applications of the method to times between 100 million and 400 million years ago, Franks finds hints of a foreboding message. During documented episodes of global warmth, he says, the method reveals relatively low CO2 values, nothing like the levels of 2000 ppm or more suggested by other proxies. If these downward revisions hold, Earth may be even more sensitive to injections of CO2 than current models predict. “Temperatures are going to climb further for less carbon and we better be mindful of that,” Franks says.
Geoscientists go to elaborate lengths to figure out Earth’s past climate. For temperatures, they typically measure oxygen isotopes in carbonate rocks made up of the shells of tiny sea creatures that once lived near the sea surface. As temperatures drop, the animals tend to incorporate more of a heavier oxygen isotope into their shells—yielding a reliable measure of sea surface temperatures, which correlate well to atmospheric temperatures. “Those estimates are pretty good,” Franks says. “CO2 is the harder problem.”
For the past 800,000 years, CO2 levels can be measured more or less directly, in air bubbles trapped in ice cores retrieved from Antarctica or Greenland. But for earlier times, scientists look to models or proxies. Models reconstruct atmospheric CO2 based on the geological processes that affect the long-term carbon cycle. Before humans came along, volcanoes were responsible for injecting most CO2, whereas CO2 was removed by the burial of organic matter—think coal beds—and the weathering of rocks and formation of limestone. By rewinding the motions of plate tectonics and tracking broad areas of volcanism, vegetation, and weathering, scientists such as the late Bob Berner of Yale University were able to chart rising and falling CO2 over hundreds of millions of years. But their curves had huge margins of error.
Proxies, in contrast, ground estimates of ancient CO2 in real observations. A handful of them have been used for the past 100 million years (see graphic, above). But for more ancient epochs, geoscientists rely on fossil soils (paleosols) or fossil leaves. Some paleosols contain nodules of precipitated calcium carbonate as well as bits of organic matter, from which atmospheric CO2 levels can be worked out. But although they are sensitive recorders of higher CO2 levels, paleosols do poorly below a few hundred ppm.
In the 1980s came the first method based on plant stomata—little openings that allow CO2 into the leaf, where it is fixed into sugars through photosynthesis. Plants tend to have fewer stomata when CO2 is plentiful, because water also escapes through these pores and plants must guard against losing too much. But the number of stomata in each species responds to rising or falling CO2 in its own way. When fossil leaves belong to existing plant lineages like ginkgos, scientists can estimate ancient CO2 levels based on studies of close contemporary relatives. But for extinct species, they have to take a best guess. And in contrast to paleosols, the stomatal technique is insensitive to high CO2.
Franks and his colleagues set out to improve it. Their leaf gas exchange technique, outlined in a 2014 paper in Geophysical Research Letters, relies on two key inputs. One is a calculation of stomatal density—not only the number, but also the size and depth of the stomata in a fossil leaf—which indicates the rate at which gas could pass in or out of the plant. The other is an analysis of organic residue in the fossil, which contains carbon isotopes that track the ratio of CO2 inside the leaf to that in the atmosphere. Together, those factors can be parlayed into a reading of the atmospheric CO2 concentration.
Jennifer McElwain, a paleobotanist at University College Dublin and a longtime advocate for the basic stomatal method, was initially a critic of the arriviste. But she has since come around and is using it alongside older techniques. “It is going to be widely adopted and it is going to be a powerful method,” she says.
If the gas exchange technique does end up supplanting the others, its lower-than-expected values for ancient CO2 offer a sobering message for the future. Climate modelers talk about climate sensitivity—how much the world will warm for a doubling of CO2 from preindustrial values of 280 ppm. (With Earth recently passing 400 ppm, it is well on its way.) Most of the models used in assembling the latest report of the Intergovernmental Panel on Climate Change, which forecasts climate change and its impacts, have sensitivities that cluster around 3°C.
But these estimates focus on “fast feedbacks”—effects that will quickly amplify small amounts of warming, such as the shrinkage of Arctic sea ice and the rise in atmospheric water vapor, itself a greenhouse gas. They ignore longer term feedbacks like the melting of land-based ice sheets and changes in vegetation, which most scientists say will contribute additional warming over decades or centuries. “They can’t take into account these large-scale, deep-time processes—that’s what we can glean from the geological history,” Franks says.
By revealing lower CO2 levels during ancient warmings, he says, the gas exchange technique suggests a climate sensitivity closer to 4°C, not 3°C. It may take several generations for that rise to kick in, but history suggests that it is built into the climate system. “I do find it worrying,” McElwain says. “Within 50–100 years the Earth’s surface temperature could rise much higher than we currently anticipate.”
Still, the technique is new, and its message is far from definitive. This March, at a workshop at Columbia University’s Lamont-Doherty Earth Observatory in Palisades, New York, the CO2 proxies will square off in a competition of sorts. Paleoclimatologists plan to weigh the different proxy techniques and come up with a consensus record of CO2 over the past 66 million years.
Franks is confident that the gas exchange technique will fare well. “There’s little argument that the uncertainty you get from this is improved,” he says. “I’m not evangelizing for this model. I think it will take care of itself.”
*Correction, 4 January, 3:38 p.m.: A previous version of this story incorrectly identified David Beerling as a biochemist. He is a biogeochemist.