BACK in 1993, a boy playing football near Nanjing, China, suddenly fell through the ground. He had inadvertently found a new cave, later named Hulu, which has turned out to be a scientific treasure chest. Besides two Homo erectus skeletons, it contains stalagmites that have helped solve one of the greatest mysteries in climate science: why the ice ages came and went when they did.
For more than 2 million years, Earth’s climate has been oscillating wildly. Immense ice sheets slowly advance across northern lands, then suddenly melt away to leave the planet basking in a relatively brief period of warmth before the ice creeps back again. Climate scientists have long suspected that these glacial cycles are triggered by changes in our planet’s orbit. Yet while this theory has had many successes, it fails to explain one critical fact: why the ice ages end every 100,000 years or so. “It’s a big problem,” says Larry Edwards of the University of Minnesota in Minneapolis.
Advertisement
Edwards is part of a group of researchers who may finally have the answer, thanks to Hulu and other nearby caves. If their conclusions are right, then the greatest ice sheets of the past were remarkably vulnerable, melting away when there was just a glimmer of extra sunlight. But what have stalagmites in China got to do with the vast ice sheets that covered much of Europe and Siberia, and North America?
By the middle of the 19th century, it was clear that there had once been a . The evidence was everywhere, from ice-carved landscapes to vast deposits of glacial debris. In fact, some geologists argued that there had been not just one ice age but as many as four. The question was, why?
Theories abounded. In 1864, , a Scottish jack-of-all-trades who had taught himself physics, proposed that periodic changes in Earth’s orbit change the amount of sunshine reaching the planet at various times of the year. Less sunshine in winter, he argued, would lead to snow accumulating. As ice sheets began to grow, the Earth would reflect more heat, amplifying the effect of the orbital changes and leading to ice ages. There were other positive feedbacks too, he suggested, such as changes in ocean currents.
While it gradually became clear that Croll was wrong about the timing of the ice ages, his orbital theory was revived early in the 20th century by a Serbian engineer called Milutin Milankovitch. Unlike Croll, Milankovitch focused on how orbital changes affect the amount of summer sunshine in the far north. Colder winters make no difference to ice sheet growth, he reasoned, but colder summers do. If the snow that falls during winter does not melt completely in the summer, ice sheets will grow; when summer melting gains the upper hand, ice sheets will shrink.
Milankovitch cycles
Milankovitch spent decades doing painstaking calculations to work out the effects of the three main orbital cycles. For instance, the tilt of Earth’s axis increases and decreases every 41,000 years, making summers hotter and winters colder (see diagram).
Milankovitch’s work was largely ignored until the 1960s and 1970s, when researchers began working out a detailed time line of the ice ages, based on isotope ratios in shells in ocean sediments. Water containing a lighter variety of oxygen evaporates more easily than that containing heavy oxygen, so when vast amounts of snow are locked away in ice sheets, the ratio of heavy oxygen-18 to lighter oxygen-16 in ocean water increases. Isotope measurements of marine sediments showed that there had been not just four ice ages, but dozens. What’s more, the waxing and waning of ice sheets usually coincided with orbital changes, a discovery .
This was far from the end of the story, though. We now know that the polar ice caps started to form around 30 million years ago, as carbon dioxide levels fell. Around 2.5 million years ago, as it got colder still, a cycle began in which more extensive ice sheets repeatedly spread across the northern hemisphere and then retreated. At first, these ice ages were relatively minor and occurred roughly every 41,000 years – just as you would expect based on the changing tilt of Earth’s axis.
But then, a little less than a million years ago, the pattern changed. A series of much more severe ice ages began that lasted 100,000 years. That is a big mystery, because although the shape of the Earth’s orbit alters slightly over periods of 95,000 and 125,000 years, this has a far weaker effect on the seasons than the other orbital cycles. Why would the deepest ice ages be driven by the smallest changes in summer sunshine?
Faced with this conundrum, some researchers began to explore alternatives to the mainstream orbital theory. One idea is that Earth sometimes passes through interplanetary dust clouds that cut off some of the sun’s heat. Or perhaps our star could be periodically getting brighter and dimmer.
Studies of ice cores from Antarctica, however, were starting to point in a different direction. The cores showed there was a close correlation between temperature and the levels of greenhouse gases in the atmosphere. This suggested a partial answer to the 100,000-year problem: small changes in sunshine might be greatly amplified by rises in CO2 levels. But there was too much uncertainty about the timing of events to say what caused what.
To find out what really happened, researchers need accurate dates, especially for the ends of the ice ages. “I’ve been after the timing of these terminations for 25 years,” says Edwards. While marine sediments and ice cores record the sequence of events, it is difficult to date those events precisely.
Edwards started by looking at coral. As tropical corals grow only in shallow water, they can reveal how the oceans rose and fell with the ice ages. And coral skeletons contain traces of uranium, which gradually decays into thorium, so ancient corals can be dated by measuring the ratio of uranium to thorium.
Edwards and others found sudden surges in sea level marking the last two terminations. But they couldn’t look further back because they couldn’t find pristine samples of older corals. And with only two terminations to go on, it is hard to establish any clear links between ice ages and orbital cycles.
In the mid-1990s, Edwards and his student Jeff Dorale turned to another kind of limestone clock: stalagmites. These grow over ten of thousands of years as dripping water deposits calcium carbonate on a cave floor, and as with coral each layer can be precisely dated from its uranium and thorium content.
The next trick was to find caves with sufficiently old and well-preserved stalagmites. After the two skeletons of Homo erectus were found in Hulu cave, Yongjin Wang of Nanjing Normal University was sent to date the fossils. In the process he found stalagmites tens of thousands of years old. Wang met Hai Cheng, a colleague of Edwards, and together they found a curious message in the rock.
The cave code
The stalagmites hold indirect clues to the climate in the form of oxygen isotopes, which record the strength of the summer monsoon. Water containing heavy oxygen condenses more easily, so the moisture-laden air of the monsoon loses most of its oxygen-18 as it moves inland. By the time it reaches central China, the rains are low in oxygen-18, and the stalagmites there record this depletion. But as the last ice age was ending, 11,000 to 17,000 years ago, the oxygen-18 content of the stalagmites increased – a sign that summer monsoon rains were much weaker than usual.
Wang then went looking for a cave with older stalagmites. He struck lucky at nearby Linzhu cave, despite a rather unusual hazard. “Bats stole our guide rope,” says Wang. “The cave has many branches, and we lost our way out.” When his team did eventually escape, around midnight, they brought out samples that held a much longer climate record. And stalagmites from another nearby cave called Sanbao provided even more precise dates.
“Bats stole our guide rope. The cave has many branches, and we lost our way”
These Chinese cave records show that the monsoons failed during all the last four terminations. “It is amazing how well they are linked,” says Edwards. The reason must be, as long suspected, that the melting of the ice sheets alters ocean circulation, producing drastic regional changes in climate. This link also means the timing of events during the past four terminations can be pinned down, allowing Edwards and his team to align the records from marine sediments, ice cores and caves, and compare these paroxysms in Earth’s climate with the changes in summer sunlight (see diagram).FIG-mg27610901.jpg
In between each ice age termination, the graph of summer sunshine wobbles up and down a few times due to the combined effect of all the orbital changes. The wobbles get weaker, and then shortly after the sunshine curve begins to rise from the fourth or fifth dip, the ice age ends. “We see exactly the same thing in all four terminations,” says Edwards. “It suggests that the ice sheets are very sensitive to changes in insolation.”
But hang on – if all it takes is a fairly small increase in summer sunlight to melt the ice sheets, why don’t they melt every time there’s more sunshine instead of waiting for the fourth or fifth wobble?
A clue comes from the sawtooth pattern of ice ages. Apart from a few fits and starts, the ice sheets keep on growing during a glacial period, reaching their greatest size just before a termination. This pattern suggests that there is something about being big that makes an ice sheet’s existence precarious.
One weakness may be weight. The bigger ice sheets grow, the lower the continental crust beneath them sinks. At lower altitudes the air temperature is higher, which would increase melting. And as the crust sinks, much of the ice sheet will end up below sea level. Ice sheets resting on the seabed – like the West Antarctic ice sheet today – are far more vulnerable to warming. Being somewhat buoyed up, for instance, the ice can flow more easily and thus disintegrate faster.
This much was already suspected, but the latest work points to another frailty of big ice sheets. While earlier studies suggested that CO2 levels start rising thousands of years before the ice sheets begin to melt, according to Edwards’s team the two begin simultaneously. The difference is crucial, because it means the melting of the ice sheets could cause the rise in CO2. The mechanism might be the change in ocean circulation.
Whenever the Laurentide ice sheet covering North America starts melting, vast amounts of water and ice pour into the North Atlantic. We know this because at the end of the ice ages, debris dropped by melting icebergs appears in marine sediments. This fresh water will reduce the density of the surface layer, stopping it sinking and thus shutting down the Atlantic overturning circulation – the great ocean current that carries heat north, then sinks and flows back along the bottom of the ocean.
In the popular imagination – thanks to Hollywood – the shutdown of this current leads to a global ice age, but in fact the main effect is a redistribution of heat. If less heat is carried north, the southern oceans warm. Since CO2 is less soluble in warm water, this leads to the release of CO2 (see “Blame the corals”).
So the new evidence points to a coherent story. Ice sheets build up until they near the brink of stability, at which point a modest rise in summer sunshine is enough to tip them over the edge. As the ice sheets melt, fresh water is released into the Atlantic, shutting down ocean circulation and pumping CO2 into the atmosphere. As long as the combined effect of extra summer sunshine and rising CO2 outweighs the regional cooling produced by the shutdown of ocean circulation, the ice keeps melting, pouring more fresh water into the Atlantic. And the melting of a really large ice sheet keeps ocean circulation shut down for a long time, eventually pumping so much CO2 into the atmosphere that the ice sheets melt away in just a few thousand years.
Other climate scientists are impressed with the results. “Their work in dating the cave record is phenomenal,” says Peter Huybers of Harvard University. “It is wonderful to have these constraints, and it gives us reassurance that the timing of climate events coincides with orbital changes.”
There are still plenty of loose ends, though. Most troubling, perhaps, are coral samples from Tahiti. If the dating of these samples is correct, sea levels began to rise a few thousand years earlier than the Chinese cave date for the ice age termination around 130,000 years ago.
Such inconsistencies have led some researchers to suspect that there is yet more complexity to uncover. “There is a lot of evidence that orbital cycles drive climate, but the precise linkage is still unclear,” says Gideon Henderson of the University of Oxford, who worked on the Tahiti corals. He thinks the focus on sunshine during northern summers may be too narrow. “People tend to look only at the insolation at 65 degrees north in summer, but there are lots of other ways the distribution of solar energy could affect climate. It could be in the tropics or the southern hemisphere.”
Changing pace
Another puzzle is why the length of climate cycles suddenly changed from 41,000 years to 100,000. It may be connected to the overall reason behind ice ages: the gradually falling levels of CO2 in our atmosphere. When the ice ages began, the climate may still have been warm enough that a little extra sunshine was enough to melt the ice sheets every time the Earth’s axial tilt swung towards a maximum. As CO2 and temperatures kept falling, we may have passed another threshold beyond which the change in the tilt was no longer enough to melt all the ice, so the ice ages started skipping one or two “beats”, only melting when they had become bloated and unstable. “But we still don’t really understand this,” Edwards admits.
Plenty of minor puzzles remain – enough to keep palaeoclimatologists busy for another century or two – but the latest evidence is all very much in Milankovitch’s favour. Indeed, even Croll has been vindicated to some extent: it’s clear that positive feedbacks play a huge role.
While orbital variations will continue to have a minor effect on climate, the epoch of ice ages is almost certainly over. With CO2 levels of 380 parts per million and climbing, the climate is currently on course to become like that of the Miocene 10 to 15 million years ago, long before the ice age cycle began, when it was 6 °C warmer and sea level was up to 40 metres higher. If the planet flips into a different climate regime rather than eventually returning to its pre-industrial state, the ice may never reconquer Europe and North America.
Blame the corals
There is no doubt that carbon dioxide is a major player in the coming and going of the ice ages. When the planet starts warming as ice ages end, atmospheric CO2 levels start rising, amplifying the effects of orbital changes. What is less clear is where the extra CO2 comes from.
CO2 is , so the warming of oceans would lead to its release. The change in solubility cannot fully explain the rises in CO2 at the ends of ice ages, though. It appears there are several sources, perhaps including a rather surprising one: coral. The formation of corals’ carbonate skeletons releases CO2, and a rise in sea level will lead to a burst of reef building, as existing reefs grow upwards and as corals colonise shallow waters where land has been submerged. So according to the , up to half of the rise in CO2 during ice age terminations might be due to coral growth.
This idea was proposed more than a decade ago, but has won little support because the rise in CO2 was thought to precede the rise in sea level. “Since we conclude that they do shift together, the coral reef mechanism could be involved,” says Larry Edwards of the University of Minnesota in Minneapolis (see main story). “In our scenario, the possible mechanisms (including the coral reef mechanism) fit together in a plausible sequence, which starts with a simple trigger of insolation rise initiating the demise of the ice sheets.”
