Ocean sediments, like the igneous rocks of Hawaii, contain magnetic minerals. As they slowly settle to the sea floor, these minerals too line up with the Earth’s field. The long cores from the oceans thus contain a continuous record of changes in the magnetic field, a built-in timescale for dating variations in sediment properties that might be related to glacial climate cycles. Many of these properties seemed to vary in a regular way, but the question was, Which would be most useful for understanding ice age climate? The answer was not immediately obvious, but one property in particular turned out to be crucial for confirming, once and for all, the link between climate and variations in the Earth’s orbit. It was the oxygen isotopic composition of fossil shells in the sediments. What in the world, you may wonder, do oxygen isotopes have to do with glaciation or the astronomical theory of climate? The connection was made by Harold Urey, a chemist who, like Willard Libby, worked at the University of Chicago, and who, again like Libby, was awarded the Nobel Prize, his for the discovery of deuterium, one of the isotopes of hydrogen. Urey was especially interested in how different isotopes of the same element behave when they take part in chemical reactions or processes such as evaporation and precipitation. Most of the elements in the periodic table have multiple isotopes; oxygen, for example, has three. All three have the same chemical properties—they are all oxygen—but they exhibit minute differences because of their different masses. In a tank of oxygen gas all of the molecules have the same energy. They whiz around, bumping into the walls of the tank and each other, but those containing oxygen-16 travel slightly faster than those containing oxygen-18, because oxygen-16 is lighter.
From theoretical considerations, Urey discovered that during chemical reactions, the oxygen isotopes are fractionated from one another because of the slight differences in their masses. One isotope is preferred over another in the products of the reaction. Furthermore, he found that the amount of fractionation depends on the temperature. In a flash of insight, he realized that oxygen isotopes could act as a natural thermometer. Many organisms that live in the oceans make their shells of calcium carbonate, an oxygen-containing compound. The oxygen comes from seawater, and when the shells are being precipitated, one oxygen isotope is preferred over another. Thus the shells end up with different proportions of the three oxygen isotopes compared to seawater, and the amount of that difference depends on the water temperature. By measuring the ratio of oxygen-16 to oxygen-18 in an ancient shell, Urey realized, he could determine the temperature of seawater in the distant past—a stunning concept.
Like most such ideas, bringing this possibility to fruition took some time. Urey and his students had to perfect the measurement techniques so that they could measure oxygen isotope proportions accurately in small amounts of calcium carbonate shell. On the basis of their calculations, the variations would probably only amount to a few tenths of 1 percent. They also faced the perennial problem of geochemists: Which samples would be most representative and provide the most important information? Their method, too, they soon learned, had its own complications. For example, they realized that both evaporation and precipitation would change the proportion of the various oxygen isotopes in seawater. How could they distinguish these variations from those caused by temperature changes?
By the early 1970s, most of these difficulties were well understood. Fortunately for research into the Pleistocene Ice Age, evaporation of water from the oceans—the process ultimately responsible for the supply of snow to the glaciers—and cold seawater temperatures both act to change the oxygen isotope proportions in carbonate shells in the same direction. Although it was still difficult to separate the two effects in a quantitative way, they at least reinforce one another, producing a stronger oxygen isotope signal than would temperature changes alone.
Research groups interested in glaciation and the Earth’s climate history scrambled to make oxygen isotope measurements on deep-sea sediments. Two papers from these early studies were especially important in the debate over the astronomical theory of climate. The first was by Wally Broecker and J. van Donk, who were working at what was then the Lamont-Doherty Geological Observatory (now the Lamont-Doherty Earth Observatory) of Columbia University. Published in 1970 in the journal Reviews of Geophysics and Space Physics, their paper had the title “Insolation Changes, Ice Volumes, and the Oxygen-18 Record in Deep-Sea Cores.” Broecker and van Donk used the magnetic properties of the sediments to determine a timescale for the cores, and they showed that when their oxygen isotope analyses were plotted against this timescale, they exhibited a smooth and systematic variation over the past 400,000 years.
What was puzzling about this graph for advocates of the Milankovitch astronomical theory was that it showed several cycles of peaks and valleys, with each cycle lasting about 100,000 years (figure 16). James Croll had predicted that the eccentricity of the Earth’s orbit, with a cycle close to 100,000 years, would be important for glaciation, but Milankovitch’s calculations of Northern Hemisphere temperatures had shown that the more important parameter is actually the tilt of the Earth’s axis of rotation. The tilt changes through a cycle lasting approximately 41,000 years. Combined high eccentricity and maximum tilt might result in especially severe glaciation, but according to Milankovitch, the tilt should be the determining factor. Why didn’t the oxygen isotopes follow the tilt cycle rather than exhibiting regular 100,000-year variations?
Broecker and van Donk’s work was not the only study that showed the 100,000-year cycles. Several other groups, some approaching the problem from different perspectives, found the same thing. Was the eccentricity really the important thing after all, or was there some unknown process at work with a 100,000-year cycle not connected in any way with the Earth’s orbit? The debate about whether or not astronomical variations could be responsible for glaciation heated up once again.
As more and more oxygen isotope data accumulated, it became clear that the same variations found by Broecker and van Donk were present in deep-sea cores from all of the world’s oceans. They were a global phenomenon, and they had to reflect the temperature changes and waxing and waning of the ice sheets that characterized the Pleistocene Ice Age. The magnetic timescale was crucial for this conclusion, because sediments accumulate at different rates in different places. The only way to be sure that the peaks and valleys in the oxygen isotope records occurred simultaneously throughout the globe was through accurate dating of each core using its magnetic properties.
The 100,000-year oxygen isotope cycles discovered by Broecker and van Donk are quite regular, but they are not perfectly smooth. Superimposed on these long cycles are many smaller wiggles. In a few cases, early workers examined cores that extended back to a million years or more, and in these they found that the prominent 100,000-year cycles seemed to die out between about 800,000 and one million years ago; before that there were also cyclical variations, but on a shorter timescale. How could all of these features of the oxygen isotope graphs be explained? Astronomical changes still seemed attractive because of the regular nature and planetwide occurrence of the variations. But once again the question of reconciling the timing of glaciation with the well-determined orbital variations became an issue.
The problem was solved in 1976, in a paper that appeared in the journal Science. Using the technique of spectral analysis—an approach that is capable of disentangling multiple, superimposed, cyclical curves and retrieving the original characteristics of each type of cycle—James Hays, John Imbrie, and Nick Shackleton showed that the oxygen isotope record is actually made up of several distinct, superimposed cycles, with timescales corresponding almost exactly to the predictions of the astronomical theory: a 100,000-year cycle that reflects changes in the eccentricity of the Earth’s orbit, another cycle of about 43,000 years, close to the timescale of changes in the tilt of the rotation axis, and one near 20,000 years that corresponds well with the wobble of the rotation axis. Hays, Imbrie, and Shackleton titled their paper “Variations in the Earth’s Orbit: Pacema
ker of the Ice Ages.” Finally, it seemed, the ideas that had been formulated first by Croll and then refined and extended by Milankovitch had been shown to be correct. Exactly how the orbital variations get translated into glacial-interglacial temperature differences was still uncertain, because virtually every analysis that had been carried out concluded that these variations result in only very small changes in the amount of solar energy received on Earth. But the close correspondence in timing between the astronomical cycles and the isotopic properties of deep-sea cores could not be denied. It was unlikely to be a coincidence. The repeated buildup and decline of the vast Pleistocene ice sheets must have been linked directly to changes in the Earth’s position relative to the sun. The Earth’s orbit is truly a pacemaker for climate.
Figure 16.A representation of climate changes over the past 550,000 years, based on oxygen isotope analyses of deep-sea sediments. The oxygen isotope “proxy” combines information about both temperature and the amount of glacial ice that exists on the continents. It is obvious that cold and warm periods alternated on a roughly 100,000-year timescale over this period.
CHAPTER EIGHT
Our Planet’s Icy Past
Oxygen isotopes in deep-sea cores, together with a few other indicators of past climates, have given us a surprisingly clear picture of the coming and going of glaciers during the Pleistocene Ice Age. But the current ice age occupies only the past few million years, an almost insignificant slice of our planet’s four-and-a-half-billion-year history. What was the climate like for the rest of that vast sweep of time? The “norm,” if it is possible to speak of such a thing, was one of warmth and little or no permanent ice. However, there is good evidence that our small (by the standards of the universe) planet has experienced sporadic ice ages for at least the past three billion years. Almost as soon as Louis Agassiz had pointed out the significance of glacial drift and other ice-produced effects in the 1830s, geologists began to find similar signs of ice ages in the more distant past. The very first such reports came from India, where, trapped within layers of sedimentary rocks, deposits of glacial drift were found lying atop scratched and grooved bedrock. Unlike Pleistocene drift in the Alps or in Canada, the ancient drift in India was not loose, but had been cemented and indurated into solid rock over hundreds of millions of years. Such drift-turned-to-rock was termed “tillite” by geologists, employing a seventeenth-century word describing chaotic rock deposits that contain fragments of a variety of sizes. Soon similar occurrences of tillites had been found in Australia, South Africa, and South America. To early geologists, one of the most startling aspects of these discoveries was that many of the places showing evidence of past ice ages were tropical or subtropical. In those pre-plate-tectonics days, when the continents were believed to be fixed and immobile, it was difficult to imagine tropical ice sheets. It seemed reasonable enough to think that glaciers in the Alps had once been more extensive, or that there might have been ice in Scotland in the past, but an Earth with glaciers near the equator was hard to grasp.
Aside from the problem of ice in the tropics, a major difficulty for those attempting to characterize ancient ice ages was that much of the evidence is missing. Even for the Pleistocene glaciation, erosion has obliterated some of the geological signs of ice action. The most recent glacial advance of the Pleistocene ended only some twenty thousand years ago, and most of its effects on the landscape are still quite obvious. But earlier Pleistocene glacial advances and retreats, even those that occurred only one or two hundred thousand years ago, are much more difficult to study, because the moraines and erratics and scratched bedrock from those episodes have not all been preserved intact. Such difficulties are compounded many times over for the ice ages that occurred in the Earth’s very distant past. But in spite of this, geologists have been able to identify at least four periods of severe glaciation that occurred long before the Pleistocene, all probably more intense than the current ice age. The timing of these is shown schematically in figure 17—the earliest known dates to about 2.9 billion years before the present, the most recent, 300 million years. Some of the early ice ages are depicted here as a series of events stretching over several hundred million years; at these distant times in the past the uncertainty in dating is such that it is not clear whether these were actually discrete ice ages, or just especially severe intervals within an overall cold period. In addition to the ones shown, several other times have been identified when the planet experienced cool periods, if not full-blown ice ages.
Figure 17.The Earth’s major ice ages, as identified from glacial drift, tillite, varves, and glacially scoured bedrock. Heights of the shaded bars give a rough indication of the intensity of these glacial periods, although the estimates are speculative for the glaciations between 2.2 and 2.4 billion years ago. In addition to the four major ice age periods before the Pleistocene discussed in the text, a further cold period that occurred 450 million years ago has been identified and is shown here. It is not possible to represent the durations of ice ages accurately on this small graph.
Traveling back into Earth’s history from the present, the first really major ice age that appears in the geologic record occurred about 300 million years ago. This is the very same ice age for which evidence was uncovered in India and other southern continents in the nineteenth century. It is worth considering for a moment what criteria must be satisfied before an ancient event can be called an ice age. How is it possible to know, for example, that a tillite, or glacial scratching, results from global glaciation and not from local mountain glaciers? Usually, at least three important characteristics must distinguish the evidence for glaciation. First, the effects of ice sheets should be widespread, usually meaning that they occur on several, well-separated continents. Secondly, the widespread glacial deposits must be contemporaneous. And, finally, there should be evidence that the glaciation took place at low elevations in most localities. This is not so difficult to establish as it might seem, because when ice sheets reach the sea, they drop glacial drift into the ocean, where it is preserved under later blankets of sediments.
The older the purported ice age, the more difficult it becomes to satisfy all of these criteria, especially the criterion of contemporaneity. If a dating method is accurate to a few percent, the uncertainty in dating a glacial advance that occurred 100,000 years ago is only a few thousand years, but for a 300-million-year-old tillite, it can be six to ten million years. That’s several times the length of the entire Pleistocene Ice Age. Furthermore, no method has yet been devised that can accurately date a glacially scratched surface, or an ancient tillite. Usually, the best that can be done is to bracket the age by dating lava flows or volcanic ash layers that occur above or below the glacial deposits. Sometimes, fossils in sedimentary rocks that accompany the glacial deposits are useful too, but usually these can only bracket the time of glaciation and do not date it directly.
The ice age that occurred near 300 million years ago is sometimes referred to as the Permo-Carboniferous Ice Age, after the two geological periods that it spans, the Permian and the Carboniferous. There is evidence that the Earth was cold for about 80 million years, from 340 until 260 million years ago. During this very long span of time, there were many cycles of glaciation and deglaciation, much as has occurred during the current ice age, although because of their great antiquity, it has not been possible to work out the timing of these cycles with any confidence. Whether or not they were influenced by astronomical cycles is also not known, but given the importance of orbital variations for the Pleistocene Ice Age, it is likely. Wegener—the same Wegener who worked together with Wladimir Köppen and Milutin Milankovitch on past climate and ice ages—was putting together his theory that the continents had moved about over the Earth’s surface, he searched for geological features that appeared to be continuous across now-separate continents. He was especially struck by the widespread evidence for the Permo-Carboniferous glaciation. Glacial deposits had been reported from all of the southern continents—India, S
outhern Africa, South America, and Australia—and as far as could be determined, the glaciation had been roughly contemporaneous. Wegener realized that a single, very large, continental ice sheet could account for all of the deposits if these now widely spread localities had once been contiguous. That would require closing up the Atlantic Ocean so that Africa and South America were joined, pushing India up against the eastern coast of Africa, and somehow attaching Australia to this group of continents. In addition, the problem of explaining how glaciers could exist in the tropics would disappear if his theory of continental drift were correct—the great agglomeration of landmasses could have been located much farther south, near the South Pole, at the time of glaciation, and only later drifted to their present, much warmer, localities.
Figure 18.At the time of the Permo-Carboniferous glaciation, the Earth’s land masses were joined in the supercontinent Pangea, which stretched from the South Pole to northern latitudes. The current southern continents—Africa, India, Australia, and South America—were clustered together with Antarctica near the South Pole, in a landmass referred to as Gondwanaland. Ice sheets spread northward to at least 40° south latitude.
Frozen Earth: The Once and Future Story of Ice Ages Page 17