Frozen Earth: The Once and Future Story of Ice Ages

by Doug Macdougall

  The presence of cap carbonates in Namibia highlighted one of the issues consistently raised by critics of the Snowball Earth hypothesis, the problem of how the Earth could ever have thawed out again once it was completely ice-covered. The high reflectivity of the snow and ice would have bounced much of the solar energy that normally warms our planet right back out into space. An entirely frozen Earth, critics of the theory claimed, would have reached a climatic point of no return and could never have recovered. But the cap carbonates in Namibia and other localities indicate that it did recover, and very rapidly at that. Had the deep freeze really been as severe as the proponents of Snowball Earth would have it? And if it was, just how had the climate changed abruptly from icy to tropical?

  A possible solution to the permanently frozen Earth problem had actually been suggested by Joe Kirschvink several years before the work of Hoffman and his colleagues. Kirschvink’s idea was that carbon dioxide, a “greenhouse” gas that traps solar energy in the form of heat in the atmosphere, would build up to high levels under Snowball Earth conditions, eventually leading to global warming and complete melting of the glaciers. His reasoning went approximately as follows. First, we know that the main source of CO2 to the atmosphere is volcanic eruptions, which spew out gases as well as lava. Secondly, there are two removal mechanisms that keep CO2 roughly in balance—one is photosynthesis (plants use CO2 to make organic tissue, and release oxygen into the atmosphere as a by-product), and the other is the chemical weathering of rocks on the surface. This process will be explored in more detail later in this book, but in the simplest terms, CO2 is removed from the atmosphere because it dissolves in rainwater to make carbonic acid, which attacks and dissolves rocks. Because there is no evidence to suggest that the level of volcanic activity—the source of carbon dioxide—was radically different in the Late Proterozoic compared with today, a buildup of atmospheric CO2 could only occur if there was a decrease in its rate of removal. That is exactly what one would expect if the oceans were frozen—most of the photosynthetic organisms living in the sea would die, greatly diminishing one of the removal mechanisms, and the cycle of evaporation from the ocean and precipitation over land that drives the other, chemical weathering of rocks, would also cease. Dry, desertlike conditions would prevail globally. With the two primary CO2 removal processes greatly diminished or shut down altogether, its concentration in the atmosphere would be expected to rise to quite high levels.

  The CO2 concentration that would be required to warm the Earth and melt a completely frozen ocean today is very high. At the time of the Late Proterozoic Ice Age, even higher contents would have been necessary. According to astronomers who investigate the life cycles of stars, our sun’s energy output has gradually increased over the Earth’s history and would have been significantly lower during the time of Snowball Earth. Some researchers believe this was a factor in the initiation of the Late Proterozoic ice ages, but whether it was or not, a CO2 concentration several hundred times that of today’s atmosphere would have been required to thaw the completely frozen planet. At those levels, melting would begin first at low latitudes, and, once begun, would proceed in a runaway fashion with the help of positive feedback. Decreasing ice cover meant that more and more of the incident solar energy warmed the ocean instead of being reflected back into space. Evaporation from the newly uncovered ocean increased atmospheric humidity, which in turn intensified the greenhouse effect, because water vapor is an even more efficient trap for the sun’s energy than CO2. If this version of the Snowball Earth theory is correct, not only did the planet suffer through periods of extreme cold in the Late Proterozoic, it also endured brief but intense “super greenhouse” episodes when global temperatures soared far above anything experienced since, before returning to more normal levels as the CO2 balance was restored.

  The work in Namibia by Hoffman and his colleagues supports the Kirschvink scenario of a buildup in atmospheric CO2. A centerpiece of their investigation is the chemical data they collected. Like those who study the effects of Pleistocene Ice Age cycles in deep-sea sediments, they used the isotopic composition of sedimentary rocks to monitor environmental changes. But instead of measuring oxygen isotopes, they examined isotopes of carbon. These, like oxygen, can be fractionated when some process prefers one isotope over another. During photosynthesis, the plankton that live in seawater extract carbon from the oceans, preferentially taking up one of the isotopes of carbon, carbon-12, relative to the other, carbon-13. This leaves behind seawater enriched in carbon-13. The Harvard researchers found that in the Namibian rocks deposited just before the ice age began, the carbon isotopes are consistent with this normal situation. But in the glacial interval, and through hundreds of meters of the cap carbonates that overlay the glacial deposits, there is no longer evidence of an enrichment in carbon-13. The logical conclusion is that photosynthesis had ceased, that the ocean was effectively “dead,” because it was frozen and had been that way through the several million years represented by the glacial and cap carbonate sediments. The CO2 that would normally be consumed in photosynthesis instead built up in the atmosphere.

  The beauty of an isotope signature of the kind described by Hoffman and his colleagues is that it can be checked in ocean-deposited rocks of the same age, no matter where they occur. They need not be associated directly with evidence for glaciation; if photosynthesis had been reduced to a low level in a frozen Snowball Earth ocean, the same change in carbon isotopes should be apparent everywhere. And for the most part, that seems to be the case. The same kinds of carbon isotope changes observed for the Namibian rocks are also seen in Northwestern Canada, in Spitzbergen in the Arctic Ocean, and in Australia. The widespread nature of the evidence indicates that regardless of the cause, the effects were global. Furthermore, the detailed analyses that have been conducted at these localities have shown that there were four, and possibly five, major glacial episodes between approximately 850 and 550 million years ago. Each one of these exhibits changes in the carbon isotope composition of seawater that are larger than anything seen elsewhere in the geologic record, and in each case both the onset and the termination of the glacial episode seem to have been very rapid. One interpretation is that each of these periods of glaciation was a separate Snowball Earth interval, with a completely frozen ocean.

  Why should a series of Snowball Earth periods occur during the Late Proterozoic, and, as far as we know, at no other times in Earth history? Critics of the idea suggest that the very uniqueness of Snowball Earth episodes is good reason to doubt their existence. But there is circumstantial evidence that may explain why they occurred at this time and not at others. First, as we have already seen, the sun’s energy output was lower than today. That alone is not a very compelling argument, however, because it was lower still prior to the Late Proterozoic, through a long period when there is no evidence of any significant glaciation on Earth. Secondly, as we have also seen, the continents at this time were clustered at low latitudes, a configuration that has not occurred since then. Oceans in tropical regions absorb and store solar energy and tend to warm the Earth, but when continents are present at low latitudes, they reflect sunlight, and this would have been especially true of the barren continents of the Late Proterozoic, which were not yet covered in vegetation. The net result would have been that considerably less solar energy than at present was retained by the Earth, causing global cooling. More speculatively, chemical weathering of the tropical and subtropical continents—one of the processes that remove CO2 from the atmosphere—may have proceeded rapidly on the low-latitude landmasses, decreasing the Earth’s ability to retain solar heat through a reduction of the greenhouse effect. Theoretical mathematical “models” of how the Earth would behave with low atmospheric CO2, low solar energy input, and the continents located near the equator indicate that the oceans would cool and then begin to freeze at the poles. Because of positive feedback—as the ice began to extend from the poles toward the equator, more and more solar energy would be reflec
ted back into space—the temperature would continue to decrease. The theoretical treatments also suggest that at some point, this would become a runaway process, and the entire ocean would freeze rapidly. This is consistent with the field evidence from Namibia and elsewhere, which indicates that the Late Proterozoic glacial episodes began very rapidly.

  The Snowball Earth hypothesis seems to be consistent with virtually all of the Late Proterozoic geological features: evidence for simultaneous widespread glaciation, ice sheets at low latitudes, the sudden appearance in the geological record of banded iron formations, large changes in the carbon isotopes in seawater, and the juxtaposition of glacial deposits with the warm-water cap carbonates. Still, that does not mean it is correct in all details. There are alternative explanations for some of these features. One school of thought suggests that both the sudden warming at the end of glaciation and the carbon isotope changes in seawater could be explained by the release of a large amount of methane gas from frozen ground as continental glaciers began to wane. An important difference between this scenario and the Kirschvink hypothesis is that it would not require a frozen ocean. Methane is a powerful greenhouse gas, even more effective than carbon dioxide at trapping the sun’s heat energy in the atmosphere. It is produced by bacterial decomposition of organic matter, and because it is a gas at most surface temperatures, it normally seeps slowly out of the ground into the atmosphere, where it is gradually destroyed by chemical reactions with other compounds. However, at low temperatures, it can be trapped underground in an icelike compound. Large amounts can be stored in the frozen ground of cold regions, as is the case in arctic permafrost today. Although the overall abundance of organic matter was much less during the Late Proterozoic compared to today, bacteria were widespread. Large quantities of methane could have been formed and stored during the long glacial intervals that characterized this time. Its rapid release could account for the “super greenhouse” episodes that apparently followed the Snowball periods. Methane produced by bacteria contains carbon with an isotropic composition consistent with the carbon isotope shifts observed in the Namibian sedimentary rocks, and those from other locations as well. Perhaps after all the Snowball Earth was not quite as frozen as some think. Perhaps there were tropical glaciers, but much of the low-latitude ocean remained open.

  As fragmentary as the geological record is for the Late Proterozoic, things get even murkier farther back in the Earth’s past. We really have no clear idea about what fraction of the preexisting rock of the Earth’s crust—our only window back into the Earth’s history—has been eroded away or caught up and transformed beyond recognition in mountain-building events. In spite of that, by piecing together information from all of the world’s continents, it is possible to determine with a high degree of confidence that there were, at a minimum, two periods of worldwide glaciation, true ice ages, prior to the Snowball Earth episodes. One of these occurred between 2.2 and 2.4 billion years ago, the other at roughly 2.9 billion years ago. Equally important, however, is the fact that in spite of careful searching, there is no hint of an ice age in the vast stretch of time, approximately 1.4 billion years, between the Late Proterozoic events and the ice age that ended 2.2 billion years ago. That gap is 30 percent of the Earth’s entire history, and its very existence must hold clues to the conditions necessary for glaciation. But it occurred so far back in the past that our knowledge of geological events then is limited. We do not have very reliable information, for example, about where the continents were situated on the globe, or how they moved about, or what their surface area was during this period. We also have little information about the concentration of greenhouse gases in the atmosphere.

  The evidence for one or more ice ages between 2.2 and 2.4 billion years ago is widespread. In central Canada and the United States there are glacially scratched and grooved bedrock exposures, finely layered varves with occasional large “dropstones” that are characteristic of glacial lakes, and tillitelike glacial sediments. Similar features are widespread in South Africa, and some also occur in northern Europe, although the European localities have been more heavily metamorphosed than those in North America or South Africa and are not as complete. There are also glacial sediments from this timespan in western Australia. Like the evidence of Late Proterozoic glaciation, that for the ice age 2.2 to 2.4 billion years ago is so far-flung that it suggests a global episode. The limited available magnetic data indicate that at least the South African glaciation, and perhaps some of the others, occurred at low latitudes. The Canadian deposits also include rocks that are remarkably similar to the Late Proterozoic cap carbonates. These features have led some to suggest that this early glaciation must also have been a Snowball Earth episode. But while such an idea is tantalizing, it is, at present, much less susceptible to rigorous testing than is the Late Proterozoic evidence. Dating such ancient events is subject to uncertainties that make it impossible to say with confidence that glaciation took place simultaneously in all localities—only that it occurred within the same overall time window. And whether the now-widespread locations where evidence for glaciation has been found were similarly distant 2.2 billion years ago, or whether they were once contiguous and were later dispersed by continental breakup, is also unknown.

  The very earliest glaciation we know about on Earth, according to the best dating that is available, occurred between 2.9 and 3.0 billion years ago, in the Archean Eon, the most ancient of the subdivisions of geological time. Once again the evidence comes from South Africa. Spread across a distance of several hundred kilometers there are several outcrops of tillitelike rocks that contain pebbles and boulders clearly scratched and faceted by glacial processes (figure 19). Similar deposits have also been observed deep underground in mines. There is little doubt that these record the workings of glaciers. The only question is, Was the glaciation a local phenomenon in a mountainous region, or is this evidence of an early ice age? With the scanty clues from South Africa, and so far no reports of glacial deposits in rocks of this age from elsewhere, the question is impossible to answer. There are, however, several discrete tillite layers, reminiscent of the multiple cycles of glaciation and deglaciation that are known from other, younger, ice ages. The existence of any decipherable record of glaciation at all from this distant time in the Earth’s past is itself quite remarkable. The scratched and grooved rocks from South Africa could be the products of the Earth’s very first ice age. On the other hand, there are also no compelling reasons to believe there were no earlier glacial periods. We may never know with certainty, because much older rocks are rare and for the most part have been buried, heated, and metamorphosed to the point where any signs of glacial activity would have been obliterated. There are, perhaps, a few secrets that Nature will not reveal.

  Figure 19.A glacially scratched and faceted boulder from the Earth’s oldest known ice age. This sample comes from a tillite in South Africa that is estimated to be between 2.9 and 3.0 billion years old. Photograph courtesy Professor John Crowell, University of California, Santa Barbara.

  CHAPTER NINE

  Coring for the Details

  The physical and chemical records of ancient ice ages stored in rocks—scratched and faceted boulders, glacial drift deposits hardened into coherent tillite, carbon isotopes, and various other signatures—allow us to trace glaciation back through almost three-quarters of our planet’s history. However, we know almost nothing about the finer details of the frigid intervals that occurred before the Pleistocene Ice Age—what the actual temperatures were, whether there were repeated 100,000-year cycles of glaciation and deglaciation paced by the Earth’s orbit, or how the climate varied across the globe. But over the past few decades, scientists have accumulated an amazingly detailed picture of all of these things, and more, for the most recent cycles of the Pleistocene Ice Age. Nearly all of this information has come from cores—cores of sediments from the seafloor, cores of ice from Antarctica and Greenland, and cores from lake beds and small mountain glaciers. Almost anythi
ng that accumulates or grows in a regular fashion has the potential to preserve a decipherable record of the environment. Even trees and coral heads have been cored, although in these cases the record usually does not extend very far into the past.

  The information about ice age climates revealed by these types of records is not very comforting. The ice cores have produced especially dramatic results, because they allow an almost year-by-year reconstruction of climate history back through several glacial-interglacial cycles. They show that in the geologically recent past, wild but long-lasting swings in average temperature have occurred over periods as short as a few years. Climate shifts during the ice age were once believed to be slow and ponderous, but the new data show that sometimes they can be very rapid. Our distant ancestors lived through such periods, although we have no record of the effects on their lives. It is fairly obvious, however, that in the finely balanced twenty-first century world, drastic changes in the frequency and intensity of storms, in patterns of precipitation, or even just a significant change in average temperature, could wreak havoc with agriculture, trade, and transportation. The possibility that mankind’s activities, such as adding the greenhouse gas CO2 to the atmosphere through the burning of fossil fuels, might affect the natural changes in unknown ways adds great urgency to the need to understand what has happened in the past. This has not been lost on some of those likely to be affected. Small island nations in the Pacific have become ardent supporters of attempts to curb greenhouse gas emissions, basing their stance on research showing that rising sea level—which would cause them great damage—accompanies increased atmospheric CO2 during interglacial periods. Some large insurance companies have been following climate change research with much interest, even to the extent of funding investigations into how global warming may affect the frequency of storms and floods. Self-interest is a potent motivator.