The ambiguity of the term "ice age" becomes clear as soon as we try to answer the question: "Are we in an ice age now?" If we look at the whole of Earth history, we realize that we live in an unusual time, because there are large ice sheets in the northern and southern polar areas. It is not clear what controls the existence of a glaciated ('ice house') or non-glaciated ('greenhouse') state of the Earth, but continental positions almost certainly play a role: ice sheets can not exist in deep oceans, so ice caps can form only when there are continents in polar position (Antarctica) or marginal seas surrounded by continents (Arctic Ocean). But glaciation did not always occur when continents were at the poles. Recently, a considerable amount of evidence has implicated variations in the concentration of CO₂ in the atmosphere.

There are many sources of information on these Cenozoic climate fluctuations, and we will discuss only a few: the record of land plants, and that of oxygen isotopes in marine limestones.

Land plant assemblages are very strongly linked to precipitation and temperature: if we know what flora occurs, we can say what the climate must have been. Interestingly, there is presently a very strong correlation between the outline of leaves and the mean annual temperature: at higher temperatures, a higher percentage of the flora has entire-margined leaves (i.e., smooth outlines, like a laurel), and a lower percentage has jagged margins (e.g., oak). Leaf margin data are time intense to collect, because many leaves have to be assessed in order to obtain statistically valid data. The data on leaf margins (especially from the US Western Interior region) clearly show a strong drop in mean annual temperature in the earliest Oligocene (about 33.5 Ma). At the same time, the floras indicate increased aridity in the American continent.

During the last 25 years, we have gained much information on the fluctuations in the size of the polar ice caps and in sea surface temperature by the study of oxygen isotopes. Remember that the oxygen isotopic composition of the shells depends upon the isotopic composition of the sea water in which they grew, as well as on the temperature at which they formed: in colder water, the shells are isotopically heavier (have relatively more ¹⁸O) than in warmer water with the same isotopic composition. During ice ages, it was colder and the ice caps were larger: both effects cooperated to make the shells enriched in ¹⁸O. Therefore, we do not know which part of the "getting isotopically heavier" resulted from changes in ice volume, which part resulted from changes in temperature. We can try to find this out by comparing oxygen isotope data for benthic and planktonic foraminifera: the effects of ice volume are valid for the whole ocean, and if changes in ice volume occur we must see a change in isotopic composition of both benthic and planktonic foraminifera.

Over the whole Cenozoic, deeper ocean waters (which sink at the poles and thus tell us what temperature the waters were at the poles) have decreased in temperature: from about 10 to 15^o C in the early Cenozoic to close to freezing now. The difference in temperatures between the equator and the poles has thus increased very much from the early Cenozoic to the present day. In addition, the difference in temperature between the deep waters at the equator and the surface waters at the equator has increased just as much.

We now explain the oxygen isotope record by the theory that a large Antarctic continental ice-sheet on East Antarctica (reaching down to sea level from the mountains) formed rapidly (in less than 100,000 years) in the earliest Oligocene (about 33.5 million years ago); it expanded considerably in the middle Miocene (14.6 Ma), possibly by formation of the West Antarctic ice sheet, also over about 100,000 years. A small northern hemisphere ice sheet may have started to develop about 6 million years ago, but a major increase in ice volume occurred only about 3.0 to 2.5 million years ago. Since that time, large ice sheets waxed and waned cyclically (Thursday's lecture).

How did we reach this ice house world?

• The overall Cenozoic climate trend was clearly one of cooling: at the end of the Cretaceous the polar regions did not have ice caps, and deciduous forests were growing at higher latitudes than the polar circles. The overall cooling trend is commonly linked to different aspects of plate tectonic processes: the position of continents (thus that of ocean currents), as well as the amount of CO₂ in the atmosphere (through volcanic emissions and mountain building, thus weathering). During the Cenozoic, the last major eruption of flood basalts (a source of carbon dioxide to the atmosphere) occurred during the opening on the North Atlantic Ocean in the late Paleocene - early Eocene (56-49 million years ago). After that, not much CO₂ was delivered by such basalts to the atmosphere.
• The collision of the Indian subcontinent with Asia, however, resulted in the formation of the Himalayan mountains. Early uplift of the Himalayas started in the Eocene (at about 38 million years ago), followed by major uplift of this mountain chain by the early Miocene (20 to 17 million years ago). The rising of the mountains may have caused more intense weathering of rocks, resulting in uptake of carbon dioxide. The increased weathering could have transported more nutrients (particularly phosphorous) into the oceans, increasing primary productivity by unicellular algae in the oceans, thus taking away more CO₂ from the atmosphere (indirect plate tectonics). This idea (major influence of weasthering of rocks in the Hiamlayas) is still being hotly debated: would more weathering be plausible when climate cooled? Would it not be more reasonable to suppose that weathering was more intense during the early Cenozoic, when high temperatures speeded up reactions, and high CO₂ levels likewise helped in rapid weathering? The increased weathering has been linked particularly to the cooling episode (possibly formation of the West Antarctic ice sheet) in the middle Miocene.
• The uplift of the Himalayas may also have had effects on climate by itself: the formation of the high mountains must have influenced the patterns that winds can take over earth. Before the formation of these mountains the strong monsoon in the Indian Ocean can not have existed. The monsoonal circulation developed probably during the late Miocene (about 8 Ma).
• In addition, evolution of plants may have played a role. During the Early Cretaceous (about 110 to120 million years ago) the plants that dominate the modern world, the Angiosperms (flowering plants) evolved. These plants include most deciduous trees. Their evolution and rise to dominance meant an increasing abundance of deciduous (rather than coniferous) forests, and a huge delivery of dead leaves every year. This delivery of dead leaves meant an increase in organic matter supplied to soils on land. The organic matter in these leaves (which rot away) is stored in soils, staying out of the atmosphere. Grasses evolved much later, becoming common in the Miocene (about 15 million years ago), and may have trapped yet more organic matter in soils.
• Evidence that atmospheric CO₂ levels decreased during the Cenozoic comes from several different sources, including modeling using information on the kind of rocks deposited. Additional evidence comes from the pattern of plant evolution: during the Cenozoic (7 to 8 Ma) the C4 plants evolved (most of the C4 plants are tropical grasses). These plants use a photosynthesis reaction that is less energy efficient than that of most plants (C3 plants), but becomes more advantageous at low CO₂ and water levels.
• It has also been argued that the opening of the passage between South America and Antarctica may have played a large role: that opened the way for the Antarctic Circumpolar Current, which roars around the Antarctic continent today, and prevents all warm currents from lower latitudes to reach its coasts. Once this current started, it cooled down the continent, then ice accumulated which caused more reflection of heat back into space, which caused more cooling: positive feedback! In this model, the cooling stopped when the continent became so very cold that the air could not hold much water vapor - thus not enough snow fell to increase the ice caps in size. The problem with this theory is that we do not have enough evidence on the timing of the opening of Drake Passage to say whether it occurred at the right time to cause the formation of the Antarctic ice cap in the earliest Oligocene.
• The closing of the oceanic gateway in the Caribbean across the Panamanian land bridge. North and South America were joined during the latest Miocene through early Pliocene. The waters across the isthmus gradually got shallower from about 8 million years ago, with the isthmus finally closing at about 4 to 3 million years ago. At the time that the two land masses became connected, the faunas from North and South America came into contact, and a mass migration followed, called 'The Great American Interchange', resulting in extinction of most the South American marsupial fauna. The closure of the isthmus of Panama must have caused a major change in ocean circulation, with warm equatorial waters no longer flowing through into the Pacific. The current of warm water from the Gulf of Mexico into the northern Atlantic (Gulfstream) thus intensified, but also became saltier because less low salinity surface waters could make it into the Pacific. The intensified currents brought more moisture (i.e.. snow) to the northernmost Atlantic. The saltier waters started to sink to the bottom of the North Atlantic, rather than being able to penetrate into the Arctic Basin, so that the Arctic Basin cooled, and glaciers could start to grow even at sea level.

The causes of the Cenozoic 'descent into the cold' are thus not clear, but we do know that it happened, when it happened, and that it happened not gradually, but stepped, but development of the eastern Antarctic ice sheet (33.5 Ma), the western Antarctic ice sheet (14 Ma) and the northern hemisphere ice sheet (2.5 Ma) at different times.