enrichment in 18O when polar glacial expansion preferentially sequesters 16O from the hydrosphere. For reasons not yet known, periodicities of ~41,000 years, reflecting the tilt cycle of the Earth's axis, dominated until about 0.8 m.y. ago, when periodicities of ~100,000 years, reflecting the shape of the Earth's orbit, began to prevail.
Cycles in some pre-Neogene marine successions appear to reflect minor changes in sea-level or biotic productivity that were forced by Milankovich controls mediated by factors that remain poorly understood but may not always have entailed changes in ice volume. Certain Mesozoic lake deposits also contain evidence for Milankovitch-driven cyclicity, perhaps related to shifting monsoons or other patterns of rainfall and evaporation.
The most profound nonperiodic cycles of global change have been long-terms oscillations between what have been termed the ''hothouse" and the "icehouse" states for oceans and atmospheres. The term hothouse is preferred to "greenhouse" because the conditions described may not always result from greenhouse warming; the hothouse states are, however, characterized by warm polar regions and warm deep oceans. In contrast, the icehouse state entails cold (usually glacial) polar conditions and a frigid deep-sea that results from the descent of cold polar waters.
The geologic record spanning the Eocene-Oligocene boundary documents the transition between a hothouse state and the icehouse state that has persisted to the present (Kennett et al., 1972). Much farther back in the geologic record, the interval spanning the Ordovician-Silurian boundary documents a similar transition, as well as the subsequent melting and retreat of glaciers and return of warmer conditions across broad regions (see Berry et al., Chapter 2).
The recent ice age in the Northern Hemisphere constitutes only an intensification of the icehouse state that our planet entered about 34 m.y. ago, at the end of the Eocene Epoch. Fossil floras and vertebrate faunas reveal that early in Eocene time, subtropical conditions extended north of the Arctic Circle and that southeastern England and the Paris Basin (45 to 50°N) supported tropical rain forests. Fossil floras are, in fact, the most valuable indicators of terrestrial climates for the past 100 m.y. Not only does the taxonomic composition of fossil floras reflect climatic conditions, but so does leaf morphology, especially the percentage of species with smooth, as opposed to jagged or lobed, leaf margins; this percentage varies linearly in the modern world with mean annual temperature (Figure 1). Leaf morphology and cuticular structure also provide a guide to precipitation conditions. Fossil floras show that the Eocene-Oligocene climatic shift was profound at middle and high latitudes in both hemispheres. As warm-adapted floral elements disappeared from these regions, other types of vegetation, adapted to colder and drier conditions, expanded (see Christophel, Chapter 10).
Climates actually did not undergo a simple shift between Early Eocene and Early Oligocene time. The tropical flora of England began to disappear at the end of Early Eocene time, as global temperatures began to cool, especially at high latitudes. By Late Eocene time, woodland savanna had already become the dominant vegetation of midcontinental North America (see Webb and Opdyke, Chapter 11). Whether the particular temporal pattern observed for North America characterized other continents remains uncertain, in part because of uncertain dating and in part because in some regions, such as Australia, a floral record is missing for much of the Eocene.
It is now widely agreed that the plate tectonic separation of Australia from Antarctica was a primary trigger of climatic changes near the end of the Eocene (and continuing separation caused further climatic changes after Eocene time). This event created the