The last interglacial, commonly understood as an interval with climate as warm or warmer than today, is represented by marine isotope stage (MIS) 5e, which is a proxy record of low global ice volume and high sea level. It is arbitrarily dated to begin at approximately 130,000 yr B.P. and end at 116,000 yr B.P. with the onset of the early glacial unit MIS 5d. The age of the stage is determined by correlation to uranium–thorium dates of raised coral reefs. The most detailed proxy record of interglacial climate is found in the Vostok ice core where the temperature reached current levels 132,000 yr ago and continued rising for another two millennia. Approximately 127,000 yr ago the Eemian mixed forests were established in Europe. They developed through a characteristic succession of tree species, probably surviving well into the early glacial stage in southern parts of Europe. After ca. 115,000 yr ago, open vegetation replaced forests in northwestern Europe and the proportion of conifers increased significantly farther south. Air temperature at Vostok dropped sharply. Pulses of cold water affected the northern North Atlantic already in late MIS 5e, but the central North Atlantic remained warm throughout most of MIS 5d. Model results show that the sea surface in the eastern tropical Pacific warmed when the ice grew and sea level dropped. The essentially interglacial conditions in southwestern Europe remained unaffected by ice buildup until late MIS 5d when the forests disappeared abruptly and cold water invaded the central North Atlantic ca. 107,000 yr ago.

Since its identification nearly fifty years ago, Marine Isotope Stage 5 (MIS 5) has been placed onto absolute time scales on the basis of three independent approaches. Cesare Emiliani, who set up the isotope stages (Emiliani, 1955), depended on uranium-series dating of the sediments, a method that today is regarded as not generally capable of yielding useful precision or accuracy. Broecker and van Donk (1970) pioneered the approach of correlating to radiometrically dated marine coral terraces; this has been much aided in recent years by improvements in the precision and accuracy of these age determinations that have flowed from the development of thermal ionization mass spectrometry (TIMS) for uranium-series dating (Edwards et al., 1986). The third approach is to use the astronomical record as a guide to the time scale. Martinson et al. (1987) generated a detailed time scale for MIS 5 using this approach. These authors suggested that the overall average error was of the order ±5000 yr, although the error would be smaller during interglacial periods with high precession-related variability, such as MIS5. At that time, the suggested confidence limits were smaller than typical values quoted for the radiometric dating of corals (typically ±6000 yr). Today the accuracy of the time scale of Martinson et al. (1987) is challenged by high-precision TIMS dates with quoted uncertainties of the order ±1000 yr or better. From the point of view of achieving a better understanding of the last interglacial period, the more serious disadvantage of the Martinson et al. (1987) time scale is the underlying hypothesis that all the proxy palaeoclimate records represent smoothly varying responses to changes in insolation; hence, there is no basis for estimating the duration of an extended interval with northern ice sheet volumes static at a size no greater than at present. From this point of view, the model of Gallée et al. (1993) is more promising, but that model is not at present sufficiently realistic to provide a reliable independent time scale. We have therefore chosen to depict the oxygen isotope record of core MD95-2042 (37°48′N, 10°10′W, water depth of 3146 m) on a time scale (Shackleton et al., 2001) that is based only on making use of selected radiometric dates obtained from fossil corals to calibrate the isotope record.

Deep-sea sediment cores provide spatially coherent evidence for the climatic and hydrographic conditions in the subpolar North Atlantic during the last interglaciation. Taken together with similarly high-resolution terrestrial sequences, these records indicate a regional climatic progression, beginning with the extreme and variable climate late in the penultimate glaciation, continuing through a relatively stable climatic optimum during the interglaciation, and concluding with the reestablishment of the markedly variable regime that characterized the last 100,000-yr glaciation. Relatively mild conditions in much of the subpolar region significantly outlasted the minimum in global ice volume, despite declining summer insolation and the cooling influence of incipient proximal glaciers. These effects were partially offset by enhanced thermohaline circulation that paradoxically increased heat transport into the region while simultaneously providing the likely moisture source for the growth of large northern ice sheets. The inception of the last glacial cycle thus provides an example of the influence of ocean circulation on regional climate. In contrast to the apparent orbital pace of the ongoing ice-sheet growth, the subsequent deterioration of surface conditions was abrupt and dramatic.

Studies of marine records from the northwestern European shelf and the northern North Atlantic suggest that last interglacial environments were less stable in this area than in the mid-latitude Atlantic. The influx of Atlantic water masses to the northern North Atlantic was generally higher, and the meridional temperature gradient was steeper, during the last interglaciation than during the Holocene. Strong north–south sea-surface-temperature gradients during the early Weichselian indicate a generally low influx of Atlantic water to the northern North Atlantic, even during interstades.

Although the link between insolation and climate is commonly thought to be in the high northern latitudes in summer, our results show that the start of the last glaciation in marine isotope stage (MIS) 5d was associated with a change of insolation during the transitional seasons in the low latitudes. A simplified coupled ocean-atmosphere model shows that changes in the seasonal cycle of insolation could have altered El Niño Southern Oscillation (ENSO) variability so that there were almost twice as many warm ENSO events in the early glacial than in the last interglacial. This indicates that ice buildup in the cooled high latitudes could have been accelerated by a warmed tropical Pacific.

Eemian lake deposits in Grande Pile and Ribains, France, correlate with marine isotope substages (MIS) 5e and 5d as delimited in two pollen-bearing deep-sea cores off Portugal. The Eemian forests in France lasted for approximately 20 millennia, from at least 126,000 to ca. 107,000 yr B.P. Oscillatory climate deteriorations began about 115,000 yr ago. An intense cold spell affected the region approximately 110,000 yr ago.

The last interglacial period has a timing and duration that can be estimated from U-series dating of emergent, coral-bearing deposits on tectonically stable coastlines. High-precision dating from Bermuda, the Bahamas, Hawaii, and Australia suggests that the last interglacial period had a sea level at least as high as present from ∼128,000 to 116,000 yr B.P. Sea level reached a near-present level more quickly after the close of the penultimate glacial period than at the close of the last glacial period and the duration of high sea level is longer than that implied by the deep-sea record.

The terms Eem and Eemian have been applied to lithostratigraphic, chronostratigraphic, and biostratigraphic aspects of the last interglacial in western Europe. Eemian vegetational successions show strong uniformity at sites from western France across the North European Plain to Poland, suggesting, by comparison with the Holocene, that major pollen zones are broadly synchronous. South of the Alps and Pyrenees, a different vegetational succession is observed with no evidence for a substage of post-temperate cooling.

Suggestions that the duration of the Eemian interglacial was about 11,000 yr, based on annually laminated sediment sequences in Germany, have been challenged in favor of a much longer interval. However, biostratigraphic evidence demonstrates why the Eemian sequences at Grande Pile and Ribains cannot be reliably used for alignment with the marine sequence, as applied by Kukla et al. (2002, this issue) to estimate the duration of this interglacial. The long chronology they propose would imply not just coniferous but, for up to 5000 yr, fully temperate forest in central France coexisting with treeless heath and steppe tundra conditions in northwestern Europe, an unlikely climatic and ecological scenario. The proposal that the Eemian Interglacial in western Europe lasted for 17,000 or even 23,000 yr is rejected. A duration of no more than 13,000 yr is preferred, at least for sites north of the Alps and Pyrenees. The duration of temperate conditions in the Mediterranean region is less certain.

Hundreds of small lakes became filled with Eemian deposits following the retreat of Saalian ice in northwestern Germany. Weakly expressed climate oscillations have been inferred from some local pollen records of Late Saalian sediments. Sea level stood 7 m below the current northwest German sea-level datum (NN) during the interglacial climate optimum and decreased afterward. A uniform vegetational succession of the Eemian mixed forests, correlated with marine isotope stage (MIS) 5e, terminated with the demise of boreal woodlands marked by a steep increase in nonarboreal pollen. This is the onset of the early Weichselian Herning Stade, correlated with MIS 5d. No sharp climate oscillations during the Eemian have been documented. In the ending phase of the interglaciation, the climate deteriorated gradually. Dominant heath and grass tundras point to a substantial decline of summer temperature during the Herning Stade. Sea level then stood at −40 m NN. Sedimentological considerations, supported by counts of partly varved deposits, indicate that the Eemian lasted about 10,000–11,000 yr and the Herning for several millennia.

A new astronomical calibration method for long pollen records from southern Europe is applied to the last interglacial interval of the Ioannina sequence, northwestern Greece. This shows that the last interglacial in this region, as defined by the presence of forest communities, lasted ca. 15,500 yr, from 127,300 to 111,800 yr B.P. Interglacial conditions developed within marine isotope substage (MIS) 5e and persisted into MIS 5d, lagging changes in global ice volume.

Paleoproductivity records from Lake Baikal indicate that the last interglaciation in southeastern Siberia lasted ca. 12,000 yr from 127,000 to 115,000 yr B.P., was punctuated by a climatic oscillation about 120,000 yr ago, and ended abruptly with a rapid onset of regional glaciation during marine isotope substage (MIS) 5d. The Siberian equivalent of the MIS 5e–5d transition appears to lead the post-Eemian cold pulses in Europe and in the central North Atlantic by 5000 to 8000 yr. The Baikal record also registers a pronounced cold event during early MIS 5c, which correlates with the Montaigu event in pollen records from France. In Siberia, the climatic deterioration during the MIS 5d was a full-scale glacial pulse unlike the short, moderate cooling during the Montaigu event.

Multiple lines of stratigraphic, geochemical, and fossil data suggest that fresh-mesohaline paleolakes were widespread in the Tengger Desert of northwestern China and underwent major fluctuations during the late Pleistocene. The paleolakes started to develop at ca. 42,000 14C yr B.P. The lake levels were the highest between 35,000 and 22,000 14C yr B.P., during which Megalake Tengger dominated the landscape. The climatic conditions at this time were unique for this area and have no modern analogue. After an episode of decline between 22,000 and 20,000 14C yr B.P. and an episode of rebound between 20,000 and 18,600 14C yr B.P., the paleolakes started to desiccate and completely disappeared around 18,000 14C yr B.P. The environmental proxy data indicate that the Megalake Tengger formed under warm–humid climates. The reconstructed climatic variations appear to be correlative with the abrupt climatic events reconstructed for the North Atlantic.

The Lake Tilla crater lake in northeastern Nigeria (10°23′N, 12°08′E) provides a ca. 17,000 14C yr multiproxy record of the environmental history of a Sudanian savanna in West Africa. Evaluation of pollen, diatoms, and sedimentary geochemistry from cores suggests that dry climatic conditions prevailed throughout the late Pleistocene. Before the onset of the Holocene, the slow rise in lake levels was interrupted by a distinct dry event between ca. 10,900 and 10,500 14C yr B.P., which may coincide with the Younger Dryas episode. The onset of the Holocene is marked by an abrupt increase in lake levels and a subsequent spread of Guinean and Sudanian tree taxa into the open grass savanna that predominated throughout the Late Pleistocene. The dominance of the mountain olive Olea hochstetteri suggests cool climatic conditions prior to ca. 8600 14C yr B.P. The early to mid-Holocene humid period culminated between ca. 8500 and 7000 14C yr B.P. with the establishment of a dense Guinean savanna during high lake levels. Frequent fires were important in promoting the open character of the vegetation. The palynological and palaeolimnological data demonstrate that the humid period terminated after ca. 7000 14C yr B.P. in a gradual decline of the precipitation/evaporation ratio and was not interrupted by abrupt climatic events. The aridification trend intensified after ca. 3800 14C yr B.P. and continued until the present.

A lacustrine sequence from Laguna Seca San Felipe, Baja California, in the southwestern Sonoran Desert, provides the first record of mid- to late-Wisconsin vegetation in one of the driest regions of North America. Pollen analysis was performed in the intermediate part of the sequence from ca. 44,000 to ca. 13,000 14C yr B.P. according to six 14C radiocarbon dates. Sedimentation of eolian sands prior to 44,000 14C yr B.P. indicates dry conditions. High pollen concentration of montane and chaparral species in the pollen assemblages contrast with the modern desert vegetation. Sixty-four pollen types were identified. The source of mid- to late-Wisconsin sedimentation must have been the plant communities in the surrounding highlands of the basin. Mid-Wisconsin pollen assemblages dominated by pines, junipers, and Artemisia reflect humid conditions in the area. By late Wisconsin time, a significant increment in junipers indicates a lowering in the altitudinal ranges of woodlands/chaparral. Pollen from wind-pollinated species is abundant while desert plants, specifically animal or insect pollinated species, are poorly represented. The San Felipe record does not identify the Last Glacial Maximum as the time of greatest effective moisture, as suggested by pluvial lake levels and other paleoclimatic evidence in the soutwestern United States.

Total organic carbon (TOC), black carbon (BC), and land-derived organic carbon (LOC) extracted from marine sediment core 17962 from the southern South China Sea and analyzed by pyrolysis gas chromatography, elemental analyzers, and mass spectroscopy can be interpreted in terms of regional land biomass burning, climatic conditions, vegetation cover, and ocean paleoproductivity. Carbon from terrestrial sources, including BC and LOC, appears to be a significant component of TOC (3.1–39.3% range, 16.5% average for BC; 4.4–56.6% range, 14.6% average for LOC). The production of BC and LOC was only slightly higher during the Last Maximum Glaciation (LGM) than during the Holocene. Therefore, dramatic changes in climate and vegetation cover do not seem to have occurred at the Pleistocene–Holocene boundary. Paleoproductivity data calculated from marine organic carbon, proved to be more accurate than those calculated from TOC and exhibit the same pattern of temporal variation. However, the glacial–interglacial contrast in the refined palaeoproductivity is less evident than that calculated from TOC.