Oxygen isotope stage 3 (OIS 3) was a mild interval between the two cold maxima of the last (Weichselian) glaciation marked by climate changes oscillating on a 100–1000 yr time scale between near-interglacial and peak-glacial conditions. During OIS 3, modern humans entered Europe, and somewhat later their Neanderthal predecessors became extinct. Our understanding of this momentous event depends on an answer to the question, Did the unstable environmental conditions of the time play a significant role in early human history? The Stage 3 Project is an interdisciplinary study with two main goals: (1) to describe with existing data and to simulate the climates and landscapes of typical warm and cold phases between 45,000 and 30,000 yr ago and (2) to compare the results with the spatial and temporal distribution of human beings in this context. This paper introduces the Stage 3 Project and provides background to a set of papers on the climate and landscape aspects of the Project that will appear in Quaternary Research and to studies of their relevance to the Early Upper Paleolithic of Europe to appear in journals yet to be determined.

Lake Lisan, the late Pleistocene precursor of the Dead Sea, existed from ∼70,000 to 15,000 yr B.P. It evolved through frequent water-level fluctuations, which reflected the regional hydrological and climatic conditions. We determined the water level of the lake for the time interval ∼55,000–15,000 cal yr B.P. by mapping offshore, nearshore, and fan-delta sediments; by application of sequence stratigraphy methods; and by dating with radiocarbon and U-series methods. During the studied time interval the lake-level fluctuated between ∼340 and 160 m below mean sea level (msl). Between 55,000 and 30,000 cal yr B.P. the lake evolved through short-term fluctuations around 280–290 m below msl, punctuated (at 48,000–43,000 cal yr B.P.) by a drop event to at least 340 m below msl. At ∼27,000 cal yr B.P. the lake began to rise sharply, reaching its maximum elevation of about 164 m below msl between 26,000 and 23,000 cal yr B.P., then it began dropping and reached 300 m below msl at ∼15,000 cal yr B.P. During the Holocene the lake, corresponding to the present Dead Sea, stabilized at ca. 400 m below msl with minor fluctuations. The hypsometric curve of the basin indicates that large changes in lake area are expected at above 403 and 385 m below msl. At these elevations the lake level is buffered. Lake Lisan was always higher than 380 m below msl, indicating a significantly large water contribution to the basin. The long and repetitious periods of stabilization at 280–290 m below msl during Lake Lisan time indicate hydrological control combined with the existence of a physical sill at this elevation. Crossing this sill could not have been achieved without a dramatic increase in the total water input to the lake, as occurred during the fast and intense lake rise from ∼280 to 160 m below msl at ∼27,000 cal yr B.P.

The Toba volcanic event, one of the largest eruptions during the Quaternary, is documented in marine sediment cores from the northeastern Arabian Sea. On the crest of the Murray Ridge and along the western Indian continental margin, we detected distinct concentration spikes and ash layers of rhyolithic volcanic shards near the marine isotope stage 5–4 boundary with the chemical composition of the “Youngest Toba Tuff.” Time series of the Uk′ 37-alkenone index, planktic foraminiferal species, magnetic susceptibility, and sediment accumulation rates from this interval show that the Toba event occurred between two warm periods lasting a few millennia. Using Toba as an instantaneous stratigraphic marker for correlation between the marine- and ice-core chronostratigraphies, these two Arabian Sea climatic events correspond to Greenland interstadials 20 and 19, respectively. Our data sets thus depict substantial interstadial/stadial fluctuations in sea-surface temperature and surface-water productivity. We show that variable terrigenous (eolian) sediment supply played a crucial role in transferring and preserving the productivity signal in the sediment record. Within the provided stratigraphic resolution of several decades to centennials, none of these proxies shows a particular impact of the Toba eruption. However, our results are additional support that Toba, despite its exceptional magnitude, had only a minor impact on the evolution of low-latitude monsoonal climate on centennial to millennial time scales.

δ18O records from the GISP2 and GRIP ice cores are widely used as benchmark paleoclimate chronologies, but there are significant differences between the two time scales. The present study shows that offsets between the two chronologies over the past 14,500 yr do not accumulate gradually, but appear over two short intervals of 100–200 yr. An initial offset of 80 yr occurs close to 3300–3400 yr B.P., and another 100 yr of slip appears at the start of the Younger Dryas. Since these discrepancies are localized, resolving them may be far easier than if large sections of both cores required reexamination.

The climate of the middle Weichselian Glaciation, Marine Oxygen Isotope Stage 3 (OIS-3), a relatively mild period compared to the glacial maxima of OIS-4 and OIS-2, consisted of long warm interstades punctuated by brief cold excursions that grew colder and more frequent with time. The OIS-4 ice sheet is generally thought to have persisted throughout OIS-3, but evidence from dated OIS-3 interstadial deposits suggests that it was swiftly reduced to small remnants which only briefly expanded and retreated. Only 30,000 years ago the deteriorating climate initiated a sustained ice advance leading toward the final glacial maximum of OIS-2. Dynamic ice-sheet models support the existence of a prolonged ice-free interval during OIS-3 induced, perhaps, by low precipitation due to extensive sea-ice cover offshore.

A 500,000-yr-long deep-sea sediment-color record from the Northeast Atlantic was investigated to reconstruct the evolution of late Pleistocene climate variability on millennial time scales. Variations of the red–green color intensity are probably caused by climatically induced changes in the ice-rafted input of red-colored iron-bearing terrigenous material to the core site. The resolution of the age model impedes the detection of distinct spectral features at sub-Milankovitch periodicities. Hence, millennial-scale climate variability is quantified as time-dependent variance of the high-pass filtered color time series. The course of the estimated variance shows distinct patterns, which can be linked to continental ice mass. During the past 500,000 yr, large-amplitude millennial-scale climate variability occurs only if continental ice mass exceeds a threshold level, equivalent to sea level at approximately 40% of the lowering during the last glacial maximum.

Rapidly accumulating Holocene sediments in estuaries commonly are difficult to sample and date. In Chesapeake Bay, we obtained sediment cores as much as 20 m in length and used numerous radiocarbon ages measured by accelerator mass spectrometry methods to provide the first detailed chronologies of Holocene sediment accumulation in the bay. Carbon in these sediments is a complex mixture of materials from a variety of sources. Analyses of different components of the sediments show that total organic carbon ages are largely unreliable, because much of the carbon (including coal) has been transported to the bay from upstream sources and is older than sediments in which it was deposited. Mollusk shells (clams, oysters) and foraminifera appear to give reliable results, although reworking and burrowing are potential problems. Analyses of museum specimens collected alive before atmospheric nuclear testing suggest that the standard reservoir correction for marine samples is appropriate for middle to lower Chesapeake Bay. The biogenic carbonate radiocarbon ages are compatible with 210Pb and 137Cs data and pollen stratigraphy from the same sites.

Post-settlement changes in sediment transport and accumulation is an important environmental issue in many estuaries, including the Chesapeake. Our data show that large variations in sediment mass accumulation rates occur among sites. At shallow water sites, local factors seem to control changes in accumulation rates with time. Our two relatively deep-water sites in the axial channel of the bay have different long-term average accumulation rates, but the history of sediment accumulation at these sites appears to reflect overall conditions in the bay. Mass accumulation rates at the two deep-water sites rapidly increased by about fourfold coincident with widespread land clearance for agriculture in the Chesapeake watershed.

Large heat storage capacity in the western equatorial Pacific has played an important role in modulating global climate. The fluctuation in pollen and spore abundances, together with organic matter (OM) and lithogenics sedimentation, was investigated to reconstruct terrestrial and marine environmental change around New Guinea during the Late Pleistocene. Although appreciable contribution from Indonesian Maritime Continent was expected, the majority of the pollen and spore grains found in core C4402 was transported from New Guinea. Fern spores accounted for 70% (46–90%) of the total pollens and spores. Positive correlation between lithogenic content and the relative abundance of fern spores suggests that lithogenics could be derived from coastal lowland.

The mass accumulation rate (MAR) of pollen and spores varied from 44 to 7,031×10−3 grains cm−2 yr−1 with maxima in oxygen isotope stages (OIS) 2, 3, 4, and around the OIS 4/5 boundary. Less rainfall during glacial times generally enhanced transport of pollen by wind to Site C4402. Their scavenging from the water column was promoted by high activity of the biological pump. Pollen record from core C4402 suggests that lower montane group vegetation was dominant relative to lowland vegetation and upper and mid-montane group during glacial times. Although appreciable contribution by terrestrial OM is expected from high correlation of MAR between organic carbon (OC) and pollen and spores, fairly low COrganic/N ratios and δ13C values (around −20‰) of OM demonstrate that OM in core C4402 is mainly of marine origin.

A Late Quaternary sea-level curve for South Africa is presented on the basis of new and published data from a range of sea level indicators and a variety of locations. Available evidence suggests that sea level in South Africa broadly follows that described from the Caribbean but that deviations occur during sea-level highstands. During the last interglaciation (oxygen isotope stage 5) and the late Holocene, coastal emergence produced higher sea levels in South Africa than those identified in the Caribbean during the same time intervals. This is tentatively ascribed to predicted lithospheric deformation in continental margin settings.

A 467-cm core (B94/3) from the Alta Babicora basin in the Chihuahuan Desert, northern Mexico (29°N, 108°W), documents lake-level and climate changes over the past ca. 65,000 yr. Chronological control is provided by four accelerator mass spectrometry 14C dates and five U-series dates on diatom silica. The core has been analyzed for magnetic susceptibility, loss-on-ignition, carbonate content, sediment chemistry and mineralogy, and pollen and diatom composition. The basin was occupied by a deep freshwater lake throughout the late Pleistocene which, based on shoreline evidence, was at least 19 m deeper than today. The lake shallowed after ca. 57,000 yr B.P. High variability typified the period between 54,600 and 38,000 yr B.P., probably with periodic desiccation and deflation. A deep-water lake was reestablished after 38,000 yr B.P. and persisted until ca. 29,000 yr B.P. Shallowing occurred through the last glacial maximum, although the lake was still deeper than at present. The Pleistocene–Holocene transition was marked by a distinctive change in the diatom flora. There is no diatom record for much of the Holocene, but other proxies indicate generally dry, stable catchment conditions. The Babicora record is more akin to those from the southwest United States than it is to Central America and northern South America. Wet conditions in the late Pleistocene are attributed to winter rainfall from midlatitudes in contrast to the modern, tropical, summer rainfall regime.

A packrat midden located in the Sierra San Francisco, Baja California Sur, Mexico, dating to ca. 10,200 14C yr B.P., contains remains of California juniper (Juniperus californica) and other taxa now associated with southern California chaparral. California juniper does not occur in the Sierra San Francisco today, although “relict” populations of a few chaparral taxa still occur at higher elevations. This midden record documents the early Holocene occurrence of Baja California coniferous woodland and chaparral vegetation far south of its present distribution or its previously known extent from other fossil records. Based on modern climatic tolerances of California juniper and other taxa, central Baja California experienced a mild Mediterranean-type climate at least 5°–6°C cooler than the climate of today, with at least twice the winter precipitation the region now receives.

Four data sets yield information about Holocene climatic change in China at different scales of space and time: (a) 120-yr ground temperature and precipitation measurements covering eastern China; (b) two NOAA 10-yr 850 hPa wind records that highlight features of data set a; (c) an 1100-year record of annual calcite accumulation on a stalagmite near Beijing, and (d) Lamb-type average wetness and temperature data from Chinese historical records back to A.D. 1470 and 1450, respectively. Dry–wet fluctuations and cold–warm oscillations are inferred using the long-term stalagmite thickness series. Quasi-70, 140, 450, and 750-yr oscillations have been detected using a wavelet transform technique. A phase relationship between temperature and precipitation oscillations has been identified based on modern observations and historical records. In northern China, relatively lower temperatures correlate with periods when precipitation shifted from above to below normal. Three colder periods during the Little Ice Age (LIA) in China are inferred, centered in the late 14th century (750-yr oscillation), the early 17th century (450-yr), and the 19th century (140-yr). The latest cool period (1950s–1970s) is found at the 70-yr oscillation. Interdecadal drought–flood and cold–warm differences are explained using modern circulation patterns. LIA climate in China was likely controlled by East Asian monsoon circulation anomalies that were affected by variations in continent–ocean thermal contrast.

We reconstructed Holocene water-level and vegetation dynamics based on pollen and plant macrofossils from a coastal lake in Upper Michigan. Our primary objective was to test the hypothesis that major fluctuations in Great Lakes water levels resulted in part from climatic changes. We also used our data to provide temporal constraints to the mid-Holocene dry period in Upper Michigan. From 9600 to 8600 cal yr B.P. a shallow, lacustrine environment characterized the Mud Lake basin. A Sphagnum-dominated wetland occupied the basin during the mid-Holocene dry period (∼8600 to 6600 cal yr B.P.). The basin flooded at 6600 cal yr B.P. as a result of rising water levels associated with the onset of the Nipissing I phase of ancestral Lake Superior. This flooding event occurred contemporaneously with a well-documented regional expansion of Tsuga. Betula pollen increased during the Nipissing II phase (4500 cal yr B.P.). Macrofossil evidence from Mud Lake suggests that Betula alleghaniensis expansion was primarily responsible for the rising Betula pollen percentages. Major regional and local vegetational changes were associated with all the major Holocene highstands of the western Great Lakes (Nipissing I, Nipissing II, and Algoma). Traditional interpretations of Great Lakes water-level history should be revised to include a major role of climate.

Geochemical, mineralogical, and palynological records from a Holocene sediment core 12.7 m long from Lake Silvana, southeastern Brazil, served to identify the source of detrital sediments within catchment soils. The lake basin was first flooded in the early Holocene (9400 14C yr B.P.), when a gibbsite-rich B soil horizon started to accumulate. Four distinct hydrologic phases are consistent with vegetational changes indicated by pollen data. Phases of slope instability (colluviation) and low lake level correspond to pollen-free intervals and point to a severe retreat of the vegetation cover. The rapid sedimentation of slope-wash sediments alternating with alluvial sediments appears to have formed a tributary fan 7.8 m thick that dammed the Lake Silvana Basin 8500 14C yr B.P., probably as a response to a drastic increase in precipitation rates.

Pollen records from perennially frozen sequences provide vegetation and climate reconstruction for the last 48,000 14C years in the central part of Taymyr Peninsula. Open larch forest with Alnus fruticosa and Betula nana grew during the Kargin (Middle Weichselian) Interstade, ca. 48,000–25,000 14C yr B.P. The climate was generally warmer and wetter than today. Open steppe-like communities with Artemisia, Poaceae, Asteraceae, and herb tundralike communities with dwarf Betula and Salix dominated during the Sartan (Late Weichselian) Stade, ca. 24,000–10,300 14C yr B.P. The statistical information method used for climate reconstruction shows that the coldest climate was ca. 20,000–17,000 14C yr B.P. A warming (Allerød Interstade?) with mean July temperature ca. 1.5°C warmer than today occurred ca. 12,000 14C yr B.P. The following cooling with temperatures about 3°–4°C cooler than present and precipitation about 100 mm lower corresponds well with the Younger Dryas Stade. Tundra–steppe vegetation changed to Betula nana–Alnus fruticosa shrub tundra ca. 10,000 14C yr B.P. Larch appeared in the area ca. 9400 14C yr B.P. and disappeared after 2900 14C yr B.P. Cooling events ca. 10,500, 9600, and 8200 14C yr B.P. characterized the first half of the Holocene. A significant warming occurred ca. 8500 14C yr B.P., but the Holocene temperature maximum was at about 6000–4500 14C yr B.P. The vegetation cover approximated modern conditions ca. 2800 14C yr B.P. Late Holocene warming events occurred at ca. 3500, 2000, and 1000 14C yr B.P. A cooling (Little Ice Age?) took place between 500 and 200 14C yr ago.

The paleo-depositional environment of inner Prydz Bay, East Antarctica, has been reconstructed for the past 21,320 14C yr B.P., using diatom assemblages and sediment facies from a short, 352-cm-long gravity core. Between 21,320 and 11,650 14C yr B.P., compact tillite and diamicton are present in the core, and diatom frustules are rare to absent. These data suggest that an ice sheet grounded over the site during the last glacial maximum. Following glacial retreat, siliceous muddy ooze was deposited, from 11,650 to 2600 14C yr B.P., in an open marine setting. During this stage, diatom frustules are abundant and well preserved, and Thalassiosira antarctica resting spores and Fragilariopsis curta dominate the assemblage. This assemblage suggests open marine deposition in an environment where the spatial and temporal distribution of sea ice is less than today. Since 2600 14C yr B.P., sea-ice and ice-edge diatom species have become more abundant, and neoglacial cooling is inferred. The assemblage is similar to that forming currently in Prydz Bay, where sea ice is absent (<10% cover) for 2–3 months of the year and permanent ice edge and/or multiyear sea ice remains in close proximity to the site.

Terrestrial mollusks, easily identified in Quaternary sediments, represent a reliable tool for quantitative estimates of environmental parameters. Our study, comparing the species distribution with meteorological parameters in Europe, shows that mean temperature of the coldest month and annual thermal magnitude are the most important forcing parameters. This survey allows us to adapt the mutual climatic range (MCR) method to terrestrial mollusk assemblages following two main steps. A set of assemblages from different European regions (northern Norway to southern France) is used to apply the method to present-day mollusks. The reconstructed values describe the latitudinal temperature gradient prevailing over Europe. However, the comparison between the reconstructed and the measured values indicates a shift, similar to that observed, with the same method applied to beetle assemblages. Thus, estimates must be calculated after the reconstruction is tuned with the observations. The results from the modern mollusk assemblages indicate that the MCR method can be safely applied to reconstructing temperatures from terrestrial mollusk assemblages in any worldwide Quaternary sequence. A trial application is made on Late Pleistocene assemblages from Achenheim (Alsace, France).